1 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test for Solaris
5 * configure: Regenerate.
7 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
10 (LIB_TLS_SPEC): Save as ld_tls_libs.
11 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
12 (HAVE_AS_IX86_TLSLDM): New test.
13 * configure, config.in: Regenerate.
14 * config/i386/i386.c (legitimize_tls_address): Fall back to
15 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
16 cannot support TLS_MODEL_LOCAL_DYNAMIC.
17 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
18 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
20 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
22 * common.opt (fira-loop-pressure): Mark as optimization.
24 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
26 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
27 an OpenMP mappable type.
29 2014-03-06 Matthias Klose <doko@ubuntu.com>
31 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
32 MULTILIB_OSDIRNAMES is not defined.
34 2014-03-06 Jakub Jelinek <jakub@redhat.com>
35 Meador Inge <meadori@codesourcery.com>
38 * config/arm/arm.c (arm_tls_symbol_p): Remove.
39 (arm_legitimize_address): Call legitimize_tls_address for any
40 arm_tls_referenced_p expression, handle constant addend. Call it
41 before testing for !TARGET_ARM.
42 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
44 2014-03-06 Richard Biener <rguenther@suse.de>
50 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
52 * tree-streamer.c (record_common_node): Assert we don't record
53 nodes with type double.
54 (preload_common_node): Skip type double, complex double and
55 double pointer since it is now frontend dependent due to
58 2014-03-06 Richard Biener <rguenther@suse.de>
60 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
61 or -fno-lto is specified and the linker has full plugin support.
62 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is
64 (main): Remove -flto processing, adjust lto_mode using
66 * lto-wrapper.c (merge_and_complain): Merge compile-time
68 (run_gcc): And pass it through to the link options.
70 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
74 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
76 * cselib.c (remove_useless_values): Skip to avoid quadratic
77 behavior if the condition moved from...
78 (cselib_process_insn): ... here holds.
80 2014-03-05 Jakub Jelinek <jakub@redhat.com>
83 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
84 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
87 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
88 (TM_H): Add x86-tune.def.
90 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
92 * config/aarch64/aarch64.c (generic_tunings):
93 Use cortexa57_extra_costs.
95 2014-03-05 Jakub Jelinek <jakub@redhat.com>
98 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
99 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
100 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
103 2014-03-04 Heiher <r@hev.cc>
105 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
106 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
108 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
110 * config/i386/predicates.md (const2356_operand): Change to ...
111 (const2367_operand): ... this.
112 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
114 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
115 (*avx512pf_scatterpf<mode>sf): Ditto.
116 (avx512pf_scatterpf<mode>df): Ditto.
117 (*avx512pf_scatterpf<mode>df_mask): Ditto.
118 (*avx512pf_scatterpf<mode>df): Ditto.
119 * config/i386/i386.c (ix86_expand_builtin): Update
120 incorrect hint operand error message.
122 2014-03-04 Richard Biener <rguenther@suse.de>
124 * lto-section-in.c (lto_get_section_data): Fix const cast.
126 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
128 * tree-streamer.c (record_common_node): Assert we don't record
129 nodes with type double.
130 (preload_common_node): Skip type double, complex double and
131 double pointer since it is now frontend dependent due to
132 fshort-double option.
134 2014-03-04 Richard Biener <rguenther@suse.de>
137 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
138 (lto_input_toplevel_asms): Likewise.
139 * lto-section-in.c (lto_get_section_data): Instead do it here
142 2014-03-04 Richard Biener <rguenther@suse.de>
144 PR tree-optimization/60382
145 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
146 dead PHIs a reduction.
148 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
150 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
152 (_mm_prefetch): Move out of GCC target("sse") pragma.
153 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
154 GCC target("prfchw") pragma.
155 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
157 * config/i386/i386.c (ix86_option_override_internal): Enable
158 -mprfchw with -mprefetchwt1.
160 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
162 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
165 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
167 * opts.h (CL_PCH_IGNORE): Define.
168 * targhooks.c (option_affects_pch_p):
169 Return false for options that have CL_PCH_IGNORE set.
170 * opt-functions.awk: Process PchIgnore.
171 * doc/options.texi: Document PchIgnore.
173 * config/arc/arc.opt (misize): Add PchIgnore property.
175 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
177 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
178 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
179 constraint on constants to permit them being loaded into
180 GENERAL_REGS or BASE_REGS.
182 2014-03-03 Nick Clifton <nickc@redhat.com>
184 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
185 anti-cacnonical alternatives.
186 (negandhi3_real): New pattern.
187 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
189 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
191 * config/avr/avr-mcus.def: Remove atxmega16x1.
192 * config/avr/avr-tables.opt: Regenerate.
193 * config/avr/t-multilib: Regenerate.
194 * doc/avr-mmcu.texi: Regenerate.
196 2014-03-03 Tobias Grosser <tobias@grosser.es>
197 Mircea Namolaru <mircea.namolaru@inria.fr>
199 PR tree-optimization/58028
200 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
203 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
205 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
206 not handled by recognizers.
208 2014-03-03 Jakub Jelinek <jakub@redhat.com>
211 * function.c (expand_function_end): Don't emit
212 clobber_return_register sequence if clobber_after is a BARRIER.
213 * cfgexpand.c (construct_exit_block): Append instructions before
214 return_label to prev_bb.
216 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
218 * config/rs6000/constraints.md: Document reserved use of "wc".
220 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
223 * ipa.c (function_and_variable_visibility): When dissolving comdat
224 group, also set all symbols to local.
226 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
231 2013-12-14 Jan Hubicka <jh@suse.cz>
233 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
235 2014-03-02 Jon Beniston <jon@beniston.com>
241 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
242 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
243 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
244 (simple_return, *simple_return): New patterns
245 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
246 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
248 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
250 * dwarf2out.c (gen_subprogram_die): Tidy.
252 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
255 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
256 (*mov_t_msb_neg_negc): ... this new insn.
258 2014-02-28 Jason Merrill <jason@redhat.com>
261 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
264 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
267 * dwarf2out.c (decltype_auto_die): New static.
268 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
269 (gen_type_die_with_usage): Handle 'decltype(auto)'.
270 (is_cxx_auto): Likewise.
272 2014-02-28 Ian Bolton <ian.bolton@arm.com>
274 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
275 we are not using general regs only.
277 2014-02-28 Richard Biener <rguenther@suse.de>
280 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
281 previous fix and only allow to remove trivial pre-headers
282 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
283 (remove_forwarder_block): Properly update the latch of a loop.
285 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
288 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
289 (cselib_preserved_hash_table): New.
290 (preserve_constants_and_equivs): Move preserved vals to it.
291 (cselib_find_slot): Look it up first.
292 (cselib_init): Initialize it.
293 (cselib_finish): Release it.
294 (dump_cselib_table): Dump it.
296 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
299 * cselib.c (remove_useless_values): Skip to avoid quadratic
300 behavior if the condition moved from...
301 (cselib_process_insn): ... here holds.
303 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
306 * var-tracking.c (vt_initialize): Apply the same condition to
307 preserve the CFA base value.
309 2014-02-28 Joey Ye <joey.ye@arm.com>
312 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
313 if reload in progress or completed.
315 2014-02-28 Tobias Burnus <burnus@net-b.de>
318 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
321 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
323 * doc/tm.texi.in (Condition Code Status): Update documention for
324 relative locations of cc0-setter and cc0-user.
326 2014-02-27 Jeff Law <law@redhat.com>
328 PR rtl-optimization/52714
329 * combine.c (try_combine): When splitting an unrecognized PARALLEL
330 into two independent simple sets, if I3 is a jump, ensure the
331 pattern we place into I3 is a (set (pc) ...).
333 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
334 Jeff Law <law@redhat.com>
336 PR rtl-optimization/49847
337 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
338 are in different blocks.
339 * doc/tm.texi (Condition Code Status): Update documention for
340 relative locations of cc0-setter and cc0-user.
342 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
345 * lra.c (lra_emit_add): Check SUBREG too.
347 2014-02-27 Andreas Schwab <schwab@suse.de>
349 * config/m68k/m68k.c (m68k_option_override): Disable
350 -flive-range-shrinkage for classic m68k.
351 (m68k_override_options_after_change): Likewise.
353 2014-02-27 Marek Polacek <polacek@redhat.com>
356 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
357 -Wmaybe-uninitialized.
359 2014-02-27 Alan Modra <amodra@gmail.com>
362 * reload1.c (emit_input_reload_insns): When reload_override_in,
363 set old to rl->in_reg when rl->in_reg is a subreg.
365 2014-02-26 Richard Biener <rguenther@suse.de>
368 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
370 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
372 * common/config/i386/predicates.md (const1256_operand): Remove.
373 (const2356_operand): New.
374 (const_1_to_2_operand): Remove.
375 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
376 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
377 (*avx512pf_gatherpf<mode>sf): Ditto.
378 (avx512pf_gatherpf<mode>df): Ditto.
379 (*avx512pf_gatherpf<mode>df_mask): Ditto.
380 (*avx512pf_gatherpf<mode>df): Ditto.
381 (avx512pf_scatterpf<mode>sf): Ditto.
382 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
383 (*avx512pf_scatterpf<mode>sf): Ditto.
384 (avx512pf_scatterpf<mode>df): Ditto.
385 (*avx512pf_scatterpf<mode>df_mask): Ditto.
386 (*avx512pf_scatterpf<mode>df): Ditto.
387 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
389 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
391 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
392 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
393 (_mm512_mask_testn_epi64_mask): Move to ...
394 * config/i386/avx512cdintrin.h: Here.
395 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
396 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
397 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
398 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
399 TARGET_AVX512F from TARGET_AVX512CD.
401 2014-02-26 Richard Biener <rguenther@suse.de>
404 * ipa.c (walk_polymorphic_call_targets): Properly guard
405 call to inline_update_overall_summary.
407 2014-02-26 Bin Cheng <bin.cheng@arm.com>
410 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
411 and latches only if requested. Fix latch if it is removed.
412 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
413 LOOPS_HAVE_PREHEADERS.
415 2014-02-25 Andrew Pinski <apinski@cavium.com>
417 * builtins.c (expand_builtin_thread_pointer): Create a new target
418 when the target is NULL.
420 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
422 PR rtl-optimization/60317
423 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
424 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
425 * lra-assigns.c: Include params.h.
426 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
427 other reload pseudos considerations.
429 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
431 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
432 to use canonical form for nor<mode>3.
434 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
437 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
440 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
442 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
443 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
444 (ix86_handle_option): Handle OPT_mprefetchwt1.
445 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
446 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
448 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
449 OPTION_MASK_ISA_PREFETCHWT1.
450 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
451 (PTA_PREFETCHWT1): New.
452 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
453 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
454 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
455 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
456 (*prefetch_avx512pf_<mode>_: Change into ...
457 (*prefetch_prefetchwt1_<mode>: This.
458 * config/i386/i386.opt (mprefetchwt1): New.
459 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
460 (_mm_prefetch): Handle intent to write.
461 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
463 2014-02-25 Richard Biener <rguenther@suse.de>
466 * emit-rtl.c (mem_attrs_htab): Remove.
467 (mem_attrs_htab_hash): Likewise.
468 (mem_attrs_htab_eq): Likewise.
469 (set_mem_attrs): Always allocate new mem-attrs when something changed.
470 (init_emit_once): Do not allocate mem_attrs_htab.
472 2014-02-25 Richard Biener <rguenther@suse.de>
475 * lto-opts.c (lto_write_options): Output non-explicit conservative
476 -fwrapv, -fno-trapv and -fno-strict-overflow.
477 * lto-wrapper.c (merge_and_complain): Handle merging those options.
478 (run_gcc): And pass them through.
480 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
482 * sel-sched.c (calculate_new_fences): New parameter ptime.
483 Calculate it as a maximum over all fence cycles.
484 (sel_sched_region_2): Adjust the call to calculate_new_fences.
485 Print the final schedule timing when sched_verbose.
487 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
489 PR rtl-optimization/60292
490 * sel-sched.c (fill_vec_av_set): Do not reset target availability
491 bit fot the fence instruction.
493 2014-02-24 Alangi Derick <alangiderick@gmail.com>
495 * calls.h: Fix typo in comment.
497 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
499 * config/pa/pa.c (pa_output_move_double): Don't valididate when
500 adjusting offsetable addresses.
502 2014-02-24 Guozhi Wei <carrot@google.com>
504 * sparseset.h (sparseset_pop): Fix the wrong index.
506 2014-02-24 Walter Lee <walt@tilera.com>
508 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
509 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
511 * common/config/tilegx/tilegx-common.c
512 (TARGET_DEFAULT_TARGET_FLAGS): Define.
513 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
515 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
516 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
517 (tilegx_gimplify_va_arg_expr): Handle big endian.
518 (tilegx_expand_unaligned_load): Ditto.
519 (tilegx_expand_unaligned_store): Ditto.
520 (TARGET_RETURN_IN_MSB): New.
521 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
522 (TARGET_ENDIAN_DEFAULT): New.
523 (TARGET_BIG_ENDIAN): Handle big endian.
524 (BYTES_BIG_ENDIAN): Ditto.
525 (WORDS_BIG_ENDIAN): Ditto.
526 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
529 * config/tilegx/tilegx.md (extv): Handle big endian.
532 (insn_st<n>_add<bitsuffix>): Ditto.
533 (insn_stnt<n>): Ditto.
534 (insn_stnt<n>_add<bitsuffix>):Ditto.
535 (vec_interleave_highv8qi): Handle big endian.
536 (vec_interleave_highv8qi_be): New.
537 (vec_interleave_highv8qi_le): New.
538 (insn_v1int_h): Handle big endian.
539 (vec_interleave_lowv8qi): Handle big endian.
540 (vec_interleave_lowv8qi_be): New.
541 (vec_interleave_lowv8qi_le): New.
542 (insn_v1int_l): Handle big endian.
543 (vec_interleave_highv4hi): Handle big endian.
544 (vec_interleave_highv4hi_be): New.
545 (vec_interleave_highv4hi_le): New.
546 (insn_v2int_h): Handle big endian.
547 (vec_interleave_lowv4hi): Handle big endian.
548 (vec_interleave_lowv4hi_be): New.
549 (vec_interleave_lowv4hi_le): New.
550 (insn_v2int_l): Handle big endian.
551 (vec_interleave_highv2si): Handle big endian.
552 (vec_interleave_highv2si_be): New.
553 (vec_interleave_highv2si_le): New.
554 (insn_v4int_h): Handle big endian.
555 (vec_interleave_lowv2si): Handle big endian.
556 (vec_interleave_lowv2si_be): New.
557 (vec_interleave_lowv2si_le): New.
558 (insn_v4int_l): Handle big endian.
559 * config/tilegx/tilegx.opt (mbig-endian): New option.
560 (mlittle-endian): New option.
561 * doc/install.texi: Document tilegxbe-linux.
562 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
564 2014-02-24 Martin Jambor <mjambor@suse.cz>
567 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
568 there are no parameter descriptors.
570 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
572 PR rtl-optimization/60268
573 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
574 initialization to ...
575 (sched_rgn_init): ... here.
