1 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
3 * config/arc/arc.c (conditionalize_nonjump): New function, broken
6 (arc_predicate_delay_insns): Use it.
8 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
10 * config/arc/predicates.md (extend_operand): During/after reload,
11 allow const_int_operand.
12 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
13 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
14 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
16 (umulsi3_highpart_i): Likewise.
18 2014-03-11 Richard Biener <rguenther@suse.de>
20 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
21 Add asserts to guard possible wrong-code bugs.
23 2014-03-11 Richard Biener <rguenther@suse.de>
25 PR tree-optimization/60429
26 PR tree-optimization/60485
27 * tree-ssa-structalias.c (set_union_with_increment): Properly
28 take into account all fields that overlap the shifted vars.
29 (do_sd_constraint): Likewise.
30 (do_ds_constraint): Likewise.
31 (get_constraint_for_ptr_offset): Likewise.
33 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
35 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
36 (nios2_compute_frame_layout):
37 Add calculation of cfun->machine->fp_save_offset.
38 (nios2_expand_prologue): Correct setting of frame pointer register
40 (nios2_expand_epilogue): Update recovery of stack pointer from
41 frame pointer accordingly.
42 (nios2_initial_elimination_offset): Update calculation of offset
43 for eliminating to HARD_FRAME_POINTER_REGNUM.
45 2014-03-10 Jakub Jelinek <jakub@redhat.com>
48 * ipa.c (symtab_remove_unreachable_nodes): Don't call
49 cgraph_get_create_node on VAR_DECLs.
51 2014-03-10 Richard Biener <rguenther@suse.de>
54 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
56 2014-03-08 Douglas B Rupp <rupp@gnat.com>
58 * config/vms/vms.opt (vms_float_format): New variable.
60 2014-03-08 Tobias Burnus <burnus@net-b.de>
62 * doc/invoke.texi (-fcilkplus): Update implementation
65 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
66 Richard Biener <rguenther@suse.de>
68 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
69 consistently accross all TUs.
70 (run_gcc): Enable -fshort-double automatically at link at link-time
71 and disallow override.
73 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
76 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
77 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
78 if they can't be used.
80 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
82 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test for Solaris
84 * configure: Regenerate.
86 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
88 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
89 (LIB_TLS_SPEC): Save as ld_tls_libs.
90 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
91 (HAVE_AS_IX86_TLSLDM): New test.
92 * configure, config.in: Regenerate.
93 * config/i386/i386.c (legitimize_tls_address): Fall back to
94 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
95 cannot support TLS_MODEL_LOCAL_DYNAMIC.
96 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
97 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
99 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
101 * common.opt (fira-loop-pressure): Mark as optimization.
103 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
105 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
106 an OpenMP mappable type.
108 2014-03-06 Matthias Klose <doko@ubuntu.com>
110 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
111 MULTILIB_OSDIRNAMES is not defined.
113 2014-03-06 Jakub Jelinek <jakub@redhat.com>
114 Meador Inge <meadori@codesourcery.com>
117 * config/arm/arm.c (arm_tls_symbol_p): Remove.
118 (arm_legitimize_address): Call legitimize_tls_address for any
119 arm_tls_referenced_p expression, handle constant addend. Call it
120 before testing for !TARGET_ARM.
121 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
123 2014-03-06 Richard Biener <rguenther@suse.de>
129 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
131 * tree-streamer.c (record_common_node): Assert we don't record
132 nodes with type double.
133 (preload_common_node): Skip type double, complex double and
134 double pointer since it is now frontend dependent due to
135 fshort-double option.
137 2014-03-06 Richard Biener <rguenther@suse.de>
139 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
140 or -fno-lto is specified and the linker has full plugin support.
141 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is
143 (main): Remove -flto processing, adjust lto_mode using
145 * lto-wrapper.c (merge_and_complain): Merge compile-time
147 (run_gcc): And pass it through to the link options.
149 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
153 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
155 * cselib.c (remove_useless_values): Skip to avoid quadratic
156 behavior if the condition moved from...
157 (cselib_process_insn): ... here holds.
159 2014-03-05 Jakub Jelinek <jakub@redhat.com>
162 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
163 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
166 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
167 (TM_H): Add x86-tune.def.
169 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
171 * config/aarch64/aarch64.c (generic_tunings):
172 Use cortexa57_extra_costs.
174 2014-03-05 Jakub Jelinek <jakub@redhat.com>
177 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
178 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
179 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
182 2014-03-04 Heiher <r@hev.cc>
184 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
185 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
187 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
189 * config/i386/predicates.md (const2356_operand): Change to ...
190 (const2367_operand): ... this.
191 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
193 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
194 (*avx512pf_scatterpf<mode>sf): Ditto.
195 (avx512pf_scatterpf<mode>df): Ditto.
196 (*avx512pf_scatterpf<mode>df_mask): Ditto.
197 (*avx512pf_scatterpf<mode>df): Ditto.
198 * config/i386/i386.c (ix86_expand_builtin): Update
199 incorrect hint operand error message.
201 2014-03-04 Richard Biener <rguenther@suse.de>
203 * lto-section-in.c (lto_get_section_data): Fix const cast.
205 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
207 * tree-streamer.c (record_common_node): Assert we don't record
208 nodes with type double.
209 (preload_common_node): Skip type double, complex double and
210 double pointer since it is now frontend dependent due to
211 fshort-double option.
213 2014-03-04 Richard Biener <rguenther@suse.de>
216 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
217 (lto_input_toplevel_asms): Likewise.
218 * lto-section-in.c (lto_get_section_data): Instead do it here
221 2014-03-04 Richard Biener <rguenther@suse.de>
223 PR tree-optimization/60382
224 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
225 dead PHIs a reduction.
227 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
229 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
231 (_mm_prefetch): Move out of GCC target("sse") pragma.
232 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
233 GCC target("prfchw") pragma.
234 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
236 * config/i386/i386.c (ix86_option_override_internal): Enable
237 -mprfchw with -mprefetchwt1.
239 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
241 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
244 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
246 * opts.h (CL_PCH_IGNORE): Define.
247 * targhooks.c (option_affects_pch_p):
248 Return false for options that have CL_PCH_IGNORE set.
249 * opt-functions.awk: Process PchIgnore.
250 * doc/options.texi: Document PchIgnore.
252 * config/arc/arc.opt (misize): Add PchIgnore property.
254 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
256 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
257 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
258 constraint on constants to permit them being loaded into
259 GENERAL_REGS or BASE_REGS.
261 2014-03-03 Nick Clifton <nickc@redhat.com>
263 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
264 anti-cacnonical alternatives.
265 (negandhi3_real): New pattern.
266 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
268 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
270 * config/avr/avr-mcus.def: Remove atxmega16x1.
271 * config/avr/avr-tables.opt: Regenerate.
272 * config/avr/t-multilib: Regenerate.
273 * doc/avr-mmcu.texi: Regenerate.
275 2014-03-03 Tobias Grosser <tobias@grosser.es>
276 Mircea Namolaru <mircea.namolaru@inria.fr>
278 PR tree-optimization/58028
279 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
282 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
284 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
285 not handled by recognizers.
287 2014-03-03 Jakub Jelinek <jakub@redhat.com>
290 * function.c (expand_function_end): Don't emit
291 clobber_return_register sequence if clobber_after is a BARRIER.
292 * cfgexpand.c (construct_exit_block): Append instructions before
293 return_label to prev_bb.
295 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
297 * config/rs6000/constraints.md: Document reserved use of "wc".
299 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
302 * ipa.c (function_and_variable_visibility): When dissolving comdat
303 group, also set all symbols to local.
305 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
310 2013-12-14 Jan Hubicka <jh@suse.cz>
312 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
314 2014-03-02 Jon Beniston <jon@beniston.com>
320 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
321 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
322 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
323 (simple_return, *simple_return): New patterns
324 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
325 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
327 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
329 * dwarf2out.c (gen_subprogram_die): Tidy.
331 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
334 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
335 (*mov_t_msb_neg_negc): ... this new insn.
337 2014-02-28 Jason Merrill <jason@redhat.com>
340 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
343 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
346 * dwarf2out.c (decltype_auto_die): New static.
347 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
348 (gen_type_die_with_usage): Handle 'decltype(auto)'.
349 (is_cxx_auto): Likewise.
351 2014-02-28 Ian Bolton <ian.bolton@arm.com>
353 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
354 we are not using general regs only.
356 2014-02-28 Richard Biener <rguenther@suse.de>
359 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
360 previous fix and only allow to remove trivial pre-headers
361 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
362 (remove_forwarder_block): Properly update the latch of a loop.
364 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
367 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
368 (cselib_preserved_hash_table): New.
369 (preserve_constants_and_equivs): Move preserved vals to it.
370 (cselib_find_slot): Look it up first.
371 (cselib_init): Initialize it.
372 (cselib_finish): Release it.
373 (dump_cselib_table): Dump it.
375 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
378 * cselib.c (remove_useless_values): Skip to avoid quadratic
379 behavior if the condition moved from...
380 (cselib_process_insn): ... here holds.
382 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
385 * var-tracking.c (vt_initialize): Apply the same condition to
386 preserve the CFA base value.
388 2014-02-28 Joey Ye <joey.ye@arm.com>
391 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
392 if reload in progress or completed.
394 2014-02-28 Tobias Burnus <burnus@net-b.de>
397 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
400 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
402 * doc/tm.texi.in (Condition Code Status): Update documention for
403 relative locations of cc0-setter and cc0-user.
405 2014-02-27 Jeff Law <law@redhat.com>
407 PR rtl-optimization/52714
408 * combine.c (try_combine): When splitting an unrecognized PARALLEL
409 into two independent simple sets, if I3 is a jump, ensure the
410 pattern we place into I3 is a (set (pc) ...).
412 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
413 Jeff Law <law@redhat.com>
415 PR rtl-optimization/49847
416 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
417 are in different blocks.
418 * doc/tm.texi (Condition Code Status): Update documention for
419 relative locations of cc0-setter and cc0-user.
421 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
424 * lra.c (lra_emit_add): Check SUBREG too.
426 2014-02-27 Andreas Schwab <schwab@suse.de>
428 * config/m68k/m68k.c (m68k_option_override): Disable
429 -flive-range-shrinkage for classic m68k.
430 (m68k_override_options_after_change): Likewise.
432 2014-02-27 Marek Polacek <polacek@redhat.com>
435 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
436 -Wmaybe-uninitialized.
438 2014-02-27 Alan Modra <amodra@gmail.com>
441 * reload1.c (emit_input_reload_insns): When reload_override_in,
442 set old to rl->in_reg when rl->in_reg is a subreg.
444 2014-02-26 Richard Biener <rguenther@suse.de>
447 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
449 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
451 * common/config/i386/predicates.md (const1256_operand): Remove.
452 (const2356_operand): New.
453 (const_1_to_2_operand): Remove.
454 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
455 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
456 (*avx512pf_gatherpf<mode>sf): Ditto.
457 (avx512pf_gatherpf<mode>df): Ditto.
458 (*avx512pf_gatherpf<mode>df_mask): Ditto.
459 (*avx512pf_gatherpf<mode>df): Ditto.
460 (avx512pf_scatterpf<mode>sf): Ditto.
461 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
462 (*avx512pf_scatterpf<mode>sf): Ditto.
463 (avx512pf_scatterpf<mode>df): Ditto.
464 (*avx512pf_scatterpf<mode>df_mask): Ditto.
465 (*avx512pf_scatterpf<mode>df): Ditto.
466 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
468 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
470 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
471 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
472 (_mm512_mask_testn_epi64_mask): Move to ...
473 * config/i386/avx512cdintrin.h: Here.
474 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
475 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
476 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
477 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
478 TARGET_AVX512F from TARGET_AVX512CD.
480 2014-02-26 Richard Biener <rguenther@suse.de>
483 * ipa.c (walk_polymorphic_call_targets): Properly guard
484 call to inline_update_overall_summary.
486 2014-02-26 Bin Cheng <bin.cheng@arm.com>
489 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
490 and latches only if requested. Fix latch if it is removed.
491 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
492 LOOPS_HAVE_PREHEADERS.
494 2014-02-25 Andrew Pinski <apinski@cavium.com>
496 * builtins.c (expand_builtin_thread_pointer): Create a new target
497 when the target is NULL.
499 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
501 PR rtl-optimization/60317
502 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
503 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
504 * lra-assigns.c: Include params.h.
505 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
506 other reload pseudos considerations.
508 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
510 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
511 to use canonical form for nor<mode>3.
513 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
516 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
519 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
521 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
522 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
523 (ix86_handle_option): Handle OPT_mprefetchwt1.
524 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
525 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
527 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
528 OPTION_MASK_ISA_PREFETCHWT1.
529 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
530 (PTA_PREFETCHWT1): New.
531 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
532 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
533 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
534 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
535 (*prefetch_avx512pf_<mode>_: Change into ...
536 (*prefetch_prefetchwt1_<mode>: This.
537 * config/i386/i386.opt (mprefetchwt1): New.
538 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
539 (_mm_prefetch): Handle intent to write.
540 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
542 2014-02-25 Richard Biener <rguenther@suse.de>
545 * emit-rtl.c (mem_attrs_htab): Remove.
546 (mem_attrs_htab_hash): Likewise.
547 (mem_attrs_htab_eq): Likewise.
548 (set_mem_attrs): Always allocate new mem-attrs when something changed.
549 (init_emit_once): Do not allocate mem_attrs_htab.
551 2014-02-25 Richard Biener <rguenther@suse.de>
554 * lto-opts.c (lto_write_options): Output non-explicit conservative
555 -fwrapv, -fno-trapv and -fno-strict-overflow.
556 * lto-wrapper.c (merge_and_complain): Handle merging those options.
557 (run_gcc): And pass them through.
559 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
561 * sel-sched.c (calculate_new_fences): New parameter ptime.
562 Calculate it as a maximum over all fence cycles.
563 (sel_sched_region_2): Adjust the call to calculate_new_fences.
564 Print the final schedule timing when sched_verbose.
566 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
568 PR rtl-optimization/60292
569 * sel-sched.c (fill_vec_av_set): Do not reset target availability
570 bit fot the fence instruction.
572 2014-02-24 Alangi Derick <alangiderick@gmail.com>
574 * calls.h: Fix typo in comment.
576 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
578 * config/pa/pa.c (pa_output_move_double): Don't valididate when
579 adjusting offsetable addresses.
581 2014-02-24 Guozhi Wei <carrot@google.com>
583 * sparseset.h (sparseset_pop): Fix the wrong index.
585 2014-02-24 Walter Lee <walt@tilera.com>
587 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
588 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
590 * common/config/tilegx/tilegx-common.c
591 (TARGET_DEFAULT_TARGET_FLAGS): Define.
