1 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
5 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
7 * cselib.c (remove_useless_values): Skip to avoid quadratic
8 behavior if the condition moved from...
9 (cselib_process_insn): ... here holds.
11 2014-03-05 Jakub Jelinek <jakub@redhat.com>
14 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
15 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
18 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
19 (TM_H): Add x86-tune.def.
21 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23 * config/aarch64/aarch64.c (generic_tunings):
24 Use cortexa57_extra_costs.
26 2014-03-05 Jakub Jelinek <jakub@redhat.com>
29 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
30 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
31 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
34 2014-03-04 Heiher <r@hev.cc>
36 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
37 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
39 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
41 * config/i386/predicates.md (const2356_operand): Change to ...
42 (const2367_operand): ... this.
43 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
45 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
46 (*avx512pf_scatterpf<mode>sf): Ditto.
47 (avx512pf_scatterpf<mode>df): Ditto.
48 (*avx512pf_scatterpf<mode>df_mask): Ditto.
49 (*avx512pf_scatterpf<mode>df): Ditto.
50 * config/i386/i386.c (ix86_expand_builtin): Update
51 incorrect hint operand error message.
53 2014-03-04 Richard Biener <rguenther@suse.de>
55 * lto-section-in.c (lto_get_section_data): Fix const cast.
57 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
59 * tree-streamer.c (record_common_node): Assert we don't record
60 nodes with type double.
61 (preload_common_node): Skip type double, complex double and
62 double pointer since it is now frontend dependent due to
65 2014-03-04 Richard Biener <rguenther@suse.de>
68 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
69 (lto_input_toplevel_asms): Likewise.
70 * lto-section-in.c (lto_get_section_data): Instead do it here
73 2014-03-04 Richard Biener <rguenther@suse.de>
75 PR tree-optimization/60382
76 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
77 dead PHIs a reduction.
79 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
81 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
83 (_mm_prefetch): Move out of GCC target("sse") pragma.
84 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
85 GCC target("prfchw") pragma.
86 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
88 * config/i386/i386.c (ix86_option_override_internal): Enable
89 -mprfchw with -mprefetchwt1.
91 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
93 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
96 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
98 * opts.h (CL_PCH_IGNORE): Define.
99 * targhooks.c (option_affects_pch_p):
100 Return false for options that have CL_PCH_IGNORE set.
101 * opt-functions.awk: Process PchIgnore.
102 * doc/options.texi: Document PchIgnore.
104 * config/arc/arc.opt (misize): Add PchIgnore property.
106 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
108 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
109 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
110 constraint on constants to permit them being loaded into
111 GENERAL_REGS or BASE_REGS.
113 2014-03-03 Nick Clifton <nickc@redhat.com>
115 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
116 anti-cacnonical alternatives.
117 (negandhi3_real): New pattern.
118 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
120 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
122 * config/avr/avr-mcus.def: Remove atxmega16x1.
123 * config/avr/avr-tables.opt: Regenerate.
124 * config/avr/t-multilib: Regenerate.
125 * doc/avr-mmcu.texi: Regenerate.
127 2014-03-03 Tobias Grosser <tobias@grosser.es>
128 Mircea Namolaru <mircea.namolaru@inria.fr>
130 PR tree-optimization/58028
131 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
134 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
136 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
137 not handled by recognizers.
139 2014-03-03 Jakub Jelinek <jakub@redhat.com>
142 * function.c (expand_function_end): Don't emit
143 clobber_return_register sequence if clobber_after is a BARRIER.
144 * cfgexpand.c (construct_exit_block): Append instructions before
145 return_label to prev_bb.
147 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
149 * config/rs6000/constraints.md: Document reserved use of "wc".
151 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
154 * ipa.c (function_and_variable_visibility): When dissolving comdat
155 group, also set all symbols to local.
157 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
162 2013-12-14 Jan Hubicka <jh@suse.cz>
164 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
166 2014-03-02 Jon Beniston <jon@beniston.com>
172 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
173 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
174 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
175 (simple_return, *simple_return): New patterns
176 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
177 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
179 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
181 * dwarf2out.c (gen_subprogram_die): Tidy.
183 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
186 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
187 (*mov_t_msb_neg_negc): ... this new insn.
189 2014-02-28 Jason Merrill <jason@redhat.com>
192 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
195 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
198 * dwarf2out.c (decltype_auto_die): New static.
199 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
200 (gen_type_die_with_usage): Handle 'decltype(auto)'.
201 (is_cxx_auto): Likewise.
203 2014-02-28 Ian Bolton <ian.bolton@arm.com>
205 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
206 we are not using general regs only.
208 2014-02-28 Richard Biener <rguenther@suse.de>
211 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
212 previous fix and only allow to remove trivial pre-headers
213 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
214 (remove_forwarder_block): Properly update the latch of a loop.
216 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
219 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
220 (cselib_preserved_hash_table): New.
221 (preserve_constants_and_equivs): Move preserved vals to it.
222 (cselib_find_slot): Look it up first.
223 (cselib_init): Initialize it.
224 (cselib_finish): Release it.
225 (dump_cselib_table): Dump it.
227 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
230 * cselib.c (remove_useless_values): Skip to avoid quadratic
231 behavior if the condition moved from...
232 (cselib_process_insn): ... here holds.
234 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
237 * var-tracking.c (vt_initialize): Apply the same condition to
238 preserve the CFA base value.
240 2014-02-28 Joey Ye <joey.ye@arm.com>
243 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
244 if reload in progress or completed.
246 2014-02-28 Tobias Burnus <burnus@net-b.de>
249 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
252 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
254 * doc/tm.texi.in (Condition Code Status): Update documention for
255 relative locations of cc0-setter and cc0-user.
257 2014-02-27 Jeff Law <law@redhat.com>
259 PR rtl-optimization/52714
260 * combine.c (try_combine): When splitting an unrecognized PARALLEL
261 into two independent simple sets, if I3 is a jump, ensure the
262 pattern we place into I3 is a (set (pc) ...).
264 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
265 Jeff Law <law@redhat.com>
267 PR rtl-optimization/49847
268 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
269 are in different blocks.
270 * doc/tm.texi (Condition Code Status): Update documention for
271 relative locations of cc0-setter and cc0-user.
273 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
276 * lra.c (lra_emit_add): Check SUBREG too.
278 2014-02-27 Andreas Schwab <schwab@suse.de>
280 * config/m68k/m68k.c (m68k_option_override): Disable
281 -flive-range-shrinkage for classic m68k.
282 (m68k_override_options_after_change): Likewise.
284 2014-02-27 Marek Polacek <polacek@redhat.com>
287 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
288 -Wmaybe-uninitialized.
290 2014-02-27 Alan Modra <amodra@gmail.com>
293 * reload1.c (emit_input_reload_insns): When reload_override_in,
294 set old to rl->in_reg when rl->in_reg is a subreg.
296 2014-02-26 Richard Biener <rguenther@suse.de>
299 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
301 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
303 * common/config/i386/predicates.md (const1256_operand): Remove.
304 (const2356_operand): New.
305 (const_1_to_2_operand): Remove.
306 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
307 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
308 (*avx512pf_gatherpf<mode>sf): Ditto.
309 (avx512pf_gatherpf<mode>df): Ditto.
310 (*avx512pf_gatherpf<mode>df_mask): Ditto.
311 (*avx512pf_gatherpf<mode>df): Ditto.
312 (avx512pf_scatterpf<mode>sf): Ditto.
313 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
314 (*avx512pf_scatterpf<mode>sf): Ditto.
315 (avx512pf_scatterpf<mode>df): Ditto.
316 (*avx512pf_scatterpf<mode>df_mask): Ditto.
317 (*avx512pf_scatterpf<mode>df): Ditto.
318 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
320 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
322 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
323 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
324 (_mm512_mask_testn_epi64_mask): Move to ...
325 * config/i386/avx512cdintrin.h: Here.
326 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
327 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
328 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
329 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
330 TARGET_AVX512F from TARGET_AVX512CD.
332 2014-02-26 Richard Biener <rguenther@suse.de>
335 * ipa.c (walk_polymorphic_call_targets): Properly guard
336 call to inline_update_overall_summary.
338 2014-02-26 Bin Cheng <bin.cheng@arm.com>
341 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
342 and latches only if requested. Fix latch if it is removed.
343 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
344 LOOPS_HAVE_PREHEADERS.
346 2014-02-25 Andrew Pinski <apinski@cavium.com>
348 * builtins.c (expand_builtin_thread_pointer): Create a new target
349 when the target is NULL.
351 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
353 PR rtl-optimization/60317
354 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
355 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
356 * lra-assigns.c: Include params.h.
357 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
358 other reload pseudos considerations.
360 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
362 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
363 to use canonical form for nor<mode>3.
365 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
368 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
371 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
373 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
374 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
375 (ix86_handle_option): Handle OPT_mprefetchwt1.
376 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
377 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
379 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
380 OPTION_MASK_ISA_PREFETCHWT1.
381 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
382 (PTA_PREFETCHWT1): New.
383 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
384 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
385 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
386 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
387 (*prefetch_avx512pf_<mode>_: Change into ...
388 (*prefetch_prefetchwt1_<mode>: This.
389 * config/i386/i386.opt (mprefetchwt1): New.
390 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
391 (_mm_prefetch): Handle intent to write.
392 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
394 2014-02-25 Richard Biener <rguenther@suse.de>
397 * emit-rtl.c (mem_attrs_htab): Remove.
398 (mem_attrs_htab_hash): Likewise.
399 (mem_attrs_htab_eq): Likewise.
400 (set_mem_attrs): Always allocate new mem-attrs when something changed.
401 (init_emit_once): Do not allocate mem_attrs_htab.
403 2014-02-25 Richard Biener <rguenther@suse.de>
406 * lto-opts.c (lto_write_options): Output non-explicit conservative
407 -fwrapv, -fno-trapv and -fno-strict-overflow.
408 * lto-wrapper.c (merge_and_complain): Handle merging those options.
409 (run_gcc): And pass them through.
411 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
413 * sel-sched.c (calculate_new_fences): New parameter ptime.
414 Calculate it as a maximum over all fence cycles.
415 (sel_sched_region_2): Adjust the call to calculate_new_fences.
416 Print the final schedule timing when sched_verbose.
418 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
420 PR rtl-optimization/60292
421 * sel-sched.c (fill_vec_av_set): Do not reset target availability
422 bit fot the fence instruction.
424 2014-02-24 Alangi Derick <alangiderick@gmail.com>
426 * calls.h: Fix typo in comment.
428 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
430 * config/pa/pa.c (pa_output_move_double): Don't valididate when
431 adjusting offsetable addresses.
433 2014-02-24 Guozhi Wei <carrot@google.com>
435 * sparseset.h (sparseset_pop): Fix the wrong index.
437 2014-02-24 Walter Lee <walt@tilera.com>
439 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
440 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
442 * common/config/tilegx/tilegx-common.c
443 (TARGET_DEFAULT_TARGET_FLAGS): Define.
444 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
446 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
447 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
448 (tilegx_gimplify_va_arg_expr): Handle big endian.
449 (tilegx_expand_unaligned_load): Ditto.
450 (tilegx_expand_unaligned_store): Ditto.
451 (TARGET_RETURN_IN_MSB): New.
452 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
453 (TARGET_ENDIAN_DEFAULT): New.
454 (TARGET_BIG_ENDIAN): Handle big endian.
455 (BYTES_BIG_ENDIAN): Ditto.
456 (WORDS_BIG_ENDIAN): Ditto.
457 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
460 * config/tilegx/tilegx.md (extv): Handle big endian.
463 (insn_st<n>_add<bitsuffix>): Ditto.
464 (insn_stnt<n>): Ditto.
465 (insn_stnt<n>_add<bitsuffix>):Ditto.
466 (vec_interleave_highv8qi): Handle big endian.
467 (vec_interleave_highv8qi_be): New.
468 (vec_interleave_highv8qi_le): New.
469 (insn_v1int_h): Handle big endian.
470 (vec_interleave_lowv8qi): Handle big endian.
471 (vec_interleave_lowv8qi_be): New.
472 (vec_interleave_lowv8qi_le): New.
473 (insn_v1int_l): Handle big endian.
474 (vec_interleave_highv4hi): Handle big endian.
475 (vec_interleave_highv4hi_be): New.
476 (vec_interleave_highv4hi_le): New.
477 (insn_v2int_h): Handle big endian.
478 (vec_interleave_lowv4hi): Handle big endian.
479 (vec_interleave_lowv4hi_be): New.
480 (vec_interleave_lowv4hi_le): New.
481 (insn_v2int_l): Handle big endian.
482 (vec_interleave_highv2si): Handle big endian.
483 (vec_interleave_highv2si_be): New.
484 (vec_interleave_highv2si_le): New.
485 (insn_v4int_h): Handle big endian.
486 (vec_interleave_lowv2si): Handle big endian.
487 (vec_interleave_lowv2si_be): New.
488 (vec_interleave_lowv2si_le): New.
489 (insn_v4int_l): Handle big endian.
490 * config/tilegx/tilegx.opt (mbig-endian): New option.
491 (mlittle-endian): New option.
492 * doc/install.texi: Document tilegxbe-linux.
493 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
495 2014-02-24 Martin Jambor <mjambor@suse.cz>
498 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
499 there are no parameter descriptors.
501 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
503 PR rtl-optimization/60268
504 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
505 initialization to ...
506 (sched_rgn_init): ... here.
507 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
509 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
511 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
514 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
516 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
519 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
521 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
522 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
524 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
526 * config/microblaze/predicates.md: Add cmp_op predicate.
527 * config/microblaze/microblaze.md: Add branch_compare instruction
528 which uses cmp_op predicate and emits cmp insn before branch.
529 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
530 to microblaze_expand_conditional_branch and consolidate logic.
531 (microblaze_expand_conditional_branch): emit branch_compare
532 insn instead of handling cmp op separate from branch insn.
534 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
536 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
539 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
541 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
542 define_insn with define_expand and new define_insn
543 *altivec_lve<VI_char>x_internal.
544 (altivec_stve<VI_char>x): Replace define_insn with define_expand
545 and new define_insn *altivec_stve<VI_char>x_internal.
546 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
548 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
550 (altivec_expand_stvex_be): New function.
552 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
554 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
555 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
556 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
557 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
559 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
562 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
563 instead of emit_move_insn.