576 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
578 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
580 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
583 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
585 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
588 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
590 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
591 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
593 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
595 * config/microblaze/predicates.md: Add cmp_op predicate.
596 * config/microblaze/microblaze.md: Add branch_compare instruction
597 which uses cmp_op predicate and emits cmp insn before branch.
598 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
599 to microblaze_expand_conditional_branch and consolidate logic.
600 (microblaze_expand_conditional_branch): emit branch_compare
601 insn instead of handling cmp op separate from branch insn.
603 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
605 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
608 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
610 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
611 define_insn with define_expand and new define_insn
612 *altivec_lve<VI_char>x_internal.
613 (altivec_stve<VI_char>x): Replace define_insn with define_expand
614 and new define_insn *altivec_stve<VI_char>x_internal.
615 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
617 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
619 (altivec_expand_stvex_be): New function.
621 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
623 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
624 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
625 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
626 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
628 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
631 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
632 instead of emit_move_insn.
634 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
636 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
638 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
641 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
643 * config/rs6000/altivec.md (altivec_lvxl): Rename as
644 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
645 (altivec_lvxl_<mode>): New define_expand incorporating
646 -maltivec=be semantics where needed.
647 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
648 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
649 semantics where needed.
650 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
651 (altivec_stvx_<mode>): New define_expand incorporating
652 -maltivec=be semantics where needed.
653 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
654 VM2 iterator instead of V4SI.
655 (altivec_stvxl_<mode>): New define_expand incorporating
656 -maltivec=be semantics where needed.
657 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
658 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
659 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
660 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
661 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
662 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
663 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
664 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
665 ALTIVEC_BUILTIN_STVXL.
666 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
667 (altivec_expand_stvx_be): Likewise.
668 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
669 (altivec_expand_lvx_be): Likewise.
670 (altivec_expand_stvx_be): Likewise.
671 (altivec_expand_builtin): Add cases for
672 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
673 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
674 (altivec_init_builtins): Add definitions for
675 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
676 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
678 2014-02-21 Catherine Moore <clm@codesourcery.com>
680 * doc/invoke.texi (mvirt, mno-virt): Document.
681 * config/mips/mips.opt (mvirt): New option.
682 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
684 2014-02-21 Richard Biener <rguenther@suse.de>
686 PR tree-optimization/60276
687 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
688 (STMT_VINFO_MIN_NEG_DIST): New macro.
689 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
690 STMT_VINFO_MIN_NEG_DIST.
691 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
692 made for negative dependence distances still hold.
694 2014-02-21 Richard Biener <rguenther@suse.de>
697 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
698 DECL_INITIAL for globals not in the current function context.
700 2014-02-21 Jakub Jelinek <jakub@redhat.com>
702 PR tree-optimization/56490
703 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
704 * tree-ssa-uninit.c: Include params.h.
705 (compute_control_dep_chain): Add num_calls argument, return false
706 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
707 num_calls to recursive call.
708 (find_predicates): Change dep_chain into normal array,
709 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
710 variable and adjust compute_control_dep_chain caller.
711 (find_def_preds): Likewise.
713 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
715 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
716 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
718 2014-02-21 Nick Clifton <nickc@redhat.com>
720 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
722 (popqi1): Add mode to pre_dec.
725 2014-02-21 Jakub Jelinek <jakub@redhat.com>
727 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
728 mode for mask of V8SFmode permutation.
730 2014-02-20 Richard Henderson <rth@redhat.com>
733 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
734 a new pseudo for OLDVAL.
736 2014-02-20 Jakub Jelinek <jakub@redhat.com>
739 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
740 gen_reg_rtx if d->testing_p.
741 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
742 if d->testing_p and we will certainly return true.
743 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
746 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
748 * emit-rtl.c (gen_reg_rtx): Assert that
749 crtl->emit.regno_pointer_align_length is non-zero.
751 2014-02-20 Richard Henderson <rth@redhat.com>
754 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
755 on failure the store back into EXPECT.
757 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
758 Sandra Loosemore <sandra@codesourcery.com>
760 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
761 * config/nios2/nios2.c (nios2_function_profiler): Add
762 -fPIC (flag_pic == 2) support.
763 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
764 (nios2_large_offset_p): New function.
765 (nios2_unspec_reloc_p): Move up position, update to use
766 nios2_large_offset_p.
767 (nios2_unspec_address): Remove function.
768 (nios2_unspec_offset): New function.
769 (nios2_large_got_address): New function.
770 (nios2_got_address): Add large offset support.
771 (nios2_legitimize_tls_address): Update usage of removed and new
773 (nios2_symbol_binds_local_p): New function.
774 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
775 (nios2_legitimize_address): Update to use nios2_large_offset_p.
776 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
777 (nios2_print_operand): Merge H/L processing, add hiadj/lo
778 processing for (const (unspec ...)).
779 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
781 2014-02-20 Richard Biener <rguenther@suse.de>
783 * tree-cfg.c (replace_uses_by): Mark altered BBs before
784 doing the substitution.
785 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
787 2014-02-20 Martin Jambor <mjambor@suse.cz>
790 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
791 info when checking whether lattices are bottom.
793 2014-02-20 Richard Biener <rguenther@suse.de>
796 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
799 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
802 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
803 parameter specifying the scaling.
804 (inline_call): Update.
805 (want_inline_recursively): Guard division by zero.
806 (recursive_inlining): Update.
807 * ipa-inline.h (clone_inlined_nodes): Update.
809 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
812 * config/i386/i386.c (classify_argument): Pass structures of size
813 64 bytes or less in register.
815 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
816 Kirill Yukhin <kirill.yukhin@intel.com>
818 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
819 (_mm_rcp28_round_ss): Ditto.
820 (_mm_rsqrt28_round_sd): Ditto.
821 (_mm_rsqrt28_round_ss): Ditto.
822 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
823 (_mm_rcp14_round_ss): Ditto.
824 (_mm_rsqrt14_round_sd): Ditto.
825 (_mm_rsqrt14_round_ss): Ditto.
826 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
827 the first input operand, get rid of match_dup.
828 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
830 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
832 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
833 operand as the first input operand, set type attribute.
834 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
836 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
837 operand as the first input operand, set type attribute.
839 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
841 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
844 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
847 * config/i386/i386.c (construct_container): Remove TFmode check
848 for X86_64_INTEGER_CLASS.
850 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
853 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
854 only when -Wpsabi is enabled.
856 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
859 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
860 passing arrays in registers are the same as for structs, so remove the
861 special case for them.
863 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
865 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
866 destination type, extract only the valid bits if the source type is not
867 integral and has a different mode.
869 2014-02-19 Richard Biener <rguenther@suse.de>
872 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
875 2014-02-19 Richard Biener <rguenther@suse.de>
878 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
879 (ipa_modify_call_arguments): Emit an argument load explicitely and
880 preserve virtual SSA form there and for the replacement call.
881 Do not update SSA form nor free dominance info.
883 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
885 * ipa.c (function_and_variable_visibility): Also clear WEAK
886 flag when disolving COMDAT_GROUP.
888 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
890 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
891 * ipa-prop.c (ipa_set_jf_known_type): Return early when
893 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
894 do more sanity checks.
895 (detect_type_change): Return true when giving up early.
896 (compute_complex_assign_jump_func): Fix type parameter of
898 (compute_complex_ancestor_jump_func): Likewise.
899 (update_jump_functions_after_inlining): Fix updating of
901 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
903 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
905 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
906 inline clones when edge disappears.
908 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
911 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
912 Split 64-bit moves into 2 patterns. Do not allow the use of
913 direct move for TDmode in little endian, since the decimal value
914 has little endian bytes within a word, but the 64-bit pieces are
915 ordered in a big endian fashion, and normal subreg's of TDmode are
917 (mov<mode>_64bit_dm): Likewise.
918 (movtd_64bit_nodm): Likewise.
920 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
922 PR tree-optimization/60174
923 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
924 statement of an SSA_NAME that occurs in an abnormal PHI node.
926 2014-02-18 Jakub Jelinek <jakub@redhat.com>
929 * final.c (SEEN_BB): Remove.
930 (SEEN_NOTE, SEEN_EMITTED): Renumber.
931 (final_scan_insn): Don't force_source_line on second
932 NOTE_INSN_BASIC_BLOCK.
934 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
937 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
938 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
939 (type_natural_mode): Warn ABI change when %zmm register is not
940 available for AVX512F vector value passing.
942 2014-02-18 Kai Tietz <ktietz@redhat.com>
945 * config/i386/i386.c (ix86_expand_prologue): Use value in
946 rax register as displacement when restoring %r10 or %rax.
947 Fix wrong offset when restoring both registers.
949 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
951 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
952 assertion with conditional return.
954 2014-02-18 Jakub Jelinek <jakub@redhat.com>
955 Uros Bizjak <ubizjak@gmail.com>
958 * config/i386/driver-i386.c (host_detect_local_cpu): If
959 YMM state is not saved by the OS, also clear has_f16c. Move
960 CPUID 0x80000001 handling before YMM state saving checking.
962 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
964 PR rtl-optimization/58960
965 * haifa-sched.c (alloc_global_sched_pressure_data): New,
966 factored out from ...
967 (sched_init): ... here.
968 (free_global_sched_pressure_data): New, factored out from ...
969 (sched_finish): ... here.
970 * sched-int.h (free_global_sched_pressure_data): Declare.
971 * sched-rgn.c (nr_regions_initial): New static global.
972 (haifa_find_rgns): Initialize it.
973 (schedule_region): Disable sched-pressure for the newly
976 2014-02-17 Richard Biener <rguenther@suse.de>
978 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
979 release SSA defs of pattern stmts.
981 2014-02-17 Richard Biener <rguenther@suse.de>
983 * tree-inline.c (expand_call_inline): Release the virtual
984 operand defined by the call we are about to inline.
986 2014-02-17 Richard Biener <rguenther@suse.de>
988 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
990 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
991 Ilya Tocar <ilya.tocar@intel.com>
993 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
994 arguments order in builtin.
995 (_mm512_permutexvar_epi64): Ditto.
996 (_mm512_mask_permutexvar_epi64): Ditto
997 (_mm512_maskz_permutexvar_epi32): Ditto
998 (_mm512_permutexvar_epi32): Ditto
999 (_mm512_mask_permutexvar_epi32): Ditto
1001 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1003 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1004 (p8_vmrgow): Likewise.
1006 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1008 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1011 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1014 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1015 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1016 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1017 using direct move instructions on ISA 2.07. Also adjust
1018 instruction length for 64-bit.
1019 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1020 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1022 2014-02-15 Alan Modra <amodra@gmail.com>
1026 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1027 find_replacement on parts of insn rtl that might be reloaded.
1029 2014-02-15 Richard Biener <rguenther@suse.de>
1031 PR tree-optimization/60183
1032 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1033 (tree_ssa_phiprop): Calculate and free post-dominators.
1035 2014-02-14 Jeff Law <law@redhat.com>
1037 PR rtl-optimization/60131
1038 * ree.c (get_extended_src_reg): New function.
1039 (combine_reaching_defs): Use it rather than assuming location of REG.
1040 (find_and_remove_re): Verify first operand of extension is
1041 a REG before adding the insns to the copy list.
1043 2014-02-14 Roland McGrath <mcgrathr@google.com>
1045 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1046 * configure: Regenerated.
1047 * config.in: Regenerated.
1048 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1049 instead of ASM_SHORT.
1051 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1052 Richard Earnshaw <rearnsha@arm.com>
1054 PR rtl-optimization/59535
1055 * lra-constraints.c (process_alt_operands): Encourage alternative
1056 when unassigned pseudo class is superset of the alternative class.
1057 (inherit_reload_reg): Don't inherit when optimizing for code size.
1058 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1059 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1060 modes not less than 4 for Thumb1.
1062 2014-02-14 Kyle McMartin <kyle@redhat.com>
1065 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1067 2014-02-14 Richard Biener <rguenther@suse.de>
1069 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1070 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1071 Do not strip INDIRECT_REFs.
1073 2014-02-14 Richard Biener <rguenther@suse.de>
1076 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1077 DECL_FUNCTION_SPECIFIC_TARGET.
1078 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1079 * tree-streamer-out.c (pack_ts_target_option): Remove.
1080 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1081 (write_ts_function_decl_tree_pointers): Do not stream
1082 DECL_FUNCTION_SPECIFIC_TARGET.
1083 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1084 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1085 (lto_input_ts_function_decl_tree_pointers): Do not stream
1086 DECL_FUNCTION_SPECIFIC_TARGET.
1088 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1090 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1091 (get_initial_def_for_induction, vectorizable_induction): Ignore
1092 debug stmts when looking for exit_phi.
1093 (vectorizable_live_operation): Fix up condition.
1095 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1097 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1098 nreverse() because it changes the content of original tree list.
1100 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1102 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1103 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1105 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1107 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1108 GNU coding standards.
1110 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1113 * dwarf2out.c (gen_subprogram_die): Don't call
1114 add_calling_convention_attribute if subr_die is old_die.
1116 2014-02-13 Sharad Singhai <singhai@google.com>
1118 * doc/optinfo.texi: Fix order of nodes.
1120 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1122 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1123 operands[2], not operands[3].
1125 2014-02-13 Richard Biener <rguenther@suse.de>
1128 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1129 mention the possibility of an in-tree build.
1130 (CLooG): Update recommended version to 0.18.1, mention the
1131 possibility of an in-tree build and clarify that the ISL
1132 bundled with CLooG does not work.
1134 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1137 * expr.c (compress_float_constant): If x is a hard register,
1138 extend into a pseudo and then move to x.
1140 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1142 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1143 caused by bad second argument to warning_at() with -mhotpatch and
1144 nested functions (e.g. with gfortran).
1146 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1148 * opts.c (option_name): Remove "enabled by default" rider.
1150 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1152 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1154 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1155 Uros Bizjak <ubizjak@gmail.com>
1158 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1159 * configure: Regenerated.
1161 2014-02-12 Richard Biener <rguenther@suse.de>
1163 * vec.c (vec_prefix::calculate_allocation): Move as
1164 inline variant to vec.h.
1165 (vec_prefix::calculate_allocation_1): New out-of-line version.
1166 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1167 (vec_prefix::m_has_auto_buf): Rename to ...
1168 (vec_prefix::m_using_auto_storage): ... this.
1169 (vec_prefix::calculate_allocation): Inline the easy cases
1170 and dispatch to calculate_allocation_1 which doesn't need the
1172 (va_heap::reserve): Use gcc_checking_assert.
1173 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1174 m_using_auto_storage.
1175 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1177 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1178 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1179 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1181 2014-02-12 Richard Biener <rguenther@suse.de>
1183 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1184 when we found a dependence.
1186 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1188 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1190 (maybe_fold_stmt): ... into this new function.
1191 * omp-low.c (lower_omp): Update comment.
1193 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1196 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1199 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1201 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1202 identifiers in comments.
1203 (cortexa53_extra_costs): Likewise.
1204 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1205 (cortexa7_extra_costs): Likewise.
1206 (cortexa12_extra_costs): Likewise.
1207 (cortexa15_extra_costs): Likewise.