592 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
594 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
595 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
596 (tilegx_gimplify_va_arg_expr): Handle big endian.
597 (tilegx_expand_unaligned_load): Ditto.
598 (tilegx_expand_unaligned_store): Ditto.
599 (TARGET_RETURN_IN_MSB): New.
600 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
601 (TARGET_ENDIAN_DEFAULT): New.
602 (TARGET_BIG_ENDIAN): Handle big endian.
603 (BYTES_BIG_ENDIAN): Ditto.
604 (WORDS_BIG_ENDIAN): Ditto.
605 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
608 * config/tilegx/tilegx.md (extv): Handle big endian.
611 (insn_st<n>_add<bitsuffix>): Ditto.
612 (insn_stnt<n>): Ditto.
613 (insn_stnt<n>_add<bitsuffix>):Ditto.
614 (vec_interleave_highv8qi): Handle big endian.
615 (vec_interleave_highv8qi_be): New.
616 (vec_interleave_highv8qi_le): New.
617 (insn_v1int_h): Handle big endian.
618 (vec_interleave_lowv8qi): Handle big endian.
619 (vec_interleave_lowv8qi_be): New.
620 (vec_interleave_lowv8qi_le): New.
621 (insn_v1int_l): Handle big endian.
622 (vec_interleave_highv4hi): Handle big endian.
623 (vec_interleave_highv4hi_be): New.
624 (vec_interleave_highv4hi_le): New.
625 (insn_v2int_h): Handle big endian.
626 (vec_interleave_lowv4hi): Handle big endian.
627 (vec_interleave_lowv4hi_be): New.
628 (vec_interleave_lowv4hi_le): New.
629 (insn_v2int_l): Handle big endian.
630 (vec_interleave_highv2si): Handle big endian.
631 (vec_interleave_highv2si_be): New.
632 (vec_interleave_highv2si_le): New.
633 (insn_v4int_h): Handle big endian.
634 (vec_interleave_lowv2si): Handle big endian.
635 (vec_interleave_lowv2si_be): New.
636 (vec_interleave_lowv2si_le): New.
637 (insn_v4int_l): Handle big endian.
638 * config/tilegx/tilegx.opt (mbig-endian): New option.
639 (mlittle-endian): New option.
640 * doc/install.texi: Document tilegxbe-linux.
641 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
643 2014-02-24 Martin Jambor <mjambor@suse.cz>
646 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
647 there are no parameter descriptors.
649 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
651 PR rtl-optimization/60268
652 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
653 initialization to ...
654 (sched_rgn_init): ... here.
655 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
657 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
659 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
662 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
664 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
667 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
669 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
670 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
672 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
674 * config/microblaze/predicates.md: Add cmp_op predicate.
675 * config/microblaze/microblaze.md: Add branch_compare instruction
676 which uses cmp_op predicate and emits cmp insn before branch.
677 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
678 to microblaze_expand_conditional_branch and consolidate logic.
679 (microblaze_expand_conditional_branch): emit branch_compare
680 insn instead of handling cmp op separate from branch insn.
682 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
684 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
687 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
689 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
690 define_insn with define_expand and new define_insn
691 *altivec_lve<VI_char>x_internal.
692 (altivec_stve<VI_char>x): Replace define_insn with define_expand
693 and new define_insn *altivec_stve<VI_char>x_internal.
694 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
696 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
698 (altivec_expand_stvex_be): New function.
700 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
702 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
703 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
704 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
705 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
707 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
710 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
711 instead of emit_move_insn.
713 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
715 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
717 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
720 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
722 * config/rs6000/altivec.md (altivec_lvxl): Rename as
723 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
724 (altivec_lvxl_<mode>): New define_expand incorporating
725 -maltivec=be semantics where needed.
726 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
727 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
728 semantics where needed.
729 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
730 (altivec_stvx_<mode>): New define_expand incorporating
731 -maltivec=be semantics where needed.
732 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
733 VM2 iterator instead of V4SI.
734 (altivec_stvxl_<mode>): New define_expand incorporating
735 -maltivec=be semantics where needed.
736 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
737 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
738 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
739 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
740 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
742 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
743 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
744 ALTIVEC_BUILTIN_STVXL.
745 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
746 (altivec_expand_stvx_be): Likewise.
747 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
748 (altivec_expand_lvx_be): Likewise.
749 (altivec_expand_stvx_be): Likewise.
750 (altivec_expand_builtin): Add cases for
751 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
752 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
753 (altivec_init_builtins): Add definitions for
754 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
755 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
757 2014-02-21 Catherine Moore <clm@codesourcery.com>
759 * doc/invoke.texi (mvirt, mno-virt): Document.
760 * config/mips/mips.opt (mvirt): New option.
761 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
763 2014-02-21 Richard Biener <rguenther@suse.de>
765 PR tree-optimization/60276
766 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
767 (STMT_VINFO_MIN_NEG_DIST): New macro.
768 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
769 STMT_VINFO_MIN_NEG_DIST.
770 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
771 made for negative dependence distances still hold.
773 2014-02-21 Richard Biener <rguenther@suse.de>
776 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
777 DECL_INITIAL for globals not in the current function context.
779 2014-02-21 Jakub Jelinek <jakub@redhat.com>
781 PR tree-optimization/56490
782 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
783 * tree-ssa-uninit.c: Include params.h.
784 (compute_control_dep_chain): Add num_calls argument, return false
785 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
786 num_calls to recursive call.
787 (find_predicates): Change dep_chain into normal array,
788 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
789 variable and adjust compute_control_dep_chain caller.
790 (find_def_preds): Likewise.
792 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
794 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
795 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
797 2014-02-21 Nick Clifton <nickc@redhat.com>
799 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
801 (popqi1): Add mode to pre_dec.
804 2014-02-21 Jakub Jelinek <jakub@redhat.com>
806 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
807 mode for mask of V8SFmode permutation.
809 2014-02-20 Richard Henderson <rth@redhat.com>
812 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
813 a new pseudo for OLDVAL.
815 2014-02-20 Jakub Jelinek <jakub@redhat.com>
818 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
819 gen_reg_rtx if d->testing_p.
820 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
821 if d->testing_p and we will certainly return true.
822 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
825 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
827 * emit-rtl.c (gen_reg_rtx): Assert that
828 crtl->emit.regno_pointer_align_length is non-zero.
830 2014-02-20 Richard Henderson <rth@redhat.com>
833 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
834 on failure the store back into EXPECT.
836 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
837 Sandra Loosemore <sandra@codesourcery.com>
839 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
840 * config/nios2/nios2.c (nios2_function_profiler): Add
841 -fPIC (flag_pic == 2) support.
842 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
843 (nios2_large_offset_p): New function.
844 (nios2_unspec_reloc_p): Move up position, update to use
845 nios2_large_offset_p.
846 (nios2_unspec_address): Remove function.
847 (nios2_unspec_offset): New function.
848 (nios2_large_got_address): New function.
849 (nios2_got_address): Add large offset support.
850 (nios2_legitimize_tls_address): Update usage of removed and new
852 (nios2_symbol_binds_local_p): New function.
853 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
854 (nios2_legitimize_address): Update to use nios2_large_offset_p.
855 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
856 (nios2_print_operand): Merge H/L processing, add hiadj/lo
857 processing for (const (unspec ...)).
858 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
860 2014-02-20 Richard Biener <rguenther@suse.de>
862 * tree-cfg.c (replace_uses_by): Mark altered BBs before
863 doing the substitution.
864 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
866 2014-02-20 Martin Jambor <mjambor@suse.cz>
869 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
870 info when checking whether lattices are bottom.
872 2014-02-20 Richard Biener <rguenther@suse.de>
875 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
878 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
881 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
882 parameter specifying the scaling.
883 (inline_call): Update.
884 (want_inline_recursively): Guard division by zero.
885 (recursive_inlining): Update.
886 * ipa-inline.h (clone_inlined_nodes): Update.
888 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
891 * config/i386/i386.c (classify_argument): Pass structures of size
892 64 bytes or less in register.
894 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
895 Kirill Yukhin <kirill.yukhin@intel.com>
897 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
898 (_mm_rcp28_round_ss): Ditto.
899 (_mm_rsqrt28_round_sd): Ditto.
900 (_mm_rsqrt28_round_ss): Ditto.
901 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
902 (_mm_rcp14_round_ss): Ditto.
903 (_mm_rsqrt14_round_sd): Ditto.
904 (_mm_rsqrt14_round_ss): Ditto.
905 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
906 the first input operand, get rid of match_dup.
907 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
909 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
911 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
912 operand as the first input operand, set type attribute.
913 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
915 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
916 operand as the first input operand, set type attribute.
918 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
920 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
923 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
926 * config/i386/i386.c (construct_container): Remove TFmode check
927 for X86_64_INTEGER_CLASS.
929 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
932 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
933 only when -Wpsabi is enabled.
935 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
938 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
939 passing arrays in registers are the same as for structs, so remove the
940 special case for them.
942 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
944 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
945 destination type, extract only the valid bits if the source type is not
946 integral and has a different mode.
948 2014-02-19 Richard Biener <rguenther@suse.de>
951 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
954 2014-02-19 Richard Biener <rguenther@suse.de>
957 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
958 (ipa_modify_call_arguments): Emit an argument load explicitely and
959 preserve virtual SSA form there and for the replacement call.
960 Do not update SSA form nor free dominance info.
962 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
964 * ipa.c (function_and_variable_visibility): Also clear WEAK
965 flag when disolving COMDAT_GROUP.
967 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
969 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
970 * ipa-prop.c (ipa_set_jf_known_type): Return early when
972 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
973 do more sanity checks.
974 (detect_type_change): Return true when giving up early.
975 (compute_complex_assign_jump_func): Fix type parameter of
977 (compute_complex_ancestor_jump_func): Likewise.
978 (update_jump_functions_after_inlining): Fix updating of
980 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
982 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
984 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
985 inline clones when edge disappears.
987 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
990 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
991 Split 64-bit moves into 2 patterns. Do not allow the use of
992 direct move for TDmode in little endian, since the decimal value
993 has little endian bytes within a word, but the 64-bit pieces are
994 ordered in a big endian fashion, and normal subreg's of TDmode are
996 (mov<mode>_64bit_dm): Likewise.
997 (movtd_64bit_nodm): Likewise.
999 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1001 PR tree-optimization/60174
1002 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1003 statement of an SSA_NAME that occurs in an abnormal PHI node.
1005 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1008 * final.c (SEEN_BB): Remove.
1009 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1010 (final_scan_insn): Don't force_source_line on second
1011 NOTE_INSN_BASIC_BLOCK.
1013 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1016 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1017 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1018 (type_natural_mode): Warn ABI change when %zmm register is not
1019 available for AVX512F vector value passing.
1021 2014-02-18 Kai Tietz <ktietz@redhat.com>
1024 * config/i386/i386.c (ix86_expand_prologue): Use value in
1025 rax register as displacement when restoring %r10 or %rax.
1026 Fix wrong offset when restoring both registers.
1028 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1030 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1031 assertion with conditional return.
1033 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1034 Uros Bizjak <ubizjak@gmail.com>
1037 * config/i386/driver-i386.c (host_detect_local_cpu): If
1038 YMM state is not saved by the OS, also clear has_f16c. Move
1039 CPUID 0x80000001 handling before YMM state saving checking.
1041 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1043 PR rtl-optimization/58960
1044 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1045 factored out from ...
1046 (sched_init): ... here.
1047 (free_global_sched_pressure_data): New, factored out from ...
1048 (sched_finish): ... here.
1049 * sched-int.h (free_global_sched_pressure_data): Declare.
1050 * sched-rgn.c (nr_regions_initial): New static global.
1051 (haifa_find_rgns): Initialize it.
1052 (schedule_region): Disable sched-pressure for the newly
1055 2014-02-17 Richard Biener <rguenther@suse.de>
1057 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1058 release SSA defs of pattern stmts.
1060 2014-02-17 Richard Biener <rguenther@suse.de>
1062 * tree-inline.c (expand_call_inline): Release the virtual
1063 operand defined by the call we are about to inline.
1065 2014-02-17 Richard Biener <rguenther@suse.de>
1067 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1069 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1070 Ilya Tocar <ilya.tocar@intel.com>
1072 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1073 arguments order in builtin.
1074 (_mm512_permutexvar_epi64): Ditto.
1075 (_mm512_mask_permutexvar_epi64): Ditto
1076 (_mm512_maskz_permutexvar_epi32): Ditto
1077 (_mm512_permutexvar_epi32): Ditto
1078 (_mm512_mask_permutexvar_epi32): Ditto
1080 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1082 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1083 (p8_vmrgow): Likewise.
1085 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1087 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1090 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1093 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1094 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1095 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1096 using direct move instructions on ISA 2.07. Also adjust
1097 instruction length for 64-bit.
1098 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1099 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1101 2014-02-15 Alan Modra <amodra@gmail.com>
1105 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1106 find_replacement on parts of insn rtl that might be reloaded.
1108 2014-02-15 Richard Biener <rguenther@suse.de>
1110 PR tree-optimization/60183
1111 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1112 (tree_ssa_phiprop): Calculate and free post-dominators.
1114 2014-02-14 Jeff Law <law@redhat.com>
1116 PR rtl-optimization/60131
1117 * ree.c (get_extended_src_reg): New function.
1118 (combine_reaching_defs): Use it rather than assuming location of REG.
1119 (find_and_remove_re): Verify first operand of extension is
1120 a REG before adding the insns to the copy list.
1122 2014-02-14 Roland McGrath <mcgrathr@google.com>
1124 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1125 * configure: Regenerated.
1126 * config.in: Regenerated.
1127 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1128 instead of ASM_SHORT.
1130 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1131 Richard Earnshaw <rearnsha@arm.com>
1133 PR rtl-optimization/59535
1134 * lra-constraints.c (process_alt_operands): Encourage alternative
1135 when unassigned pseudo class is superset of the alternative class.
1136 (inherit_reload_reg): Don't inherit when optimizing for code size.
1137 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1138 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1139 modes not less than 4 for Thumb1.
1141 2014-02-14 Kyle McMartin <kyle@redhat.com>
1144 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1146 2014-02-14 Richard Biener <rguenther@suse.de>
1148 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1149 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1150 Do not strip INDIRECT_REFs.
1152 2014-02-14 Richard Biener <rguenther@suse.de>
1155 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1156 DECL_FUNCTION_SPECIFIC_TARGET.
1157 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1158 * tree-streamer-out.c (pack_ts_target_option): Remove.
1159 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1160 (write_ts_function_decl_tree_pointers): Do not stream
1161 DECL_FUNCTION_SPECIFIC_TARGET.
1162 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1163 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1164 (lto_input_ts_function_decl_tree_pointers): Do not stream
1165 DECL_FUNCTION_SPECIFIC_TARGET.
1167 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1169 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1170 (get_initial_def_for_induction, vectorizable_induction): Ignore
1171 debug stmts when looking for exit_phi.