565 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
567 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
569 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
572 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
574 * config/rs6000/altivec.md (altivec_lvxl): Rename as
575 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
576 (altivec_lvxl_<mode>): New define_expand incorporating
577 -maltivec=be semantics where needed.
578 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
579 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
580 semantics where needed.
581 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
582 (altivec_stvx_<mode>): New define_expand incorporating
583 -maltivec=be semantics where needed.
584 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
585 VM2 iterator instead of V4SI.
586 (altivec_stvxl_<mode>): New define_expand incorporating
587 -maltivec=be semantics where needed.
588 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
589 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
590 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
591 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
592 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
593 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
594 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
595 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
596 ALTIVEC_BUILTIN_STVXL.
597 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
598 (altivec_expand_stvx_be): Likewise.
599 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
600 (altivec_expand_lvx_be): Likewise.
601 (altivec_expand_stvx_be): Likewise.
602 (altivec_expand_builtin): Add cases for
603 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
604 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
605 (altivec_init_builtins): Add definitions for
606 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
607 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
609 2014-02-21 Catherine Moore <clm@codesourcery.com>
611 * doc/invoke.texi (mvirt, mno-virt): Document.
612 * config/mips/mips.opt (mvirt): New option.
613 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
615 2014-02-21 Richard Biener <rguenther@suse.de>
617 PR tree-optimization/60276
618 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
619 (STMT_VINFO_MIN_NEG_DIST): New macro.
620 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
621 STMT_VINFO_MIN_NEG_DIST.
622 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
623 made for negative dependence distances still hold.
625 2014-02-21 Richard Biener <rguenther@suse.de>
628 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
629 DECL_INITIAL for globals not in the current function context.
631 2014-02-21 Jakub Jelinek <jakub@redhat.com>
633 PR tree-optimization/56490
634 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
635 * tree-ssa-uninit.c: Include params.h.
636 (compute_control_dep_chain): Add num_calls argument, return false
637 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
638 num_calls to recursive call.
639 (find_predicates): Change dep_chain into normal array,
640 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
641 variable and adjust compute_control_dep_chain caller.
642 (find_def_preds): Likewise.
644 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
646 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
647 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
649 2014-02-21 Nick Clifton <nickc@redhat.com>
651 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
653 (popqi1): Add mode to pre_dec.
656 2014-02-21 Jakub Jelinek <jakub@redhat.com>
658 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
659 mode for mask of V8SFmode permutation.
661 2014-02-20 Richard Henderson <rth@redhat.com>
664 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
665 a new pseudo for OLDVAL.
667 2014-02-20 Jakub Jelinek <jakub@redhat.com>
670 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
671 gen_reg_rtx if d->testing_p.
672 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
673 if d->testing_p and we will certainly return true.
674 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
677 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
679 * emit-rtl.c (gen_reg_rtx): Assert that
680 crtl->emit.regno_pointer_align_length is non-zero.
682 2014-02-20 Richard Henderson <rth@redhat.com>
685 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
686 on failure the store back into EXPECT.
688 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
689 Sandra Loosemore <sandra@codesourcery.com>
691 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
692 * config/nios2/nios2.c (nios2_function_profiler): Add
693 -fPIC (flag_pic == 2) support.
694 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
695 (nios2_large_offset_p): New function.
696 (nios2_unspec_reloc_p): Move up position, update to use
697 nios2_large_offset_p.
698 (nios2_unspec_address): Remove function.
699 (nios2_unspec_offset): New function.
700 (nios2_large_got_address): New function.
701 (nios2_got_address): Add large offset support.
702 (nios2_legitimize_tls_address): Update usage of removed and new
704 (nios2_symbol_binds_local_p): New function.
705 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
706 (nios2_legitimize_address): Update to use nios2_large_offset_p.
707 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
708 (nios2_print_operand): Merge H/L processing, add hiadj/lo
709 processing for (const (unspec ...)).
710 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
712 2014-02-20 Richard Biener <rguenther@suse.de>
714 * tree-cfg.c (replace_uses_by): Mark altered BBs before
715 doing the substitution.
716 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
718 2014-02-20 Martin Jambor <mjambor@suse.cz>
721 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
722 info when checking whether lattices are bottom.
724 2014-02-20 Richard Biener <rguenther@suse.de>
727 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
730 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
733 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
734 parameter specifying the scaling.
735 (inline_call): Update.
736 (want_inline_recursively): Guard division by zero.
737 (recursive_inlining): Update.
738 * ipa-inline.h (clone_inlined_nodes): Update.
740 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
743 * config/i386/i386.c (classify_argument): Pass structures of size
744 64 bytes or less in register.
746 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
747 Kirill Yukhin <kirill.yukhin@intel.com>
749 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
750 (_mm_rcp28_round_ss): Ditto.
751 (_mm_rsqrt28_round_sd): Ditto.
752 (_mm_rsqrt28_round_ss): Ditto.
753 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
754 (_mm_rcp14_round_ss): Ditto.
755 (_mm_rsqrt14_round_sd): Ditto.
756 (_mm_rsqrt14_round_ss): Ditto.
757 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
758 the first input operand, get rid of match_dup.
759 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
761 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
763 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
764 operand as the first input operand, set type attribute.
765 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
767 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
768 operand as the first input operand, set type attribute.
770 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
772 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
775 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
778 * config/i386/i386.c (construct_container): Remove TFmode check
779 for X86_64_INTEGER_CLASS.
781 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
784 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
785 only when -Wpsabi is enabled.
787 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
790 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
791 passing arrays in registers are the same as for structs, so remove the
792 special case for them.
794 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
796 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
797 destination type, extract only the valid bits if the source type is not
798 integral and has a different mode.
800 2014-02-19 Richard Biener <rguenther@suse.de>
803 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
806 2014-02-19 Richard Biener <rguenther@suse.de>
809 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
810 (ipa_modify_call_arguments): Emit an argument load explicitely and
811 preserve virtual SSA form there and for the replacement call.
812 Do not update SSA form nor free dominance info.
814 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
816 * ipa.c (function_and_variable_visibility): Also clear WEAK
817 flag when disolving COMDAT_GROUP.
819 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
821 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
822 * ipa-prop.c (ipa_set_jf_known_type): Return early when
824 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
825 do more sanity checks.
826 (detect_type_change): Return true when giving up early.
827 (compute_complex_assign_jump_func): Fix type parameter of
829 (compute_complex_ancestor_jump_func): Likewise.
830 (update_jump_functions_after_inlining): Fix updating of
832 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
834 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
836 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
837 inline clones when edge disappears.
839 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
842 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
843 Split 64-bit moves into 2 patterns. Do not allow the use of
844 direct move for TDmode in little endian, since the decimal value
845 has little endian bytes within a word, but the 64-bit pieces are
846 ordered in a big endian fashion, and normal subreg's of TDmode are
848 (mov<mode>_64bit_dm): Likewise.
849 (movtd_64bit_nodm): Likewise.
851 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
853 PR tree-optimization/60174
854 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
855 statement of an SSA_NAME that occurs in an abnormal PHI node.
857 2014-02-18 Jakub Jelinek <jakub@redhat.com>
860 * final.c (SEEN_BB): Remove.
861 (SEEN_NOTE, SEEN_EMITTED): Renumber.
862 (final_scan_insn): Don't force_source_line on second
863 NOTE_INSN_BASIC_BLOCK.
865 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
868 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
869 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
870 (type_natural_mode): Warn ABI change when %zmm register is not
871 available for AVX512F vector value passing.
873 2014-02-18 Kai Tietz <ktietz@redhat.com>
876 * config/i386/i386.c (ix86_expand_prologue): Use value in
877 rax register as displacement when restoring %r10 or %rax.
878 Fix wrong offset when restoring both registers.
880 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
882 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
883 assertion with conditional return.
885 2014-02-18 Jakub Jelinek <jakub@redhat.com>
886 Uros Bizjak <ubizjak@gmail.com>
889 * config/i386/driver-i386.c (host_detect_local_cpu): If
890 YMM state is not saved by the OS, also clear has_f16c. Move
891 CPUID 0x80000001 handling before YMM state saving checking.
893 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
895 PR rtl-optimization/58960
896 * haifa-sched.c (alloc_global_sched_pressure_data): New,
897 factored out from ...
898 (sched_init): ... here.
899 (free_global_sched_pressure_data): New, factored out from ...
900 (sched_finish): ... here.
901 * sched-int.h (free_global_sched_pressure_data): Declare.
902 * sched-rgn.c (nr_regions_initial): New static global.
903 (haifa_find_rgns): Initialize it.
904 (schedule_region): Disable sched-pressure for the newly
907 2014-02-17 Richard Biener <rguenther@suse.de>
909 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
910 release SSA defs of pattern stmts.
912 2014-02-17 Richard Biener <rguenther@suse.de>
914 * tree-inline.c (expand_call_inline): Release the virtual
915 operand defined by the call we are about to inline.
917 2014-02-17 Richard Biener <rguenther@suse.de>
919 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
921 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
922 Ilya Tocar <ilya.tocar@intel.com>
924 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
925 arguments order in builtin.
926 (_mm512_permutexvar_epi64): Ditto.
927 (_mm512_mask_permutexvar_epi64): Ditto
928 (_mm512_maskz_permutexvar_epi32): Ditto
929 (_mm512_permutexvar_epi32): Ditto
930 (_mm512_mask_permutexvar_epi32): Ditto
932 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
934 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
935 (p8_vmrgow): Likewise.
937 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
939 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
942 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
945 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
946 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
947 into 64-bit and 32-bit moves. On 64-bit moves, add support for
948 using direct move instructions on ISA 2.07. Also adjust
949 instruction length for 64-bit.
950 (mov<mode>_64bit, TFmode/TDmode): Likewise.
951 (mov<mode>_32bit, TFmode/TDmode): Likewise.
953 2014-02-15 Alan Modra <amodra@gmail.com>
957 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
958 find_replacement on parts of insn rtl that might be reloaded.
960 2014-02-15 Richard Biener <rguenther@suse.de>
962 PR tree-optimization/60183
963 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
964 (tree_ssa_phiprop): Calculate and free post-dominators.
966 2014-02-14 Jeff Law <law@redhat.com>
968 PR rtl-optimization/60131
969 * ree.c (get_extended_src_reg): New function.
970 (combine_reaching_defs): Use it rather than assuming location of REG.
971 (find_and_remove_re): Verify first operand of extension is
972 a REG before adding the insns to the copy list.
974 2014-02-14 Roland McGrath <mcgrathr@google.com>
976 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
977 * configure: Regenerated.
978 * config.in: Regenerated.
979 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
980 instead of ASM_SHORT.
982 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
983 Richard Earnshaw <rearnsha@arm.com>
985 PR rtl-optimization/59535
986 * lra-constraints.c (process_alt_operands): Encourage alternative
987 when unassigned pseudo class is superset of the alternative class.
988 (inherit_reload_reg): Don't inherit when optimizing for code size.
989 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
990 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
991 modes not less than 4 for Thumb1.
993 2014-02-14 Kyle McMartin <kyle@redhat.com>
996 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
998 2014-02-14 Richard Biener <rguenther@suse.de>
1000 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1001 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1002 Do not strip INDIRECT_REFs.
1004 2014-02-14 Richard Biener <rguenther@suse.de>
1007 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1008 DECL_FUNCTION_SPECIFIC_TARGET.
1009 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1010 * tree-streamer-out.c (pack_ts_target_option): Remove.
1011 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1012 (write_ts_function_decl_tree_pointers): Do not stream
1013 DECL_FUNCTION_SPECIFIC_TARGET.
1014 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1015 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1016 (lto_input_ts_function_decl_tree_pointers): Do not stream
1017 DECL_FUNCTION_SPECIFIC_TARGET.
1019 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1021 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1022 (get_initial_def_for_induction, vectorizable_induction): Ignore
1023 debug stmts when looking for exit_phi.
1024 (vectorizable_live_operation): Fix up condition.
1026 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1028 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1029 nreverse() because it changes the content of original tree list.
1031 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1033 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1034 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1036 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1038 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1039 GNU coding standards.
1041 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1044 * dwarf2out.c (gen_subprogram_die): Don't call
1045 add_calling_convention_attribute if subr_die is old_die.
1047 2014-02-13 Sharad Singhai <singhai@google.com>
1049 * doc/optinfo.texi: Fix order of nodes.
1051 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1053 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1054 operands[2], not operands[3].
1056 2014-02-13 Richard Biener <rguenther@suse.de>
1059 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1060 mention the possibility of an in-tree build.
1061 (CLooG): Update recommended version to 0.18.1, mention the
1062 possibility of an in-tree build and clarify that the ISL
1063 bundled with CLooG does not work.
1065 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1068 * expr.c (compress_float_constant): If x is a hard register,
1069 extend into a pseudo and then move to x.
1071 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1073 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1074 caused by bad second argument to warning_at() with -mhotpatch and
1075 nested functions (e.g. with gfortran).
1077 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1079 * opts.c (option_name): Remove "enabled by default" rider.
1081 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1083 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1085 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1086 Uros Bizjak <ubizjak@gmail.com>
1089 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1090 * configure: Regenerated.
1092 2014-02-12 Richard Biener <rguenther@suse.de>
1094 * vec.c (vec_prefix::calculate_allocation): Move as
1095 inline variant to vec.h.
1096 (vec_prefix::calculate_allocation_1): New out-of-line version.
1097 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1098 (vec_prefix::m_has_auto_buf): Rename to ...
1099 (vec_prefix::m_using_auto_storage): ... this.
1100 (vec_prefix::calculate_allocation): Inline the easy cases
1101 and dispatch to calculate_allocation_1 which doesn't need the
1103 (va_heap::reserve): Use gcc_checking_assert.
1104 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1105 m_using_auto_storage.
1106 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1108 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1109 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1110 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1112 2014-02-12 Richard Biener <rguenther@suse.de>
1114 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1115 when we found a dependence.
1117 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1119 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1121 (maybe_fold_stmt): ... into this new function.
1122 * omp-low.c (lower_omp): Update comment.
1124 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1127 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1130 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1132 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1133 identifiers in comments.
1134 (cortexa53_extra_costs): Likewise.
1135 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1136 (cortexa7_extra_costs): Likewise.
1137 (cortexa12_extra_costs): Likewise.
1138 (cortexa15_extra_costs): Likewise.