1208 (v7m_extra_costs): Likewise.
1210 2014-02-12 Richard Biener <rguenther@suse.de>
1213 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1214 of posix_memalign being successful.
1215 (lower_stmt): Restrict lowering of posix_memalign to when
1216 -ftree-bit-ccp is enabled.
1218 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1220 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1222 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1224 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1226 PR rtl-optimization/60116
1227 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1228 other_insn once the combination has been validated.
1230 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1233 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1235 * ipa-devirt.c: Include demangle.h
1236 (odr_violation_reported): New static variable.
1237 (add_type_duplicate): Update odr_violations.
1238 (maybe_record_node): Add completep parameter; update it.
1239 (record_target_from_binfo): Add COMPLETEP parameter;
1240 update it as needed.
1241 (possible_polymorphic_call_targets_1): Likewise.
1242 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1243 rename FINAL to COMPLETE.
1244 (record_targets_from_bases): Sanity check we found the binfo;
1245 fix COMPLETEP updating.
1246 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1247 parameter, fix computing of COMPLETEP.
1248 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1249 at LTO time do demangling.
1250 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1251 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1253 (gimple_get_virt_method_for_binfo): Likewise.
1254 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1255 gimple_get_virt_method_for_vtable): Update prototypes.
1257 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1260 * genautomata.c (add_presence_absence): Fix typo with
1261 {final_}presence_list.
1263 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1266 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1267 for VSX/Altivec vectors that land in GPR registers.
1269 2014-02-11 Richard Henderson <rth@redhat.com>
1270 Jakub Jelinek <jakub@redhat.com>
1273 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
1274 around drhs if type conversion to lacc->type is not useless.
1276 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1278 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
1280 (cortex-a57.cortex-a53): Likewise.
1281 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
1283 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1285 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
1288 2014-02-11 Renlin Li <Renlin.Li@arm.com>
1290 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
1291 add_options_for_arm_vfp3.
1293 2014-02-11 Jeff Law <law@redhat.com>
1296 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
1297 object with an undesirable mode.
1299 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1302 * config/i386/sol2-9.h: New file.
1303 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
1304 *-*-solaris2.9*): Use it.
1306 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1308 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
1309 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
1311 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
1313 * config/microblaze/microblaze.c: Extend mcpu version format
1315 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
1317 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
1319 2014-02-10 Richard Henderson <rth@redhat.com>
1322 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
1323 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
1324 ms-abi vs -mno-accumulate-outgoing-args.
1325 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
1326 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
1329 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
1332 * cfgexpand.c (expand_asm_operands): Attach source location to
1333 ASM_INPUT rtx objects.
1334 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
1336 2014-02-10 Nick Clifton <nickc@redhat.com>
1338 * config/mn10300/mn10300.c (popcount): New function.
1339 (mn10300_expand_prologue): Include saved registers in stack usage
1342 2014-02-10 Jeff Law <law@redhat.com>
1345 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
1346 when changing the SET_DEST of a prior insn to avoid an input reload.
1348 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
1350 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
1351 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
1352 -mcall-openbsd, or -mcall-linux.
1353 (CC1_ENDIAN_BIG_SPEC): Remove.
1354 (CC1_ENDIAN_LITTLE_SPEC): Remove.
1355 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1356 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
1357 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
1358 and %cc1_endian_default.
1359 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
1361 2014-02-10 Richard Biener <rguenther@suse.de>
1363 PR tree-optimization/60115
1364 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
1365 MEM_REF handling. Properly verify that the accesses are not
1366 out of the objects bound.
1368 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1370 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
1373 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
1375 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
1376 proper constants and fix formatting.
1377 (possible_polymorphic_call_targets): Fix formatting.
1379 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
1380 Ilya Tocar <ilya.tocar@intel.com>
1382 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
1383 (_mm512_loadu_epi32): Renamed into...
1384 (_mm512_loadu_si512): This.
1385 (_mm512_storeu_epi32): Renamed into...
1386 (_mm512_storeu_si512): This.
1387 (_mm512_maskz_ceil_ps): Removed.
1388 (_mm512_maskz_ceil_pd): Ditto.
1389 (_mm512_maskz_floor_ps): Ditto.
1390 (_mm512_maskz_floor_pd): Ditto.
1391 (_mm512_floor_round_ps): Ditto.
1392 (_mm512_floor_round_pd): Ditto.
1393 (_mm512_ceil_round_ps): Ditto.
1394 (_mm512_ceil_round_pd): Ditto.
1395 (_mm512_mask_floor_round_ps): Ditto.
1396 (_mm512_mask_floor_round_pd): Ditto.
1397 (_mm512_mask_ceil_round_ps): Ditto.
1398 (_mm512_mask_ceil_round_pd): Ditto.
1399 (_mm512_maskz_floor_round_ps): Ditto.
1400 (_mm512_maskz_floor_round_pd): Ditto.
1401 (_mm512_maskz_ceil_round_ps): Ditto.
1402 (_mm512_maskz_ceil_round_pd): Ditto.
1403 (_mm512_expand_pd): Ditto.
1404 (_mm512_expand_ps): Ditto.
1405 * config/i386/i386.c (ix86_builtins): Remove
1406 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
1407 (bdesc_args): Ditto.
1408 * config/i386/predicates.md (const1256_operand): New.
1409 (const_1_to_2_operand): Ditto.
1410 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1411 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1412 (*avx512pf_gatherpf<mode>sf): Ditto.
1413 (avx512pf_gatherpf<mode>df): Ditto.
1414 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1415 (*avx512pf_gatherpf<mode>df): Ditto.
1416 (avx512pf_scatterpf<mode>sf): Ditto.
1417 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1418 (*avx512pf_scatterpf<mode>sf): Ditto.
1419 (avx512pf_scatterpf<mode>df): Ditto.
1420 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1421 (*avx512pf_scatterpf<mode>df): Ditto.
1422 (avx512f_expand<mode>): Removed.
1423 (<shift_insn><mode>3<mask_name>): Change predicate type.
1425 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1427 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
1428 not at the end of datarefs vector use ordered_remove to avoid
1429 reordering datarefs vector.
1432 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
1433 mark local addressable non-static vars as GOVD_PRIVATE
1434 instead of GOVD_LOCAL.
1435 * omp-low.c (lower_omp_for): Move gimple_bind_vars
1436 and BLOCK_VARS of gimple_bind_block to new_stmt rather
1440 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
1441 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
1442 assume_aligned or alloc_align attributes.
1443 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
1444 arguments. Handle also assume_aligned and alloc_align attributes.
1445 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
1446 calls to functions with assume_aligned or alloc_align attributes.
1447 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
1449 2014-02-08 Terry Guo <terry.guo@arm.com>
1451 * doc/invoke.texi: Document ARM -march=armv7e-m.
1453 2014-02-08 Jakub Jelinek <jakub@redhat.com>
1455 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
1456 flag on __cilkrts_rethrow builtin.
1459 * ipa-cp.c (determine_versionability): Fail at -O0
1460 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
1461 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
1464 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1467 * tree-inline.c (copy_forbidden): Fail for
1468 __attribute__((optimize (0))) functions.
1470 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1472 * varpool.c: Include pointer-set.h.
1473 (varpool_remove_unreferenced_decls): Variables in other partitions
1474 will not be output; be however careful to not lose information
1477 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
1479 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
1480 lookup in the vtable constructor.
1482 2014-02-07 Jeff Law <law@redhat.com>
1485 * config/m68k/m68k.md (ashldi_extsi): Turn into a
1486 define_insn_and_split.
1488 * ipa-inline.c (inline_small_functions): Fix typos.
1490 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1492 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
1493 (s390_can_use_return_insn): Declare.
1494 * config/s390/s390.h (EPILOGUE_USES): Define.
1495 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
1497 (s390_chunkify_start): Handle return JUMP_LABELs.
1498 (s390_early_mach): Emit a main_pool instruction on the entry edge.
1499 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
1500 (s390_can_use_return_insn): New functions.
1501 (s390_fix_long_loop_prediction): Handle conditional returns.
1502 (TARGET_SET_UP_BY_PROLOGUE): Define.
1503 * config/s390/s390.md (ANY_RETURN): New code iterator.
1504 (*creturn, *csimple_return, return, simple_return): New patterns.
1506 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1508 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
1509 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
1510 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
1511 REG_CFA_RESTORE list when deciding not to restore a register.
1513 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1515 * config/s390/s390.c: Include tree-pass.h and context.h.
1516 (s390_early_mach): New function, split out from...
1517 (s390_emit_prologue): ...here.
1518 (pass_data_s390_early_mach): New pass structure.
1519 (pass_s390_early_mach): New class.
1520 (s390_option_override): Create and register early_mach pass.
1521 Move to end of file.
1523 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
1525 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
1526 to match for the exit block.
1528 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1530 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
1531 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
1532 Reject misaligned operands.
1534 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1536 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
1538 2014-02-07 Richard Biener <rguenther@suse.de>
1541 * gimple-low.c (lower_builtin_posix_memalign): New function.
1542 (lower_stmt): Call it to lower posix_memalign in a way
1543 to make alignment info accessible.
1545 2014-02-07 Jakub Jelinek <jakub@redhat.com>
1548 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
1549 __builtin_setjmp_receiver.
1551 2014-02-07 Richard Biener <rguenther@suse.de>
1554 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
1555 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
1556 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
1557 Handle BUILT_IN_POSIX_MEMALIGN.
1558 (find_func_clobbers): Likewise.
1559 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1560 (call_may_clobber_ref_p_1): Likewise.
1562 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1565 * ipa-devirt.c (record_target_from_binfo): Remove overactive
1568 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1571 * lto-cgraph.c (lto_output_node): Use
1572 symtab_get_symbol_partitioning_class.
1573 (lto_output_varpool_node): likewise.
1574 (symtab_get_symbol_partitioning_class): Move here from
1576 * cgraph.h (symbol_partitioning_class): Likewise.
1577 (symtab_get_symbol_partitioning_class): Declare.
1579 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1581 * ggc.h (ggc_internal_cleared_alloc): New macro.
1582 * vec.h (vec_safe_copy): Handle memory stats.
1583 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
1584 * target-globals.c (save_target_globals): Likewise.
1586 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
1589 * expr.c (emit_move_resolve_push): Export; be bit more selective
1590 on when to clear alias set.
1591 * expr.h (emit_move_resolve_push): Declare.
1592 * function.h (struct function): Add tail_call_marked.
1593 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
1594 * config/i386/i386-protos.h (ix86_expand_push): Remove.
1595 * config/i386/i386.md (TImode move expander): De not call
1597 (FP push expanders): Preserve memory attributes.
1598 * config/i386/sse.md (push<mode>1): Remove.
1599 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
1600 (ix86_expand_push): Remove.
1601 * config/i386/mmx.md (push<mode>1): Remove.
1603 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1605 PR rtl-optimization/60030
1606 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
1607 lopart with paradoxical subreg before shifting it up by hprec.
1609 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1611 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
1612 Remove extra newline at end of file.
1613 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
1614 (arm_issue_rate): Handle cortexa57.
1615 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
1616 (cortex-a57.cortex-a53): Likewise.
1618 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1621 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
1622 don't record in REG_FRAME_RELATED_EXPR registers not set in that
1624 (arm_expand_prologue): Adjust all callers.
1625 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
1626 info, registers also at the lowest numbered registers side. Use
1627 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
1631 * var-tracking.c (adjust_mems): Before adding a SET to
1632 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
1634 2014-02-06 Alan Modra <amodra@gmail.com>
1637 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
1638 change SDmode to DDmode when lra_in_progress.
1640 2014-02-06 Jakub Jelinek <jakub@redhat.com>
1643 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
1644 free_data_ref on the dr first, and before goto again also set dr
1645 to the next dr. For simd_lane_access, free old datarefs[i] before
1646 overwriting it. For get_vectype_for_scalar_type failure, don't
1647 free_data_ref if simd_lane_access.
1649 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
1652 * tree.h (opts_for_fn): New inline function.
1653 (opt_for_fn): Define.
1654 * config/i386/i386.c (ix86_function_regparm): Use
1655 opt_for_fn (decl, optimize) instead of optimize.
1657 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
1659 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
1660 for SYMBOL_REF in large memory model.
1662 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1664 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
1666 (cortex-a57): Likewise.
1667 (cortex-a57.cortex-a53): Likewise.
1669 2014-02-06 Yury Gribov <y.gribov@samsung.com>
1670 Kugan Vivekanandarajah <kuganv@linaro.org>
1672 * config/arm/arm.c (arm_vector_alignment_reachable): Check
1674 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
1676 2014-02-06 Richard Biener <rguenther@suse.de>
1678 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
1679 set_loop_copy and initialize_original_copy_tables.
1681 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
1683 * config/aarch64/aarch64-simd.md
1684 (aarch64_ashr_simddi): Change QI to SI.
1686 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1687 Jakub Jelinek <jakub@redhat.com>
1690 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
1693 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1695 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
1696 CODE_FOR_altivec_vpku[hw]um to
1697 CODE_FOR_altivec_vpku[hw]um_direct.
1698 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
1699 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
1700 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
1701 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
1703 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1705 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
1706 generation for -maltivec=be.
1707 (altivec_vsumsws): Simplify redundant test.
1709 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1711 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
1712 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
1713 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
1714 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
1715 gen_altivec_vpkuwum.
1716 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
1718 (altivec_vpks<VI_char>ss): Likewise.
1719 (altivec_vpks<VI_char>us): Likewise.
1720 (altivec_vpku<VI_char>us): Likewise.
1721 (altivec_vpku<VI_char>um): Likewise.
1722 (altivec_vpku<VI_char>um_direct): New (copy of
1723 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
1725 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
1726 target is little endian and -maltivec=be is not specified.
1727 (*altivec_vupkhs<VU_char>_direct): New (copy of
1728 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
1729 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
1730 target is little endian and -maltivec=be is not specified.
1731 (*altivec_vupkls<VU_char>_direct): New (copy of
1732 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
1733 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
1734 little endian and -maltivec=be is not specified.
1735 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
1736 little endian and -maltivec=be is not specified.
1738 2014-02-05 Richard Henderson <rth@redhat.com>
1741 * combine-stack-adj.c: Revert r206943.
1742 * sched-int.h (struct deps_desc): Add last_args_size.
1743 * sched-deps.c (init_deps): Initialize it.
1744 (sched_analyze_insn): Add OUTPUT dependencies between insns that
1745 contain REG_ARGS_SIZE notes.
1747 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
1749 * lto-cgraph.c (asm_nodes_output): Make global.
1750 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
1751 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
1752 (driver_handle_option): Handle OPT_fwpa.
1754 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1757 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1758 a comment typo and formatting issue. If odr_hash hasn't been
1759 created, return vNULL and set *completep to false.
1762 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1763 bb with no successors.
1765 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1768 * doc/invoke.texi (-march): Clarify documentation for ARM.
1772 2014-02-05 Richard Biener <rguenther@suse.de>
1774 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1775 when not vectorizing because of too many alias checks.
1776 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1777 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1779 2014-02-05 Nick Clifton <nickc@redhat.com>
1781 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1782 accept extended registers in any mode when compiling for the MN10300.