1172 (vectorizable_live_operation): Fix up condition.
1174 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1176 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1177 nreverse() because it changes the content of original tree list.
1179 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1181 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1182 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1184 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1186 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1187 GNU coding standards.
1189 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1192 * dwarf2out.c (gen_subprogram_die): Don't call
1193 add_calling_convention_attribute if subr_die is old_die.
1195 2014-02-13 Sharad Singhai <singhai@google.com>
1197 * doc/optinfo.texi: Fix order of nodes.
1199 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1201 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1202 operands[2], not operands[3].
1204 2014-02-13 Richard Biener <rguenther@suse.de>
1207 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1208 mention the possibility of an in-tree build.
1209 (CLooG): Update recommended version to 0.18.1, mention the
1210 possibility of an in-tree build and clarify that the ISL
1211 bundled with CLooG does not work.
1213 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1216 * expr.c (compress_float_constant): If x is a hard register,
1217 extend into a pseudo and then move to x.
1219 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1221 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1222 caused by bad second argument to warning_at() with -mhotpatch and
1223 nested functions (e.g. with gfortran).
1225 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1227 * opts.c (option_name): Remove "enabled by default" rider.
1229 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1231 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1233 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1234 Uros Bizjak <ubizjak@gmail.com>
1237 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1238 * configure: Regenerated.
1240 2014-02-12 Richard Biener <rguenther@suse.de>
1242 * vec.c (vec_prefix::calculate_allocation): Move as
1243 inline variant to vec.h.
1244 (vec_prefix::calculate_allocation_1): New out-of-line version.
1245 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1246 (vec_prefix::m_has_auto_buf): Rename to ...
1247 (vec_prefix::m_using_auto_storage): ... this.
1248 (vec_prefix::calculate_allocation): Inline the easy cases
1249 and dispatch to calculate_allocation_1 which doesn't need the
1251 (va_heap::reserve): Use gcc_checking_assert.
1252 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1253 m_using_auto_storage.
1254 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1256 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1257 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1258 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1260 2014-02-12 Richard Biener <rguenther@suse.de>
1262 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1263 when we found a dependence.
1265 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1267 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1269 (maybe_fold_stmt): ... into this new function.
1270 * omp-low.c (lower_omp): Update comment.
1272 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1275 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1278 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1280 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1281 identifiers in comments.
1282 (cortexa53_extra_costs): Likewise.
1283 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1284 (cortexa7_extra_costs): Likewise.
1285 (cortexa12_extra_costs): Likewise.
1286 (cortexa15_extra_costs): Likewise.
1287 (v7m_extra_costs): Likewise.
1289 2014-02-12 Richard Biener <rguenther@suse.de>
1292 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1293 of posix_memalign being successful.
1294 (lower_stmt): Restrict lowering of posix_memalign to when
1295 -ftree-bit-ccp is enabled.
1297 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1299 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1301 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1303 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1305 PR rtl-optimization/60116
1306 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1307 other_insn once the combination has been validated.
1309 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1312 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1314 * ipa-devirt.c: Include demangle.h
1315 (odr_violation_reported): New static variable.
1316 (add_type_duplicate): Update odr_violations.
1317 (maybe_record_node): Add completep parameter; update it.
1318 (record_target_from_binfo): Add COMPLETEP parameter;
1319 update it as needed.
1320 (possible_polymorphic_call_targets_1): Likewise.
1321 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1322 rename FINAL to COMPLETE.
1323 (record_targets_from_bases): Sanity check we found the binfo;
1324 fix COMPLETEP updating.
1325 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1326 parameter, fix computing of COMPLETEP.
1327 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1328 at LTO time do demangling.
1329 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1330 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1332 (gimple_get_virt_method_for_binfo): Likewise.
1333 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1334 gimple_get_virt_method_for_vtable): Update prototypes.
1336 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1339 * genautomata.c (add_presence_absence): Fix typo with
1340 {final_}presence_list.
1342 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1345 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1346 for VSX/Altivec vectors that land in GPR registers.
1348 2014-02-11 Richard Henderson <rth@redhat.com>
1349 Jakub Jelinek <jakub@redhat.com>
1352 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1353 around drhs if type conversion to lacc->type is not useless.
1355 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1357 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1359 (cortex-a57.cortex-a53): Likewise.
1360 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1362 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1364 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1367 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1369 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1370 add_options_for_arm_vfp3.
1372 2014-02-11 Jeff Law <law@redhat.com>
1375 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1376 object with an undesirable mode.
1378 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1381 * config/i386/sol2-9.h: New file.
1382 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1383 *-*-solaris2.9*): Use it.
1385 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1387 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1388 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1390 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1392 * config/microblaze/microblaze.c: Extend mcpu version format
1394 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1396 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1398 2014-02-10 Richard Henderson <rth@redhat.com>
1401 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1402 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1403 ms-abi vs -mno-accumulate-outgoing-args.
1404 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1405 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1408 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1411 * cfgexpand.c (expand_asm_operands): Attach source location to
1412 ASM_INPUT rtx objects.
1413 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1415 2014-02-10 Nick Clifton <nickc@redhat.com>
1417 * config/mn10300/mn10300.c (popcount): New function.
1418 (mn10300_expand_prologue): Include saved registers in stack usage
1421 2014-02-10 Jeff Law <law@redhat.com>
1424 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1425 when changing the SET_DEST of a prior insn to avoid an input reload.
1427 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1429 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1430 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1431 -mcall-openbsd, or -mcall-linux.
1432 (CC1_ENDIAN_BIG_SPEC): Remove.
1433 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1434 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1435 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1436 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1437 and %cc1_endian_default.
1438 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1440 2014-02-10 Richard Biener <rguenther@suse.de>
1442 PR tree-optimization/60115
1443 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1444 MEM_REF handling. Properly verify that the accesses are not
1445 out of the objects bound.
1447 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1449 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1452 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1454 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1455 proper constants and fix formatting.
1456 (possible_polymorphic_call_targets): Fix formatting.
1458 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1459 Ilya Tocar <ilya.tocar@intel.com>
1461 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1462 (_mm512_loadu_epi32): Renamed into...
1463 (_mm512_loadu_si512): This.
1464 (_mm512_storeu_epi32): Renamed into...
1465 (_mm512_storeu_si512): This.
1466 (_mm512_maskz_ceil_ps): Removed.
1467 (_mm512_maskz_ceil_pd): Ditto.
1468 (_mm512_maskz_floor_ps): Ditto.
1469 (_mm512_maskz_floor_pd): Ditto.
1470 (_mm512_floor_round_ps): Ditto.
1471 (_mm512_floor_round_pd): Ditto.
1472 (_mm512_ceil_round_ps): Ditto.
1473 (_mm512_ceil_round_pd): Ditto.
1474 (_mm512_mask_floor_round_ps): Ditto.
1475 (_mm512_mask_floor_round_pd): Ditto.
1476 (_mm512_mask_ceil_round_ps): Ditto.
1477 (_mm512_mask_ceil_round_pd): Ditto.
1478 (_mm512_maskz_floor_round_ps): Ditto.
1479 (_mm512_maskz_floor_round_pd): Ditto.
1480 (_mm512_maskz_ceil_round_ps): Ditto.
1481 (_mm512_maskz_ceil_round_pd): Ditto.
1482 (_mm512_expand_pd): Ditto.
1483 (_mm512_expand_ps): Ditto.
1484 * config/i386/i386.c (ix86_builtins): Remove
1485 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1486 (bdesc_args): Ditto.
1487 * config/i386/predicates.md (const1256_operand): New.
1488 (const_1_to_2_operand): Ditto.
1489 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1490 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1491 (*avx512pf_gatherpf<mode>sf): Ditto.
1492 (avx512pf_gatherpf<mode>df): Ditto.
1493 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1494 (*avx512pf_gatherpf<mode>df): Ditto.
1495 (avx512pf_scatterpf<mode>sf): Ditto.
1496 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1497 (*avx512pf_scatterpf<mode>sf): Ditto.
1498 (avx512pf_scatterpf<mode>df): Ditto.
1499 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1500 (*avx512pf_scatterpf<mode>df): Ditto.
1501 (avx512f_expand<mode>): Removed.
1502 (<shift_insn><mode>3<mask_name>): Change predicate type.
1504 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1506 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1507 not at the end of datarefs vector use ordered_remove to avoid
1508 reordering datarefs vector.
1511 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1512 mark local addressable non-static vars as GOVD_PRIVATE
1513 instead of GOVD_LOCAL.
1514 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1515 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1519 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1520 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1521 assume_aligned or alloc_align attributes.
1522 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1523 arguments. Handle also assume_aligned and alloc_align attributes.
1524 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1525 calls to functions with assume_aligned or alloc_align attributes.
1526 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1528 2014-02-08 Terry Guo <terry.guo@arm.com>
1530 * doc/invoke.texi: Document ARM -march=armv7e-m.
1532 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1534 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1535 flag on __cilkrts_rethrow builtin.
1538 * ipa-cp.c (determine_versionability): Fail at -O0
1539 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1540 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1543 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1546 * tree-inline.c (copy_forbidden): Fail for
1547 __attribute__((optimize (0))) functions.
1549 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1551 * varpool.c: Include pointer-set.h.
1552 (varpool_remove_unreferenced_decls): Variables in other partitions
1553 will not be output; be however careful to not lose information
1556 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1558 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1559 lookup in the vtable constructor.
1561 2014-02-07 Jeff Law <law@redhat.com>
1564 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1565 define_insn_and_split.
1567 * ipa-inline.c (inline_small_functions): Fix typos.
1569 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1571 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1572 (s390_can_use_return_insn): Declare.
1573 * config/s390/s390.h (EPILOGUE_USES): Define.
1574 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1576 (s390_chunkify_start): Handle return JUMP_LABELs.
1577 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1578 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1579 (s390_can_use_return_insn): New functions.
1580 (s390_fix_long_loop_prediction): Handle conditional returns.
1581 (TARGET_SET_UP_BY_PROLOGUE): Define.
1582 * config/s390/s390.md (ANY_RETURN): New code iterator.
1583 (*creturn, *csimple_return, return, simple_return): New patterns.
1585 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1587 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1588 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1589 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1590 REG_CFA_RESTORE list when deciding not to restore a register.
1592 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1594 * config/s390/s390.c: Include tree-pass.h and context.h.
1595 (s390_early_mach): New function, split out from...
1596 (s390_emit_prologue): ...here.
1597 (pass_data_s390_early_mach): New pass structure.
1598 (pass_s390_early_mach): New class.
1599 (s390_option_override): Create and register early_mach pass.
1600 Move to end of file.
1602 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1604 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1605 to match for the exit block.
1607 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1609 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1610 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1611 Reject misaligned operands.
1613 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1615 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1617 2014-02-07 Richard Biener <rguenther@suse.de>
1620 * gimple-low.c (lower_builtin_posix_memalign): New function.
1621 (lower_stmt): Call it to lower posix_memalign in a way
1622 to make alignment info accessible.
1624 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1627 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1628 __builtin_setjmp_receiver.
1630 2014-02-07 Richard Biener <rguenther@suse.de>
1633 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1634 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1635 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1636 Handle BUILT_IN_POSIX_MEMALIGN.
1637 (find_func_clobbers): Likewise.
1638 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1639 (call_may_clobber_ref_p_1): Likewise.
1641 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1644 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1647 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1650 * lto-cgraph.c (lto_output_node): Use
1651 symtab_get_symbol_partitioning_class.
1652 (lto_output_varpool_node): likewise.
1653 (symtab_get_symbol_partitioning_class): Move here from
1655 * cgraph.h (symbol_partitioning_class): Likewise.
1656 (symtab_get_symbol_partitioning_class): Declare.
1658 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1660 * ggc.h (ggc_internal_cleared_alloc): New macro.
1661 * vec.h (vec_safe_copy): Handle memory stats.
1662 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1663 * target-globals.c (save_target_globals): Likewise.
1665 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1668 * expr.c (emit_move_resolve_push): Export; be bit more selective
1669 on when to clear alias set.
1670 * expr.h (emit_move_resolve_push): Declare.
1671 * function.h (struct function): Add tail_call_marked.
1672 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1673 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1674 * config/i386/i386.md (TImode move expander): De not call
1676 (FP push expanders): Preserve memory attributes.
1677 * config/i386/sse.md (push<mode>1): Remove.
1678 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1679 (ix86_expand_push): Remove.
1680 * config/i386/mmx.md (push<mode>1): Remove.
1682 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1684 PR rtl-optimization/60030
1685 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1686 lopart with paradoxical subreg before shifting it up by hprec.
1688 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1690 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1691 Remove extra newline at end of file.
1692 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1693 (arm_issue_rate): Handle cortexa57.
1694 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1695 (cortex-a57.cortex-a53): Likewise.
1697 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1700 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1701 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1703 (arm_expand_prologue): Adjust all callers.
1704 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1705 info, registers also at the lowest numbered registers side. Use
1706 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1710 * var-tracking.c (adjust_mems): Before adding a SET to
1711 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1713 2014-02-06 Alan Modra <amodra@gmail.com>
1716 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1717 change SDmode to DDmode when lra_in_progress.
1719 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1722 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1723 free_data_ref on the dr first, and before goto again also set dr
1724 to the next dr. For simd_lane_access, free old datarefs[i] before
1725 overwriting it. For get_vectype_for_scalar_type failure, don't
1726 free_data_ref if simd_lane_access.
1728 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1731 * tree.h (opts_for_fn): New inline function.
1732 (opt_for_fn): Define.
1733 * config/i386/i386.c (ix86_function_regparm): Use
1734 opt_for_fn (decl, optimize) instead of optimize.
1736 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1738 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1739 for SYMBOL_REF in large memory model.
1741 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1743 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1745 (cortex-a57): Likewise.
1746 (cortex-a57.cortex-a53): Likewise.
1748 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1749 Kugan Vivekanandarajah <kuganv@linaro.org>
1751 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1753 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1755 2014-02-06 Richard Biener <rguenther@suse.de>
1757 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1758 set_loop_copy and initialize_original_copy_tables.
1760 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1762 * config/aarch64/aarch64-simd.md
1763 (aarch64_ashr_simddi): Change QI to SI.
1765 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1766 Jakub Jelinek <jakub@redhat.com>
1769 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1772 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1774 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1775 CODE_FOR_altivec_vpku[hw]um to
1776 CODE_FOR_altivec_vpku[hw]um_direct.
1777 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1778 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1779 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1780 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1782 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1784 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1785 generation for -maltivec=be.
1786 (altivec_vsumsws): Simplify redundant test.
1788 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1790 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1791 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1792 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1793 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1794 gen_altivec_vpkuwum.