1139 (v7m_extra_costs): Likewise.
1141 2014-02-12 Richard Biener <rguenther@suse.de>
1144 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1145 of posix_memalign being successful.
1146 (lower_stmt): Restrict lowering of posix_memalign to when
1147 -ftree-bit-ccp is enabled.
1149 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1151 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1153 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1155 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1157 PR rtl-optimization/60116
1158 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1159 other_insn once the combination has been validated.
1161 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1164 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1166 * ipa-devirt.c: Include demangle.h
1167 (odr_violation_reported): New static variable.
1168 (add_type_duplicate): Update odr_violations.
1169 (maybe_record_node): Add completep parameter; update it.
1170 (record_target_from_binfo): Add COMPLETEP parameter;
1171 update it as needed.
1172 (possible_polymorphic_call_targets_1): Likewise.
1173 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1174 rename FINAL to COMPLETE.
1175 (record_targets_from_bases): Sanity check we found the binfo;
1176 fix COMPLETEP updating.
1177 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1178 parameter, fix computing of COMPLETEP.
1179 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1180 at LTO time do demangling.
1181 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1182 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1184 (gimple_get_virt_method_for_binfo): Likewise.
1185 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1186 gimple_get_virt_method_for_vtable): Update prototypes.
1188 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1191 * genautomata.c (add_presence_absence): Fix typo with
1192 {final_}presence_list.
1194 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1197 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1198 for VSX/Altivec vectors that land in GPR registers.
1200 2014-02-11 Richard Henderson <rth@redhat.com>
1201 Jakub Jelinek <jakub@redhat.com>
1204 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1205 around drhs if type conversion to lacc->type is not useless.
1207 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1209 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1211 (cortex-a57.cortex-a53): Likewise.
1212 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1214 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1216 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1219 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1221 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1222 add_options_for_arm_vfp3.
1224 2014-02-11 Jeff Law <law@redhat.com>
1227 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1228 object with an undesirable mode.
1230 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1233 * config/i386/sol2-9.h: New file.
1234 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1235 *-*-solaris2.9*): Use it.
1237 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1239 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1240 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1242 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1244 * config/microblaze/microblaze.c: Extend mcpu version format
1246 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1248 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1250 2014-02-10 Richard Henderson <rth@redhat.com>
1253 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1254 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1255 ms-abi vs -mno-accumulate-outgoing-args.
1256 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1257 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1260 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1263 * cfgexpand.c (expand_asm_operands): Attach source location to
1264 ASM_INPUT rtx objects.
1265 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1267 2014-02-10 Nick Clifton <nickc@redhat.com>
1269 * config/mn10300/mn10300.c (popcount): New function.
1270 (mn10300_expand_prologue): Include saved registers in stack usage
1273 2014-02-10 Jeff Law <law@redhat.com>
1276 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1277 when changing the SET_DEST of a prior insn to avoid an input reload.
1279 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1281 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1282 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1283 -mcall-openbsd, or -mcall-linux.
1284 (CC1_ENDIAN_BIG_SPEC): Remove.
1285 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1286 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1287 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1288 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1289 and %cc1_endian_default.
1290 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1292 2014-02-10 Richard Biener <rguenther@suse.de>
1294 PR tree-optimization/60115
1295 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1296 MEM_REF handling. Properly verify that the accesses are not
1297 out of the objects bound.
1299 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1301 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1304 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1306 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1307 proper constants and fix formatting.
1308 (possible_polymorphic_call_targets): Fix formatting.
1310 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1311 Ilya Tocar <ilya.tocar@intel.com>
1313 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1314 (_mm512_loadu_epi32): Renamed into...
1315 (_mm512_loadu_si512): This.
1316 (_mm512_storeu_epi32): Renamed into...
1317 (_mm512_storeu_si512): This.
1318 (_mm512_maskz_ceil_ps): Removed.
1319 (_mm512_maskz_ceil_pd): Ditto.
1320 (_mm512_maskz_floor_ps): Ditto.
1321 (_mm512_maskz_floor_pd): Ditto.
1322 (_mm512_floor_round_ps): Ditto.
1323 (_mm512_floor_round_pd): Ditto.
1324 (_mm512_ceil_round_ps): Ditto.
1325 (_mm512_ceil_round_pd): Ditto.
1326 (_mm512_mask_floor_round_ps): Ditto.
1327 (_mm512_mask_floor_round_pd): Ditto.
1328 (_mm512_mask_ceil_round_ps): Ditto.
1329 (_mm512_mask_ceil_round_pd): Ditto.
1330 (_mm512_maskz_floor_round_ps): Ditto.
1331 (_mm512_maskz_floor_round_pd): Ditto.
1332 (_mm512_maskz_ceil_round_ps): Ditto.
1333 (_mm512_maskz_ceil_round_pd): Ditto.
1334 (_mm512_expand_pd): Ditto.
1335 (_mm512_expand_ps): Ditto.
1336 * config/i386/i386.c (ix86_builtins): Remove
1337 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1338 (bdesc_args): Ditto.
1339 * config/i386/predicates.md (const1256_operand): New.
1340 (const_1_to_2_operand): Ditto.
1341 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1342 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1343 (*avx512pf_gatherpf<mode>sf): Ditto.
1344 (avx512pf_gatherpf<mode>df): Ditto.
1345 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1346 (*avx512pf_gatherpf<mode>df): Ditto.
1347 (avx512pf_scatterpf<mode>sf): Ditto.
1348 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1349 (*avx512pf_scatterpf<mode>sf): Ditto.
1350 (avx512pf_scatterpf<mode>df): Ditto.
1351 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1352 (*avx512pf_scatterpf<mode>df): Ditto.
1353 (avx512f_expand<mode>): Removed.
1354 (<shift_insn><mode>3<mask_name>): Change predicate type.
1356 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1358 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1359 not at the end of datarefs vector use ordered_remove to avoid
1360 reordering datarefs vector.
1363 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1364 mark local addressable non-static vars as GOVD_PRIVATE
1365 instead of GOVD_LOCAL.
1366 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1367 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1371 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1372 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1373 assume_aligned or alloc_align attributes.
1374 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1375 arguments. Handle also assume_aligned and alloc_align attributes.
1376 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1377 calls to functions with assume_aligned or alloc_align attributes.
1378 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1380 2014-02-08 Terry Guo <terry.guo@arm.com>
1382 * doc/invoke.texi: Document ARM -march=armv7e-m.
1384 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1386 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1387 flag on __cilkrts_rethrow builtin.
1390 * ipa-cp.c (determine_versionability): Fail at -O0
1391 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1392 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1395 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1398 * tree-inline.c (copy_forbidden): Fail for
1399 __attribute__((optimize (0))) functions.
1401 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1403 * varpool.c: Include pointer-set.h.
1404 (varpool_remove_unreferenced_decls): Variables in other partitions
1405 will not be output; be however careful to not lose information
1408 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1410 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1411 lookup in the vtable constructor.
1413 2014-02-07 Jeff Law <law@redhat.com>
1416 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1417 define_insn_and_split.
1419 * ipa-inline.c (inline_small_functions): Fix typos.
1421 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1423 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1424 (s390_can_use_return_insn): Declare.
1425 * config/s390/s390.h (EPILOGUE_USES): Define.
1426 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1428 (s390_chunkify_start): Handle return JUMP_LABELs.
1429 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1430 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1431 (s390_can_use_return_insn): New functions.
1432 (s390_fix_long_loop_prediction): Handle conditional returns.
1433 (TARGET_SET_UP_BY_PROLOGUE): Define.
1434 * config/s390/s390.md (ANY_RETURN): New code iterator.
1435 (*creturn, *csimple_return, return, simple_return): New patterns.
1437 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1439 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1440 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1441 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1442 REG_CFA_RESTORE list when deciding not to restore a register.
1444 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1446 * config/s390/s390.c: Include tree-pass.h and context.h.
1447 (s390_early_mach): New function, split out from...
1448 (s390_emit_prologue): ...here.
1449 (pass_data_s390_early_mach): New pass structure.
1450 (pass_s390_early_mach): New class.
1451 (s390_option_override): Create and register early_mach pass.
1452 Move to end of file.
1454 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1456 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1457 to match for the exit block.
1459 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1461 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1462 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1463 Reject misaligned operands.
1465 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1467 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1469 2014-02-07 Richard Biener <rguenther@suse.de>
1472 * gimple-low.c (lower_builtin_posix_memalign): New function.
1473 (lower_stmt): Call it to lower posix_memalign in a way
1474 to make alignment info accessible.
1476 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1479 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1480 __builtin_setjmp_receiver.
1482 2014-02-07 Richard Biener <rguenther@suse.de>
1485 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1486 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1487 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1488 Handle BUILT_IN_POSIX_MEMALIGN.
1489 (find_func_clobbers): Likewise.
1490 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1491 (call_may_clobber_ref_p_1): Likewise.
1493 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1496 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1499 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1502 * lto-cgraph.c (lto_output_node): Use
1503 symtab_get_symbol_partitioning_class.
1504 (lto_output_varpool_node): likewise.
1505 (symtab_get_symbol_partitioning_class): Move here from
1507 * cgraph.h (symbol_partitioning_class): Likewise.
1508 (symtab_get_symbol_partitioning_class): Declare.
1510 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1512 * ggc.h (ggc_internal_cleared_alloc): New macro.
1513 * vec.h (vec_safe_copy): Handle memory stats.
1514 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1515 * target-globals.c (save_target_globals): Likewise.
1517 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1520 * expr.c (emit_move_resolve_push): Export; be bit more selective
1521 on when to clear alias set.
1522 * expr.h (emit_move_resolve_push): Declare.
1523 * function.h (struct function): Add tail_call_marked.
1524 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1525 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1526 * config/i386/i386.md (TImode move expander): De not call
1528 (FP push expanders): Preserve memory attributes.
1529 * config/i386/sse.md (push<mode>1): Remove.
1530 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1531 (ix86_expand_push): Remove.
1532 * config/i386/mmx.md (push<mode>1): Remove.
1534 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1536 PR rtl-optimization/60030
1537 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1538 lopart with paradoxical subreg before shifting it up by hprec.
1540 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1542 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1543 Remove extra newline at end of file.
1544 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1545 (arm_issue_rate): Handle cortexa57.
1546 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1547 (cortex-a57.cortex-a53): Likewise.
1549 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1552 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1553 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1555 (arm_expand_prologue): Adjust all callers.
1556 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1557 info, registers also at the lowest numbered registers side. Use
1558 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1562 * var-tracking.c (adjust_mems): Before adding a SET to
1563 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1565 2014-02-06 Alan Modra <amodra@gmail.com>
1568 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1569 change SDmode to DDmode when lra_in_progress.
1571 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1574 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1575 free_data_ref on the dr first, and before goto again also set dr
1576 to the next dr. For simd_lane_access, free old datarefs[i] before
1577 overwriting it. For get_vectype_for_scalar_type failure, don't
1578 free_data_ref if simd_lane_access.
1580 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1583 * tree.h (opts_for_fn): New inline function.
1584 (opt_for_fn): Define.
1585 * config/i386/i386.c (ix86_function_regparm): Use
1586 opt_for_fn (decl, optimize) instead of optimize.
1588 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1590 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1591 for SYMBOL_REF in large memory model.
1593 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1595 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1597 (cortex-a57): Likewise.
1598 (cortex-a57.cortex-a53): Likewise.
1600 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1601 Kugan Vivekanandarajah <kuganv@linaro.org>
1603 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1605 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1607 2014-02-06 Richard Biener <rguenther@suse.de>
1609 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1610 set_loop_copy and initialize_original_copy_tables.
1612 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1614 * config/aarch64/aarch64-simd.md
1615 (aarch64_ashr_simddi): Change QI to SI.
1617 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1618 Jakub Jelinek <jakub@redhat.com>
1621 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1624 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1626 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1627 CODE_FOR_altivec_vpku[hw]um to
1628 CODE_FOR_altivec_vpku[hw]um_direct.
1629 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1630 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1631 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1632 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1634 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1636 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1637 generation for -maltivec=be.
1638 (altivec_vsumsws): Simplify redundant test.
1640 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1642 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1643 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1644 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1645 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1646 gen_altivec_vpkuwum.
1647 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1649 (altivec_vpks<VI_char>ss): Likewise.
1650 (altivec_vpks<VI_char>us): Likewise.
1651 (altivec_vpku<VI_char>us): Likewise.
1652 (altivec_vpku<VI_char>um): Likewise.
1653 (altivec_vpku<VI_char>um_direct): New (copy of
1654 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1656 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1657 target is little endian and -maltivec=be is not specified.
1658 (*altivec_vupkhs<VU_char>_direct): New (copy of
1659 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1660 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1661 target is little endian and -maltivec=be is not specified.
1662 (*altivec_vupkls<VU_char>_direct): New (copy of
1663 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1664 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1665 little endian and -maltivec=be is not specified.
1666 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1667 little endian and -maltivec=be is not specified.
1669 2014-02-05 Richard Henderson <rth@redhat.com>
1672 * combine-stack-adj.c: Revert r206943.
1673 * sched-int.h (struct deps_desc): Add last_args_size.
1674 * sched-deps.c (init_deps): Initialize it.
1675 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1676 contain REG_ARGS_SIZE notes.
1678 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1680 * lto-cgraph.c (asm_nodes_output): Make global.
1681 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1682 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1683 (driver_handle_option): Handle OPT_fwpa.
1685 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1688 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1689 a comment typo and formatting issue. If odr_hash hasn't been
1690 created, return vNULL and set *completep to false.
1693 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1694 bb with no successors.
1696 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1699 * doc/invoke.texi (-march): Clarify documentation for ARM.
1703 2014-02-05 Richard Biener <rguenther@suse.de>
1705 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1706 when not vectorizing because of too many alias checks.
1707 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1708 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1710 2014-02-05 Nick Clifton <nickc@redhat.com>
1712 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1713 accept extended registers in any mode when compiling for the MN10300.
1715 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1717 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1718 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1719 sanitization attributes.
1720 (can_inline_edge_p): Likewise.
1721 (sanitize_attrs_match_for_inline_p): New function.
1723 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1725 * ipa-prop.c (detect_type_change): Shor circuit testing of
1726 type changes on THIS pointer.
1728 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1731 * config/pa/pa.c (legitimize_tls_address): Return original address
1732 if not passed a SYMBOL_REF rtx.