1784 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1786 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1787 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1788 sanitization attributes.
1789 (can_inline_edge_p): Likewise.
1790 (sanitize_attrs_match_for_inline_p): New function.
1792 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1794 * ipa-prop.c (detect_type_change): Shor circuit testing of
1795 type changes on THIS pointer.
1797 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1800 * config/pa/pa.c (legitimize_tls_address): Return original address
1801 if not passed a SYMBOL_REF rtx.
1802 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1804 (pa_emit_move_sequence): Simplify TLS source operands.
1805 (pa_legitimate_constant_p): Reject all TLS constants.
1806 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1807 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1809 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1811 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1812 groups when we know they are controlled by LTO.
1813 * varasm.c (default_binds_local_p_1): If object is in other partition,
1814 it will be resolved locally.
1816 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1818 * config/host-linux.c (linux_gt_pch_use_address): Don't
1819 use SSIZE_MAX because it is not always defined.
1821 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1824 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1825 threshold for pseudo splitting.
1826 (update_ebb_live_info): Process call argument hard registers and
1827 hard registers from insn definition too.
1828 (max_small_class_regs_num): New constant.
1829 (inherit_in_ebb): Update live hard regs through EBBs. Update
1830 reloads_num only for small register classes. Don't split for
1833 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1836 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1839 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1841 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1844 2014-02-04 Marek Polacek <polacek@redhat.com>
1846 * gdbinit.in (pel): Define.
1848 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1850 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1853 2014-02-04 Richard Biener <rguenther@suse.de>
1856 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1857 in function context local.
1858 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1859 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1860 similar to LTO_imported_decl_ref.
1862 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1864 PR tree-optimization/60002
1865 * cgraphclones.c (build_function_decl_skip_args): Clear
1868 PR tree-optimization/60023
1869 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1870 false to gsi_replace.
1871 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1872 has been in some EH region and vec_stmt could throw, add
1873 vec_stmt into the same EH region.
1874 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1875 has no lhs, ignore it.
1876 * internal-fn.c (expand_MASK_LOAD): Likewise.
1879 * tree-inline.c (copy_forbidden): Fail for
1880 __attribute__((optimize (0))) functions.
1883 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1884 is set, copy one less argument.
1885 (expand_simd_clones): Don't subtract clone_info->inbranch
1886 from simd_clone_struct_alloc argument.
1888 PR rtl-optimization/57915
1889 * recog.c (simplify_while_replacing): If all unary/binary/relational
1890 operation arguments are constant, attempt to simplify those.
1893 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1894 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1896 2014-02-04 Richard Biener <rguenther@suse.de>
1898 PR tree-optimization/60012
1899 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1900 TBAA disambiguation to all DDRs.
1902 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1905 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1906 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1908 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1911 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1913 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1915 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1916 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1918 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1921 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1922 to figure out targets of polymorphic calls with known decl.
1923 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1924 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1925 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1926 (get_polymorphic_call_info): ... here.
1927 (get_polymorphic_call_info_from_invariant): New function.
1929 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1931 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1932 lookup via vtable pointer; check for type consistency
1933 and turn inconsitent facts into UNREACHABLE.
1934 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1935 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1936 type inconsistent querries; return UNREACHABLE instead.
1938 2014-02-03 Richard Henderson <rth@twiddle.net>
1941 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1942 already processed this node.
1943 (normalize_one_pred_1): Pass along mark_set.
1944 (normalize_one_pred): Create and destroy a pointer_set_t.
1945 (normalize_one_pred_chain): Likewise.
1947 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1949 PR gcov-profile/58602
1950 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1952 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1955 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1956 -fno-devirtualize; try to devirtualize by the knowledge of
1957 virtual table pointer given by aggregate propagation.
1958 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1959 (ipa_print_node_jump_functions): Dump also offset that
1960 is relevant for polymorphic calls.
1961 (determine_known_aggregate_parts): Add arg_type parameter; use it
1962 instead of determining the type from pointer type.
1963 (ipa_compute_jump_functions_for_edge): Update call of
1964 determine_known_aggregate_parts.
1965 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1966 (gimple_get_virt_method_for_binfo): ... here; simplify using
1967 vtable_pointer_value_to_vtable.
1968 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1969 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1970 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1971 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1973 (vtable_pointer_value_to_binfo): ... here.
1974 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1976 2014-02-03 Teresa Johnson <tejohnson@google.com>
1978 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1979 redef of outer loop index variable.
1981 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1985 * doc/extend.texi (Function Attributes): Typo.
1987 2014-02-03 Cong Hou <congh@google.com>
1989 PR tree-optimization/60000
1990 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1991 if the vectorized statement is a store. A store statement can only
1992 appear at the end of pattern statements.
1994 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1996 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1997 (ix86_option_override_internal): Default long double to 64-bit for
1998 32-bit Bionic and to 128-bit for 64-bit Bionic.
2000 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2001 TARGET_LONG_DOUBLE_128 is true.
2002 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2004 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2005 (mlong-double-64): Negate -mlong-double-128.
2006 (mlong-double-128): New option.
2008 * config/i386/i386-c.c (ix86_target_macros): Define
2009 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2011 * doc/invoke.texi: Document -mlong-double-128.
2013 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2015 PR rtl-optimization/60024
2016 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2018 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2020 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2022 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2024 PR rtl-optimization/57662
2025 * sel-sched.c (code_motion_path_driver): Do not mark already not
2026 existing blocks in the visiting bitmap.
2028 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2030 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2031 on the insn being emitted.
2033 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2034 Will Deacon <will.deacon@arm.com>
2036 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2038 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2040 * config/arm/arm-tables.opt: Regenerate.
2042 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2044 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2045 for vector types other than V16QImode.
2046 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2047 define_expand, and call altivec_expand_vec_perm_le when producing
2048 code with little endian element order.
2049 (*altivec_vperm_<mode>_internal): New insn having previous
2050 behavior of altivec_vperm_<mode>.
2051 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2052 altivec_expand_vec_perm_le when producing code with little endian
2054 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2055 behavior of altivec_vperm_<mode>_uns.
2057 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2059 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2060 (altivec_vsumsws): Add handling for -maltivec=be with a little
2062 (altivec_vsumsws_direct): New.
2063 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2064 gen_altivec_vsumsws.
2066 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2068 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2069 vtable_pointer_value_to_binfo): New functions.
2070 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2071 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2073 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2075 * config/nios2/nios2.md (load_got_register): Initialize GOT
2076 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2077 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2079 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2081 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2082 preserverd by passthrough, do not propagate the type.
2084 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2086 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2087 (mips_atomic_assign_expand_fenv): New function.
2088 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2090 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2092 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2093 (__builtin_mips_set_fcsr): Likewise.
2094 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2095 MIPS_USI_FTYPE_VOID.
2096 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2097 (mips16_expand_set_fcsr): Likewise.
2098 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2099 (mips16_set_fcsr_stub): Likewise.
2100 (mips16_get_fcsr_one_only_stub): New class.
2101 (mips16_set_fcsr_one_only_stub): Likewise.
2102 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2103 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2104 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2105 (hard_float): New availability predicate.
2106 (mips_builtins): Add get_fcsr and set_fcsr.
2107 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2108 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2109 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2110 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2111 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2114 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2116 * config/mips/mips.c (mips_one_only_stub): New class.
2117 (mips_need_mips16_rdhwr_p): Replace with...
2118 (mips16_rdhwr_stub): ...this new variable.
2119 (mips16_stub_call_address): New function.
2120 (mips16_rdhwr_one_only_stub): New class.
2121 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2122 (mips_output_mips16_rdhwr): Delete.
2123 (mips_finish_stub): New function.
2124 (mips_code_end): Use it to handle rdhwr stubs.
2126 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2129 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2130 when calculating size of integer atomic types.
2132 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2134 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2136 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2138 PR tree-optimization/60003
2139 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2140 * profile.c (branch_prob): Use gimple_call_builtin_p
2141 to check for BUILT_IN_SETJMP_RECEIVER.
2142 * tree-inline.c (copy_bb): Call notice_special_calls.
2144 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2147 * lra-constraints.c (process_alt_operands): Update reload_sum only
2150 2014-01-31 Richard Henderson <rth@redhat.com>
2153 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2154 until after else_eh is processed.
2156 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2158 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2159 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2160 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2161 in smmintrin.h, remove them.
2162 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2163 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2164 * config/i386/i386.md (ROUND_SAE): Fix value.
2165 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2166 (const48_operand): New.
2167 * config/i386/subst.md (round), (round_expand): Use
2168 const_4_or_8_to_11_operand.
2169 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2171 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2173 * config/i386/constraints.md (Yk): Swap meaning with k.
2174 * config/i386/i386.md (movhi_internal): Change Yk to k.
2175 (movqi_internal): Ditto.
2176 (*k<logic><mode>): Ditto.
2179 (kandn<mode>): Ditto.
2180 (*<code>hi_1): Ditto.
2181 (*<code>qi_1): Ditto.
2182 (kxnor<mode>): Ditto.
2183 (kortestzhi): Ditto.
2184 (kortestchi): Ditto.
2186 (*one_cmplhi2_1): Ditto.
2187 (*one_cmplqi2_1): Ditto.
2188 * config/i386/sse.md (): Change k to Yk.
2189 (avx512f_load<mode>_mask): Ditto.
2190 (avx512f_blendm<mode>): Ditto.
2191 (avx512f_store<mode>_mask): Ditto.
2192 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2193 (avx512f_storedqu<mode>_mask): Ditto.
2194 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2196 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2197 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2198 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2199 (avx512f_maskcmp<mode>3): Ditto.
2200 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2201 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2202 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2203 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2204 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2205 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2206 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2207 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2208 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2209 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2210 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2211 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2212 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2213 (vec_extract_lo_<mode>_maskm): Ditto.
2214 (vec_extract_hi_<mode>_maskm): Ditto.
2215 (avx512f_vternlog<mode>_mask): Ditto.
2216 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2217 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2218 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2219 (avx512f_<code>v8div16qi2_mask): Ditto.
2220 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2221 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2222 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2223 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2224 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2225 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2226 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2227 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2228 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2229 (avx512cd_maskb_vec_dupv8di): Ditto.
2230 (avx512cd_maskw_vec_dupv16si): Ditto.
2231 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2232 (avx512f_vpermi2var<mode>3_mask): Ditto.
2233 (avx512f_vpermi2var<mode>3_mask): Ditto.
2234 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2235 (*avx512f_gathersi<mode>): Ditto.
2236 (*avx512f_gathersi<mode>_2): Ditto.
2237 (*avx512f_gatherdi<mode>): Ditto.
2238 (*avx512f_gatherdi<mode>_2): Ditto.
2239 (*avx512f_scattersi<mode>): Ditto.
2240 (*avx512f_scatterdi<mode>): Ditto.
2241 (avx512f_compress<mode>_mask): Ditto.
2242 (avx512f_compressstore<mode>_mask): Ditto.
2243 (avx512f_expand<mode>_mask): Ditto.
2244 * config/i386/subst.md (mask): Change k to Yk.
2245 (mask_scalar_merge): Ditto.
2248 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2250 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2252 2014-01-31 Richard Biener <rguenther@suse.de>
2255 * builtins.c (fold_builtin_memory_op): Make sure to not
2256 use a floating-point mode or a boolean or enumeral type for
2259 2014-01-30 DJ Delorie <dj@redhat.com>
2261 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2262 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2263 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2264 whenever main() has an epilogue.
2266 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2268 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
2269 unused variable "field".
2270 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
2271 (vsx_mergeh_<mode>): Likewise.
2272 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
2273 (altivec_vmrghh): Likewise.
2274 (altivec_vmrghw): Likewise.
2275 (altivec_vmrglb): Likewise.
2276 (altivec_vmrglh): Likewise.
2277 (altivec_vmrglw): Likewise.
2278 (altivec_vspltb): Add missing uses.
2279 (altivec_vsplth): Likewise.
2280 (altivec_vspltw): Likewise.
2281 (altivec_vspltsf): Likewise.
2283 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2286 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
2287 frame related instructions.
2289 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
2291 PR rtl-optimization/59959
2292 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
2293 any reload of register whose subreg is invalid.
2295 2014-01-30 Jakub Jelinek <jakub@redhat.com>
2297 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
2298 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
2299 Add missing return type - void.
2301 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2303 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
2304 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
2305 remove element index adjustment for endian (now handled in vsx.md
2307 (altivec_expand_vec_perm_const): Use
2308 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
2309 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
2310 (vsx_xxspltw_<mode>): Adjust element index for little endian.
2311 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
2312 define_expand and a new define_insn *altivec_vspltb_internal;
2313 adjust for -maltivec=be on a little endian target.
2314 (altivec_vspltb_direct): New.
2315 (altivec_vsplth): Divide into a define_expand and a new
2316 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
2317 little endian target.
2318 (altivec_vsplth_direct): New.
2319 (altivec_vspltw): Divide into a define_expand and a new
2320 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
2321 little endian target.
2322 (altivec_vspltw_direct): New.
2323 (altivec_vspltsf): Divide into a define_expand and a new
2324 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
2325 a little endian target.
2327 2014-01-30 Richard Biener <rguenther@suse.de>
2329 PR tree-optimization/59993
2330 * tree-ssa-forwprop.c (associate_pointerplus): Check we
2331 can propagate form the earlier stmt and avoid the transform
2332 when the intermediate result is needed.
2334 2014-01-30 Alangi Derick <alangiderick@gmail.com>
2336 * README.Portability: Fix typo.
2338 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
2340 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
2341 comparison_operator with ordered_comparison_operator.
2343 2014-01-30 Nick Clifton <nickc@redhat.com>
2345 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
2346 Rename to mn10300_store_multiple_regs.
2347 * config/mn10300/mn10300.c: Likewise.
2348 * config/mn10300/mn10300.md (store_movm): Fix typo: call
2349 store_multiple_regs.
2350 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
2351 Call mn10300_store_multiple_regs.
2353 2014-01-30 Nick Clifton <nickc@redhat.com>
2354 DJ Delorie <dj@redhat.com>
2356 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
2357 %fp 2 to keep registers after it properly word-aligned.
2358 (rl78_alloc_physical_registers_umul): Handle the case where both
2359 input operands are the same.
2361 2014-01-30 Richard Biener <rguenther@suse.de>
2363 PR tree-optimization/59903
2364 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
2367 2014-01-30 Jason Merrill <jason@redhat.com>
2370 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
2373 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
2375 2014-01-30 Richard Biener <rguenther@suse.de>
2377 PR tree-optimization/59951
2378 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
2380 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
2383 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
2384 SFmode to DFmode case.
2386 2014-01-29 DJ Delorie <dj@redhat.com>
2388 * config/msp430/msp430.opt (-minrt): New.
2389 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
2391 (ENDFILE_SPEC): Likewise.
2393 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
2395 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
2396 (estimate_function_body_sizes): Use it.
2398 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
2401 * dwarf2out.c (is_cxx_auto): New.
2402 (is_base_type): Use it.
2403 (gen_type_die_with_usage): Likewise.