1795 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1797 (altivec_vpks<VI_char>ss): Likewise.
1798 (altivec_vpks<VI_char>us): Likewise.
1799 (altivec_vpku<VI_char>us): Likewise.
1800 (altivec_vpku<VI_char>um): Likewise.
1801 (altivec_vpku<VI_char>um_direct): New (copy of
1802 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1804 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1805 target is little endian and -maltivec=be is not specified.
1806 (*altivec_vupkhs<VU_char>_direct): New (copy of
1807 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1808 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1809 target is little endian and -maltivec=be is not specified.
1810 (*altivec_vupkls<VU_char>_direct): New (copy of
1811 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1812 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1813 little endian and -maltivec=be is not specified.
1814 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1815 little endian and -maltivec=be is not specified.
1817 2014-02-05 Richard Henderson <rth@redhat.com>
1820 * combine-stack-adj.c: Revert r206943.
1821 * sched-int.h (struct deps_desc): Add last_args_size.
1822 * sched-deps.c (init_deps): Initialize it.
1823 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1824 contain REG_ARGS_SIZE notes.
1826 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1828 * lto-cgraph.c (asm_nodes_output): Make global.
1829 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1830 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1831 (driver_handle_option): Handle OPT_fwpa.
1833 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1836 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1837 a comment typo and formatting issue. If odr_hash hasn't been
1838 created, return vNULL and set *completep to false.
1841 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1842 bb with no successors.
1844 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1847 * doc/invoke.texi (-march): Clarify documentation for ARM.
1851 2014-02-05 Richard Biener <rguenther@suse.de>
1853 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1854 when not vectorizing because of too many alias checks.
1855 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1856 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1858 2014-02-05 Nick Clifton <nickc@redhat.com>
1860 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1861 accept extended registers in any mode when compiling for the MN10300.
1863 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1865 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1866 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1867 sanitization attributes.
1868 (can_inline_edge_p): Likewise.
1869 (sanitize_attrs_match_for_inline_p): New function.
1871 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1873 * ipa-prop.c (detect_type_change): Shor circuit testing of
1874 type changes on THIS pointer.
1876 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1879 * config/pa/pa.c (legitimize_tls_address): Return original address
1880 if not passed a SYMBOL_REF rtx.
1881 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1883 (pa_emit_move_sequence): Simplify TLS source operands.
1884 (pa_legitimate_constant_p): Reject all TLS constants.
1885 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1886 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1888 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1890 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1891 groups when we know they are controlled by LTO.
1892 * varasm.c (default_binds_local_p_1): If object is in other partition,
1893 it will be resolved locally.
1895 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1897 * config/host-linux.c (linux_gt_pch_use_address): Don't
1898 use SSIZE_MAX because it is not always defined.
1900 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1903 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1904 threshold for pseudo splitting.
1905 (update_ebb_live_info): Process call argument hard registers and
1906 hard registers from insn definition too.
1907 (max_small_class_regs_num): New constant.
1908 (inherit_in_ebb): Update live hard regs through EBBs. Update
1909 reloads_num only for small register classes. Don't split for
1912 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1915 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1918 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1920 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1923 2014-02-04 Marek Polacek <polacek@redhat.com>
1925 * gdbinit.in (pel): Define.
1927 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1929 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1932 2014-02-04 Richard Biener <rguenther@suse.de>
1935 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1936 in function context local.
1937 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1938 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1939 similar to LTO_imported_decl_ref.
1941 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1943 PR tree-optimization/60002
1944 * cgraphclones.c (build_function_decl_skip_args): Clear
1947 PR tree-optimization/60023
1948 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1949 false to gsi_replace.
1950 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1951 has been in some EH region and vec_stmt could throw, add
1952 vec_stmt into the same EH region.
1953 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1954 has no lhs, ignore it.
1955 * internal-fn.c (expand_MASK_LOAD): Likewise.
1958 * tree-inline.c (copy_forbidden): Fail for
1959 __attribute__((optimize (0))) functions.
1962 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1963 is set, copy one less argument.
1964 (expand_simd_clones): Don't subtract clone_info->inbranch
1965 from simd_clone_struct_alloc argument.
1967 PR rtl-optimization/57915
1968 * recog.c (simplify_while_replacing): If all unary/binary/relational
1969 operation arguments are constant, attempt to simplify those.
1972 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1973 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1975 2014-02-04 Richard Biener <rguenther@suse.de>
1977 PR tree-optimization/60012
1978 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1979 TBAA disambiguation to all DDRs.
1981 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1984 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1985 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1987 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1990 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1992 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1994 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1995 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1997 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2000 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2001 to figure out targets of polymorphic calls with known decl.
2002 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2003 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2004 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2005 (get_polymorphic_call_info): ... here.
2006 (get_polymorphic_call_info_from_invariant): New function.
2008 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2010 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2011 lookup via vtable pointer; check for type consistency
2012 and turn inconsitent facts into UNREACHABLE.
2013 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2014 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2015 type inconsistent querries; return UNREACHABLE instead.
2017 2014-02-03 Richard Henderson <rth@twiddle.net>
2020 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2021 already processed this node.
2022 (normalize_one_pred_1): Pass along mark_set.
2023 (normalize_one_pred): Create and destroy a pointer_set_t.
2024 (normalize_one_pred_chain): Likewise.
2026 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2028 PR gcov-profile/58602
2029 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2031 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2034 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2035 -fno-devirtualize; try to devirtualize by the knowledge of
2036 virtual table pointer given by aggregate propagation.
2037 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2038 (ipa_print_node_jump_functions): Dump also offset that
2039 is relevant for polymorphic calls.
2040 (determine_known_aggregate_parts): Add arg_type parameter; use it
2041 instead of determining the type from pointer type.
2042 (ipa_compute_jump_functions_for_edge): Update call of
2043 determine_known_aggregate_parts.
2044 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2045 (gimple_get_virt_method_for_binfo): ... here; simplify using
2046 vtable_pointer_value_to_vtable.
2047 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2048 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2049 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2050 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2052 (vtable_pointer_value_to_binfo): ... here.
2053 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2055 2014-02-03 Teresa Johnson <tejohnson@google.com>
2057 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2058 redef of outer loop index variable.
2060 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2064 * doc/extend.texi (Function Attributes): Typo.
2066 2014-02-03 Cong Hou <congh@google.com>
2068 PR tree-optimization/60000
2069 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2070 if the vectorized statement is a store. A store statement can only
2071 appear at the end of pattern statements.
2073 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2075 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2076 (ix86_option_override_internal): Default long double to 64-bit for
2077 32-bit Bionic and to 128-bit for 64-bit Bionic.
2079 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2080 TARGET_LONG_DOUBLE_128 is true.
2081 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2083 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2084 (mlong-double-64): Negate -mlong-double-128.
2085 (mlong-double-128): New option.
2087 * config/i386/i386-c.c (ix86_target_macros): Define
2088 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2090 * doc/invoke.texi: Document -mlong-double-128.
2092 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2094 PR rtl-optimization/60024
2095 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2097 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2099 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2101 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2103 PR rtl-optimization/57662
2104 * sel-sched.c (code_motion_path_driver): Do not mark already not
2105 existing blocks in the visiting bitmap.
2107 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2109 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2110 on the insn being emitted.
2112 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2113 Will Deacon <will.deacon@arm.com>
2115 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2117 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2119 * config/arm/arm-tables.opt: Regenerate.
2121 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2123 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2124 for vector types other than V16QImode.
2125 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2126 define_expand, and call altivec_expand_vec_perm_le when producing
2127 code with little endian element order.
2128 (*altivec_vperm_<mode>_internal): New insn having previous
2129 behavior of altivec_vperm_<mode>.
2130 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2131 altivec_expand_vec_perm_le when producing code with little endian
2133 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2134 behavior of altivec_vperm_<mode>_uns.
2136 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2138 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2139 (altivec_vsumsws): Add handling for -maltivec=be with a little
2141 (altivec_vsumsws_direct): New.
2142 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2143 gen_altivec_vsumsws.
2145 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2147 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2148 vtable_pointer_value_to_binfo): New functions.
2149 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2150 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2152 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2154 * config/nios2/nios2.md (load_got_register): Initialize GOT
2155 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2156 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2158 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2160 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2161 preserverd by passthrough, do not propagate the type.
2163 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2165 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2166 (mips_atomic_assign_expand_fenv): New function.
2167 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2169 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2171 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2172 (__builtin_mips_set_fcsr): Likewise.
2173 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2174 MIPS_USI_FTYPE_VOID.
2175 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2176 (mips16_expand_set_fcsr): Likewise.
2177 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2178 (mips16_set_fcsr_stub): Likewise.
2179 (mips16_get_fcsr_one_only_stub): New class.
2180 (mips16_set_fcsr_one_only_stub): Likewise.
2181 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2182 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2183 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2184 (hard_float): New availability predicate.
2185 (mips_builtins): Add get_fcsr and set_fcsr.
2186 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2187 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2188 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2189 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2190 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2193 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2195 * config/mips/mips.c (mips_one_only_stub): New class.
2196 (mips_need_mips16_rdhwr_p): Replace with...
2197 (mips16_rdhwr_stub): ...this new variable.
2198 (mips16_stub_call_address): New function.
2199 (mips16_rdhwr_one_only_stub): New class.
2200 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2201 (mips_output_mips16_rdhwr): Delete.
2202 (mips_finish_stub): New function.
2203 (mips_code_end): Use it to handle rdhwr stubs.
2205 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2208 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2209 when calculating size of integer atomic types.
2211 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2213 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2215 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2217 PR tree-optimization/60003
2218 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2219 * profile.c (branch_prob): Use gimple_call_builtin_p
2220 to check for BUILT_IN_SETJMP_RECEIVER.
2221 * tree-inline.c (copy_bb): Call notice_special_calls.
2223 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2226 * lra-constraints.c (process_alt_operands): Update reload_sum only
2229 2014-01-31 Richard Henderson <rth@redhat.com>
2232 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2233 until after else_eh is processed.
2235 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2237 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2238 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2239 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2240 in smmintrin.h, remove them.
2241 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2242 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2243 * config/i386/i386.md (ROUND_SAE): Fix value.
2244 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2245 (const48_operand): New.
2246 * config/i386/subst.md (round), (round_expand): Use
2247 const_4_or_8_to_11_operand.
2248 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2250 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2252 * config/i386/constraints.md (Yk): Swap meaning with k.
2253 * config/i386/i386.md (movhi_internal): Change Yk to k.
2254 (movqi_internal): Ditto.
2255 (*k<logic><mode>): Ditto.
2258 (kandn<mode>): Ditto.
2259 (*<code>hi_1): Ditto.
2260 (*<code>qi_1): Ditto.
2261 (kxnor<mode>): Ditto.
2262 (kortestzhi): Ditto.
2263 (kortestchi): Ditto.
2265 (*one_cmplhi2_1): Ditto.
2266 (*one_cmplqi2_1): Ditto.
2267 * config/i386/sse.md (): Change k to Yk.
2268 (avx512f_load<mode>_mask): Ditto.
2269 (avx512f_blendm<mode>): Ditto.
2270 (avx512f_store<mode>_mask): Ditto.
2271 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2272 (avx512f_storedqu<mode>_mask): Ditto.
2273 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2275 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2276 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2277 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2278 (avx512f_maskcmp<mode>3): Ditto.
2279 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2280 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2281 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2282 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2283 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2284 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2285 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2286 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2287 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2288 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2289 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2290 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2291 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2292 (vec_extract_lo_<mode>_maskm): Ditto.
2293 (vec_extract_hi_<mode>_maskm): Ditto.
2294 (avx512f_vternlog<mode>_mask): Ditto.
2295 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2296 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2297 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2298 (avx512f_<code>v8div16qi2_mask): Ditto.
2299 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2300 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2301 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2302 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2303 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2304 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2305 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2306 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2307 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2308 (avx512cd_maskb_vec_dupv8di): Ditto.
2309 (avx512cd_maskw_vec_dupv16si): Ditto.
2310 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2311 (avx512f_vpermi2var<mode>3_mask): Ditto.
2312 (avx512f_vpermi2var<mode>3_mask): Ditto.
2313 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2314 (*avx512f_gathersi<mode>): Ditto.
2315 (*avx512f_gathersi<mode>_2): Ditto.
2316 (*avx512f_gatherdi<mode>): Ditto.
2317 (*avx512f_gatherdi<mode>_2): Ditto.
2318 (*avx512f_scattersi<mode>): Ditto.
2319 (*avx512f_scatterdi<mode>): Ditto.
2320 (avx512f_compress<mode>_mask): Ditto.
2321 (avx512f_compressstore<mode>_mask): Ditto.
2322 (avx512f_expand<mode>_mask): Ditto.
2323 * config/i386/subst.md (mask): Change k to Yk.
2324 (mask_scalar_merge): Ditto.
2327 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2329 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2331 2014-01-31 Richard Biener <rguenther@suse.de>
2334 * builtins.c (fold_builtin_memory_op): Make sure to not
2335 use a floating-point mode or a boolean or enumeral type for
2338 2014-01-30 DJ Delorie <dj@redhat.com>
2340 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2341 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2342 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2343 whenever main() has an epilogue.
2345 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2347 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2348 unused variable "field".
2349 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2350 (vsx_mergeh_<mode>): Likewise.
2351 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2352 (altivec_vmrghh): Likewise.
2353 (altivec_vmrghw): Likewise.
2354 (altivec_vmrglb): Likewise.
2355 (altivec_vmrglh): Likewise.
2356 (altivec_vmrglw): Likewise.
2357 (altivec_vspltb): Add missing uses.
2358 (altivec_vsplth): Likewise.
2359 (altivec_vspltw): Likewise.
2360 (altivec_vspltsf): Likewise.
2362 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2365 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2366 frame related instructions.
2368 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2370 PR rtl-optimization/59959
2371 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2372 any reload of register whose subreg is invalid.
2374 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2376 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2377 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2378 Add missing return type - void.
2380 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2382 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2383 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2384 remove element index adjustment for endian (now handled in vsx.md
2386 (altivec_expand_vec_perm_const): Use
2387 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2388 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2389 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2390 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2391 define_expand and a new define_insn *altivec_vspltb_internal;
2392 adjust for -maltivec=be on a little endian target.
2393 (altivec_vspltb_direct): New.
2394 (altivec_vsplth): Divide into a define_expand and a new
2395 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2396 little endian target.
2397 (altivec_vsplth_direct): New.
2398 (altivec_vspltw): Divide into a define_expand and a new
2399 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2400 little endian target.
2401 (altivec_vspltw_direct): New.
2402 (altivec_vspltsf): Divide into a define_expand and a new
2403 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2404 a little endian target.