1733 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1735 (pa_emit_move_sequence): Simplify TLS source operands.
1736 (pa_legitimate_constant_p): Reject all TLS constants.
1737 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1738 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1740 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1742 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1743 groups when we know they are controlled by LTO.
1744 * varasm.c (default_binds_local_p_1): If object is in other partition,
1745 it will be resolved locally.
1747 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1749 * config/host-linux.c (linux_gt_pch_use_address): Don't
1750 use SSIZE_MAX because it is not always defined.
1752 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1755 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1756 threshold for pseudo splitting.
1757 (update_ebb_live_info): Process call argument hard registers and
1758 hard registers from insn definition too.
1759 (max_small_class_regs_num): New constant.
1760 (inherit_in_ebb): Update live hard regs through EBBs. Update
1761 reloads_num only for small register classes. Don't split for
1764 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1767 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1770 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1772 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1775 2014-02-04 Marek Polacek <polacek@redhat.com>
1777 * gdbinit.in (pel): Define.
1779 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1781 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1784 2014-02-04 Richard Biener <rguenther@suse.de>
1787 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1788 in function context local.
1789 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1790 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1791 similar to LTO_imported_decl_ref.
1793 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1795 PR tree-optimization/60002
1796 * cgraphclones.c (build_function_decl_skip_args): Clear
1799 PR tree-optimization/60023
1800 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1801 false to gsi_replace.
1802 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1803 has been in some EH region and vec_stmt could throw, add
1804 vec_stmt into the same EH region.
1805 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1806 has no lhs, ignore it.
1807 * internal-fn.c (expand_MASK_LOAD): Likewise.
1810 * tree-inline.c (copy_forbidden): Fail for
1811 __attribute__((optimize (0))) functions.
1814 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1815 is set, copy one less argument.
1816 (expand_simd_clones): Don't subtract clone_info->inbranch
1817 from simd_clone_struct_alloc argument.
1819 PR rtl-optimization/57915
1820 * recog.c (simplify_while_replacing): If all unary/binary/relational
1821 operation arguments are constant, attempt to simplify those.
1824 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1825 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1827 2014-02-04 Richard Biener <rguenther@suse.de>
1829 PR tree-optimization/60012
1830 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1831 TBAA disambiguation to all DDRs.
1833 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1836 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1837 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1839 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1842 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1844 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1846 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1847 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1849 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1852 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1853 to figure out targets of polymorphic calls with known decl.
1854 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1855 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1856 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1857 (get_polymorphic_call_info): ... here.
1858 (get_polymorphic_call_info_from_invariant): New function.
1860 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1862 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1863 lookup via vtable pointer; check for type consistency
1864 and turn inconsitent facts into UNREACHABLE.
1865 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1866 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1867 type inconsistent querries; return UNREACHABLE instead.
1869 2014-02-03 Richard Henderson <rth@twiddle.net>
1872 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1873 already processed this node.
1874 (normalize_one_pred_1): Pass along mark_set.
1875 (normalize_one_pred): Create and destroy a pointer_set_t.
1876 (normalize_one_pred_chain): Likewise.
1878 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1880 PR gcov-profile/58602
1881 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1883 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1886 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1887 -fno-devirtualize; try to devirtualize by the knowledge of
1888 virtual table pointer given by aggregate propagation.
1889 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1890 (ipa_print_node_jump_functions): Dump also offset that
1891 is relevant for polymorphic calls.
1892 (determine_known_aggregate_parts): Add arg_type parameter; use it
1893 instead of determining the type from pointer type.
1894 (ipa_compute_jump_functions_for_edge): Update call of
1895 determine_known_aggregate_parts.
1896 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1897 (gimple_get_virt_method_for_binfo): ... here; simplify using
1898 vtable_pointer_value_to_vtable.
1899 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1900 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1901 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1902 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1904 (vtable_pointer_value_to_binfo): ... here.
1905 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1907 2014-02-03 Teresa Johnson <tejohnson@google.com>
1909 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1910 redef of outer loop index variable.
1912 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1916 * doc/extend.texi (Function Attributes): Typo.
1918 2014-02-03 Cong Hou <congh@google.com>
1920 PR tree-optimization/60000
1921 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1922 if the vectorized statement is a store. A store statement can only
1923 appear at the end of pattern statements.
1925 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1927 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1928 (ix86_option_override_internal): Default long double to 64-bit for
1929 32-bit Bionic and to 128-bit for 64-bit Bionic.
1931 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1932 TARGET_LONG_DOUBLE_128 is true.
1933 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1935 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1936 (mlong-double-64): Negate -mlong-double-128.
1937 (mlong-double-128): New option.
1939 * config/i386/i386-c.c (ix86_target_macros): Define
1940 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1942 * doc/invoke.texi: Document -mlong-double-128.
1944 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1946 PR rtl-optimization/60024
1947 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1949 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1951 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1953 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1955 PR rtl-optimization/57662
1956 * sel-sched.c (code_motion_path_driver): Do not mark already not
1957 existing blocks in the visiting bitmap.
1959 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1961 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1962 on the insn being emitted.
1964 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1965 Will Deacon <will.deacon@arm.com>
1967 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1969 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1971 * config/arm/arm-tables.opt: Regenerate.
1973 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1975 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1976 for vector types other than V16QImode.
1977 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1978 define_expand, and call altivec_expand_vec_perm_le when producing
1979 code with little endian element order.
1980 (*altivec_vperm_<mode>_internal): New insn having previous
1981 behavior of altivec_vperm_<mode>.
1982 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1983 altivec_expand_vec_perm_le when producing code with little endian
1985 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1986 behavior of altivec_vperm_<mode>_uns.
1988 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1990 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1991 (altivec_vsumsws): Add handling for -maltivec=be with a little
1993 (altivec_vsumsws_direct): New.
1994 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1995 gen_altivec_vsumsws.
1997 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1999 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2000 vtable_pointer_value_to_binfo): New functions.
2001 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2002 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2004 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2006 * config/nios2/nios2.md (load_got_register): Initialize GOT
2007 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2008 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2010 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2012 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2013 preserverd by passthrough, do not propagate the type.
2015 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2017 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2018 (mips_atomic_assign_expand_fenv): New function.
2019 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2021 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2023 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2024 (__builtin_mips_set_fcsr): Likewise.
2025 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2026 MIPS_USI_FTYPE_VOID.
2027 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2028 (mips16_expand_set_fcsr): Likewise.
2029 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2030 (mips16_set_fcsr_stub): Likewise.
2031 (mips16_get_fcsr_one_only_stub): New class.
2032 (mips16_set_fcsr_one_only_stub): Likewise.
2033 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2034 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2035 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2036 (hard_float): New availability predicate.
2037 (mips_builtins): Add get_fcsr and set_fcsr.
2038 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2039 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2040 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2041 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2042 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2045 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2047 * config/mips/mips.c (mips_one_only_stub): New class.
2048 (mips_need_mips16_rdhwr_p): Replace with...
2049 (mips16_rdhwr_stub): ...this new variable.
2050 (mips16_stub_call_address): New function.
2051 (mips16_rdhwr_one_only_stub): New class.
2052 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2053 (mips_output_mips16_rdhwr): Delete.
2054 (mips_finish_stub): New function.
2055 (mips_code_end): Use it to handle rdhwr stubs.
2057 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2060 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2061 when calculating size of integer atomic types.
2063 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2065 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2067 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2069 PR tree-optimization/60003
2070 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2071 * profile.c (branch_prob): Use gimple_call_builtin_p
2072 to check for BUILT_IN_SETJMP_RECEIVER.
2073 * tree-inline.c (copy_bb): Call notice_special_calls.
2075 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2078 * lra-constraints.c (process_alt_operands): Update reload_sum only
2081 2014-01-31 Richard Henderson <rth@redhat.com>
2084 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2085 until after else_eh is processed.
2087 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2089 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2090 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2091 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2092 in smmintrin.h, remove them.
2093 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2094 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2095 * config/i386/i386.md (ROUND_SAE): Fix value.
2096 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2097 (const48_operand): New.
2098 * config/i386/subst.md (round), (round_expand): Use
2099 const_4_or_8_to_11_operand.
2100 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2102 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2104 * config/i386/constraints.md (Yk): Swap meaning with k.
2105 * config/i386/i386.md (movhi_internal): Change Yk to k.
2106 (movqi_internal): Ditto.
2107 (*k<logic><mode>): Ditto.
2110 (kandn<mode>): Ditto.
2111 (*<code>hi_1): Ditto.
2112 (*<code>qi_1): Ditto.
2113 (kxnor<mode>): Ditto.
2114 (kortestzhi): Ditto.
2115 (kortestchi): Ditto.
2117 (*one_cmplhi2_1): Ditto.
2118 (*one_cmplqi2_1): Ditto.
2119 * config/i386/sse.md (): Change k to Yk.
2120 (avx512f_load<mode>_mask): Ditto.
2121 (avx512f_blendm<mode>): Ditto.
2122 (avx512f_store<mode>_mask): Ditto.
2123 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2124 (avx512f_storedqu<mode>_mask): Ditto.
2125 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2127 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2128 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2129 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2130 (avx512f_maskcmp<mode>3): Ditto.
2131 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2132 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2133 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2134 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2135 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2136 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2137 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2138 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2139 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2140 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2141 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2142 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2143 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2144 (vec_extract_lo_<mode>_maskm): Ditto.
2145 (vec_extract_hi_<mode>_maskm): Ditto.
2146 (avx512f_vternlog<mode>_mask): Ditto.
2147 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2148 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2149 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2150 (avx512f_<code>v8div16qi2_mask): Ditto.
2151 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2152 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2153 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2154 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2155 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2156 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2157 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2158 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2159 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2160 (avx512cd_maskb_vec_dupv8di): Ditto.
2161 (avx512cd_maskw_vec_dupv16si): Ditto.
2162 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2163 (avx512f_vpermi2var<mode>3_mask): Ditto.
2164 (avx512f_vpermi2var<mode>3_mask): Ditto.
2165 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2166 (*avx512f_gathersi<mode>): Ditto.
2167 (*avx512f_gathersi<mode>_2): Ditto.
2168 (*avx512f_gatherdi<mode>): Ditto.
2169 (*avx512f_gatherdi<mode>_2): Ditto.
2170 (*avx512f_scattersi<mode>): Ditto.
2171 (*avx512f_scatterdi<mode>): Ditto.
2172 (avx512f_compress<mode>_mask): Ditto.
2173 (avx512f_compressstore<mode>_mask): Ditto.
2174 (avx512f_expand<mode>_mask): Ditto.
2175 * config/i386/subst.md (mask): Change k to Yk.
2176 (mask_scalar_merge): Ditto.
2179 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2181 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2183 2014-01-31 Richard Biener <rguenther@suse.de>
2186 * builtins.c (fold_builtin_memory_op): Make sure to not
2187 use a floating-point mode or a boolean or enumeral type for
2190 2014-01-30 DJ Delorie <dj@redhat.com>
2192 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2193 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2194 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2195 whenever main() has an epilogue.
2197 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2199 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2200 unused variable "field".
2201 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2202 (vsx_mergeh_<mode>): Likewise.
2203 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2204 (altivec_vmrghh): Likewise.
2205 (altivec_vmrghw): Likewise.
2206 (altivec_vmrglb): Likewise.
2207 (altivec_vmrglh): Likewise.
2208 (altivec_vmrglw): Likewise.
2209 (altivec_vspltb): Add missing uses.
2210 (altivec_vsplth): Likewise.
2211 (altivec_vspltw): Likewise.
2212 (altivec_vspltsf): Likewise.
2214 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2217 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2218 frame related instructions.
2220 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2222 PR rtl-optimization/59959
2223 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2224 any reload of register whose subreg is invalid.
2226 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2228 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2229 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2230 Add missing return type - void.
2232 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2234 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2235 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2236 remove element index adjustment for endian (now handled in vsx.md
2238 (altivec_expand_vec_perm_const): Use
2239 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2240 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2241 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2242 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2243 define_expand and a new define_insn *altivec_vspltb_internal;
2244 adjust for -maltivec=be on a little endian target.
2245 (altivec_vspltb_direct): New.
2246 (altivec_vsplth): Divide into a define_expand and a new
2247 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2248 little endian target.
2249 (altivec_vsplth_direct): New.
2250 (altivec_vspltw): Divide into a define_expand and a new
2251 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2252 little endian target.
2253 (altivec_vspltw_direct): New.
2254 (altivec_vspltsf): Divide into a define_expand and a new
2255 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2256 a little endian target.
2258 2014-01-30 Richard Biener <rguenther@suse.de>
2260 PR tree-optimization/59993
2261 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2262 can propagate form the earlier stmt and avoid the transform
2263 when the intermediate result is needed.
2265 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2267 * README.Portability: Fix typo.
2269 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2271 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2272 comparison_operator with ordered_comparison_operator.
2274 2014-01-30 Nick Clifton <nickc@redhat.com>
2276 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2277 Rename to mn10300_store_multiple_regs.
2278 * config/mn10300/mn10300.c: Likewise.
2279 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2280 store_multiple_regs.
2281 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2282 Call mn10300_store_multiple_regs.
2284 2014-01-30 Nick Clifton <nickc@redhat.com>
2285 DJ Delorie <dj@redhat.com>
2287 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2288 %fp 2 to keep registers after it properly word-aligned.
2289 (rl78_alloc_physical_registers_umul): Handle the case where both
2290 input operands are the same.
2292 2014-01-30 Richard Biener <rguenther@suse.de>
2294 PR tree-optimization/59903
2295 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2298 2014-01-30 Jason Merrill <jason@redhat.com>
2301 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2304 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2306 2014-01-30 Richard Biener <rguenther@suse.de>
2308 PR tree-optimization/59951
2309 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2311 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2314 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2315 SFmode to DFmode case.
2317 2014-01-29 DJ Delorie <dj@redhat.com>
2319 * config/msp430/msp430.opt (-minrt): New.
2320 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2322 (ENDFILE_SPEC): Likewise.
2324 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2326 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2327 (estimate_function_body_sizes): Use it.
2329 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2332 * dwarf2out.c (is_cxx_auto): New.
2333 (is_base_type): Use it.
2334 (gen_type_die_with_usage): Likewise.
2336 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2338 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2339 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2340 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2341 -maltivec=be with LE targets.