2405 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2407 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
2408 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
2409 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
2410 -maltivec=be with LE targets.
2411 (vsx_mergeh_<mode>): Likewise.
2412 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
2413 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
2414 (altivec_vmrghb): Replace with define_expand and new
2415 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
2416 (altivec_vmrghb_direct): New define_insn.
2417 (altivec_vmrghh): Replace with define_expand and new
2418 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
2419 (altivec_vmrghh_direct): New define_insn.
2420 (altivec_vmrghw): Replace with define_expand and new
2421 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
2422 (altivec_vmrghw_direct): New define_insn.
2423 (*altivec_vmrghsf): Adjust for endianness.
2424 (altivec_vmrglb): Replace with define_expand and new
2425 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
2426 (altivec_vmrglb_direct): New define_insn.
2427 (altivec_vmrglh): Replace with define_expand and new
2428 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
2429 (altivec_vmrglh_direct): New define_insn.
2430 (altivec_vmrglw): Replace with define_expand and new
2431 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
2432 (altivec_vmrglw_direct): New define_insn.
2433 (*altivec_vmrglsf): Adjust for endianness.
2434 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2435 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2436 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
2437 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
2438 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2439 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2440 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
2441 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
2443 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
2445 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
2446 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
2449 2014-01-29 Richard Biener <rguenther@suse.de>
2451 PR tree-optimization/58742
2452 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
2453 associate_pointerplus_align.
2454 (associate_pointerplus_diff): New function.
2455 (associate_pointerplus): Likewise. Call associate_pointerplus_align
2456 and associate_pointerplus_diff.
2458 2014-01-29 Richard Biener <rguenther@suse.de>
2460 * lto-streamer.h (LTO_major_version): Bump to 3.
2461 (LTO_minor_version): Reset to 0.
2463 2014-01-29 Renlin Li <Renlin.Li@arm.com>
2465 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
2466 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
2467 (arm_file_start): Generate correct asm header for armv7ve.
2468 * config/arm/bpabi.h: Add multilib support for armv7ve.
2469 * config/arm/driver-arm.c: Change the architectures of cortex-a7
2470 and cortex-a15 to armv7ve.
2471 * config/arm/t-aprofile: Add multilib support for armv7ve.
2472 * doc/invoke.texi: Document -march=armv7ve.
2474 2014-01-29 Richard Biener <rguenther@suse.de>
2476 PR tree-optimization/58742
2477 * tree-ssa-forwprop.c (associate_plusminus): Return true
2478 if we changed sth, defer EH cleanup to ...
2479 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
2480 (simplify_mult): New function.
2482 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2485 PR tree-optimization/59920
2486 * tree.c (build_common_builtin_nodes): Remove
2487 __builtin_setjmp_dispatcher initialization.
2488 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
2489 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
2490 instead of gsi_after_labels + manually skipping debug stmts.
2491 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
2492 ignore bbs with IFN_ABNORMAL_DISPATCHER.
2493 * tree-inline.c (copy_edges_for_bb): Remove
2494 can_make_abnormal_goto argument, instead add abnormal_goto_dest
2495 argument. Ignore computed_goto_p stmts. Don't call
2496 make_abnormal_goto_edges. If a call might need abnormal edges
2497 for non-local gotos, see if it already has an edge to
2498 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
2499 with true argument, don't do anything then, otherwise add
2500 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
2501 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
2503 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
2504 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
2505 (lower_stmt): Don't set data->calls_builtin_setjmp.
2506 (lower_builtin_setjmp): Adjust comment.
2507 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
2508 * tree-cfg.c (found_computed_goto): Remove.
2509 (factor_computed_gotos): Remove.
2510 (make_goto_expr_edges): Return bool, true for computed gotos.
2511 Don't call make_abnormal_goto_edges.
2512 (build_gimple_cfg): Don't set found_computed_goto, don't call
2513 factor_computed_gotos.
2514 (computed_goto_p): No longer static.
2515 (make_blocks): Don't set found_computed_goto.
2516 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
2517 (make_edges): If make_goto_expr_edges returns true, push bb
2518 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
2519 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
2520 vector. Record mapping between bbs and OpenMP regions if there
2521 are any, adjust make_gimple_omp_edges caller. Call
2522 handle_abnormal_edges.
2523 (make_abnormal_goto_edges): Remove.
2524 * tree-cfg.h (make_abnormal_goto_edges): Remove.
2525 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
2526 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
2527 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
2528 * internal-fn.def (ABNORMAL_DISPATCHER): New.
2529 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
2530 filling *region also set *region_idx to (*region)->entry->index.
2533 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
2534 For REGs set ORIGINAL_REGNO.
2536 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
2538 * doc/md.texi: Mention that a target shouldn't implement
2539 vec_widen_(s|u)mul_even/odd pair if it is less efficient
2542 2014-01-29 Jakub Jelinek <jakub@redhat.com>
2544 PR tree-optimization/59594
2545 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
2546 a copy of the datarefs vector rather than the vector itself.
2548 2014-01-28 Jason Merrill <jason@redhat.com>
2551 * dwarf2out.c (auto_die): New static.
2552 (gen_type_die_with_usage): Handle C++1y 'auto'.
2553 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
2556 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
2559 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
2560 (SPEC_X32): Likewise.
2561 (SPEC_64): Likewise.
2562 * config/i386/i386.c (ix86_option_override_internal): Turn off
2563 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
2565 (x86_file_start): Output .code16gcc for TARGET_16BIT.
2566 * config/i386/i386.h (TARGET_16BIT): New macro.
2567 (TARGET_16BIT_P): Likewise.
2568 * config/i386/i386.opt: Add m16.
2569 * doc/invoke.texi: Document -m16.
2571 2014-01-28 Jakub Jelinek <jakub@redhat.com>
2573 PR preprocessor/59935
2574 * input.c (location_get_source_line): Bail out on when line number
2575 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
2577 2014-01-28 Richard Biener <rguenther@suse.de>
2579 PR tree-optimization/58742
2580 * tree-ssa-forwprop.c (associate_plusminus): Handle
2581 pointer subtraction of the form (T)(P + A) - (T)P.
2583 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2585 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
2588 2014-01-28 Richard Biener <rguenther@suse.de>
2591 2014-01-28 Richard Biener <rguenther@suse.de>
2593 PR rtl-optimization/45364
2594 PR rtl-optimization/59890
2595 * var-tracking.c (local_get_addr_clear_given_value): Handle
2596 already cleared slot.
2597 (val_reset): Handle not allocated local_get_addr_cache.
2598 (vt_find_locations): Use post-order on the inverted CFG.
2600 2014-01-28 Richard Biener <rguenther@suse.de>
2602 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
2604 2014-01-28 Richard Biener <rguenther@suse.de>
2606 PR rtl-optimization/45364
2607 PR rtl-optimization/59890
2608 * var-tracking.c (local_get_addr_clear_given_value): Handle
2609 already cleared slot.
2610 (val_reset): Handle not allocated local_get_addr_cache.
2611 (vt_find_locations): Use post-order on the inverted CFG.
2613 2014-01-28 Alan Modra <amodra@gmail.com>
2615 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
2616 * configure.ac <recursive call for build != host>: Define
2617 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
2618 and LD_FOR_BUILD too.
2619 * configure: Regenerate.
2621 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
2623 * config/i386/i386.c (get_builtin_code_for_version): Separate
2624 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
2625 Broadwell from Haswell.
2627 2014-01-27 Steve Ellcey <sellcey@mips.com>
2629 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
2630 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
2631 * config/mips/mips.c (mips_option_override): Change setting
2633 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
2634 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
2635 Change from Mask to Var.
2637 2014-01-27 Jeff Law <law@redhat.com>
2639 * ipa-inline.c (inline_small_functions): Fix typo.
2641 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
2643 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
2644 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
2645 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
2646 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
2647 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
2648 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
2649 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
2650 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
2651 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
2652 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
2653 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
2654 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
2655 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
2656 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
2657 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
2658 (_mm512_storeu_epi64): Ditto.
2659 (_mm512_cmpge_epi32_mask): Ditto.
2660 (_mm512_cmpge_epu32_mask): Ditto.
2661 (_mm512_cmpge_epi64_mask): Ditto.
2662 (_mm512_cmpge_epu64_mask): Ditto.
2663 (_mm512_cmple_epi32_mask): Ditto.
2664 (_mm512_cmple_epu32_mask): Ditto.
2665 (_mm512_cmple_epi64_mask): Ditto.
2666 (_mm512_cmple_epu64_mask): Ditto.
2667 (_mm512_cmplt_epi32_mask): Ditto.
2668 (_mm512_cmplt_epu32_mask): Ditto.
2669 (_mm512_cmplt_epi64_mask): Ditto.
2670 (_mm512_cmplt_epu64_mask): Ditto.
2671 (_mm512_cmpneq_epi32_mask): Ditto.
2672 (_mm512_cmpneq_epu32_mask): Ditto.
2673 (_mm512_cmpneq_epi64_mask): Ditto.
2674 (_mm512_cmpneq_epu64_mask): Ditto.
2675 (_mm512_expand_pd): Ditto.
2676 (_mm512_expand_ps): Ditto.
2677 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
2678 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
2679 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
2680 * config/i386/i386.c (ix86_builtins): Add
2681 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
2682 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
2683 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
2684 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
2685 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
2686 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
2687 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
2688 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
2689 IX86_BUILTIN_PMOVUSQW512_MEM.
2690 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
2691 __builtin_ia32_pmovsqd512mem_mask,
2692 __builtin_ia32_pmovqd512mem_mask,
2693 __builtin_ia32_pmovusqw512mem_mask,
2694 __builtin_ia32_pmovsqw512mem_mask,
2695 __builtin_ia32_pmovqw512mem_mask,
2696 __builtin_ia32_pmovusdw512mem_mask,
2697 __builtin_ia32_pmovsdw512mem_mask,
2698 __builtin_ia32_pmovdw512mem_mask,
2699 __builtin_ia32_pmovqb512mem_mask,
2700 __builtin_ia32_pmovusqb512mem_mask,
2701 __builtin_ia32_pmovsqb512mem_mask,
2702 __builtin_ia32_pmovusdb512mem_mask,
2703 __builtin_ia32_pmovsdb512mem_mask,
2704 __builtin_ia32_pmovdb512mem_mask.
2705 (bdesc_args): Add __builtin_ia32_expanddf512,
2706 __builtin_ia32_expandsf512.
2707 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
2708 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
2709 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
2710 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
2711 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
2712 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
2713 (avx512f_<code>v8div16qi2_mask_store): This.
2714 (avx512f_expand<mode>): New.
2716 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
2718 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
2720 (_mm512_mask_prefetch_i64gather_pd): Ditto.
2721 (_mm512_prefetch_i32scatter_pd): Ditto.
2722 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
2723 (_mm512_prefetch_i64scatter_pd): Ditto.
2724 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
2725 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
2726 (_mm512_mask_prefetch_i64gather_ps): Ditto.
2727 (_mm512_prefetch_i32scatter_ps): Ditto.
2728 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
2729 (_mm512_prefetch_i64scatter_ps): Ditto.
2730 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
2731 * config/i386/i386-builtin-types.def: Define
2732 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
2733 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
2734 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
2735 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
2736 IX86_BUILTIN_SCATTERPFQPD.
2737 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
2738 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
2739 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
2740 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
2741 __builtin_ia32_scatterpfqps.
2742 (ix86_expand_builtin): Expand new built-ins.
2743 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
2744 fix memory access data type.
2745 (*avx512pf_gatherpf<mode>_mask): Ditto.
2746 (*avx512pf_gatherpf<mode>): Ditto.
2747 (avx512pf_scatterpf<mode>): Ditto.
2748 (*avx512pf_scatterpf<mode>_mask): Ditto.
2749 (*avx512pf_scatterpf<mode>): Ditto.
2750 (GATHER_SCATTER_SF_MEM_MODE): New.
2751 (avx512pf_gatherpf<mode>df): Ditto.
2752 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2753 (*avx512pf_scatterpf<mode>df): Ditto.
2755 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2758 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2759 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2762 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2764 * common/config/arm/arm-common.c
2765 (arm_rewrite_mcpu): Handle multiple names.
2767 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2769 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2771 * gimple-builder.h (create_gimple_tmp): Delete.
2773 2014-01-27 Christian Bruel <christian.bruel@st.com>
2775 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2778 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2780 * config/pa/pa.md (call): Generate indirect long calls to non-local
2781 functions when outputing 32-bit code.
2782 (call_value): Likewise except for special call to buggy powf function.
2784 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2785 portable runtime and PIC indirect calls.
2786 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2787 and PIC call sequences. Use ldo instead of blr to set return register
2788 in PIC call sequence.
2790 2014-01-25 Walter Lee <walt@tilera.com>
2792 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2793 avoid clobbering a live register.
2795 2014-01-25 Walter Lee <walt@tilera.com>
2797 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2798 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2799 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2800 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2802 2014-01-25 Walter Lee <walt@tilera.com>
2804 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2805 arguments on even registers.
2806 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2808 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2809 (BIGGEST_ALIGNMENT): Ditto.
2810 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2812 2014-01-25 Walter Lee <walt@tilera.com>
2814 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2815 insns before bundling.
2816 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2818 2014-01-25 Walter Lee <walt@tilera.com>
2820 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2821 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2822 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2824 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2826 * config/mips/constraints.md (kl): Delete.
2827 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2828 define expands, using...
2829 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2830 instructions for MIPS16.
2831 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2832 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2834 2014-01-25 Walter Lee <walt@tilera.com>
2836 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2840 2014-01-25 Walter Lee <walt@tilera.com>
2842 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2843 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2845 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2847 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2850 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2852 * print-rtl.c (in_call_function_usage): New var.
2853 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2854 EXPR_LIST mode as mode and not as reg note name.
2857 * cfgloopmanip.c (copy_loop_info): If
2858 loop->warned_aggressive_loop_optimizations, make sure
2859 the flag is set in target loop too.
2861 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2863 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2865 * builtins.def: Likewise.
2866 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2867 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2868 * ira.c (ira_setup_eliminable_regset): Likewise.
2869 * omp-low.c (gate_expand_omp): Likewise.
2870 (execute_lower_omp): Likewise.
2871 (diagnose_sb_0): Likewise.
2872 (gate_diagnose_omp_blocks): Likewise.
2873 (simd_clone_clauses_extract): Likewise.
2876 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2878 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2879 correction for little endian...
2880 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2883 2014-01-24 Jeff Law <law@redhat.com>
2885 PR tree-optimization/59919
2886 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2887 for non-returning calls.
2889 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2891 * common/config/aarch64/aarch64-common.c
2892 (aarch64_rewrite_mcpu): Handle multiple names.
2893 * config/aarch64/aarch64.h
2894 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2896 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2898 * input.c (add_file_to_cache_tab): Handle the case where fopen
2901 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2904 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2905 from push operand if code of push isn't PRE_DEC.
2907 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2910 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2911 -mquad-memory-atomic. Update -mquad-memory documentation to say
2912 it is only used for non-atomic loads/stores.
2914 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2915 -mquad-memory or -mquad-memory-atomic switches.