2406 2014-01-30 Richard Biener <rguenther@suse.de>
2408 PR tree-optimization/59993
2409 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2410 can propagate form the earlier stmt and avoid the transform
2411 when the intermediate result is needed.
2413 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2415 * README.Portability: Fix typo.
2417 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2419 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2420 comparison_operator with ordered_comparison_operator.
2422 2014-01-30 Nick Clifton <nickc@redhat.com>
2424 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2425 Rename to mn10300_store_multiple_regs.
2426 * config/mn10300/mn10300.c: Likewise.
2427 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2428 store_multiple_regs.
2429 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2430 Call mn10300_store_multiple_regs.
2432 2014-01-30 Nick Clifton <nickc@redhat.com>
2433 DJ Delorie <dj@redhat.com>
2435 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2436 %fp 2 to keep registers after it properly word-aligned.
2437 (rl78_alloc_physical_registers_umul): Handle the case where both
2438 input operands are the same.
2440 2014-01-30 Richard Biener <rguenther@suse.de>
2442 PR tree-optimization/59903
2443 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2446 2014-01-30 Jason Merrill <jason@redhat.com>
2449 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2452 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2454 2014-01-30 Richard Biener <rguenther@suse.de>
2456 PR tree-optimization/59951
2457 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2459 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2462 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2463 SFmode to DFmode case.
2465 2014-01-29 DJ Delorie <dj@redhat.com>
2467 * config/msp430/msp430.opt (-minrt): New.
2468 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2470 (ENDFILE_SPEC): Likewise.
2472 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2474 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2475 (estimate_function_body_sizes): Use it.
2477 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2480 * dwarf2out.c (is_cxx_auto): New.
2481 (is_base_type): Use it.
2482 (gen_type_die_with_usage): Likewise.
2484 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2486 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2487 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2488 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2489 -maltivec=be with LE targets.
2490 (vsx_mergeh_<mode>): Likewise.
2491 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2492 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2493 (altivec_vmrghb): Replace with define_expand and new
2494 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2495 (altivec_vmrghb_direct): New define_insn.
2496 (altivec_vmrghh): Replace with define_expand and new
2497 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2498 (altivec_vmrghh_direct): New define_insn.
2499 (altivec_vmrghw): Replace with define_expand and new
2500 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2501 (altivec_vmrghw_direct): New define_insn.
2502 (*altivec_vmrghsf): Adjust for endianness.
2503 (altivec_vmrglb): Replace with define_expand and new
2504 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2505 (altivec_vmrglb_direct): New define_insn.
2506 (altivec_vmrglh): Replace with define_expand and new
2507 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2508 (altivec_vmrglh_direct): New define_insn.
2509 (altivec_vmrglw): Replace with define_expand and new
2510 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2511 (altivec_vmrglw_direct): New define_insn.
2512 (*altivec_vmrglsf): Adjust for endianness.
2513 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2514 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2515 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2516 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2517 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2518 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2519 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2520 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2522 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2524 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2525 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2528 2014-01-29 Richard Biener <rguenther@suse.de>
2530 PR tree-optimization/58742
2531 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2532 associate_pointerplus_align.
2533 (associate_pointerplus_diff): New function.
2534 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2535 and associate_pointerplus_diff.
2537 2014-01-29 Richard Biener <rguenther@suse.de>
2539 * lto-streamer.h (LTO_major_version): Bump to 3.
2540 (LTO_minor_version): Reset to 0.
2542 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2544 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2545 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2546 (arm_file_start): Generate correct asm header for armv7ve.
2547 * config/arm/bpabi.h: Add multilib support for armv7ve.
2548 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2549 and cortex-a15 to armv7ve.
2550 * config/arm/t-aprofile: Add multilib support for armv7ve.
2551 * doc/invoke.texi: Document -march=armv7ve.
2553 2014-01-29 Richard Biener <rguenther@suse.de>
2555 PR tree-optimization/58742
2556 * tree-ssa-forwprop.c (associate_plusminus): Return true
2557 if we changed sth, defer EH cleanup to ...
2558 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2559 (simplify_mult): New function.
2561 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2564 PR tree-optimization/59920
2565 * tree.c (build_common_builtin_nodes): Remove
2566 __builtin_setjmp_dispatcher initialization.
2567 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2568 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2569 instead of gsi_after_labels + manually skipping debug stmts.
2570 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2571 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2572 * tree-inline.c (copy_edges_for_bb): Remove
2573 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2574 argument. Ignore computed_goto_p stmts. Don't call
2575 make_abnormal_goto_edges. If a call might need abnormal edges
2576 for non-local gotos, see if it already has an edge to
2577 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2578 with true argument, don't do anything then, otherwise add
2579 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2580 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2582 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2583 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2584 (lower_stmt): Don't set data->calls_builtin_setjmp.
2585 (lower_builtin_setjmp): Adjust comment.
2586 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2587 * tree-cfg.c (found_computed_goto): Remove.
2588 (factor_computed_gotos): Remove.
2589 (make_goto_expr_edges): Return bool, true for computed gotos.
2590 Don't call make_abnormal_goto_edges.
2591 (build_gimple_cfg): Don't set found_computed_goto, don't call
2592 factor_computed_gotos.
2593 (computed_goto_p): No longer static.
2594 (make_blocks): Don't set found_computed_goto.
2595 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2596 (make_edges): If make_goto_expr_edges returns true, push bb
2597 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2598 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2599 vector. Record mapping between bbs and OpenMP regions if there
2600 are any, adjust make_gimple_omp_edges caller. Call
2601 handle_abnormal_edges.
2602 (make_abnormal_goto_edges): Remove.
2603 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2604 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2605 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2606 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2607 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2608 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2609 filling *region also set *region_idx to (*region)->entry->index.
2612 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2613 For REGs set ORIGINAL_REGNO.
2615 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2617 * doc/md.texi: Mention that a target shouldn't implement
2618 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2621 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2623 PR tree-optimization/59594
2624 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2625 a copy of the datarefs vector rather than the vector itself.
2627 2014-01-28 Jason Merrill <jason@redhat.com>
2630 * dwarf2out.c (auto_die): New static.
2631 (gen_type_die_with_usage): Handle C++1y 'auto'.
2632 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2635 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2638 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2639 (SPEC_X32): Likewise.
2640 (SPEC_64): Likewise.
2641 * config/i386/i386.c (ix86_option_override_internal): Turn off
2642 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2644 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2645 * config/i386/i386.h (TARGET_16BIT): New macro.
2646 (TARGET_16BIT_P): Likewise.
2647 * config/i386/i386.opt: Add m16.
2648 * doc/invoke.texi: Document -m16.
2650 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2652 PR preprocessor/59935
2653 * input.c (location_get_source_line): Bail out on when line number
2654 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2656 2014-01-28 Richard Biener <rguenther@suse.de>
2658 PR tree-optimization/58742
2659 * tree-ssa-forwprop.c (associate_plusminus): Handle
2660 pointer subtraction of the form (T)(P + A) - (T)P.
2662 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2664 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2667 2014-01-28 Richard Biener <rguenther@suse.de>
2670 2014-01-28 Richard Biener <rguenther@suse.de>
2672 PR rtl-optimization/45364
2673 PR rtl-optimization/59890
2674 * var-tracking.c (local_get_addr_clear_given_value): Handle
2675 already cleared slot.
2676 (val_reset): Handle not allocated local_get_addr_cache.
2677 (vt_find_locations): Use post-order on the inverted CFG.
2679 2014-01-28 Richard Biener <rguenther@suse.de>
2681 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2683 2014-01-28 Richard Biener <rguenther@suse.de>
2685 PR rtl-optimization/45364
2686 PR rtl-optimization/59890
2687 * var-tracking.c (local_get_addr_clear_given_value): Handle
2688 already cleared slot.
2689 (val_reset): Handle not allocated local_get_addr_cache.
2690 (vt_find_locations): Use post-order on the inverted CFG.
2692 2014-01-28 Alan Modra <amodra@gmail.com>
2694 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2695 * configure.ac <recursive call for build != host>: Define
2696 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2697 and LD_FOR_BUILD too.
2698 * configure: Regenerate.
2700 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2702 * config/i386/i386.c (get_builtin_code_for_version): Separate
2703 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2704 Broadwell from Haswell.
2706 2014-01-27 Steve Ellcey <sellcey@mips.com>
2708 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2709 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2710 * config/mips/mips.c (mips_option_override): Change setting
2712 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2713 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2714 Change from Mask to Var.
2716 2014-01-27 Jeff Law <law@redhat.com>
2718 * ipa-inline.c (inline_small_functions): Fix typo.
2720 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2722 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2723 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2724 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2725 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2726 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2727 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2728 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2729 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2730 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2731 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2732 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2733 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2734 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2735 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2736 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2737 (_mm512_storeu_epi64): Ditto.
2738 (_mm512_cmpge_epi32_mask): Ditto.
2739 (_mm512_cmpge_epu32_mask): Ditto.
2740 (_mm512_cmpge_epi64_mask): Ditto.
2741 (_mm512_cmpge_epu64_mask): Ditto.
2742 (_mm512_cmple_epi32_mask): Ditto.
2743 (_mm512_cmple_epu32_mask): Ditto.
2744 (_mm512_cmple_epi64_mask): Ditto.
2745 (_mm512_cmple_epu64_mask): Ditto.
2746 (_mm512_cmplt_epi32_mask): Ditto.
2747 (_mm512_cmplt_epu32_mask): Ditto.
2748 (_mm512_cmplt_epi64_mask): Ditto.
2749 (_mm512_cmplt_epu64_mask): Ditto.
2750 (_mm512_cmpneq_epi32_mask): Ditto.
2751 (_mm512_cmpneq_epu32_mask): Ditto.
2752 (_mm512_cmpneq_epi64_mask): Ditto.
2753 (_mm512_cmpneq_epu64_mask): Ditto.
2754 (_mm512_expand_pd): Ditto.
2755 (_mm512_expand_ps): Ditto.
2756 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2757 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2758 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2759 * config/i386/i386.c (ix86_builtins): Add
2760 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2761 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2762 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2763 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2764 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2765 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2766 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2767 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2768 IX86_BUILTIN_PMOVUSQW512_MEM.
2769 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2770 __builtin_ia32_pmovsqd512mem_mask,
2771 __builtin_ia32_pmovqd512mem_mask,
2772 __builtin_ia32_pmovusqw512mem_mask,
2773 __builtin_ia32_pmovsqw512mem_mask,
2774 __builtin_ia32_pmovqw512mem_mask,
2775 __builtin_ia32_pmovusdw512mem_mask,
2776 __builtin_ia32_pmovsdw512mem_mask,
2777 __builtin_ia32_pmovdw512mem_mask,
2778 __builtin_ia32_pmovqb512mem_mask,
2779 __builtin_ia32_pmovusqb512mem_mask,
2780 __builtin_ia32_pmovsqb512mem_mask,
2781 __builtin_ia32_pmovusdb512mem_mask,
2782 __builtin_ia32_pmovsdb512mem_mask,
2783 __builtin_ia32_pmovdb512mem_mask.
2784 (bdesc_args): Add __builtin_ia32_expanddf512,
2785 __builtin_ia32_expandsf512.
2786 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2787 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2788 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2789 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2790 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2791 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2792 (avx512f_<code>v8div16qi2_mask_store): This.
2793 (avx512f_expand<mode>): New.
2795 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2797 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2799 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2800 (_mm512_prefetch_i32scatter_pd): Ditto.
2801 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2802 (_mm512_prefetch_i64scatter_pd): Ditto.
2803 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2804 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2805 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2806 (_mm512_prefetch_i32scatter_ps): Ditto.
2807 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2808 (_mm512_prefetch_i64scatter_ps): Ditto.
2809 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2810 * config/i386/i386-builtin-types.def: Define
2811 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2812 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2813 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2814 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2815 IX86_BUILTIN_SCATTERPFQPD.
2816 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2817 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2818 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2819 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2820 __builtin_ia32_scatterpfqps.
2821 (ix86_expand_builtin): Expand new built-ins.
2822 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2823 fix memory access data type.
2824 (*avx512pf_gatherpf<mode>_mask): Ditto.
2825 (*avx512pf_gatherpf<mode>): Ditto.
2826 (avx512pf_scatterpf<mode>): Ditto.
2827 (*avx512pf_scatterpf<mode>_mask): Ditto.
2828 (*avx512pf_scatterpf<mode>): Ditto.
2829 (GATHER_SCATTER_SF_MEM_MODE): New.
2830 (avx512pf_gatherpf<mode>df): Ditto.
2831 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2832 (*avx512pf_scatterpf<mode>df): Ditto.
2834 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2837 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2838 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2841 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2843 * common/config/arm/arm-common.c
2844 (arm_rewrite_mcpu): Handle multiple names.
2846 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2848 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2850 * gimple-builder.h (create_gimple_tmp): Delete.
2852 2014-01-27 Christian Bruel <christian.bruel@st.com>
2854 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2857 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2859 * config/pa/pa.md (call): Generate indirect long calls to non-local
2860 functions when outputing 32-bit code.
2861 (call_value): Likewise except for special call to buggy powf function.
2863 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2864 portable runtime and PIC indirect calls.
2865 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2866 and PIC call sequences. Use ldo instead of blr to set return register
2867 in PIC call sequence.
2869 2014-01-25 Walter Lee <walt@tilera.com>
2871 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2872 avoid clobbering a live register.
2874 2014-01-25 Walter Lee <walt@tilera.com>
2876 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2877 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2878 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2879 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2881 2014-01-25 Walter Lee <walt@tilera.com>
2883 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2884 arguments on even registers.
2885 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2887 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2888 (BIGGEST_ALIGNMENT): Ditto.
2889 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2891 2014-01-25 Walter Lee <walt@tilera.com>
2893 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2894 insns before bundling.
2895 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2897 2014-01-25 Walter Lee <walt@tilera.com>
2899 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2900 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2901 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2903 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2905 * config/mips/constraints.md (kl): Delete.
2906 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2907 define expands, using...
2908 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2909 instructions for MIPS16.
2910 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2911 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2913 2014-01-25 Walter Lee <walt@tilera.com>
2915 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2919 2014-01-25 Walter Lee <walt@tilera.com>
2921 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2922 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2924 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2926 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2929 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2931 * print-rtl.c (in_call_function_usage): New var.
2932 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2933 EXPR_LIST mode as mode and not as reg note name.
2936 * cfgloopmanip.c (copy_loop_info): If
2937 loop->warned_aggressive_loop_optimizations, make sure
2938 the flag is set in target loop too.
2940 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2942 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2944 * builtins.def: Likewise.
2945 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2946 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2947 * ira.c (ira_setup_eliminable_regset): Likewise.
2948 * omp-low.c (gate_expand_omp): Likewise.
2949 (execute_lower_omp): Likewise.
2950 (diagnose_sb_0): Likewise.