2342 (vsx_mergeh_<mode>): Likewise.
2343 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2344 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2345 (altivec_vmrghb): Replace with define_expand and new
2346 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2347 (altivec_vmrghb_direct): New define_insn.
2348 (altivec_vmrghh): Replace with define_expand and new
2349 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2350 (altivec_vmrghh_direct): New define_insn.
2351 (altivec_vmrghw): Replace with define_expand and new
2352 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2353 (altivec_vmrghw_direct): New define_insn.
2354 (*altivec_vmrghsf): Adjust for endianness.
2355 (altivec_vmrglb): Replace with define_expand and new
2356 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2357 (altivec_vmrglb_direct): New define_insn.
2358 (altivec_vmrglh): Replace with define_expand and new
2359 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2360 (altivec_vmrglh_direct): New define_insn.
2361 (altivec_vmrglw): Replace with define_expand and new
2362 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2363 (altivec_vmrglw_direct): New define_insn.
2364 (*altivec_vmrglsf): Adjust for endianness.
2365 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2366 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2367 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2368 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2369 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2370 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2371 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2372 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2374 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2376 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2377 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2380 2014-01-29 Richard Biener <rguenther@suse.de>
2382 PR tree-optimization/58742
2383 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2384 associate_pointerplus_align.
2385 (associate_pointerplus_diff): New function.
2386 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2387 and associate_pointerplus_diff.
2389 2014-01-29 Richard Biener <rguenther@suse.de>
2391 * lto-streamer.h (LTO_major_version): Bump to 3.
2392 (LTO_minor_version): Reset to 0.
2394 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2396 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2397 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2398 (arm_file_start): Generate correct asm header for armv7ve.
2399 * config/arm/bpabi.h: Add multilib support for armv7ve.
2400 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2401 and cortex-a15 to armv7ve.
2402 * config/arm/t-aprofile: Add multilib support for armv7ve.
2403 * doc/invoke.texi: Document -march=armv7ve.
2405 2014-01-29 Richard Biener <rguenther@suse.de>
2407 PR tree-optimization/58742
2408 * tree-ssa-forwprop.c (associate_plusminus): Return true
2409 if we changed sth, defer EH cleanup to ...
2410 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2411 (simplify_mult): New function.
2413 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2416 PR tree-optimization/59920
2417 * tree.c (build_common_builtin_nodes): Remove
2418 __builtin_setjmp_dispatcher initialization.
2419 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2420 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2421 instead of gsi_after_labels + manually skipping debug stmts.
2422 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2423 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2424 * tree-inline.c (copy_edges_for_bb): Remove
2425 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2426 argument. Ignore computed_goto_p stmts. Don't call
2427 make_abnormal_goto_edges. If a call might need abnormal edges
2428 for non-local gotos, see if it already has an edge to
2429 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2430 with true argument, don't do anything then, otherwise add
2431 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2432 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2434 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2435 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2436 (lower_stmt): Don't set data->calls_builtin_setjmp.
2437 (lower_builtin_setjmp): Adjust comment.
2438 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2439 * tree-cfg.c (found_computed_goto): Remove.
2440 (factor_computed_gotos): Remove.
2441 (make_goto_expr_edges): Return bool, true for computed gotos.
2442 Don't call make_abnormal_goto_edges.
2443 (build_gimple_cfg): Don't set found_computed_goto, don't call
2444 factor_computed_gotos.
2445 (computed_goto_p): No longer static.
2446 (make_blocks): Don't set found_computed_goto.
2447 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2448 (make_edges): If make_goto_expr_edges returns true, push bb
2449 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2450 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2451 vector. Record mapping between bbs and OpenMP regions if there
2452 are any, adjust make_gimple_omp_edges caller. Call
2453 handle_abnormal_edges.
2454 (make_abnormal_goto_edges): Remove.
2455 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2456 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2457 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2458 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2459 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2460 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2461 filling *region also set *region_idx to (*region)->entry->index.
2464 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2465 For REGs set ORIGINAL_REGNO.
2467 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2469 * doc/md.texi: Mention that a target shouldn't implement
2470 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2473 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2475 PR tree-optimization/59594
2476 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2477 a copy of the datarefs vector rather than the vector itself.
2479 2014-01-28 Jason Merrill <jason@redhat.com>
2482 * dwarf2out.c (auto_die): New static.
2483 (gen_type_die_with_usage): Handle C++1y 'auto'.
2484 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2487 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2490 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2491 (SPEC_X32): Likewise.
2492 (SPEC_64): Likewise.
2493 * config/i386/i386.c (ix86_option_override_internal): Turn off
2494 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2496 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2497 * config/i386/i386.h (TARGET_16BIT): New macro.
2498 (TARGET_16BIT_P): Likewise.
2499 * config/i386/i386.opt: Add m16.
2500 * doc/invoke.texi: Document -m16.
2502 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2504 PR preprocessor/59935
2505 * input.c (location_get_source_line): Bail out on when line number
2506 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2508 2014-01-28 Richard Biener <rguenther@suse.de>
2510 PR tree-optimization/58742
2511 * tree-ssa-forwprop.c (associate_plusminus): Handle
2512 pointer subtraction of the form (T)(P + A) - (T)P.
2514 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2516 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2519 2014-01-28 Richard Biener <rguenther@suse.de>
2522 2014-01-28 Richard Biener <rguenther@suse.de>
2524 PR rtl-optimization/45364
2525 PR rtl-optimization/59890
2526 * var-tracking.c (local_get_addr_clear_given_value): Handle
2527 already cleared slot.
2528 (val_reset): Handle not allocated local_get_addr_cache.
2529 (vt_find_locations): Use post-order on the inverted CFG.
2531 2014-01-28 Richard Biener <rguenther@suse.de>
2533 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2535 2014-01-28 Richard Biener <rguenther@suse.de>
2537 PR rtl-optimization/45364
2538 PR rtl-optimization/59890
2539 * var-tracking.c (local_get_addr_clear_given_value): Handle
2540 already cleared slot.
2541 (val_reset): Handle not allocated local_get_addr_cache.
2542 (vt_find_locations): Use post-order on the inverted CFG.
2544 2014-01-28 Alan Modra <amodra@gmail.com>
2546 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2547 * configure.ac <recursive call for build != host>: Define
2548 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2549 and LD_FOR_BUILD too.
2550 * configure: Regenerate.
2552 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2554 * config/i386/i386.c (get_builtin_code_for_version): Separate
2555 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2556 Broadwell from Haswell.
2558 2014-01-27 Steve Ellcey <sellcey@mips.com>
2560 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2561 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2562 * config/mips/mips.c (mips_option_override): Change setting
2564 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2565 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2566 Change from Mask to Var.
2568 2014-01-27 Jeff Law <law@redhat.com>
2570 * ipa-inline.c (inline_small_functions): Fix typo.
2572 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2574 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2575 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2576 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2577 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2578 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2579 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2580 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2581 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2582 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2583 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2584 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2585 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2586 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2587 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2588 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2589 (_mm512_storeu_epi64): Ditto.
2590 (_mm512_cmpge_epi32_mask): Ditto.
2591 (_mm512_cmpge_epu32_mask): Ditto.
2592 (_mm512_cmpge_epi64_mask): Ditto.
2593 (_mm512_cmpge_epu64_mask): Ditto.
2594 (_mm512_cmple_epi32_mask): Ditto.
2595 (_mm512_cmple_epu32_mask): Ditto.
2596 (_mm512_cmple_epi64_mask): Ditto.
2597 (_mm512_cmple_epu64_mask): Ditto.
2598 (_mm512_cmplt_epi32_mask): Ditto.
2599 (_mm512_cmplt_epu32_mask): Ditto.
2600 (_mm512_cmplt_epi64_mask): Ditto.
2601 (_mm512_cmplt_epu64_mask): Ditto.
2602 (_mm512_cmpneq_epi32_mask): Ditto.
2603 (_mm512_cmpneq_epu32_mask): Ditto.
2604 (_mm512_cmpneq_epi64_mask): Ditto.
2605 (_mm512_cmpneq_epu64_mask): Ditto.
2606 (_mm512_expand_pd): Ditto.
2607 (_mm512_expand_ps): Ditto.
2608 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2609 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2610 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2611 * config/i386/i386.c (ix86_builtins): Add
2612 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2613 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2614 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2615 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2616 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2617 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2618 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2619 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2620 IX86_BUILTIN_PMOVUSQW512_MEM.
2621 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2622 __builtin_ia32_pmovsqd512mem_mask,
2623 __builtin_ia32_pmovqd512mem_mask,
2624 __builtin_ia32_pmovusqw512mem_mask,
2625 __builtin_ia32_pmovsqw512mem_mask,
2626 __builtin_ia32_pmovqw512mem_mask,
2627 __builtin_ia32_pmovusdw512mem_mask,
2628 __builtin_ia32_pmovsdw512mem_mask,
2629 __builtin_ia32_pmovdw512mem_mask,
2630 __builtin_ia32_pmovqb512mem_mask,
2631 __builtin_ia32_pmovusqb512mem_mask,
2632 __builtin_ia32_pmovsqb512mem_mask,
2633 __builtin_ia32_pmovusdb512mem_mask,
2634 __builtin_ia32_pmovsdb512mem_mask,
2635 __builtin_ia32_pmovdb512mem_mask.
2636 (bdesc_args): Add __builtin_ia32_expanddf512,
2637 __builtin_ia32_expandsf512.
2638 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2639 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2640 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2641 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2642 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2643 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2644 (avx512f_<code>v8div16qi2_mask_store): This.
2645 (avx512f_expand<mode>): New.
2647 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2649 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2651 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2652 (_mm512_prefetch_i32scatter_pd): Ditto.
2653 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2654 (_mm512_prefetch_i64scatter_pd): Ditto.
2655 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2656 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2657 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2658 (_mm512_prefetch_i32scatter_ps): Ditto.
2659 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2660 (_mm512_prefetch_i64scatter_ps): Ditto.
2661 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2662 * config/i386/i386-builtin-types.def: Define
2663 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2664 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2665 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2666 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2667 IX86_BUILTIN_SCATTERPFQPD.
2668 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2669 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2670 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2671 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2672 __builtin_ia32_scatterpfqps.
2673 (ix86_expand_builtin): Expand new built-ins.
2674 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2675 fix memory access data type.
2676 (*avx512pf_gatherpf<mode>_mask): Ditto.
2677 (*avx512pf_gatherpf<mode>): Ditto.
2678 (avx512pf_scatterpf<mode>): Ditto.
2679 (*avx512pf_scatterpf<mode>_mask): Ditto.
2680 (*avx512pf_scatterpf<mode>): Ditto.
2681 (GATHER_SCATTER_SF_MEM_MODE): New.
2682 (avx512pf_gatherpf<mode>df): Ditto.
2683 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2684 (*avx512pf_scatterpf<mode>df): Ditto.
2686 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2689 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2690 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2693 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2695 * common/config/arm/arm-common.c
2696 (arm_rewrite_mcpu): Handle multiple names.
2698 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2700 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2702 * gimple-builder.h (create_gimple_tmp): Delete.
2704 2014-01-27 Christian Bruel <christian.bruel@st.com>
2706 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2709 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2711 * config/pa/pa.md (call): Generate indirect long calls to non-local
2712 functions when outputing 32-bit code.
2713 (call_value): Likewise except for special call to buggy powf function.
2715 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2716 portable runtime and PIC indirect calls.
2717 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2718 and PIC call sequences. Use ldo instead of blr to set return register
2719 in PIC call sequence.
2721 2014-01-25 Walter Lee <walt@tilera.com>
2723 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2724 avoid clobbering a live register.
2726 2014-01-25 Walter Lee <walt@tilera.com>
2728 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2729 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2730 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2731 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2733 2014-01-25 Walter Lee <walt@tilera.com>
2735 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2736 arguments on even registers.
2737 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2739 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2740 (BIGGEST_ALIGNMENT): Ditto.
2741 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2743 2014-01-25 Walter Lee <walt@tilera.com>
2745 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2746 insns before bundling.
2747 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2749 2014-01-25 Walter Lee <walt@tilera.com>
2751 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2752 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2753 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2755 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2757 * config/mips/constraints.md (kl): Delete.
2758 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2759 define expands, using...
2760 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2761 instructions for MIPS16.
2762 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2763 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2765 2014-01-25 Walter Lee <walt@tilera.com>
2767 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2771 2014-01-25 Walter Lee <walt@tilera.com>
2773 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2774 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2776 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2778 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2781 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2783 * print-rtl.c (in_call_function_usage): New var.
2784 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2785 EXPR_LIST mode as mode and not as reg note name.
2788 * cfgloopmanip.c (copy_loop_info): If
2789 loop->warned_aggressive_loop_optimizations, make sure
2790 the flag is set in target loop too.
2792 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2794 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2796 * builtins.def: Likewise.
2797 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2798 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2799 * ira.c (ira_setup_eliminable_regset): Likewise.
2800 * omp-low.c (gate_expand_omp): Likewise.
2801 (execute_lower_omp): Likewise.
2802 (diagnose_sb_0): Likewise.
2803 (gate_diagnose_omp_blocks): Likewise.
2804 (simd_clone_clauses_extract): Likewise.
2807 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2809 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2810 correction for little endian...
2811 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2814 2014-01-24 Jeff Law <law@redhat.com>
2816 PR tree-optimization/59919
2817 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2818 for non-returning calls.
2820 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2822 * common/config/aarch64/aarch64-common.c
2823 (aarch64_rewrite_mcpu): Handle multiple names.
2824 * config/aarch64/aarch64.h
2825 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2827 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2829 * input.c (add_file_to_cache_tab): Handle the case where fopen
2832 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2835 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2836 from push operand if code of push isn't PRE_DEC.
2838 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2841 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2842 -mquad-memory-atomic. Update -mquad-memory documentation to say
2843 it is only used for non-atomic loads/stores.
2845 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2846 -mquad-memory or -mquad-memory-atomic switches.
2848 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2849 -mquad-memory-atomic to ISA 2.07 support.
2851 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2852 to separate support of normal quad word memory operations (ldq, stq)
2853 from the atomic quad word memory operations.
2855 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2856 support to separate non-atomic quad word operations from atomic
2857 quad word operations. Disable non-atomic quad word operations in
2858 little endian mode so that we don't have to swap words after the
2859 load and before the store.