2917 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2918 -mquad-memory-atomic to ISA 2.07 support.
2920 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2921 to separate support of normal quad word memory operations (ldq, stq)
2922 from the atomic quad word memory operations.
2924 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2925 support to separate non-atomic quad word operations from atomic
2926 quad word operations. Disable non-atomic quad word operations in
2927 little endian mode so that we don't have to swap words after the
2928 load and before the store.
2929 (quad_load_store_p): Add comment about atomic quad word support.
2930 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2931 options printed with -mdebug=reg.
2933 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2934 -mquad-memory-atomic as the test for whether we have quad word
2935 atomic instructions.
2936 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2937 or -mp8-vector are used, allow byte/half-word atomic operations.
2939 * config/rs6000/sync.md (load_lockedti): Insure that the address
2940 is a proper indexed or indirect address for the lqarx instruction.
2941 On little endian systems, swap the hi/lo registers after the lqarx
2943 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2944 insure the address is valid for the lqarx instruction.
2945 (store_conditionalti): Insure that the address is a proper indexed
2946 or indirect address for the stqcrx. instruction. On little endian
2947 systems, swap the hi/lo registers before doing the stqcrx.
2949 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2950 insure the address is valid for the stqcrx. instruction.
2952 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2953 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2954 type of quad memory support is available.
2956 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2959 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2960 there is a danger of looping.
2962 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2964 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2965 force flag_ira_loop_pressure if set via command line.
2967 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2969 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2970 (ashr_simd): New builtin handling DI mode.
2971 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2972 (aarch64_sshr_simddi): New match pattern.
2973 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2974 (vshrd_n_s64): Likewise.
2975 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2977 2014-01-23 Nick Clifton <nickc@redhat.com>
2979 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2980 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2981 favour of mcu specific scripts.
2982 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2985 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2986 Alex Velenko <Alex.Velenko@arm.com>
2988 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2989 (vaddv_s16): Likewise.
2990 (vaddv_s32): Likewise.
2991 (vaddv_u8): Likewise.
2992 (vaddv_u16): Likewise.
2993 (vaddv_u32): Likewise.
2994 (vaddvq_s8): Likewise.
2995 (vaddvq_s16): Likewise.
2996 (vaddvq_s32): Likewise.
2997 (vaddvq_s64): Likewise.
2998 (vaddvq_u8): Likewise.
2999 (vaddvq_u16): Likewise.
3000 (vaddvq_u32): Likewise.
3001 (vaddvq_u64): Likewise.
3002 (vaddv_f32): Likewise.
3003 (vaddvq_f32): Likewise.
3004 (vaddvq_f64): Likewise.
3005 (vmaxv_f32): Likewise.
3006 (vmaxv_s8): Likewise.
3007 (vmaxv_s16): Likewise.
3008 (vmaxv_s32): Likewise.
3009 (vmaxv_u8): Likewise.
3010 (vmaxv_u16): Likewise.
3011 (vmaxv_u32): Likewise.
3012 (vmaxvq_f32): Likewise.
3013 (vmaxvq_f64): Likewise.
3014 (vmaxvq_s8): Likewise.
3015 (vmaxvq_s16): Likewise.
3016 (vmaxvq_s32): Likewise.
3017 (vmaxvq_u8): Likewise.
3018 (vmaxvq_u16): Likewise.
3019 (vmaxvq_u32): Likewise.
3020 (vmaxnmv_f32): Likewise.
3021 (vmaxnmvq_f32): Likewise.
3022 (vmaxnmvq_f64): Likewise.
3023 (vminv_f32): Likewise.
3024 (vminv_s8): Likewise.
3025 (vminv_s16): Likewise.
3026 (vminv_s32): Likewise.
3027 (vminv_u8): Likewise.
3028 (vminv_u16): Likewise.
3029 (vminv_u32): Likewise.
3030 (vminvq_f32): Likewise.
3031 (vminvq_f64): Likewise.
3032 (vminvq_s8): Likewise.
3033 (vminvq_s16): Likewise.
3034 (vminvq_s32): Likewise.
3035 (vminvq_u8): Likewise.
3036 (vminvq_u16): Likewise.
3037 (vminvq_u32): Likewise.
3038 (vminnmv_f32): Likewise.
3039 (vminnmvq_f32): Likewise.
3040 (vminnmvq_f64): Likewise.
3042 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3044 * config/aarch64/aarch64-simd.md
3045 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3046 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3047 (*aarch64_mul3_elt<mode>): Likewise.
3048 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3049 (*aarch64_mul3_elt_to_64v2df): Likewise.
3050 (*aarch64_mla_elt<mode>): Likewise.
3051 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3052 (*aarch64_mls_elt<mode>): Likewise.
3053 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3054 (*aarch64_fma4_elt<mode>): Likewise.
3055 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3056 (*aarch64_fma4_elt_to_64v2df): Likewise.
3057 (*aarch64_fnma4_elt<mode>): Likewise.
3058 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3059 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3060 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3061 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3062 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3063 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3064 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3065 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3066 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3068 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3070 * config/aarch64/aarch64-simd.md
3071 (aarch64_be_checked_get_lane<mode>): New define_expand.
3072 * config/aarch64/aarch64-simd-builtins.def
3073 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3074 New builtin definition.
3075 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3076 Use new safe be builtin.
3078 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3080 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3082 (aarch64_be_st1<mode>): Likewise.
3083 (aarch_ld1<VALL:mode>): Define_expand modified.
3084 (aarch_st1<VALL:mode>): Likewise.
3085 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3086 (UNSPEC_ST1): Likewise.
3088 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3090 * config/microblaze/microblaze.md: Add trap insn and attribute
3092 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3094 PR preprocessor/58580
3095 * input.h (location_get_source_line): Take an additional line_size
3097 (void diagnostics_file_cache_fini): Declare new function.
3098 * input.c (struct fcache): New type.
3099 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3100 New static constants.
3101 (diagnostic_file_cache_init, total_lines_num)
3102 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3103 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3104 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3105 (get_next_line, read_next_line, goto_next_line, read_line_num):
3106 New static function definitions.
3107 (diagnostic_file_cache_fini): New function.
3108 (location_get_source_line): Take an additional output line_len
3109 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3111 * diagnostic.c (diagnostic_finish): Call
3112 diagnostic_file_cache_fini.
3113 (adjust_line): Take an additional input parameter for the length
3114 of the line, rather than calculating it with strlen.
3115 (diagnostic_show_locus): Adjust the use of
3116 location_get_source_line and adjust_line with respect to their new
3117 signature. While displaying a line now, do not stop at the first
3118 null byte. Rather, display the zero byte as a space and keep
3119 going until we reach the size of the line.
3120 * Makefile.in: Add vec.o to OBJS-libcommon
3122 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3123 Ilya Tocar <ilya.tocar@intel.com>
3125 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3126 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3127 (__builtin_ia32_kmov16): Ditto.
3128 * config/i386/i386.md (UNSPEC_KMOV): New.
3131 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3133 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3134 (_mm512_storeu_si512): Ditto.
3136 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3139 * rtl.h (get_referenced_operands): Declare.
3140 * recog.c (get_referenced_operands): New function.
3141 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3142 operands have been referenced when recording LO_SUM references.
3144 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3146 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3148 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3150 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3151 Enable for generic and recent AMD targets.
3153 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3155 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3156 ARG_SIZE note when adjustment was eliminated.
3158 2014-01-22 Jeff Law <law@redhat.com>
3160 PR tree-optimization/59597
3161 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3162 in file. Accept new argument REGISTERING and use it to modify
3163 dump output appropriately.
3164 (register_jump_thread): Corresponding changes.
3165 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3166 thread paths involving joiner blocks. Add code to dump cancelled
3167 jump threading paths.
3169 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3171 PR rtl-optimization/59477
3172 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3173 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3175 2014-01-22 Tom Tromey <tromey@redhat.com>
3177 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3179 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3181 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3183 PR rtl-optimization/59896
3184 * lra-constraints.c (process_alt_operands): Check unused note for
3185 matched operands of insn with no output reloads.
3187 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3189 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3190 (mips_move_from_gpr_cost): Likewise.
3192 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3194 PR rtl-optimization/59858
3195 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3196 ira_class_hard_regs_num.
3197 (process_alt_operands): Increase reject for dying matched operand.
3199 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3202 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3203 smaller than size, perform several stores or loads and stores
3204 at dst + count - size to store or copy all of size bytes, rather
3205 than just last modesize bytes.
3207 2014-01-20 DJ Delorie <dj@redhat.com>
3209 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3210 that CLOBBERs are REGs before propogating their values.
3212 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3215 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3216 (cgraph_inline_failed_type): New function.
3217 * cgraph.h (DEFCIFCODE): Add type.
3218 (cgraph_inline_failed_type_t): New enum.
3219 (cgraph_inline_failed_type): New prototype.
3220 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3221 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3222 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3223 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3224 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3225 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3226 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3227 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3228 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3229 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3230 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3231 OPTIMIZATION_MISMATCH.
3232 * tree-inline.c (expand_call_inline): Emit errors during
3233 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3235 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3238 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3239 mode attribute in insn output.
3241 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3243 * output.h (output_constant): Delete.
3244 * varasm.c (output_constant): Make private.
3246 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3248 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3250 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3253 * tree.h (fold_builtin_strcat): New prototype.
3254 * builtins.c (fold_builtin_strcat): No longer static. Add len
3255 argument, if non-NULL, don't call c_strlen. Optimize
3256 directly into __builtin_memcpy instead of __builtin_strcpy.
3257 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3258 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3260 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3262 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3263 for SImode_address_operand operands, having only a REG argument.
3265 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3267 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
3268 loader name using mbig-endian.
3269 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
3271 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3273 * doc/invoke.texi (-march): Clarify documentation for AArch64.
3277 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
3279 * config/aarch64/aarch64-protos.h
3280 (aarch64_cannot_change_mode_class_ptr): Declare.
3281 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
3282 aarch64_cannot_change_mode_class_ptr): New.
3283 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
3284 backend hook aarch64_cannot_change_mode_class.
3286 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
3288 * common/config/aarch64/aarch64-common.c
3289 (aarch64_handle_option): Don't handle any option order logic here.
3290 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
3291 selected_cpu, warn on architecture version mismatch.
3292 (aarch64_override_options): Fix parsing order for option strings.
3294 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3295 Iain Sandoe <iain@codesourcery.com>
3298 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
3299 warning. Amend comment to reflect current functionality.
3301 2014-01-20 Richard Biener <rguenther@suse.de>
3304 * builtins.c (fold_builtin_strcat): Remove case better handled
3305 by tree-ssa-strlen.c.
3307 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
3309 * config/aarch64/aarch64.opt
3310 (mcpu, march, mtune): Make case-insensitive.
3312 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3315 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3316 if operands[1] is a REG or ZERO_EXTEND of a REG.
3318 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
3320 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
3322 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
3324 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
3325 long non-pic millicode calls.
3327 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3329 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
3331 2014-01-19 Kito Cheng <kito@0xlab.org>
3333 * builtins.c (expand_movstr): Check movstr expand done or fail.
3335 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3336 H.J. Lu <hongjiu.lu@intel.com>
3339 * config/i386/i386.md (*lea<mode>): Zero-extend return register
3340 to DImode for zero-extended addresses.
3342 2014-01-19 Jakub Jelinek <jakub@redhat.com>
3344 PR rtl-optimization/57763
3345 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
3346 on the new indirect jump_insn and increment LABEL_NUSES (label).
3348 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
3352 * config.gcc (x86_archs): New variable.
3353 (x86_64_archs): Likewise.
3354 (x86_cpus): Likewise.
3355 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
3356 --with-arch/--with-cpu= options.
3357 Support --with-arch=/--with-cpu={nehalem,westmere,
3358 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
3360 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3362 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
3363 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
3365 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
3367 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
3369 2014-01-18 Jakub Jelinek <jakub@redhat.com>
3372 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
3373 clear cpp_get_options (parse_in)->warn_unused_macros for
3374 ix86_target_macros_internal with cpp_define.
3376 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
3378 * jump.c (delete_related_insns): Keep (use (insn))s.
3379 * reorg.c (redundant_insn): Check for barriers too.
3381 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3383 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
3385 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
3387 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
3388 call to $$dyncall when TARGET_LONG_CALLS is true.
3390 2014-01-17 Jeff Law <law@redhat.com>
3392 * ree.c (combine_set_extension): Temporarily disable test for
3393 changing number of hard registers.
3395 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3398 * ipa-inline-analysis.c (inline_free_summary):
3399 Do not free summary of aliases.
3401 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3404 * gimplify.c (gimplify_expr): Use create_tmp_var
3405 instead of create_tmp_var_raw. If cond doesn't have
3406 integral type, don't add the IFN_ANNOTATE builtin at all.
3408 2014-01-17 Martin Jambor <mjambor@suse.cz>
3411 * ipa-cp.c (prev_edge_clone): New variable.
3412 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
3413 Also resize prev_edge_clone vector.
3414 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
3415 (ipcp_edge_removal_hook): New function.
3416 (ipcp_driver): Register ipcp_edge_removal_hook.
3418 2014-01-17 Andrew Pinski <apinski@cavium.com>
3419 Steve Ellcey <sellcey@mips.com>
3422 * config/mips/mips.c (mips_print_operand): Check operand mode instead
3425 2014-01-17 Jeff Law <law@redhat.com>
3428 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
3429 so that pass_ccp runs first.
3431 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3433 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
3434 (ix86_adjust_cost): Use !TARGET_XXX.
3435 (do_reorder_for_imul): Likewise.
3436 (swap_top_of_ready_list): Likewise.
3437 (ix86_sched_reorder): Likewise.
3439 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
3441 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3442 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
3443 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
3444 (intel_memset): New. Duplicate slm_memset.
3445 (intel_cost): New. Duplicate slm_cost.
3446 (m_INTEL): New macro.
3447 (processor_target_table): Add "intel".
3448 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
3449 with PROCESSOR_INTEL for "intel".
3450 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
3451 PROCESSOR_SILVERMONT.
3452 (ix86_issue_rate): Likewise.
3453 (ix86_adjust_cost): Likewise.
3454 (ia32_multipass_dfa_lookahead): Likewise.
3455 (swap_top_of_ready_list): Likewise.
3456 (ix86_sched_reorder): Likewise.
3457 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
3458 instead of TARGET_OPT_AGU.
3459 * config/i386/i386.h (TARGET_INTEL): New.
3460 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
3461 (processor_type): Add PROCESSOR_INTEL.
3462 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
3463 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
3465 2014-01-17 Marek Polacek <polacek@redhat.com>
3468 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
3471 2014-01-17 Richard Biener <rguenther@suse.de>
3473 PR tree-optimization/46590
3474 * opts.c (default_options_table): Add entries for
3475 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
3476 all enabled at -O1 but not for -Og.
3477 * common.opt (fbranch-count-reg): Remove Init(1).
3478 (fmove-loop-invariants): Likewise.
3479 (ftree-pta): Likewise.