2951 (gate_diagnose_omp_blocks): Likewise.
2952 (simd_clone_clauses_extract): Likewise.
2955 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2957 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2958 correction for little endian...
2959 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2962 2014-01-24 Jeff Law <law@redhat.com>
2964 PR tree-optimization/59919
2965 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2966 for non-returning calls.
2968 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2970 * common/config/aarch64/aarch64-common.c
2971 (aarch64_rewrite_mcpu): Handle multiple names.
2972 * config/aarch64/aarch64.h
2973 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2975 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2977 * input.c (add_file_to_cache_tab): Handle the case where fopen
2980 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2983 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2984 from push operand if code of push isn't PRE_DEC.
2986 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2989 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2990 -mquad-memory-atomic. Update -mquad-memory documentation to say
2991 it is only used for non-atomic loads/stores.
2993 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2994 -mquad-memory or -mquad-memory-atomic switches.
2996 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2997 -mquad-memory-atomic to ISA 2.07 support.
2999 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3000 to separate support of normal quad word memory operations (ldq, stq)
3001 from the atomic quad word memory operations.
3003 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3004 support to separate non-atomic quad word operations from atomic
3005 quad word operations. Disable non-atomic quad word operations in
3006 little endian mode so that we don't have to swap words after the
3007 load and before the store.
3008 (quad_load_store_p): Add comment about atomic quad word support.
3009 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3010 options printed with -mdebug=reg.
3012 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3013 -mquad-memory-atomic as the test for whether we have quad word
3014 atomic instructions.
3015 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3016 or -mp8-vector are used, allow byte/half-word atomic operations.
3018 * config/rs6000/sync.md (load_lockedti): Insure that the address
3019 is a proper indexed or indirect address for the lqarx instruction.
3020 On little endian systems, swap the hi/lo registers after the lqarx
3022 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3023 insure the address is valid for the lqarx instruction.
3024 (store_conditionalti): Insure that the address is a proper indexed
3025 or indirect address for the stqcrx. instruction. On little endian
3026 systems, swap the hi/lo registers before doing the stqcrx.
3028 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3029 insure the address is valid for the stqcrx. instruction.
3031 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3032 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3033 type of quad memory support is available.
3035 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3038 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3039 there is a danger of looping.
3041 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3043 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3044 force flag_ira_loop_pressure if set via command line.
3046 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3048 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3049 (ashr_simd): New builtin handling DI mode.
3050 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3051 (aarch64_sshr_simddi): New match pattern.
3052 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3053 (vshrd_n_s64): Likewise.
3054 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3056 2014-01-23 Nick Clifton <nickc@redhat.com>
3058 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3059 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3060 favour of mcu specific scripts.
3061 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3064 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3065 Alex Velenko <Alex.Velenko@arm.com>
3067 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3068 (vaddv_s16): Likewise.
3069 (vaddv_s32): Likewise.
3070 (vaddv_u8): Likewise.
3071 (vaddv_u16): Likewise.
3072 (vaddv_u32): Likewise.
3073 (vaddvq_s8): Likewise.
3074 (vaddvq_s16): Likewise.
3075 (vaddvq_s32): Likewise.
3076 (vaddvq_s64): Likewise.
3077 (vaddvq_u8): Likewise.
3078 (vaddvq_u16): Likewise.
3079 (vaddvq_u32): Likewise.
3080 (vaddvq_u64): Likewise.
3081 (vaddv_f32): Likewise.
3082 (vaddvq_f32): Likewise.
3083 (vaddvq_f64): Likewise.
3084 (vmaxv_f32): Likewise.
3085 (vmaxv_s8): Likewise.
3086 (vmaxv_s16): Likewise.
3087 (vmaxv_s32): Likewise.
3088 (vmaxv_u8): Likewise.
3089 (vmaxv_u16): Likewise.
3090 (vmaxv_u32): Likewise.
3091 (vmaxvq_f32): Likewise.
3092 (vmaxvq_f64): Likewise.
3093 (vmaxvq_s8): Likewise.
3094 (vmaxvq_s16): Likewise.
3095 (vmaxvq_s32): Likewise.
3096 (vmaxvq_u8): Likewise.
3097 (vmaxvq_u16): Likewise.
3098 (vmaxvq_u32): Likewise.
3099 (vmaxnmv_f32): Likewise.
3100 (vmaxnmvq_f32): Likewise.
3101 (vmaxnmvq_f64): Likewise.
3102 (vminv_f32): Likewise.
3103 (vminv_s8): Likewise.
3104 (vminv_s16): Likewise.
3105 (vminv_s32): Likewise.
3106 (vminv_u8): Likewise.
3107 (vminv_u16): Likewise.
3108 (vminv_u32): Likewise.
3109 (vminvq_f32): Likewise.
3110 (vminvq_f64): Likewise.
3111 (vminvq_s8): Likewise.
3112 (vminvq_s16): Likewise.
3113 (vminvq_s32): Likewise.
3114 (vminvq_u8): Likewise.
3115 (vminvq_u16): Likewise.
3116 (vminvq_u32): Likewise.
3117 (vminnmv_f32): Likewise.
3118 (vminnmvq_f32): Likewise.
3119 (vminnmvq_f64): Likewise.
3121 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3123 * config/aarch64/aarch64-simd.md
3124 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3125 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3126 (*aarch64_mul3_elt<mode>): Likewise.
3127 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3128 (*aarch64_mul3_elt_to_64v2df): Likewise.
3129 (*aarch64_mla_elt<mode>): Likewise.
3130 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3131 (*aarch64_mls_elt<mode>): Likewise.
3132 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3133 (*aarch64_fma4_elt<mode>): Likewise.
3134 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3135 (*aarch64_fma4_elt_to_64v2df): Likewise.
3136 (*aarch64_fnma4_elt<mode>): Likewise.
3137 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3138 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3139 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3140 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3141 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3142 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3143 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3144 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3145 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3147 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3149 * config/aarch64/aarch64-simd.md
3150 (aarch64_be_checked_get_lane<mode>): New define_expand.
3151 * config/aarch64/aarch64-simd-builtins.def
3152 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3153 New builtin definition.
3154 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3155 Use new safe be builtin.
3157 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3159 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3161 (aarch64_be_st1<mode>): Likewise.
3162 (aarch_ld1<VALL:mode>): Define_expand modified.
3163 (aarch_st1<VALL:mode>): Likewise.
3164 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3165 (UNSPEC_ST1): Likewise.
3167 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3169 * config/microblaze/microblaze.md: Add trap insn and attribute
3171 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3173 PR preprocessor/58580
3174 * input.h (location_get_source_line): Take an additional line_size
3176 (void diagnostics_file_cache_fini): Declare new function.
3177 * input.c (struct fcache): New type.
3178 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3179 New static constants.
3180 (diagnostic_file_cache_init, total_lines_num)
3181 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3182 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3183 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3184 (get_next_line, read_next_line, goto_next_line, read_line_num):
3185 New static function definitions.
3186 (diagnostic_file_cache_fini): New function.
3187 (location_get_source_line): Take an additional output line_len
3188 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3190 * diagnostic.c (diagnostic_finish): Call
3191 diagnostic_file_cache_fini.
3192 (adjust_line): Take an additional input parameter for the length
3193 of the line, rather than calculating it with strlen.
3194 (diagnostic_show_locus): Adjust the use of
3195 location_get_source_line and adjust_line with respect to their new
3196 signature. While displaying a line now, do not stop at the first
3197 null byte. Rather, display the zero byte as a space and keep
3198 going until we reach the size of the line.
3199 * Makefile.in: Add vec.o to OBJS-libcommon
3201 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3202 Ilya Tocar <ilya.tocar@intel.com>
3204 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3205 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3206 (__builtin_ia32_kmov16): Ditto.
3207 * config/i386/i386.md (UNSPEC_KMOV): New.
3210 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3212 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3213 (_mm512_storeu_si512): Ditto.
3215 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3218 * rtl.h (get_referenced_operands): Declare.
3219 * recog.c (get_referenced_operands): New function.
3220 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3221 operands have been referenced when recording LO_SUM references.
3223 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3225 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3227 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3229 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3230 Enable for generic and recent AMD targets.
3232 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3234 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3235 ARG_SIZE note when adjustment was eliminated.
3237 2014-01-22 Jeff Law <law@redhat.com>
3239 PR tree-optimization/59597
3240 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3241 in file. Accept new argument REGISTERING and use it to modify
3242 dump output appropriately.
3243 (register_jump_thread): Corresponding changes.
3244 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3245 thread paths involving joiner blocks. Add code to dump cancelled
3246 jump threading paths.
3248 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3250 PR rtl-optimization/59477
3251 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3252 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3254 2014-01-22 Tom Tromey <tromey@redhat.com>
3256 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3258 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3260 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3262 PR rtl-optimization/59896
3263 * lra-constraints.c (process_alt_operands): Check unused note for
3264 matched operands of insn with no output reloads.
3266 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3268 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3269 (mips_move_from_gpr_cost): Likewise.
3271 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3273 PR rtl-optimization/59858
3274 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3275 ira_class_hard_regs_num.
3276 (process_alt_operands): Increase reject for dying matched operand.
3278 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3281 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3282 smaller than size, perform several stores or loads and stores
3283 at dst + count - size to store or copy all of size bytes, rather
3284 than just last modesize bytes.
3286 2014-01-20 DJ Delorie <dj@redhat.com>
3288 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3289 that CLOBBERs are REGs before propogating their values.
3291 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3294 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3295 (cgraph_inline_failed_type): New function.
3296 * cgraph.h (DEFCIFCODE): Add type.
3297 (cgraph_inline_failed_type_t): New enum.
3298 (cgraph_inline_failed_type): New prototype.
3299 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3300 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3301 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3302 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3303 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3304 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3305 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3306 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3307 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3308 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3309 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3310 OPTIMIZATION_MISMATCH.
3311 * tree-inline.c (expand_call_inline): Emit errors during
3312 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3314 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3317 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3318 mode attribute in insn output.
3320 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3322 * output.h (output_constant): Delete.
3323 * varasm.c (output_constant): Make private.
3325 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3327 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3329 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3332 * tree.h (fold_builtin_strcat): New prototype.
3333 * builtins.c (fold_builtin_strcat): No longer static. Add len
3334 argument, if non-NULL, don't call c_strlen. Optimize
3335 directly into __builtin_memcpy instead of __builtin_strcpy.
3336 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3337 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3339 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3341 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3342 for SImode_address_operand operands, having only a REG argument.
3344 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3346 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3347 loader name using mbig-endian.
3348 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3350 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3352 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3356 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3358 * config/aarch64/aarch64-protos.h
3359 (aarch64_cannot_change_mode_class_ptr): Declare.
3360 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3361 aarch64_cannot_change_mode_class_ptr): New.
3362 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3363 backend hook aarch64_cannot_change_mode_class.
3365 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3367 * common/config/aarch64/aarch64-common.c
3368 (aarch64_handle_option): Don't handle any option order logic here.
3369 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3370 selected_cpu, warn on architecture version mismatch.
3371 (aarch64_override_options): Fix parsing order for option strings.
3373 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3374 Iain Sandoe <iain@codesourcery.com>
3377 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3378 warning. Amend comment to reflect current functionality.
3380 2014-01-20 Richard Biener <rguenther@suse.de>
3383 * builtins.c (fold_builtin_strcat): Remove case better handled
3384 by tree-ssa-strlen.c.
3386 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3388 * config/aarch64/aarch64.opt
3389 (mcpu, march, mtune): Make case-insensitive.
3391 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3394 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3395 if operands[1] is a REG or ZERO_EXTEND of a REG.
3397 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3399 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3401 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3403 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3404 long non-pic millicode calls.
3406 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3408 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3410 2014-01-19 Kito Cheng <kito@0xlab.org>
3412 * builtins.c (expand_movstr): Check movstr expand done or fail.
3414 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3415 H.J. Lu <hongjiu.lu@intel.com>
3418 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3419 to DImode for zero-extended addresses.
3421 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3423 PR rtl-optimization/57763
3424 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3425 on the new indirect jump_insn and increment LABEL_NUSES (label).
3427 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3431 * config.gcc (x86_archs): New variable.
3432 (x86_64_archs): Likewise.
3433 (x86_cpus): Likewise.
3434 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3435 --with-arch/--with-cpu= options.
3436 Support --with-arch=/--with-cpu={nehalem,westmere,
3437 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3439 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3441 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3442 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3444 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3446 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3448 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3451 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3452 clear cpp_get_options (parse_in)->warn_unused_macros for
3453 ix86_target_macros_internal with cpp_define.
3455 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3457 * jump.c (delete_related_insns): Keep (use (insn))s.
3458 * reorg.c (redundant_insn): Check for barriers too.
3460 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3462 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3464 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3466 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3467 call to $$dyncall when TARGET_LONG_CALLS is true.
3469 2014-01-17 Jeff Law <law@redhat.com>
3471 * ree.c (combine_set_extension): Temporarily disable test for
3472 changing number of hard registers.
3474 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3477 * ipa-inline-analysis.c (inline_free_summary):
3478 Do not free summary of aliases.
3480 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3483 * gimplify.c (gimplify_expr): Use create_tmp_var
3484 instead of create_tmp_var_raw. If cond doesn't have
3485 integral type, don't add the IFN_ANNOTATE builtin at all.
3487 2014-01-17 Martin Jambor <mjambor@suse.cz>
3490 * ipa-cp.c (prev_edge_clone): New variable.
3491 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3492 Also resize prev_edge_clone vector.
3493 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3494 (ipcp_edge_removal_hook): New function.
3495 (ipcp_driver): Register ipcp_edge_removal_hook.
3497 2014-01-17 Andrew Pinski <apinski@cavium.com>
3498 Steve Ellcey <sellcey@mips.com>
3501 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3504 2014-01-17 Jeff Law <law@redhat.com>
3507 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3508 so that pass_ccp runs first.
3510 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3512 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3513 (ix86_adjust_cost): Use !TARGET_XXX.
3514 (do_reorder_for_imul): Likewise.
3515 (swap_top_of_ready_list): Likewise.
3516 (ix86_sched_reorder): Likewise.
3518 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3520 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3521 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3522 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3523 (intel_memset): New. Duplicate slm_memset.
3524 (intel_cost): New. Duplicate slm_cost.
3525 (m_INTEL): New macro.
3526 (processor_target_table): Add "intel".
3527 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3528 with PROCESSOR_INTEL for "intel".
3529 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3530 PROCESSOR_SILVERMONT.
3531 (ix86_issue_rate): Likewise.
3532 (ix86_adjust_cost): Likewise.
3533 (ia32_multipass_dfa_lookahead): Likewise.
3534 (swap_top_of_ready_list): Likewise.
3535 (ix86_sched_reorder): Likewise.