2860 (quad_load_store_p): Add comment about atomic quad word support.
2861 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2862 options printed with -mdebug=reg.
2864 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2865 -mquad-memory-atomic as the test for whether we have quad word
2866 atomic instructions.
2867 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2868 or -mp8-vector are used, allow byte/half-word atomic operations.
2870 * config/rs6000/sync.md (load_lockedti): Insure that the address
2871 is a proper indexed or indirect address for the lqarx instruction.
2872 On little endian systems, swap the hi/lo registers after the lqarx
2874 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2875 insure the address is valid for the lqarx instruction.
2876 (store_conditionalti): Insure that the address is a proper indexed
2877 or indirect address for the stqcrx. instruction. On little endian
2878 systems, swap the hi/lo registers before doing the stqcrx.
2880 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2881 insure the address is valid for the stqcrx. instruction.
2883 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2884 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2885 type of quad memory support is available.
2887 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2890 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2891 there is a danger of looping.
2893 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2896 force flag_ira_loop_pressure if set via command line.
2898 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2900 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2901 (ashr_simd): New builtin handling DI mode.
2902 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2903 (aarch64_sshr_simddi): New match pattern.
2904 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2905 (vshrd_n_s64): Likewise.
2906 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2908 2014-01-23 Nick Clifton <nickc@redhat.com>
2910 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2911 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2912 favour of mcu specific scripts.
2913 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2916 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2917 Alex Velenko <Alex.Velenko@arm.com>
2919 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2920 (vaddv_s16): Likewise.
2921 (vaddv_s32): Likewise.
2922 (vaddv_u8): Likewise.
2923 (vaddv_u16): Likewise.
2924 (vaddv_u32): Likewise.
2925 (vaddvq_s8): Likewise.
2926 (vaddvq_s16): Likewise.
2927 (vaddvq_s32): Likewise.
2928 (vaddvq_s64): Likewise.
2929 (vaddvq_u8): Likewise.
2930 (vaddvq_u16): Likewise.
2931 (vaddvq_u32): Likewise.
2932 (vaddvq_u64): Likewise.
2933 (vaddv_f32): Likewise.
2934 (vaddvq_f32): Likewise.
2935 (vaddvq_f64): Likewise.
2936 (vmaxv_f32): Likewise.
2937 (vmaxv_s8): Likewise.
2938 (vmaxv_s16): Likewise.
2939 (vmaxv_s32): Likewise.
2940 (vmaxv_u8): Likewise.
2941 (vmaxv_u16): Likewise.
2942 (vmaxv_u32): Likewise.
2943 (vmaxvq_f32): Likewise.
2944 (vmaxvq_f64): Likewise.
2945 (vmaxvq_s8): Likewise.
2946 (vmaxvq_s16): Likewise.
2947 (vmaxvq_s32): Likewise.
2948 (vmaxvq_u8): Likewise.
2949 (vmaxvq_u16): Likewise.
2950 (vmaxvq_u32): Likewise.
2951 (vmaxnmv_f32): Likewise.
2952 (vmaxnmvq_f32): Likewise.
2953 (vmaxnmvq_f64): Likewise.
2954 (vminv_f32): Likewise.
2955 (vminv_s8): Likewise.
2956 (vminv_s16): Likewise.
2957 (vminv_s32): Likewise.
2958 (vminv_u8): Likewise.
2959 (vminv_u16): Likewise.
2960 (vminv_u32): Likewise.
2961 (vminvq_f32): Likewise.
2962 (vminvq_f64): Likewise.
2963 (vminvq_s8): Likewise.
2964 (vminvq_s16): Likewise.
2965 (vminvq_s32): Likewise.
2966 (vminvq_u8): Likewise.
2967 (vminvq_u16): Likewise.
2968 (vminvq_u32): Likewise.
2969 (vminnmv_f32): Likewise.
2970 (vminnmvq_f32): Likewise.
2971 (vminnmvq_f64): Likewise.
2973 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2975 * config/aarch64/aarch64-simd.md
2976 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2977 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2978 (*aarch64_mul3_elt<mode>): Likewise.
2979 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2980 (*aarch64_mul3_elt_to_64v2df): Likewise.
2981 (*aarch64_mla_elt<mode>): Likewise.
2982 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2983 (*aarch64_mls_elt<mode>): Likewise.
2984 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2985 (*aarch64_fma4_elt<mode>): Likewise.
2986 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2987 (*aarch64_fma4_elt_to_64v2df): Likewise.
2988 (*aarch64_fnma4_elt<mode>): Likewise.
2989 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2990 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2991 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2992 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2993 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2994 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2995 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2996 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2997 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2999 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3001 * config/aarch64/aarch64-simd.md
3002 (aarch64_be_checked_get_lane<mode>): New define_expand.
3003 * config/aarch64/aarch64-simd-builtins.def
3004 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3005 New builtin definition.
3006 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3007 Use new safe be builtin.
3009 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3011 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3013 (aarch64_be_st1<mode>): Likewise.
3014 (aarch_ld1<VALL:mode>): Define_expand modified.
3015 (aarch_st1<VALL:mode>): Likewise.
3016 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3017 (UNSPEC_ST1): Likewise.
3019 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3021 * config/microblaze/microblaze.md: Add trap insn and attribute
3023 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3025 PR preprocessor/58580
3026 * input.h (location_get_source_line): Take an additional line_size
3028 (void diagnostics_file_cache_fini): Declare new function.
3029 * input.c (struct fcache): New type.
3030 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3031 New static constants.
3032 (diagnostic_file_cache_init, total_lines_num)
3033 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3034 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3035 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3036 (get_next_line, read_next_line, goto_next_line, read_line_num):
3037 New static function definitions.
3038 (diagnostic_file_cache_fini): New function.
3039 (location_get_source_line): Take an additional output line_len
3040 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3042 * diagnostic.c (diagnostic_finish): Call
3043 diagnostic_file_cache_fini.
3044 (adjust_line): Take an additional input parameter for the length
3045 of the line, rather than calculating it with strlen.
3046 (diagnostic_show_locus): Adjust the use of
3047 location_get_source_line and adjust_line with respect to their new
3048 signature. While displaying a line now, do not stop at the first
3049 null byte. Rather, display the zero byte as a space and keep
3050 going until we reach the size of the line.
3051 * Makefile.in: Add vec.o to OBJS-libcommon
3053 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3054 Ilya Tocar <ilya.tocar@intel.com>
3056 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3057 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3058 (__builtin_ia32_kmov16): Ditto.
3059 * config/i386/i386.md (UNSPEC_KMOV): New.
3062 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3064 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3065 (_mm512_storeu_si512): Ditto.
3067 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3070 * rtl.h (get_referenced_operands): Declare.
3071 * recog.c (get_referenced_operands): New function.
3072 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3073 operands have been referenced when recording LO_SUM references.
3075 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3077 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3079 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3081 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3082 Enable for generic and recent AMD targets.
3084 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3086 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3087 ARG_SIZE note when adjustment was eliminated.
3089 2014-01-22 Jeff Law <law@redhat.com>
3091 PR tree-optimization/59597
3092 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3093 in file. Accept new argument REGISTERING and use it to modify
3094 dump output appropriately.
3095 (register_jump_thread): Corresponding changes.
3096 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3097 thread paths involving joiner blocks. Add code to dump cancelled
3098 jump threading paths.
3100 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3102 PR rtl-optimization/59477
3103 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3104 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3106 2014-01-22 Tom Tromey <tromey@redhat.com>
3108 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3110 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3112 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3114 PR rtl-optimization/59896
3115 * lra-constraints.c (process_alt_operands): Check unused note for
3116 matched operands of insn with no output reloads.
3118 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3120 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3121 (mips_move_from_gpr_cost): Likewise.
3123 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3125 PR rtl-optimization/59858
3126 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3127 ira_class_hard_regs_num.
3128 (process_alt_operands): Increase reject for dying matched operand.
3130 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3133 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3134 smaller than size, perform several stores or loads and stores
3135 at dst + count - size to store or copy all of size bytes, rather
3136 than just last modesize bytes.
3138 2014-01-20 DJ Delorie <dj@redhat.com>
3140 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3141 that CLOBBERs are REGs before propogating their values.
3143 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3146 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3147 (cgraph_inline_failed_type): New function.
3148 * cgraph.h (DEFCIFCODE): Add type.
3149 (cgraph_inline_failed_type_t): New enum.
3150 (cgraph_inline_failed_type): New prototype.
3151 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3152 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3153 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3154 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3155 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3156 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3157 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3158 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3159 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3160 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3161 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3162 OPTIMIZATION_MISMATCH.
3163 * tree-inline.c (expand_call_inline): Emit errors during
3164 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3166 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3169 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3170 mode attribute in insn output.
3172 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3174 * output.h (output_constant): Delete.
3175 * varasm.c (output_constant): Make private.
3177 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3179 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3181 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3184 * tree.h (fold_builtin_strcat): New prototype.
3185 * builtins.c (fold_builtin_strcat): No longer static. Add len
3186 argument, if non-NULL, don't call c_strlen. Optimize
3187 directly into __builtin_memcpy instead of __builtin_strcpy.
3188 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3189 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3191 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3193 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3194 for SImode_address_operand operands, having only a REG argument.
3196 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3198 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3199 loader name using mbig-endian.
3200 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3202 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3204 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3208 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3210 * config/aarch64/aarch64-protos.h
3211 (aarch64_cannot_change_mode_class_ptr): Declare.
3212 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3213 aarch64_cannot_change_mode_class_ptr): New.
3214 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3215 backend hook aarch64_cannot_change_mode_class.
3217 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3219 * common/config/aarch64/aarch64-common.c
3220 (aarch64_handle_option): Don't handle any option order logic here.
3221 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3222 selected_cpu, warn on architecture version mismatch.
3223 (aarch64_override_options): Fix parsing order for option strings.
3225 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3226 Iain Sandoe <iain@codesourcery.com>
3229 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3230 warning. Amend comment to reflect current functionality.
3232 2014-01-20 Richard Biener <rguenther@suse.de>
3235 * builtins.c (fold_builtin_strcat): Remove case better handled
3236 by tree-ssa-strlen.c.
3238 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3240 * config/aarch64/aarch64.opt
3241 (mcpu, march, mtune): Make case-insensitive.
3243 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3246 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3247 if operands[1] is a REG or ZERO_EXTEND of a REG.
3249 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3251 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3253 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3255 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3256 long non-pic millicode calls.
3258 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3260 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3262 2014-01-19 Kito Cheng <kito@0xlab.org>
3264 * builtins.c (expand_movstr): Check movstr expand done or fail.
3266 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3267 H.J. Lu <hongjiu.lu@intel.com>
3270 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3271 to DImode for zero-extended addresses.
3273 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3275 PR rtl-optimization/57763
3276 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3277 on the new indirect jump_insn and increment LABEL_NUSES (label).
3279 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3283 * config.gcc (x86_archs): New variable.
3284 (x86_64_archs): Likewise.
3285 (x86_cpus): Likewise.
3286 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3287 --with-arch/--with-cpu= options.
3288 Support --with-arch=/--with-cpu={nehalem,westmere,
3289 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3291 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3293 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3294 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3296 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3298 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3300 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3303 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3304 clear cpp_get_options (parse_in)->warn_unused_macros for
3305 ix86_target_macros_internal with cpp_define.
3307 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3309 * jump.c (delete_related_insns): Keep (use (insn))s.
3310 * reorg.c (redundant_insn): Check for barriers too.
3312 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3314 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3316 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3318 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3319 call to $$dyncall when TARGET_LONG_CALLS is true.
3321 2014-01-17 Jeff Law <law@redhat.com>
3323 * ree.c (combine_set_extension): Temporarily disable test for
3324 changing number of hard registers.
3326 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3329 * ipa-inline-analysis.c (inline_free_summary):
3330 Do not free summary of aliases.
3332 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3335 * gimplify.c (gimplify_expr): Use create_tmp_var
3336 instead of create_tmp_var_raw. If cond doesn't have
3337 integral type, don't add the IFN_ANNOTATE builtin at all.
3339 2014-01-17 Martin Jambor <mjambor@suse.cz>
3342 * ipa-cp.c (prev_edge_clone): New variable.
3343 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3344 Also resize prev_edge_clone vector.
3345 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3346 (ipcp_edge_removal_hook): New function.
3347 (ipcp_driver): Register ipcp_edge_removal_hook.
3349 2014-01-17 Andrew Pinski <apinski@cavium.com>
3350 Steve Ellcey <sellcey@mips.com>
3353 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3356 2014-01-17 Jeff Law <law@redhat.com>
3359 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3360 so that pass_ccp runs first.
3362 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3364 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3365 (ix86_adjust_cost): Use !TARGET_XXX.
3366 (do_reorder_for_imul): Likewise.
3367 (swap_top_of_ready_list): Likewise.
3368 (ix86_sched_reorder): Likewise.
3370 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3372 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3373 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3374 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3375 (intel_memset): New. Duplicate slm_memset.
3376 (intel_cost): New. Duplicate slm_cost.
3377 (m_INTEL): New macro.
3378 (processor_target_table): Add "intel".
3379 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3380 with PROCESSOR_INTEL for "intel".
3381 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3382 PROCESSOR_SILVERMONT.
3383 (ix86_issue_rate): Likewise.
3384 (ix86_adjust_cost): Likewise.
3385 (ia32_multipass_dfa_lookahead): Likewise.
3386 (swap_top_of_ready_list): Likewise.
3387 (ix86_sched_reorder): Likewise.
3388 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3389 instead of TARGET_OPT_AGU.
3390 * config/i386/i386.h (TARGET_INTEL): New.
3391 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3392 (processor_type): Add PROCESSOR_INTEL.
3393 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3394 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3396 2014-01-17 Marek Polacek <polacek@redhat.com>
3399 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3402 2014-01-17 Richard Biener <rguenther@suse.de>
3404 PR tree-optimization/46590
3405 * opts.c (default_options_table): Add entries for
3406 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3407 all enabled at -O1 but not for -Og.
3408 * common.opt (fbranch-count-reg): Remove Init(1).
3409 (fmove-loop-invariants): Likewise.
3410 (ftree-pta): Likewise.
3412 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3414 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3415 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3416 decls to at least the GCC 4.8 used alignments.
3419 * tree-nested.c (convert_nonlocal_reference_stmt,
3420 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3421 of GIMPLE_BIND stmts, adjust associated decls.