3481 2014-01-17 Jakub Jelinek <jakub@redhat.com>
3483 * config/i386/i386.c (ix86_data_alignment): For compatibility with
3484 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
3485 decls to at least the GCC 4.8 used alignments.
3488 * tree-nested.c (convert_nonlocal_reference_stmt,
3489 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
3490 of GIMPLE_BIND stmts, adjust associated decls.
3492 2014-01-17 Richard Biener <rguenther@suse.de>
3494 PR tree-optimization/46590
3495 * vec.h (vec<>::bseach): New member function implementing
3496 binary search according to C89 bsearch.
3497 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
3498 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
3499 bitmap pointer again. Make accesses_in_loop a flat array.
3500 (mem_ref_obstack): New global.
3501 (outermost_indep_loop): Adjust for mem_ref->stored changes.
3502 (mark_ref_stored): Likewise.
3503 (ref_indep_loop_p_2): Likewise.
3504 (set_ref_stored_in_loop): New helper function.
3505 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
3506 (memref_free): Adjust.
3507 (record_mem_ref_loc): Simplify.
3508 (gather_mem_refs_stmt): Adjust.
3509 (sort_locs_in_loop_postorder_cmp): New function.
3510 (analyze_memory_references): Sort accesses_in_loop after
3511 loop postorder number.
3512 (find_ref_loc_in_loop_cmp): New function.
3513 (for_all_locs_in_loop): Find relevant cluster of locs in
3514 accesses_in_loop and iterate without recursion.
3515 (execute_sm): Avoid uninit warning.
3516 (struct ref_always_accessed): Simplify.
3517 (ref_always_accessed::operator ()): Likewise.
3518 (ref_always_accessed_p): Likewise.
3519 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
3520 loop postorder numbers here.
3521 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
3524 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
3527 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
3528 on decls for which assemble_alias has been called.
3530 2014-01-17 Nick Clifton <nickc@redhat.com>
3532 * config/msp430/msp430.opt: (mcpu): New option.
3533 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
3534 (msp430_option_override): Parse target_cpu. If the MCU name
3535 matches a generic string, clear target_mcu.
3536 (msp430_attr): Allow numeric interrupt values up to 63.
3537 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
3538 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
3540 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
3542 * config/msp430/msp430.md (popm): Use %J rather than %I.
3543 (addsi3): Use msp430_nonimmediate_operand for operand 2.
3544 (addhi_cy_i): Use immediate_operand for operand 2.
3545 * doc/invoke.texi: Document -mcpu option.
3547 2014-01-17 Richard Biener <rguenther@suse.de>
3549 PR rtl-optimization/38518
3550 * df.h (df_analyze_loop): Declare.
3551 * df-core.c: Include cfgloop.h.
3552 (df_analyze_1): Split out main part of df_analyze.
3553 (df_analyze): Adjust.
3554 (loop_inverted_post_order_compute): New function.
3555 (loop_post_order_compute): Likewise.
3556 (df_analyze_loop): New function avoiding whole-function
3558 * loop-invariant.c (find_defs): Use df_analyze_loop.
3559 (find_invariants): Adjust.
3560 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
3562 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
3564 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
3565 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
3567 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
3569 * ipa-ref.c (ipa_remove_stmt_references): Fix references
3570 traversal when removing references.
3572 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
3575 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
3577 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
3580 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
3581 pushing a reload for an autoinc when we had previously reloaded an
3582 inner part of the address.
3584 2014-01-16 Jakub Jelinek <jakub@redhat.com>
3586 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
3588 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
3589 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
3590 when not giving up or versioning for alias only because of
3592 (vect_analyze_data_ref_dependences): Set to true.
3593 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
3595 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
3596 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
3600 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
3603 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
3604 operand 0 predicate for gathers, use a new pseudo as subtarget.
3606 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3609 * lra-constraints.c (process_alt_operands): Add printing debug info.
3610 Check absence of input/output reloads for matched operands too.
3612 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
3614 PR rtl-optimization/59835
3615 * ira.c (ira_init_register_move_cost): Increase cost for
3618 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
3620 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
3622 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
3625 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
3626 non-register objects. Use gen_(high/low)part more consistently.
3629 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3632 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
3633 endian support, remove tests for WORDS_BIG_ENDIAN.
3634 (p8_mfvsrd_3_<mode>): Likewise.
3635 (reload_gpr_from_vsx<mode>): Likewise.
3636 (reload_gpr_from_vsxsf): Likewise.
3637 (p8_mfvsrd_4_disf): Likewise.
3639 2014-01-16 Richard Biener <rguenther@suse.de>
3641 PR rtl-optimization/46590
3642 * lcm.c (compute_antinout_edge): Use postorder iteration.
3643 (compute_laterin): Use inverted postorder iteration.
3645 2014-01-16 Nick Clifton <nickc@redhat.com>
3648 * varasm.c (output_constant): Return the number of bytes actually
3650 (output_constructor_array_range): Update the field size with the
3651 number of bytes emitted by output_constant.
3652 (output_constructor_regular_field): Likewise. Also do not
3653 complain if the total number of bytes emitted is now greater
3654 than the expected fieldpos.
3655 * output.h (output_constant): Update prototype and descriptive comment.
3657 2014-01-16 Marek Polacek <polacek@redhat.com>
3660 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
3661 it is error_mark_node.
3663 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
3665 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
3666 VALID_AVX256_REG_OR_OI_MODE.
3668 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
3670 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
3671 current procedure should be profiled.
3673 2014-01-15 Andrew Pinski <apinski@cavium.com>
3675 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
3676 of moving from/to the STACK_REG register class.
3678 2014-01-15 Richard Henderson <rth@redhat.com>
3681 * reginfo.c (global_regs_decl): Globalize.
3682 * rtl.h (global_regs_decl): Declare.
3683 * ira.c (do_reload): Diagnose frame_pointer_needed and it
3684 reserved via global_regs.
3686 2014-01-15 Teresa Johnson <tejohnson@google.com>
3688 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
3690 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3692 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
3693 and vmulosh rather than call gen_vec_widen_smult_*.
3694 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
3695 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
3696 (vec_widen_smult_even_v16qi): Likewise.
3697 (vec_widen_umult_even_v8hi): Likewise.
3698 (vec_widen_smult_even_v8hi): Likewise.
3699 (vec_widen_umult_odd_v16qi): Likewise.
3700 (vec_widen_smult_odd_v16qi): Likewise.
3701 (vec_widen_umult_odd_v8hi): Likewise.
3702 (vec_widen_smult_odd_v8hi): Likewise.
3703 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
3704 vmuloub rather than call gen_vec_widen_umult_*.
3705 (vec_widen_umult_lo_v16qi): Likewise.
3706 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
3707 vmulosb rather than call gen_vec_widen_smult_*.
3708 (vec_widen_smult_lo_v16qi): Likewise.
3709 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
3710 rather than call gen_vec_widen_umult_*.
3711 (vec_widen_umult_lo_v8hi): Likewise.
3712 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
3713 rather than call gen_vec_widen_smult_*.
3714 (vec_widen_smult_lo_v8hi): Likewise.
3716 2014-01-15 Jeff Law <law@redhat.com>
3718 PR tree-optimization/59747
3719 * ree.c (find_and_remove_re): Properly handle case where a second
3720 eliminated extension requires widening a copy created for elimination
3721 of a prior extension.
3722 (combine_set_extension): Ensure that the number of hard regs needed
3723 for a destination register does not change when we widen it.
3725 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3727 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
3728 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
3729 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
3730 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
3731 (avr-*-rtems*): Likewise.
3732 (bfin*-rtems*): Likewise.
3733 (moxie-*-rtems*): Likewise.
3734 (h8300-*-rtems*): Likewise.
3735 (i[34567]86-*-rtems*): Likewise.
3736 (lm32-*-rtems*): Likewise.
3737 (m32r-*-rtems*): Likewise.
3738 (m68k-*-rtems*): Likewise.
3739 (microblaze*-*-rtems*): Likewise.
3740 (mips*-*-rtems*): Likewise.
3741 (powerpc-*-rtems*): Likewise.
3742 (sh-*-rtems*): Likewise.
3743 (sparc-*-rtems*): Likewise.
3744 (sparc64-*-rtems*): Likewise.
3745 (v850-*-rtems*): Likewise.
3746 (m32c-*-rtems*): Likewise.
3748 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
3750 PR rtl-optimization/59511
3751 * ira.c (ira_init_register_move_cost): Use memory costs for some
3752 cases of register move cost calculations.
3753 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
3754 instead of BB frequency.
3755 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3756 * lra-assigns.c (find_hard_regno_for): Ditto.
3758 2014-01-15 Richard Biener <rguenther@suse.de>
3760 PR tree-optimization/59822
3761 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3762 (vectorizable_load): Use it to hoist defs of uses of invariant
3763 loads out of the loop.
3765 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3766 Kugan Vivekanandarajah <kuganv@linaro.org>
3769 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3772 2014-01-15 Richard Biener <rguenther@suse.de>
3774 PR rtl-optimization/59802
3775 * lcm.c (compute_available): Use inverted postorder to seed
3776 the initial worklist.
3778 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3781 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3782 ADDR_REGS for invalid symrefs in non-PIC code.
3784 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3787 * builtins.c (determine_block_size): Initialize *probable_max_size
3788 even if len_rtx is CONST_INT.
3790 2014-01-14 Andrew Pinski <apinski@cavium.com>
3792 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3793 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3794 (cortexa53_tunings): Likewise.
3795 (aarch64_sched_issue_rate): New function.
3796 (TARGET_SCHED_ISSUE_RATE): Define.
3798 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3800 * ira-costs.c (find_costs_and_classes): Add missed
3801 ira_init_register_move_cost_if_necessary.
3803 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3806 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3808 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3811 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3812 to indicate if type is used for function return value. Warn ABI
3813 change if the vector mode isn't available for function return value.
3814 (ix86_function_arg_advance): Pass false to type_natural_mode.
3815 (ix86_function_arg): Likewise.
3816 (ix86_gimplify_va_arg): Likewise.
3817 (function_arg_32): Don't warn ABI change.
3818 (ix86_function_value): Pass true to type_natural_mode.
3819 (ix86_return_in_memory): Likewise.
3820 (ix86_struct_value_rtx): Removed.
3821 (TARGET_STRUCT_VALUE_RTX): Likewise.
3823 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3825 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3826 converting a conditional jump into a conditional return.
3828 2014-01-14 Richard Biener <rguenther@suse.de>
3830 PR tree-optimization/58921
3831 PR tree-optimization/59006
3832 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3833 hoisting invariant stmts.
3834 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3835 invariant loads on the preheader edge if possible.
3837 2014-01-14 Joey Ye <joey.ye@arm.com>
3839 * doc/plugin.texi (Building GCC plugins): Update to C++.
3841 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3843 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3844 (_mm_rcp28_round_ss): Ditto.
3845 (_mm_rsqrt28_round_sd): Ditto.
3846 (_mm_rsqrt28_round_ss): Ditto.
3847 (_mm_rcp28_sd): Ditto.
3848 (_mm_rcp28_ss): Ditto.
3849 (_mm_rsqrt28_sd): Ditto.
3850 (_mm_rsqrt28_ss): Ditto.
3851 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3852 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3853 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3854 (IX86_BUILTIN_RCP28SD): Ditto.
3855 (IX86_BUILTIN_RCP28SS): Ditto.
3856 (IX86_BUILTIN_RSQRT28SD): Ditto.
3857 (IX86_BUILTIN_RSQRT28SS): Ditto.
3858 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3859 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3860 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3861 (ix86_expand_special_args_builtin): Expand new FTYPE.
3862 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3863 (srcp14<mode>): Make insn unary.
3864 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3865 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3866 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3867 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3868 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3869 Fix rounding: make it SAE only.
3870 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3872 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3874 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3875 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3876 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3877 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3878 (round_saeonly_mask_scalar_operand4): Ditto.
3879 (round_saeonly_mask_scalar_op3): Ditto.
3880 (round_saeonly_mask_scalar_op4): Ditto.
3882 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3884 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3885 Implement -maltivec=be for vec_insert and vec_extract.
3887 2014-01-10 DJ Delorie <dj@redhat.com>
3889 * config/msp430/msp430.md (call_internal): Don't allow memory
3890 references with SP as the base register.
3891 (call_value_internal): Likewise.
3892 * config/msp430/constraints.md (Yc): New. For memory references
3893 that don't use SP as a base register.
3895 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3896 "an integer without a # prefix"
3897 * config/msp430/msp430.md (epilogue_helper): Use it.
3899 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3902 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3903 AVX512F gather builtins.
3904 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3905 on gather decls with INTEGER_TYPE masktype.
3906 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3907 directly into the builtin rather than hoisting it before loop.
3909 PR tree-optimization/59387
3910 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3911 (scev_const_prop): If folded_casts and type has undefined overflow,
3912 use force_gimple_operand instead of force_gimple_operand_gsi and
3913 for each added stmt if it is assign with
3914 arith_code_with_undefined_signed_overflow, call
3915 rewrite_to_defined_overflow.
3916 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3917 gimple-fold.h instead.
3918 (arith_code_with_undefined_signed_overflow,
3919 rewrite_to_defined_overflow): Moved to ...
3920 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3921 rewrite_to_defined_overflow): ... here. No longer static.
3922 Include gimplify-me.h.
3923 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3924 rewrite_to_defined_overflow): New prototypes.
3926 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3928 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3930 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3932 * builtins.c (get_object_alignment_2): Minor tweak.
3933 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3935 2014-01-13 Christian Bruel <christian.bruel@st.com>
3937 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3938 optimized non constant lengths.
3940 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3943 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3944 load as __atomic_load_N if possible.
3946 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3948 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3950 (rs6000_expand_builtin): Adjust call.
3952 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3955 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3956 * config/rs6000/rs6000.c: Include target-globals.h.
3957 (rs6000_set_current_function): Instead of doing target_reinit
3958 unconditionally, use save_target_globals_default_opts and
3959 restore_target_globals.
3961 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3963 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3964 (rs6000_expand_builtin): Handle mffs and mtfsf.
3965 (rs6000_init_builtins): Define mffs and mtfsf.
3966 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3967 (rs6000_mffs): New pattern.
3968 (rs6000_mtfsf): New pattern.
3970 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3972 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3973 Start narrowing with START. Apply candidate-use pair
3974 and check overall cost in narrowing.
3975 (iv_ca_prune): Pass new argument.
3977 2014-01-10 Jeff Law <law@redhat.com>
3980 * ree.c (combine_reaching_defs): Ensure the defining statement
3981 occurs before the extension when optimizing extensions with
3982 different source and destination hard registers.
3984 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3987 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3988 vtables into the type inheritance graph.
3990 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3992 PR rtl-optimization/59754
3993 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3994 modes in the REGNO != REGNO case.
3996 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3998 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4000 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4002 PR tree-optimization/59745
4003 * tree-predcom.c (tree_predictive_commoning_loop): Call
4004 free_affine_expand_cache if giving up because components is NULL.
4006 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4007 GC in payload of target_globals struct instead of allocating them on
4008 the heap and the larger structs separately using GC.
4009 * target-globals.h (struct target_globals): Make regs, hard_regs,
4010 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4011 of GTY((skip)) and change type to void *.