3536 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3537 instead of TARGET_OPT_AGU.
3538 * config/i386/i386.h (TARGET_INTEL): New.
3539 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3540 (processor_type): Add PROCESSOR_INTEL.
3541 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3542 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3544 2014-01-17 Marek Polacek <polacek@redhat.com>
3547 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3550 2014-01-17 Richard Biener <rguenther@suse.de>
3552 PR tree-optimization/46590
3553 * opts.c (default_options_table): Add entries for
3554 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3555 all enabled at -O1 but not for -Og.
3556 * common.opt (fbranch-count-reg): Remove Init(1).
3557 (fmove-loop-invariants): Likewise.
3558 (ftree-pta): Likewise.
3560 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3562 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3563 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3564 decls to at least the GCC 4.8 used alignments.
3567 * tree-nested.c (convert_nonlocal_reference_stmt,
3568 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3569 of GIMPLE_BIND stmts, adjust associated decls.
3571 2014-01-17 Richard Biener <rguenther@suse.de>
3573 PR tree-optimization/46590
3574 * vec.h (vec<>::bseach): New member function implementing
3575 binary search according to C89 bsearch.
3576 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3577 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3578 bitmap pointer again. Make accesses_in_loop a flat array.
3579 (mem_ref_obstack): New global.
3580 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3581 (mark_ref_stored): Likewise.
3582 (ref_indep_loop_p_2): Likewise.
3583 (set_ref_stored_in_loop): New helper function.
3584 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3585 (memref_free): Adjust.
3586 (record_mem_ref_loc): Simplify.
3587 (gather_mem_refs_stmt): Adjust.
3588 (sort_locs_in_loop_postorder_cmp): New function.
3589 (analyze_memory_references): Sort accesses_in_loop after
3590 loop postorder number.
3591 (find_ref_loc_in_loop_cmp): New function.
3592 (for_all_locs_in_loop): Find relevant cluster of locs in
3593 accesses_in_loop and iterate without recursion.
3594 (execute_sm): Avoid uninit warning.
3595 (struct ref_always_accessed): Simplify.
3596 (ref_always_accessed::operator ()): Likewise.
3597 (ref_always_accessed_p): Likewise.
3598 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3599 loop postorder numbers here.
3600 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3603 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3606 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3607 on decls for which assemble_alias has been called.
3609 2014-01-17 Nick Clifton <nickc@redhat.com>
3611 * config/msp430/msp430.opt: (mcpu): New option.
3612 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3613 (msp430_option_override): Parse target_cpu. If the MCU name
3614 matches a generic string, clear target_mcu.
3615 (msp430_attr): Allow numeric interrupt values up to 63.
3616 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3617 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3619 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3621 * config/msp430/msp430.md (popm): Use %J rather than %I.
3622 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3623 (addhi_cy_i): Use immediate_operand for operand 2.
3624 * doc/invoke.texi: Document -mcpu option.
3626 2014-01-17 Richard Biener <rguenther@suse.de>
3628 PR rtl-optimization/38518
3629 * df.h (df_analyze_loop): Declare.
3630 * df-core.c: Include cfgloop.h.
3631 (df_analyze_1): Split out main part of df_analyze.
3632 (df_analyze): Adjust.
3633 (loop_inverted_post_order_compute): New function.
3634 (loop_post_order_compute): Likewise.
3635 (df_analyze_loop): New function avoiding whole-function
3637 * loop-invariant.c (find_defs): Use df_analyze_loop.
3638 (find_invariants): Adjust.
3639 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3641 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3643 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3644 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3646 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3648 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3649 traversal when removing references.
3651 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3654 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3656 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3659 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3660 pushing a reload for an autoinc when we had previously reloaded an
3661 inner part of the address.
3663 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3665 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3667 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3668 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3669 when not giving up or versioning for alias only because of
3671 (vect_analyze_data_ref_dependences): Set to true.
3672 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3674 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3675 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3679 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3682 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3683 operand 0 predicate for gathers, use a new pseudo as subtarget.
3685 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3688 * lra-constraints.c (process_alt_operands): Add printing debug info.
3689 Check absence of input/output reloads for matched operands too.
3691 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3693 PR rtl-optimization/59835
3694 * ira.c (ira_init_register_move_cost): Increase cost for
3697 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3699 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3701 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3704 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3705 non-register objects. Use gen_(high/low)part more consistently.
3708 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3711 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3712 endian support, remove tests for WORDS_BIG_ENDIAN.
3713 (p8_mfvsrd_3_<mode>): Likewise.
3714 (reload_gpr_from_vsx<mode>): Likewise.
3715 (reload_gpr_from_vsxsf): Likewise.
3716 (p8_mfvsrd_4_disf): Likewise.
3718 2014-01-16 Richard Biener <rguenther@suse.de>
3720 PR rtl-optimization/46590
3721 * lcm.c (compute_antinout_edge): Use postorder iteration.
3722 (compute_laterin): Use inverted postorder iteration.
3724 2014-01-16 Nick Clifton <nickc@redhat.com>
3727 * varasm.c (output_constant): Return the number of bytes actually
3729 (output_constructor_array_range): Update the field size with the
3730 number of bytes emitted by output_constant.
3731 (output_constructor_regular_field): Likewise. Also do not
3732 complain if the total number of bytes emitted is now greater
3733 than the expected fieldpos.
3734 * output.h (output_constant): Update prototype and descriptive comment.
3736 2014-01-16 Marek Polacek <polacek@redhat.com>
3739 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3740 it is error_mark_node.
3742 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3744 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3745 VALID_AVX256_REG_OR_OI_MODE.
3747 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3749 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3750 current procedure should be profiled.
3752 2014-01-15 Andrew Pinski <apinski@cavium.com>
3754 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3755 of moving from/to the STACK_REG register class.
3757 2014-01-15 Richard Henderson <rth@redhat.com>
3760 * reginfo.c (global_regs_decl): Globalize.
3761 * rtl.h (global_regs_decl): Declare.
3762 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3763 reserved via global_regs.
3765 2014-01-15 Teresa Johnson <tejohnson@google.com>
3767 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3769 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3771 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3772 and vmulosh rather than call gen_vec_widen_smult_*.
3773 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3774 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3775 (vec_widen_smult_even_v16qi): Likewise.
3776 (vec_widen_umult_even_v8hi): Likewise.
3777 (vec_widen_smult_even_v8hi): Likewise.
3778 (vec_widen_umult_odd_v16qi): Likewise.
3779 (vec_widen_smult_odd_v16qi): Likewise.
3780 (vec_widen_umult_odd_v8hi): Likewise.
3781 (vec_widen_smult_odd_v8hi): Likewise.
3782 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3783 vmuloub rather than call gen_vec_widen_umult_*.
3784 (vec_widen_umult_lo_v16qi): Likewise.
3785 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3786 vmulosb rather than call gen_vec_widen_smult_*.
3787 (vec_widen_smult_lo_v16qi): Likewise.
3788 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3789 rather than call gen_vec_widen_umult_*.
3790 (vec_widen_umult_lo_v8hi): Likewise.
3791 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3792 rather than call gen_vec_widen_smult_*.
3793 (vec_widen_smult_lo_v8hi): Likewise.
3795 2014-01-15 Jeff Law <law@redhat.com>
3797 PR tree-optimization/59747
3798 * ree.c (find_and_remove_re): Properly handle case where a second
3799 eliminated extension requires widening a copy created for elimination
3800 of a prior extension.
3801 (combine_set_extension): Ensure that the number of hard regs needed
3802 for a destination register does not change when we widen it.
3804 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3806 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3807 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3808 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3809 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3810 (avr-*-rtems*): Likewise.
3811 (bfin*-rtems*): Likewise.
3812 (moxie-*-rtems*): Likewise.
3813 (h8300-*-rtems*): Likewise.
3814 (i[34567]86-*-rtems*): Likewise.
3815 (lm32-*-rtems*): Likewise.
3816 (m32r-*-rtems*): Likewise.
3817 (m68k-*-rtems*): Likewise.
3818 (microblaze*-*-rtems*): Likewise.
3819 (mips*-*-rtems*): Likewise.
3820 (powerpc-*-rtems*): Likewise.
3821 (sh-*-rtems*): Likewise.
3822 (sparc-*-rtems*): Likewise.
3823 (sparc64-*-rtems*): Likewise.
3824 (v850-*-rtems*): Likewise.
3825 (m32c-*-rtems*): Likewise.
3827 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3829 PR rtl-optimization/59511
3830 * ira.c (ira_init_register_move_cost): Use memory costs for some
3831 cases of register move cost calculations.
3832 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3833 instead of BB frequency.
3834 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3835 * lra-assigns.c (find_hard_regno_for): Ditto.
3837 2014-01-15 Richard Biener <rguenther@suse.de>
3839 PR tree-optimization/59822
3840 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3841 (vectorizable_load): Use it to hoist defs of uses of invariant
3842 loads out of the loop.
3844 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3845 Kugan Vivekanandarajah <kuganv@linaro.org>
3848 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3851 2014-01-15 Richard Biener <rguenther@suse.de>
3853 PR rtl-optimization/59802
3854 * lcm.c (compute_available): Use inverted postorder to seed
3855 the initial worklist.
3857 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3860 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3861 ADDR_REGS for invalid symrefs in non-PIC code.
3863 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3866 * builtins.c (determine_block_size): Initialize *probable_max_size
3867 even if len_rtx is CONST_INT.
3869 2014-01-14 Andrew Pinski <apinski@cavium.com>
3871 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3872 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3873 (cortexa53_tunings): Likewise.
3874 (aarch64_sched_issue_rate): New function.
3875 (TARGET_SCHED_ISSUE_RATE): Define.
3877 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3879 * ira-costs.c (find_costs_and_classes): Add missed
3880 ira_init_register_move_cost_if_necessary.
3882 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3885 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3887 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3890 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3891 to indicate if type is used for function return value. Warn ABI
3892 change if the vector mode isn't available for function return value.
3893 (ix86_function_arg_advance): Pass false to type_natural_mode.
3894 (ix86_function_arg): Likewise.
3895 (ix86_gimplify_va_arg): Likewise.
3896 (function_arg_32): Don't warn ABI change.
3897 (ix86_function_value): Pass true to type_natural_mode.
3898 (ix86_return_in_memory): Likewise.
3899 (ix86_struct_value_rtx): Removed.
3900 (TARGET_STRUCT_VALUE_RTX): Likewise.
3902 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3904 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3905 converting a conditional jump into a conditional return.
3907 2014-01-14 Richard Biener <rguenther@suse.de>
3909 PR tree-optimization/58921
3910 PR tree-optimization/59006
3911 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3912 hoisting invariant stmts.
3913 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3914 invariant loads on the preheader edge if possible.
3916 2014-01-14 Joey Ye <joey.ye@arm.com>
3918 * doc/plugin.texi (Building GCC plugins): Update to C++.
3920 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3922 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3923 (_mm_rcp28_round_ss): Ditto.
3924 (_mm_rsqrt28_round_sd): Ditto.
3925 (_mm_rsqrt28_round_ss): Ditto.
3926 (_mm_rcp28_sd): Ditto.
3927 (_mm_rcp28_ss): Ditto.
3928 (_mm_rsqrt28_sd): Ditto.
3929 (_mm_rsqrt28_ss): Ditto.
3930 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3931 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3932 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3933 (IX86_BUILTIN_RCP28SD): Ditto.
3934 (IX86_BUILTIN_RCP28SS): Ditto.
3935 (IX86_BUILTIN_RSQRT28SD): Ditto.
3936 (IX86_BUILTIN_RSQRT28SS): Ditto.
3937 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3938 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3939 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3940 (ix86_expand_special_args_builtin): Expand new FTYPE.
3941 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3942 (srcp14<mode>): Make insn unary.
3943 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3944 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3945 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3946 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3947 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3948 Fix rounding: make it SAE only.
3949 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3951 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3953 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3954 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3955 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3956 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3957 (round_saeonly_mask_scalar_operand4): Ditto.
3958 (round_saeonly_mask_scalar_op3): Ditto.
3959 (round_saeonly_mask_scalar_op4): Ditto.
3961 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3963 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3964 Implement -maltivec=be for vec_insert and vec_extract.
3966 2014-01-10 DJ Delorie <dj@redhat.com>
3968 * config/msp430/msp430.md (call_internal): Don't allow memory
3969 references with SP as the base register.
3970 (call_value_internal): Likewise.
3971 * config/msp430/constraints.md (Yc): New. For memory references
3972 that don't use SP as a base register.
3974 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3975 "an integer without a # prefix"
3976 * config/msp430/msp430.md (epilogue_helper): Use it.
3978 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3981 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3982 AVX512F gather builtins.
3983 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3984 on gather decls with INTEGER_TYPE masktype.
3985 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3986 directly into the builtin rather than hoisting it before loop.
3988 PR tree-optimization/59387
3989 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3990 (scev_const_prop): If folded_casts and type has undefined overflow,
3991 use force_gimple_operand instead of force_gimple_operand_gsi and
3992 for each added stmt if it is assign with
3993 arith_code_with_undefined_signed_overflow, call
3994 rewrite_to_defined_overflow.
3995 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3996 gimple-fold.h instead.
3997 (arith_code_with_undefined_signed_overflow,
3998 rewrite_to_defined_overflow): Moved to ...
3999 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4000 rewrite_to_defined_overflow): ... here. No longer static.
4001 Include gimplify-me.h.
4002 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4003 rewrite_to_defined_overflow): New prototypes.
4005 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4007 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4009 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4011 * builtins.c (get_object_alignment_2): Minor tweak.
4012 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4014 2014-01-13 Christian Bruel <christian.bruel@st.com>
4016 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4017 optimized non constant lengths.
4019 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4022 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4023 load as __atomic_load_N if possible.
4025 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4027 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4029 (rs6000_expand_builtin): Adjust call.
4031 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4034 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4035 * config/rs6000/rs6000.c: Include target-globals.h.
4036 (rs6000_set_current_function): Instead of doing target_reinit
4037 unconditionally, use save_target_globals_default_opts and
4038 restore_target_globals.
4040 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4042 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4043 (rs6000_expand_builtin): Handle mffs and mtfsf.
4044 (rs6000_init_builtins): Define mffs and mtfsf.
4045 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4046 (rs6000_mffs): New pattern.
4047 (rs6000_mtfsf): New pattern.
4049 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4051 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4052 Start narrowing with START. Apply candidate-use pair
4053 and check overall cost in narrowing.
4054 (iv_ca_prune): Pass new argument.
4056 2014-01-10 Jeff Law <law@redhat.com>
4059 * ree.c (combine_reaching_defs): Ensure the defining statement
4060 occurs before the extension when optimizing extensions with
4061 different source and destination hard registers.
4063 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4066 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4067 vtables into the type inheritance graph.