3423 2014-01-17 Richard Biener <rguenther@suse.de>
3425 PR tree-optimization/46590
3426 * vec.h (vec<>::bseach): New member function implementing
3427 binary search according to C89 bsearch.
3428 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3429 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3430 bitmap pointer again. Make accesses_in_loop a flat array.
3431 (mem_ref_obstack): New global.
3432 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3433 (mark_ref_stored): Likewise.
3434 (ref_indep_loop_p_2): Likewise.
3435 (set_ref_stored_in_loop): New helper function.
3436 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3437 (memref_free): Adjust.
3438 (record_mem_ref_loc): Simplify.
3439 (gather_mem_refs_stmt): Adjust.
3440 (sort_locs_in_loop_postorder_cmp): New function.
3441 (analyze_memory_references): Sort accesses_in_loop after
3442 loop postorder number.
3443 (find_ref_loc_in_loop_cmp): New function.
3444 (for_all_locs_in_loop): Find relevant cluster of locs in
3445 accesses_in_loop and iterate without recursion.
3446 (execute_sm): Avoid uninit warning.
3447 (struct ref_always_accessed): Simplify.
3448 (ref_always_accessed::operator ()): Likewise.
3449 (ref_always_accessed_p): Likewise.
3450 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3451 loop postorder numbers here.
3452 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3455 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3458 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3459 on decls for which assemble_alias has been called.
3461 2014-01-17 Nick Clifton <nickc@redhat.com>
3463 * config/msp430/msp430.opt: (mcpu): New option.
3464 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3465 (msp430_option_override): Parse target_cpu. If the MCU name
3466 matches a generic string, clear target_mcu.
3467 (msp430_attr): Allow numeric interrupt values up to 63.
3468 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3469 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3471 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3473 * config/msp430/msp430.md (popm): Use %J rather than %I.
3474 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3475 (addhi_cy_i): Use immediate_operand for operand 2.
3476 * doc/invoke.texi: Document -mcpu option.
3478 2014-01-17 Richard Biener <rguenther@suse.de>
3480 PR rtl-optimization/38518
3481 * df.h (df_analyze_loop): Declare.
3482 * df-core.c: Include cfgloop.h.
3483 (df_analyze_1): Split out main part of df_analyze.
3484 (df_analyze): Adjust.
3485 (loop_inverted_post_order_compute): New function.
3486 (loop_post_order_compute): Likewise.
3487 (df_analyze_loop): New function avoiding whole-function
3489 * loop-invariant.c (find_defs): Use df_analyze_loop.
3490 (find_invariants): Adjust.
3491 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3493 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3495 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3496 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3498 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3500 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3501 traversal when removing references.
3503 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3506 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3508 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3511 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3512 pushing a reload for an autoinc when we had previously reloaded an
3513 inner part of the address.
3515 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3517 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3519 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3520 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3521 when not giving up or versioning for alias only because of
3523 (vect_analyze_data_ref_dependences): Set to true.
3524 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3526 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3527 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3531 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3534 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3535 operand 0 predicate for gathers, use a new pseudo as subtarget.
3537 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3540 * lra-constraints.c (process_alt_operands): Add printing debug info.
3541 Check absence of input/output reloads for matched operands too.
3543 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3545 PR rtl-optimization/59835
3546 * ira.c (ira_init_register_move_cost): Increase cost for
3549 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3551 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3553 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3556 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3557 non-register objects. Use gen_(high/low)part more consistently.
3560 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3563 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3564 endian support, remove tests for WORDS_BIG_ENDIAN.
3565 (p8_mfvsrd_3_<mode>): Likewise.
3566 (reload_gpr_from_vsx<mode>): Likewise.
3567 (reload_gpr_from_vsxsf): Likewise.
3568 (p8_mfvsrd_4_disf): Likewise.
3570 2014-01-16 Richard Biener <rguenther@suse.de>
3572 PR rtl-optimization/46590
3573 * lcm.c (compute_antinout_edge): Use postorder iteration.
3574 (compute_laterin): Use inverted postorder iteration.
3576 2014-01-16 Nick Clifton <nickc@redhat.com>
3579 * varasm.c (output_constant): Return the number of bytes actually
3581 (output_constructor_array_range): Update the field size with the
3582 number of bytes emitted by output_constant.
3583 (output_constructor_regular_field): Likewise. Also do not
3584 complain if the total number of bytes emitted is now greater
3585 than the expected fieldpos.
3586 * output.h (output_constant): Update prototype and descriptive comment.
3588 2014-01-16 Marek Polacek <polacek@redhat.com>
3591 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3592 it is error_mark_node.
3594 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3596 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3597 VALID_AVX256_REG_OR_OI_MODE.
3599 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3601 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3602 current procedure should be profiled.
3604 2014-01-15 Andrew Pinski <apinski@cavium.com>
3606 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3607 of moving from/to the STACK_REG register class.
3609 2014-01-15 Richard Henderson <rth@redhat.com>
3612 * reginfo.c (global_regs_decl): Globalize.
3613 * rtl.h (global_regs_decl): Declare.
3614 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3615 reserved via global_regs.
3617 2014-01-15 Teresa Johnson <tejohnson@google.com>
3619 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3621 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3623 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3624 and vmulosh rather than call gen_vec_widen_smult_*.
3625 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3626 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3627 (vec_widen_smult_even_v16qi): Likewise.
3628 (vec_widen_umult_even_v8hi): Likewise.
3629 (vec_widen_smult_even_v8hi): Likewise.
3630 (vec_widen_umult_odd_v16qi): Likewise.
3631 (vec_widen_smult_odd_v16qi): Likewise.
3632 (vec_widen_umult_odd_v8hi): Likewise.
3633 (vec_widen_smult_odd_v8hi): Likewise.
3634 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3635 vmuloub rather than call gen_vec_widen_umult_*.
3636 (vec_widen_umult_lo_v16qi): Likewise.
3637 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3638 vmulosb rather than call gen_vec_widen_smult_*.
3639 (vec_widen_smult_lo_v16qi): Likewise.
3640 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3641 rather than call gen_vec_widen_umult_*.
3642 (vec_widen_umult_lo_v8hi): Likewise.
3643 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3644 rather than call gen_vec_widen_smult_*.
3645 (vec_widen_smult_lo_v8hi): Likewise.
3647 2014-01-15 Jeff Law <law@redhat.com>
3649 PR tree-optimization/59747
3650 * ree.c (find_and_remove_re): Properly handle case where a second
3651 eliminated extension requires widening a copy created for elimination
3652 of a prior extension.
3653 (combine_set_extension): Ensure that the number of hard regs needed
3654 for a destination register does not change when we widen it.
3656 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3658 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3659 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3660 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3661 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3662 (avr-*-rtems*): Likewise.
3663 (bfin*-rtems*): Likewise.
3664 (moxie-*-rtems*): Likewise.
3665 (h8300-*-rtems*): Likewise.
3666 (i[34567]86-*-rtems*): Likewise.
3667 (lm32-*-rtems*): Likewise.
3668 (m32r-*-rtems*): Likewise.
3669 (m68k-*-rtems*): Likewise.
3670 (microblaze*-*-rtems*): Likewise.
3671 (mips*-*-rtems*): Likewise.
3672 (powerpc-*-rtems*): Likewise.
3673 (sh-*-rtems*): Likewise.
3674 (sparc-*-rtems*): Likewise.
3675 (sparc64-*-rtems*): Likewise.
3676 (v850-*-rtems*): Likewise.
3677 (m32c-*-rtems*): Likewise.
3679 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3681 PR rtl-optimization/59511
3682 * ira.c (ira_init_register_move_cost): Use memory costs for some
3683 cases of register move cost calculations.
3684 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3685 instead of BB frequency.
3686 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3687 * lra-assigns.c (find_hard_regno_for): Ditto.
3689 2014-01-15 Richard Biener <rguenther@suse.de>
3691 PR tree-optimization/59822
3692 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3693 (vectorizable_load): Use it to hoist defs of uses of invariant
3694 loads out of the loop.
3696 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3697 Kugan Vivekanandarajah <kuganv@linaro.org>
3700 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3703 2014-01-15 Richard Biener <rguenther@suse.de>
3705 PR rtl-optimization/59802
3706 * lcm.c (compute_available): Use inverted postorder to seed
3707 the initial worklist.
3709 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3712 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3713 ADDR_REGS for invalid symrefs in non-PIC code.
3715 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3718 * builtins.c (determine_block_size): Initialize *probable_max_size
3719 even if len_rtx is CONST_INT.
3721 2014-01-14 Andrew Pinski <apinski@cavium.com>
3723 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3724 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3725 (cortexa53_tunings): Likewise.
3726 (aarch64_sched_issue_rate): New function.
3727 (TARGET_SCHED_ISSUE_RATE): Define.
3729 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3731 * ira-costs.c (find_costs_and_classes): Add missed
3732 ira_init_register_move_cost_if_necessary.
3734 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3737 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3739 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3742 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3743 to indicate if type is used for function return value. Warn ABI
3744 change if the vector mode isn't available for function return value.
3745 (ix86_function_arg_advance): Pass false to type_natural_mode.
3746 (ix86_function_arg): Likewise.
3747 (ix86_gimplify_va_arg): Likewise.
3748 (function_arg_32): Don't warn ABI change.
3749 (ix86_function_value): Pass true to type_natural_mode.
3750 (ix86_return_in_memory): Likewise.
3751 (ix86_struct_value_rtx): Removed.
3752 (TARGET_STRUCT_VALUE_RTX): Likewise.
3754 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3756 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3757 converting a conditional jump into a conditional return.
3759 2014-01-14 Richard Biener <rguenther@suse.de>
3761 PR tree-optimization/58921
3762 PR tree-optimization/59006
3763 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3764 hoisting invariant stmts.
3765 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3766 invariant loads on the preheader edge if possible.
3768 2014-01-14 Joey Ye <joey.ye@arm.com>
3770 * doc/plugin.texi (Building GCC plugins): Update to C++.
3772 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3774 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3775 (_mm_rcp28_round_ss): Ditto.
3776 (_mm_rsqrt28_round_sd): Ditto.
3777 (_mm_rsqrt28_round_ss): Ditto.
3778 (_mm_rcp28_sd): Ditto.
3779 (_mm_rcp28_ss): Ditto.
3780 (_mm_rsqrt28_sd): Ditto.
3781 (_mm_rsqrt28_ss): Ditto.
3782 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3783 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3784 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3785 (IX86_BUILTIN_RCP28SD): Ditto.
3786 (IX86_BUILTIN_RCP28SS): Ditto.
3787 (IX86_BUILTIN_RSQRT28SD): Ditto.
3788 (IX86_BUILTIN_RSQRT28SS): Ditto.
3789 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3790 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3791 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3792 (ix86_expand_special_args_builtin): Expand new FTYPE.
3793 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3794 (srcp14<mode>): Make insn unary.
3795 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3796 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3797 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3798 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3799 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3800 Fix rounding: make it SAE only.
3801 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3803 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3805 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3806 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3807 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3808 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3809 (round_saeonly_mask_scalar_operand4): Ditto.
3810 (round_saeonly_mask_scalar_op3): Ditto.
3811 (round_saeonly_mask_scalar_op4): Ditto.
3813 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3815 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3816 Implement -maltivec=be for vec_insert and vec_extract.
3818 2014-01-10 DJ Delorie <dj@redhat.com>
3820 * config/msp430/msp430.md (call_internal): Don't allow memory
3821 references with SP as the base register.
3822 (call_value_internal): Likewise.
3823 * config/msp430/constraints.md (Yc): New. For memory references
3824 that don't use SP as a base register.
3826 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3827 "an integer without a # prefix"
3828 * config/msp430/msp430.md (epilogue_helper): Use it.
3830 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3833 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3834 AVX512F gather builtins.
3835 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3836 on gather decls with INTEGER_TYPE masktype.
3837 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3838 directly into the builtin rather than hoisting it before loop.
3840 PR tree-optimization/59387
3841 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3842 (scev_const_prop): If folded_casts and type has undefined overflow,
3843 use force_gimple_operand instead of force_gimple_operand_gsi and
3844 for each added stmt if it is assign with
3845 arith_code_with_undefined_signed_overflow, call
3846 rewrite_to_defined_overflow.
3847 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3848 gimple-fold.h instead.
3849 (arith_code_with_undefined_signed_overflow,
3850 rewrite_to_defined_overflow): Moved to ...
3851 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3852 rewrite_to_defined_overflow): ... here. No longer static.
3853 Include gimplify-me.h.
3854 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3855 rewrite_to_defined_overflow): New prototypes.
3857 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3859 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3861 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3863 * builtins.c (get_object_alignment_2): Minor tweak.
3864 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3866 2014-01-13 Christian Bruel <christian.bruel@st.com>
3868 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3869 optimized non constant lengths.
3871 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3874 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3875 load as __atomic_load_N if possible.
3877 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3879 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3881 (rs6000_expand_builtin): Adjust call.
3883 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3886 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3887 * config/rs6000/rs6000.c: Include target-globals.h.
3888 (rs6000_set_current_function): Instead of doing target_reinit
3889 unconditionally, use save_target_globals_default_opts and
3890 restore_target_globals.
3892 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3894 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3895 (rs6000_expand_builtin): Handle mffs and mtfsf.
3896 (rs6000_init_builtins): Define mffs and mtfsf.
3897 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3898 (rs6000_mffs): New pattern.
3899 (rs6000_mtfsf): New pattern.
3901 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3903 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3904 Start narrowing with START. Apply candidate-use pair
3905 and check overall cost in narrowing.
3906 (iv_ca_prune): Pass new argument.
3908 2014-01-10 Jeff Law <law@redhat.com>
3911 * ree.c (combine_reaching_defs): Ensure the defining statement
3912 occurs before the extension when optimizing extensions with
3913 different source and destination hard registers.
3915 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3918 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3919 vtables into the type inheritance graph.
3921 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3923 PR rtl-optimization/59754
3924 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3925 modes in the REGNO != REGNO case.
3927 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3929 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3931 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3933 PR tree-optimization/59745
3934 * tree-predcom.c (tree_predictive_commoning_loop): Call
3935 free_affine_expand_cache if giving up because components is NULL.
3937 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3938 GC in payload of target_globals struct instead of allocating them on
3939 the heap and the larger structs separately using GC.
3940 * target-globals.h (struct target_globals): Make regs, hard_regs,
3941 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3942 of GTY((skip)) and change type to void *.
3943 (reset_target_globals): Cast loads from those fields to corresponding
3946 2014-01-10 Steve Ellcey <sellcey@mips.com>
3949 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3950 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3951 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3953 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3956 * aarch64-modes.def (CC_Zmode): New flags mode.
3957 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3958 represents an equality.
3959 (aarch64_get_condition_code): Handle CC_Zmode.
3960 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3962 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3964 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3965 extraction in good case.
3967 2014-01-10 Richard Biener <rguenther@suse.de>
3969 PR tree-optimization/59374
3970 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3971 checking after SLP discovery. Mark stmts not participating
3972 in any SLP instance properly.
3974 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3976 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3977 when handling a SET rtx.
3979 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3981 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3982 (cortex-a57): Likewise.
3983 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3985 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3987 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3988 non-iwmmxt builtins.
3990 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3994 * ipa-devirt.c record_target_from_binfo): Take as argument
3995 stack of binfos and lookup matching one for virtual inheritance.
3996 (possible_polymorphic_call_targets_1): Update.
3998 2014-01-10 Huacai Chen <chenhc@lemote.com>
4000 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4001 kernel strings for Loongson-2E/2F/3A.
4003 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4006 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4007 is_gimple_call before calling gimple_call_internal_p.
4009 2014-01-09 Steve Ellcey <sellcey@mips.com>
4011 * Makefile.in (TREE_FLOW_H): Remove.
4012 (TREE_SSA_H): Add file names from tree-flow.h.
4013 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4014 * tree.h: Remove tree-flow.h reference.
4015 * hash-table.h: Remove tree-flow.h reference.
4016 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4017 reference with tree-ssa-loop.h.
4019 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4021 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4022 default element-order behavior for -maltivec.
4023 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4024 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4025 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4026 when targeting big endian, at least for now.
4027 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4029 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4032 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4033 var satisfies use_register_for_decl, just take into account type
4034 alignment, rather than decl alignment.
4036 PR tree-optimization/59622
4037 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4038 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4039 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4040 Don't devirtualize for inplace at all. For targets.length () == 1,
4041 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4043 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4045 * config/i386/i386.md (cpu): Remove the unused btver1.
4047 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4049 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4051 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4054 * tree-core.h (struct target_globals): New forward declaration.
4055 (struct tree_target_option): Add globals field.
4056 * tree.h (TREE_TARGET_GLOBALS): Define.
4057 (prepare_target_option_nodes_for_pch): New prototype.
4058 * target-globals.h (struct target_globals): Define even if
4060 * tree.c (prepare_target_option_node_for_pch,
4061 prepare_target_option_nodes_for_pch): New functions.
4062 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4063 * config/i386/i386.c: Include target-globals.h.
4064 (ix86_set_current_function): Instead of doing target_reinit
4065 unconditionally, use save_target_globals_default_opts and
4066 restore_target_globals.
4068 2014-01-09 Richard Biener <rguenther@suse.de>
4070 PR tree-optimization/59715
4071 * tree-cfg.h (split_critical_edges): Declare.
4072 * tree-cfg.c (split_critical_edges): Export.
4073 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4075 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4077 * cfgexpand.c (expand_stack_vars): Optionally disable
4078 asan stack protection.
4079 (expand_used_vars): Likewise.
4080 (partition_stack_vars): Likewise.
4081 * asan.c (asan_emit_stack_protection): Optionally disable
4082 after return stack usage.
4083 (instrument_derefs): Optionally disable memory access instrumentation.
4084 (instrument_builtin_call): Likewise.
4085 (instrument_strlen_call): Likewise.
4086 (asan_protect_global): Optionally disable global variables protection.
4087 * doc/invoke.texi: Added doc for new options.
4088 * params.def: Added new options.
4089 * params.h: Likewise.
4091 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4093 PR rtl-optimization/59724
4094 * ifcvt.c (cond_exec_process_if_block): Don't call
4095 flow_find_head_matching_sequence with 0 longest_match.
4096 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4097 non-active insns if !stop_after.
4098 (try_head_merge_bb): Revert 2014-01-07 changes.
4100 2014-01-08 Jeff Law <law@redhat.com>
4102 * ree.c (get_sub_rtx): New function, extracted from...
4103 (merge_def_and_ext): Here.
4104 (combine_reaching_defs): Use get_sub_rtx.
4106 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4108 * cgraph.h (varpool_variable_node): Do not choke on null node.
4110 2014-01-08 Catherine Moore <clm@codesourcery.com>
4112 * config/mips/mips.md (simple_return): Attempt to use JRC
4114 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4116 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4118 PR rtl-optimization/59137
4119 * reorg.c (steal_delay_list_from_target): Call update_block for
4121 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4123 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4125 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4126 two duplicate entries.
4128 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4131 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4133 * config/mips/mips.c (mips_truncated_op_cost): New function.
4134 (mips_rtx_costs): Adjust test for BADDU.
4135 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4137 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4139 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4140 (*baddu_si): ...this new pattern.
4142 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4145 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4147 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4150 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4152 (expand_expr, expand_normal): Adjust.
4153 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4154 inner_reference_p. Use inner_reference_p to expand inner references.
4155 (store_expr): Adjust.
4156 * cfgexpand.c (expand_call_stmt): Adjust.
4158 2014-01-08 Rong Xu <xur@google.com>
4160 * gcov-io.c (gcov_var): Move from gcov-io.h.
4161 (gcov_position): Ditto.
4162 (gcov_is_error): Ditto.
4163 (gcov_rewrite): Ditto.
4164 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4165 only part to libgcc/libgcov.h.
4167 2014-01-08 Marek Polacek <polacek@redhat.com>
4170 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4172 2014-01-08 Marek Polacek <polacek@redhat.com>
4175 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4177 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4179 PR rtl-optimization/59649
4180 * stor-layout.c (get_mode_bounds): For BImode return
4181 0 and STORE_FLAG_VALUE.
4183 2014-01-08 Richard Biener <rguenther@suse.de>
4186 * gimple.h (is_gimple_builtin_call): Remove.
4187 (gimple_builtin_call_types_compatible_p): New.
4188 (gimple_call_builtin_p): New overload.
4189 * gimple.c (is_gimple_builtin_call): Remove.
4190 (validate_call): Rename to ...
4191 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4193 (validate_type): New static function.
4194 (gimple_call_builtin_p): New overload and adjust.
4195 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4196 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4197 (gimple_fold_stmt_to_constant_1): Likewise.
4198 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4200 2014-01-08 Richard Biener <rguenther@suse.de>
4203 * gimplify.c (gimplify_expr): Gimplify register-register type
4204 VIEW_CONVERT_EXPRs to separate stmts.
4206 2014-01-07 Jeff Law <law@redhat.com>
4209 * ree.c (combine_set_extension): Handle case where source
4210 and destination registers in an extension insn are different.
4211 (combine_reaching_defs): Allow source and destination registers
4212 in extension to be different under limited circumstances.
4213 (add_removable_extension): Remove restriction that the
4214 source and destination registers in the extension are the same.
4215 (find_and_remove_re): Emit a copy from the extension's
4216 destination to its source after the defining insn if
4217 the source and destination registers are different.
4220 * ifcvt.c (merge_if_block): If we are merging a block with more than
4221 one successor with a block with no successors, remove any BARRIER
4222 after the second block.
4224 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4226 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4228 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4231 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4232 for 14-bit register offsets when INT14_OK_STRICT is false.
4234 2014-01-07 Roland Stigge <stigge@antcom.de>
4235 Michael Meissner <meissner@linux.vnet.ibm.com>
4238 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4239 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4241 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4243 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4246 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4248 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4249 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4250 rtx is const0_rtx or not.
4252 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4255 * target-globals.c (save_target_globals): Remove this_fn_optab
4257 * toplev.c: Include optabs.h.
4258 (target_reinit): Temporarily restore the global options if another
4259 set of options are in force.
4261 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4263 PR rtl-optimization/58668
4264 * cfgcleanup.c (flow_find_cross_jump): Don't count
4265 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4266 to determine what is counted.
4267 (flow_find_head_matching_sequence): Use active_insn_p to determine
4269 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4271 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4272 determine what is counted.
4274 PR tree-optimization/59643
4275 * tree-predcom.c (split_data_refs_to_components): If one dr is
4276 read and one write, determine_offset fails and the write isn't
4277 in the bad component, just put the read into the bad component.
4279 2014-01-07 Mike Stump <mikestump@comcast.net>
4280 Jakub Jelinek <jakub@redhat.com>
4283 * tree-core.h (struct tree_optimization_option): Change optabs
4284 type from unsigned char * to void *.
4285 * optabs.c (init_tree_optimization_optabs): Adjust
4286 TREE_OPTIMIZATION_OPTABS initialization.
4288 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4291 * config/i386/i386.h (struct machine_function): Add
4292 no_drap_save_restore field.
4293 * config/i386/i386.c (ix86_save_reg): Use
4294 !cfun->machine->no_drap_save_restore instead of
4295 crtl->stack_realign_needed.
4296 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4297 this function clears frame_pointer_needed. Set
4298 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4299 and DRAP reg is needed.
4301 2014-01-06 Marek Polacek <polacek@redhat.com>
4304 * doc/implement-c.texi: Mention that other integer types are
4305 permitted as bit-field types in strictly conforming mode.
4307 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4309 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4312 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4314 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4316 2014-01-06 Martin Jambor <mjambor@suse.cz>
4319 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4321 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4323 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4327 * var-tracking.c (add_stores): Preserve the value of the source even if
4328 we don't record the store.
4330 2014-01-06 Terry Guo <terry.guo@arm.com>
4332 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4334 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4337 * config/darwin.c (darwin_function_section): Adjust return values to
4338 correspond to optimisation changes made in r206070.
4340 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4342 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4343 from prefetch_block tune setting.
4344 (nocona_cost): Correct size of prefetch block to 64.
4346 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4348 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4349 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4350 used to save the static chain register in the computation of the offset
4351 from which the FP registers need to be restored.
4353 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4355 PR tree-optimization/59519
4356 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4357 ICE if get_current_def (current_new_name) is already non-NULL, as long
4358 as it is a phi result of some other phi in *new_exit_bb that has
4361 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4362 or vmovdqu* for misaligned_operand.
4363 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4364 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4365 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4366 aligned_mem for AVX512F masked aligned load and store builtins and for
4369 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4371 PR tree-optimization/59651
4372 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4373 Address range for negative step should be added by TYPE_SIZE_UNIT.
4375 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4377 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4378 overlapping registers also for registers other than the stack pointer.
4380 2014-01-03 Marek Polacek <polacek@redhat.com>
4383 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4386 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4389 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4392 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4393 (push splitter): Use <P:MODE_SIZE> instead of
4394 GET_MODE_SIZE (<P:MODE>mode).
4395 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4396 (mov -1, reg peephole2): Likewise.
4397 * config/i386/sse.md (*mov<mode>_internal,
4398 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4399 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4400 *<code><mode>3, *andnot<mode>3<mask_name>,
4401 <mask_codefor><code><mode>3<mask_name>): Likewise.
4402 * config/i386/subst.md (mask_mode512bit_condition,
4403 sd_mask_mode512bit_condition): Likewise.
4405 2014-01-02 Xinliang David Li <davidxl@google.com>
4407 PR tree-optimization/59303
4408 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4409 (dump_predicates): Better output format.
4410 (pred_equal_p): New function.
4411 (is_neq_relop_p): Ditto.
4412 (is_neq_zero_form_p): Ditto.
4413 (pred_expr_equal_p): Ditto.
4414 (pred_neg_p): Ditto.
4415 (simplify_pred): Ditto.
4416 (simplify_preds_2): Ditto.
4417 (simplify_preds_3): Ditto.
4418 (simplify_preds_4): Ditto.
4419 (simplify_preds): Ditto.
4421 (push_to_worklist): Ditto.
4422 (get_pred_info_from_cmp): Ditto.
4423 (is_degenerated_phi): Ditto.
4424 (normalize_one_pred_1): Ditto.
4425 (normalize_one_pred): Ditto.
4426 (normalize_one_pred_chain): Ditto.
4427 (normalize_preds): Ditto.
4428 (normalize_cond_1): Remove function.
4429 (normalize_cond): Ditto.
4430 (is_gcond_subset_of): Ditto.
4431 (is_subset_of_any): Ditto.
4432 (is_or_set_subset_of): Ditto.
4433 (is_and_set_subset_of): Ditto.
4434 (is_norm_cond_subset_of): Ditto.
4435 (pred_chain_length_cmp): Ditto.
4436 (convert_control_dep_chain_into_preds): Type change.
4437 (find_predicates): Ditto.
4438 (find_def_preds): Ditto.
4439 (destroy_predicates_vecs): Ditto.
4440 (find_matching_predicates_in_rest_chains): Ditto.
4441 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4442 (is_pred_expr_subset): Ditto.
4443 (is_pred_chain_subset_of): Ditto.
4444 (is_included_in): Ditto.
4445 (is_superset_of): Ditto.
4447 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4449 Update copyright years.
4451 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4453 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4454 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4455 config/arc/arc.md, config/arc/arc.opt,
4456 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4457 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4458 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4459 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4460 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4461 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4462 vtable-verify.c, vtable-verify.h: Use the standard form for the
4465 2014-01-02 Tobias Burnus <burnus@net-b.de>
4467 * gcc.c (process_command): Update copyright notice dates.
4468 * gcov-dump.c: Ditto.
4470 * doc/cpp.texi: Bump @copying's copyright year.
4471 * doc/cppinternals.texi: Ditto.
4472 * doc/gcc.texi: Ditto.
4473 * doc/gccint.texi: Ditto.
4474 * doc/gcov.texi: Ditto.
4475 * doc/install.texi: Ditto.
4476 * doc/invoke.texi: Ditto.
4478 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4480 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4482 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4484 * config/i386/sse.md (*mov<mode>_internal): Guard
4485 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4487 PR rtl-optimization/59647
4488 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4489 new_rtx into UNSIGNED_FLOAT rtxes.
4491 Copyright (C) 2014 Free Software Foundation, Inc.
4493 Copying and distribution of this file, with or without modification,
4494 are permitted in any medium without royalty provided the copyright
4495 notice and this notice are preserved.