4069 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4071 PR rtl-optimization/59754
4072 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4073 modes in the REGNO != REGNO case.
4075 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4077 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4079 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4081 PR tree-optimization/59745
4082 * tree-predcom.c (tree_predictive_commoning_loop): Call
4083 free_affine_expand_cache if giving up because components is NULL.
4085 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4086 GC in payload of target_globals struct instead of allocating them on
4087 the heap and the larger structs separately using GC.
4088 * target-globals.h (struct target_globals): Make regs, hard_regs,
4089 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4090 of GTY((skip)) and change type to void *.
4091 (reset_target_globals): Cast loads from those fields to corresponding
4094 2014-01-10 Steve Ellcey <sellcey@mips.com>
4097 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4098 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4099 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4101 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4104 * aarch64-modes.def (CC_Zmode): New flags mode.
4105 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4106 represents an equality.
4107 (aarch64_get_condition_code): Handle CC_Zmode.
4108 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4110 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4112 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4113 extraction in good case.
4115 2014-01-10 Richard Biener <rguenther@suse.de>
4117 PR tree-optimization/59374
4118 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4119 checking after SLP discovery. Mark stmts not participating
4120 in any SLP instance properly.
4122 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4124 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4125 when handling a SET rtx.
4127 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4129 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4130 (cortex-a57): Likewise.
4131 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4133 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4135 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4136 non-iwmmxt builtins.
4138 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4142 * ipa-devirt.c record_target_from_binfo): Take as argument
4143 stack of binfos and lookup matching one for virtual inheritance.
4144 (possible_polymorphic_call_targets_1): Update.
4146 2014-01-10 Huacai Chen <chenhc@lemote.com>
4148 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4149 kernel strings for Loongson-2E/2F/3A.
4151 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4154 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4155 is_gimple_call before calling gimple_call_internal_p.
4157 2014-01-09 Steve Ellcey <sellcey@mips.com>
4159 * Makefile.in (TREE_FLOW_H): Remove.
4160 (TREE_SSA_H): Add file names from tree-flow.h.
4161 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4162 * tree.h: Remove tree-flow.h reference.
4163 * hash-table.h: Remove tree-flow.h reference.
4164 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4165 reference with tree-ssa-loop.h.
4167 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4169 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4170 default element-order behavior for -maltivec.
4171 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4172 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4173 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4174 when targeting big endian, at least for now.
4175 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4177 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4180 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4181 var satisfies use_register_for_decl, just take into account type
4182 alignment, rather than decl alignment.
4184 PR tree-optimization/59622
4185 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4186 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4187 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4188 Don't devirtualize for inplace at all. For targets.length () == 1,
4189 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4191 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4193 * config/i386/i386.md (cpu): Remove the unused btver1.
4195 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4197 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4199 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4202 * tree-core.h (struct target_globals): New forward declaration.
4203 (struct tree_target_option): Add globals field.
4204 * tree.h (TREE_TARGET_GLOBALS): Define.
4205 (prepare_target_option_nodes_for_pch): New prototype.
4206 * target-globals.h (struct target_globals): Define even if
4208 * tree.c (prepare_target_option_node_for_pch,
4209 prepare_target_option_nodes_for_pch): New functions.
4210 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4211 * config/i386/i386.c: Include target-globals.h.
4212 (ix86_set_current_function): Instead of doing target_reinit
4213 unconditionally, use save_target_globals_default_opts and
4214 restore_target_globals.
4216 2014-01-09 Richard Biener <rguenther@suse.de>
4218 PR tree-optimization/59715
4219 * tree-cfg.h (split_critical_edges): Declare.
4220 * tree-cfg.c (split_critical_edges): Export.
4221 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4223 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4225 * cfgexpand.c (expand_stack_vars): Optionally disable
4226 asan stack protection.
4227 (expand_used_vars): Likewise.
4228 (partition_stack_vars): Likewise.
4229 * asan.c (asan_emit_stack_protection): Optionally disable
4230 after return stack usage.
4231 (instrument_derefs): Optionally disable memory access instrumentation.
4232 (instrument_builtin_call): Likewise.
4233 (instrument_strlen_call): Likewise.
4234 (asan_protect_global): Optionally disable global variables protection.
4235 * doc/invoke.texi: Added doc for new options.
4236 * params.def: Added new options.
4237 * params.h: Likewise.
4239 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4241 PR rtl-optimization/59724
4242 * ifcvt.c (cond_exec_process_if_block): Don't call
4243 flow_find_head_matching_sequence with 0 longest_match.
4244 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4245 non-active insns if !stop_after.
4246 (try_head_merge_bb): Revert 2014-01-07 changes.
4248 2014-01-08 Jeff Law <law@redhat.com>
4250 * ree.c (get_sub_rtx): New function, extracted from...
4251 (merge_def_and_ext): Here.
4252 (combine_reaching_defs): Use get_sub_rtx.
4254 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4256 * cgraph.h (varpool_variable_node): Do not choke on null node.
4258 2014-01-08 Catherine Moore <clm@codesourcery.com>
4260 * config/mips/mips.md (simple_return): Attempt to use JRC
4262 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4264 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4266 PR rtl-optimization/59137
4267 * reorg.c (steal_delay_list_from_target): Call update_block for
4269 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4271 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4273 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4274 two duplicate entries.
4276 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4279 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4281 * config/mips/mips.c (mips_truncated_op_cost): New function.
4282 (mips_rtx_costs): Adjust test for BADDU.
4283 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4285 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4287 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4288 (*baddu_si): ...this new pattern.
4290 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4293 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4295 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4298 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4300 (expand_expr, expand_normal): Adjust.
4301 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4302 inner_reference_p. Use inner_reference_p to expand inner references.
4303 (store_expr): Adjust.
4304 * cfgexpand.c (expand_call_stmt): Adjust.
4306 2014-01-08 Rong Xu <xur@google.com>
4308 * gcov-io.c (gcov_var): Move from gcov-io.h.
4309 (gcov_position): Ditto.
4310 (gcov_is_error): Ditto.
4311 (gcov_rewrite): Ditto.
4312 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4313 only part to libgcc/libgcov.h.
4315 2014-01-08 Marek Polacek <polacek@redhat.com>
4318 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4320 2014-01-08 Marek Polacek <polacek@redhat.com>
4323 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4325 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4327 PR rtl-optimization/59649
4328 * stor-layout.c (get_mode_bounds): For BImode return
4329 0 and STORE_FLAG_VALUE.
4331 2014-01-08 Richard Biener <rguenther@suse.de>
4334 * gimple.h (is_gimple_builtin_call): Remove.
4335 (gimple_builtin_call_types_compatible_p): New.
4336 (gimple_call_builtin_p): New overload.
4337 * gimple.c (is_gimple_builtin_call): Remove.
4338 (validate_call): Rename to ...
4339 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4341 (validate_type): New static function.
4342 (gimple_call_builtin_p): New overload and adjust.
4343 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4344 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4345 (gimple_fold_stmt_to_constant_1): Likewise.
4346 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4348 2014-01-08 Richard Biener <rguenther@suse.de>
4351 * gimplify.c (gimplify_expr): Gimplify register-register type
4352 VIEW_CONVERT_EXPRs to separate stmts.
4354 2014-01-07 Jeff Law <law@redhat.com>
4357 * ree.c (combine_set_extension): Handle case where source
4358 and destination registers in an extension insn are different.
4359 (combine_reaching_defs): Allow source and destination registers
4360 in extension to be different under limited circumstances.
4361 (add_removable_extension): Remove restriction that the
4362 source and destination registers in the extension are the same.
4363 (find_and_remove_re): Emit a copy from the extension's
4364 destination to its source after the defining insn if
4365 the source and destination registers are different.
4368 * ifcvt.c (merge_if_block): If we are merging a block with more than
4369 one successor with a block with no successors, remove any BARRIER
4370 after the second block.
4372 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4374 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4376 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4379 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4380 for 14-bit register offsets when INT14_OK_STRICT is false.
4382 2014-01-07 Roland Stigge <stigge@antcom.de>
4383 Michael Meissner <meissner@linux.vnet.ibm.com>
4386 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4387 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4389 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4391 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4394 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4396 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4397 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4398 rtx is const0_rtx or not.
4400 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4403 * target-globals.c (save_target_globals): Remove this_fn_optab
4405 * toplev.c: Include optabs.h.
4406 (target_reinit): Temporarily restore the global options if another
4407 set of options are in force.
4409 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4411 PR rtl-optimization/58668
4412 * cfgcleanup.c (flow_find_cross_jump): Don't count
4413 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4414 to determine what is counted.
4415 (flow_find_head_matching_sequence): Use active_insn_p to determine
4417 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4419 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4420 determine what is counted.
4422 PR tree-optimization/59643
4423 * tree-predcom.c (split_data_refs_to_components): If one dr is
4424 read and one write, determine_offset fails and the write isn't
4425 in the bad component, just put the read into the bad component.
4427 2014-01-07 Mike Stump <mikestump@comcast.net>
4428 Jakub Jelinek <jakub@redhat.com>
4431 * tree-core.h (struct tree_optimization_option): Change optabs
4432 type from unsigned char * to void *.
4433 * optabs.c (init_tree_optimization_optabs): Adjust
4434 TREE_OPTIMIZATION_OPTABS initialization.
4436 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4439 * config/i386/i386.h (struct machine_function): Add
4440 no_drap_save_restore field.
4441 * config/i386/i386.c (ix86_save_reg): Use
4442 !cfun->machine->no_drap_save_restore instead of
4443 crtl->stack_realign_needed.
4444 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4445 this function clears frame_pointer_needed. Set
4446 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4447 and DRAP reg is needed.
4449 2014-01-06 Marek Polacek <polacek@redhat.com>
4452 * doc/implement-c.texi: Mention that other integer types are
4453 permitted as bit-field types in strictly conforming mode.
4455 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4457 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4460 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4462 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4464 2014-01-06 Martin Jambor <mjambor@suse.cz>
4467 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4469 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4471 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4475 * var-tracking.c (add_stores): Preserve the value of the source even if
4476 we don't record the store.
4478 2014-01-06 Terry Guo <terry.guo@arm.com>
4480 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4482 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4485 * config/darwin.c (darwin_function_section): Adjust return values to
4486 correspond to optimisation changes made in r206070.
4488 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4490 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4491 from prefetch_block tune setting.
4492 (nocona_cost): Correct size of prefetch block to 64.
4494 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4496 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4497 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4498 used to save the static chain register in the computation of the offset
4499 from which the FP registers need to be restored.
4501 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4503 PR tree-optimization/59519
4504 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4505 ICE if get_current_def (current_new_name) is already non-NULL, as long
4506 as it is a phi result of some other phi in *new_exit_bb that has
4509 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4510 or vmovdqu* for misaligned_operand.
4511 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4512 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4513 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4514 aligned_mem for AVX512F masked aligned load and store builtins and for
4517 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4519 PR tree-optimization/59651
4520 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4521 Address range for negative step should be added by TYPE_SIZE_UNIT.
4523 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4525 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4526 overlapping registers also for registers other than the stack pointer.
4528 2014-01-03 Marek Polacek <polacek@redhat.com>
4531 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4534 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4537 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4540 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4541 (push splitter): Use <P:MODE_SIZE> instead of
4542 GET_MODE_SIZE (<P:MODE>mode).
4543 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4544 (mov -1, reg peephole2): Likewise.
4545 * config/i386/sse.md (*mov<mode>_internal,
4546 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4547 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4548 *<code><mode>3, *andnot<mode>3<mask_name>,
4549 <mask_codefor><code><mode>3<mask_name>): Likewise.
4550 * config/i386/subst.md (mask_mode512bit_condition,
4551 sd_mask_mode512bit_condition): Likewise.
4553 2014-01-02 Xinliang David Li <davidxl@google.com>
4555 PR tree-optimization/59303
4556 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4557 (dump_predicates): Better output format.
4558 (pred_equal_p): New function.
4559 (is_neq_relop_p): Ditto.
4560 (is_neq_zero_form_p): Ditto.
4561 (pred_expr_equal_p): Ditto.
4562 (pred_neg_p): Ditto.
4563 (simplify_pred): Ditto.
4564 (simplify_preds_2): Ditto.
4565 (simplify_preds_3): Ditto.
4566 (simplify_preds_4): Ditto.
4567 (simplify_preds): Ditto.
4569 (push_to_worklist): Ditto.
4570 (get_pred_info_from_cmp): Ditto.
4571 (is_degenerated_phi): Ditto.
4572 (normalize_one_pred_1): Ditto.
4573 (normalize_one_pred): Ditto.
4574 (normalize_one_pred_chain): Ditto.
4575 (normalize_preds): Ditto.
4576 (normalize_cond_1): Remove function.
4577 (normalize_cond): Ditto.
4578 (is_gcond_subset_of): Ditto.
4579 (is_subset_of_any): Ditto.
4580 (is_or_set_subset_of): Ditto.
4581 (is_and_set_subset_of): Ditto.
4582 (is_norm_cond_subset_of): Ditto.
4583 (pred_chain_length_cmp): Ditto.
4584 (convert_control_dep_chain_into_preds): Type change.
4585 (find_predicates): Ditto.
4586 (find_def_preds): Ditto.
4587 (destroy_predicates_vecs): Ditto.
4588 (find_matching_predicates_in_rest_chains): Ditto.
4589 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4590 (is_pred_expr_subset): Ditto.
4591 (is_pred_chain_subset_of): Ditto.
4592 (is_included_in): Ditto.
4593 (is_superset_of): Ditto.
4595 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4597 Update copyright years.
4599 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4601 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4602 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4603 config/arc/arc.md, config/arc/arc.opt,
4604 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4605 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4606 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4607 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4608 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4609 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4610 vtable-verify.c, vtable-verify.h: Use the standard form for the
4613 2014-01-02 Tobias Burnus <burnus@net-b.de>
4615 * gcc.c (process_command): Update copyright notice dates.
4616 * gcov-dump.c: Ditto.
4618 * doc/cpp.texi: Bump @copying's copyright year.
4619 * doc/cppinternals.texi: Ditto.
4620 * doc/gcc.texi: Ditto.
4621 * doc/gccint.texi: Ditto.
4622 * doc/gcov.texi: Ditto.
4623 * doc/install.texi: Ditto.
4624 * doc/invoke.texi: Ditto.
4626 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4628 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4630 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4632 * config/i386/sse.md (*mov<mode>_internal): Guard
4633 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4635 PR rtl-optimization/59647
4636 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4637 new_rtx into UNSIGNED_FLOAT rtxes.
4639 Copyright (C) 2014 Free Software Foundation, Inc.
4641 Copying and distribution of this file, with or without modification,
4642 are permitted in any medium without royalty provided the copyright
4643 notice and this notice are preserved.