4012 (reset_target_globals): Cast loads from those fields to corresponding
4015 2014-01-10 Steve Ellcey <sellcey@mips.com>
4018 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4019 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4020 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4022 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4025 * aarch64-modes.def (CC_Zmode): New flags mode.
4026 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4027 represents an equality.
4028 (aarch64_get_condition_code): Handle CC_Zmode.
4029 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4031 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4033 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4034 extraction in good case.
4036 2014-01-10 Richard Biener <rguenther@suse.de>
4038 PR tree-optimization/59374
4039 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4040 checking after SLP discovery. Mark stmts not participating
4041 in any SLP instance properly.
4043 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4045 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4046 when handling a SET rtx.
4048 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4050 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4051 (cortex-a57): Likewise.
4052 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4054 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4056 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4057 non-iwmmxt builtins.
4059 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4063 * ipa-devirt.c record_target_from_binfo): Take as argument
4064 stack of binfos and lookup matching one for virtual inheritance.
4065 (possible_polymorphic_call_targets_1): Update.
4067 2014-01-10 Huacai Chen <chenhc@lemote.com>
4069 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4070 kernel strings for Loongson-2E/2F/3A.
4072 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4075 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4076 is_gimple_call before calling gimple_call_internal_p.
4078 2014-01-09 Steve Ellcey <sellcey@mips.com>
4080 * Makefile.in (TREE_FLOW_H): Remove.
4081 (TREE_SSA_H): Add file names from tree-flow.h.
4082 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4083 * tree.h: Remove tree-flow.h reference.
4084 * hash-table.h: Remove tree-flow.h reference.
4085 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4086 reference with tree-ssa-loop.h.
4088 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4090 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4091 default element-order behavior for -maltivec.
4092 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4093 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4094 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4095 when targeting big endian, at least for now.
4096 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4098 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4101 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4102 var satisfies use_register_for_decl, just take into account type
4103 alignment, rather than decl alignment.
4105 PR tree-optimization/59622
4106 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4107 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4108 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4109 Don't devirtualize for inplace at all. For targets.length () == 1,
4110 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4112 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4114 * config/i386/i386.md (cpu): Remove the unused btver1.
4116 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4118 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4120 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4123 * tree-core.h (struct target_globals): New forward declaration.
4124 (struct tree_target_option): Add globals field.
4125 * tree.h (TREE_TARGET_GLOBALS): Define.
4126 (prepare_target_option_nodes_for_pch): New prototype.
4127 * target-globals.h (struct target_globals): Define even if
4129 * tree.c (prepare_target_option_node_for_pch,
4130 prepare_target_option_nodes_for_pch): New functions.
4131 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4132 * config/i386/i386.c: Include target-globals.h.
4133 (ix86_set_current_function): Instead of doing target_reinit
4134 unconditionally, use save_target_globals_default_opts and
4135 restore_target_globals.
4137 2014-01-09 Richard Biener <rguenther@suse.de>
4139 PR tree-optimization/59715
4140 * tree-cfg.h (split_critical_edges): Declare.
4141 * tree-cfg.c (split_critical_edges): Export.
4142 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4144 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4146 * cfgexpand.c (expand_stack_vars): Optionally disable
4147 asan stack protection.
4148 (expand_used_vars): Likewise.
4149 (partition_stack_vars): Likewise.
4150 * asan.c (asan_emit_stack_protection): Optionally disable
4151 after return stack usage.
4152 (instrument_derefs): Optionally disable memory access instrumentation.
4153 (instrument_builtin_call): Likewise.
4154 (instrument_strlen_call): Likewise.
4155 (asan_protect_global): Optionally disable global variables protection.
4156 * doc/invoke.texi: Added doc for new options.
4157 * params.def: Added new options.
4158 * params.h: Likewise.
4160 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4162 PR rtl-optimization/59724
4163 * ifcvt.c (cond_exec_process_if_block): Don't call
4164 flow_find_head_matching_sequence with 0 longest_match.
4165 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4166 non-active insns if !stop_after.
4167 (try_head_merge_bb): Revert 2014-01-07 changes.
4169 2014-01-08 Jeff Law <law@redhat.com>
4171 * ree.c (get_sub_rtx): New function, extracted from...
4172 (merge_def_and_ext): Here.
4173 (combine_reaching_defs): Use get_sub_rtx.
4175 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4177 * cgraph.h (varpool_variable_node): Do not choke on null node.
4179 2014-01-08 Catherine Moore <clm@codesourcery.com>
4181 * config/mips/mips.md (simple_return): Attempt to use JRC
4183 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4185 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4187 PR rtl-optimization/59137
4188 * reorg.c (steal_delay_list_from_target): Call update_block for
4190 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4192 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4194 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4195 two duplicate entries.
4197 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4200 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4202 * config/mips/mips.c (mips_truncated_op_cost): New function.
4203 (mips_rtx_costs): Adjust test for BADDU.
4204 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4206 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4208 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4209 (*baddu_si): ...this new pattern.
4211 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4214 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4216 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4219 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4221 (expand_expr, expand_normal): Adjust.
4222 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4223 inner_reference_p. Use inner_reference_p to expand inner references.
4224 (store_expr): Adjust.
4225 * cfgexpand.c (expand_call_stmt): Adjust.
4227 2014-01-08 Rong Xu <xur@google.com>
4229 * gcov-io.c (gcov_var): Move from gcov-io.h.
4230 (gcov_position): Ditto.
4231 (gcov_is_error): Ditto.
4232 (gcov_rewrite): Ditto.
4233 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4234 only part to libgcc/libgcov.h.
4236 2014-01-08 Marek Polacek <polacek@redhat.com>
4239 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4241 2014-01-08 Marek Polacek <polacek@redhat.com>
4244 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4246 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4248 PR rtl-optimization/59649
4249 * stor-layout.c (get_mode_bounds): For BImode return
4250 0 and STORE_FLAG_VALUE.
4252 2014-01-08 Richard Biener <rguenther@suse.de>
4255 * gimple.h (is_gimple_builtin_call): Remove.
4256 (gimple_builtin_call_types_compatible_p): New.
4257 (gimple_call_builtin_p): New overload.
4258 * gimple.c (is_gimple_builtin_call): Remove.
4259 (validate_call): Rename to ...
4260 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4262 (validate_type): New static function.
4263 (gimple_call_builtin_p): New overload and adjust.
4264 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4265 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4266 (gimple_fold_stmt_to_constant_1): Likewise.
4267 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
4269 2014-01-08 Richard Biener <rguenther@suse.de>
4272 * gimplify.c (gimplify_expr): Gimplify register-register type
4273 VIEW_CONVERT_EXPRs to separate stmts.
4275 2014-01-07 Jeff Law <law@redhat.com>
4278 * ree.c (combine_set_extension): Handle case where source
4279 and destination registers in an extension insn are different.
4280 (combine_reaching_defs): Allow source and destination registers
4281 in extension to be different under limited circumstances.
4282 (add_removable_extension): Remove restriction that the
4283 source and destination registers in the extension are the same.
4284 (find_and_remove_re): Emit a copy from the extension's
4285 destination to its source after the defining insn if
4286 the source and destination registers are different.
4289 * ifcvt.c (merge_if_block): If we are merging a block with more than
4290 one successor with a block with no successors, remove any BARRIER
4291 after the second block.
4293 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
4295 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
4297 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
4300 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
4301 for 14-bit register offsets when INT14_OK_STRICT is false.
4303 2014-01-07 Roland Stigge <stigge@antcom.de>
4304 Michael Meissner <meissner@linux.vnet.ibm.com>
4307 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
4308 Only check TFmode for SPE constants. Don't check TImode or TDmode.
4310 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
4312 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
4315 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
4317 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
4318 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
4319 rtx is const0_rtx or not.
4321 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
4324 * target-globals.c (save_target_globals): Remove this_fn_optab
4326 * toplev.c: Include optabs.h.
4327 (target_reinit): Temporarily restore the global options if another
4328 set of options are in force.
4330 2014-01-07 Jakub Jelinek <jakub@redhat.com>
4332 PR rtl-optimization/58668
4333 * cfgcleanup.c (flow_find_cross_jump): Don't count
4334 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
4335 to determine what is counted.
4336 (flow_find_head_matching_sequence): Use active_insn_p to determine
4338 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
4340 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
4341 determine what is counted.
4343 PR tree-optimization/59643
4344 * tree-predcom.c (split_data_refs_to_components): If one dr is
4345 read and one write, determine_offset fails and the write isn't
4346 in the bad component, just put the read into the bad component.
4348 2014-01-07 Mike Stump <mikestump@comcast.net>
4349 Jakub Jelinek <jakub@redhat.com>
4352 * tree-core.h (struct tree_optimization_option): Change optabs
4353 type from unsigned char * to void *.
4354 * optabs.c (init_tree_optimization_optabs): Adjust
4355 TREE_OPTIMIZATION_OPTABS initialization.
4357 2014-01-06 Jakub Jelinek <jakub@redhat.com>
4360 * config/i386/i386.h (struct machine_function): Add
4361 no_drap_save_restore field.
4362 * config/i386/i386.c (ix86_save_reg): Use
4363 !cfun->machine->no_drap_save_restore instead of
4364 crtl->stack_realign_needed.
4365 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
4366 this function clears frame_pointer_needed. Set
4367 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
4368 and DRAP reg is needed.
4370 2014-01-06 Marek Polacek <polacek@redhat.com>
4373 * doc/implement-c.texi: Mention that other integer types are
4374 permitted as bit-field types in strictly conforming mode.
4376 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
4378 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
4381 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
4383 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
4385 2014-01-06 Martin Jambor <mjambor@suse.cz>
4388 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
4390 * ipa-prop.c (ipa_print_node_params): Fix indentation.
4392 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
4396 * var-tracking.c (add_stores): Preserve the value of the source even if
4397 we don't record the store.
4399 2014-01-06 Terry Guo <terry.guo@arm.com>
4401 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
4403 2014-01-05 Iain Sandoe <iain@codesourcery.com>
4406 * config/darwin.c (darwin_function_section): Adjust return values to
4407 correspond to optimisation changes made in r206070.
4409 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
4411 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
4412 from prefetch_block tune setting.
4413 (nocona_cost): Correct size of prefetch block to 64.
4415 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
4417 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
4418 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
4419 used to save the static chain register in the computation of the offset
4420 from which the FP registers need to be restored.
4422 2014-01-04 Jakub Jelinek <jakub@redhat.com>
4424 PR tree-optimization/59519
4425 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
4426 ICE if get_current_def (current_new_name) is already non-NULL, as long
4427 as it is a phi result of some other phi in *new_exit_bb that has
4430 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
4431 or vmovdqu* for misaligned_operand.
4432 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
4433 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
4434 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
4435 aligned_mem for AVX512F masked aligned load and store builtins and for
4438 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
4440 PR tree-optimization/59651
4441 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
4442 Address range for negative step should be added by TYPE_SIZE_UNIT.
4444 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
4446 * config/m68k/m68k.c (handle_move_double): Handle pushes with
4447 overlapping registers also for registers other than the stack pointer.
4449 2014-01-03 Marek Polacek <polacek@redhat.com>
4452 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
4455 2014-01-03 Jakub Jelinek <jakub@redhat.com>
4458 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
4461 * config/i386/i386.md (MODE_SIZE): New mode attribute.
4462 (push splitter): Use <P:MODE_SIZE> instead of
4463 GET_MODE_SIZE (<P:MODE>mode).
4464 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
4465 (mov -1, reg peephole2): Likewise.
4466 * config/i386/sse.md (*mov<mode>_internal,
4467 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
4468 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
4469 *<code><mode>3, *andnot<mode>3<mask_name>,
4470 <mask_codefor><code><mode>3<mask_name>): Likewise.
4471 * config/i386/subst.md (mask_mode512bit_condition,
4472 sd_mask_mode512bit_condition): Likewise.
4474 2014-01-02 Xinliang David Li <davidxl@google.com>
4476 PR tree-optimization/59303
4477 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
4478 (dump_predicates): Better output format.
4479 (pred_equal_p): New function.
4480 (is_neq_relop_p): Ditto.
4481 (is_neq_zero_form_p): Ditto.
4482 (pred_expr_equal_p): Ditto.
4483 (pred_neg_p): Ditto.
4484 (simplify_pred): Ditto.
4485 (simplify_preds_2): Ditto.
4486 (simplify_preds_3): Ditto.
4487 (simplify_preds_4): Ditto.
4488 (simplify_preds): Ditto.
4490 (push_to_worklist): Ditto.
4491 (get_pred_info_from_cmp): Ditto.
4492 (is_degenerated_phi): Ditto.
4493 (normalize_one_pred_1): Ditto.
4494 (normalize_one_pred): Ditto.
4495 (normalize_one_pred_chain): Ditto.
4496 (normalize_preds): Ditto.
4497 (normalize_cond_1): Remove function.
4498 (normalize_cond): Ditto.
4499 (is_gcond_subset_of): Ditto.
4500 (is_subset_of_any): Ditto.
4501 (is_or_set_subset_of): Ditto.
4502 (is_and_set_subset_of): Ditto.
4503 (is_norm_cond_subset_of): Ditto.
4504 (pred_chain_length_cmp): Ditto.
4505 (convert_control_dep_chain_into_preds): Type change.
4506 (find_predicates): Ditto.
4507 (find_def_preds): Ditto.
4508 (destroy_predicates_vecs): Ditto.
4509 (find_matching_predicates_in_rest_chains): Ditto.
4510 (use_pred_not_overlap_with_undef_path_pred): Ditto.
4511 (is_pred_expr_subset): Ditto.
4512 (is_pred_chain_subset_of): Ditto.
4513 (is_included_in): Ditto.
4514 (is_superset_of): Ditto.
4516 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4518 Update copyright years.
4520 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
4522 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
4523 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
4524 config/arc/arc.md, config/arc/arc.opt,
4525 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
4526 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
4527 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
4528 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
4529 config/linux-protos.h, config/linux.c, config/winnt-c.c,
4530 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
4531 vtable-verify.c, vtable-verify.h: Use the standard form for the
4534 2014-01-02 Tobias Burnus <burnus@net-b.de>
4536 * gcc.c (process_command): Update copyright notice dates.
4537 * gcov-dump.c: Ditto.
4539 * doc/cpp.texi: Bump @copying's copyright year.
4540 * doc/cppinternals.texi: Ditto.
4541 * doc/gcc.texi: Ditto.
4542 * doc/gccint.texi: Ditto.
4543 * doc/gcov.texi: Ditto.
4544 * doc/install.texi: Ditto.
4545 * doc/invoke.texi: Ditto.
4547 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4549 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
4551 2014-01-01 Jakub Jelinek <jakub@redhat.com>
4553 * config/i386/sse.md (*mov<mode>_internal): Guard
4554 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
4556 PR rtl-optimization/59647
4557 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
4558 new_rtx into UNSIGNED_FLOAT rtxes.
4560 Copyright (C) 2014 Free Software Foundation, Inc.
4562 Copying and distribution of this file, with or without modification,
4563 are permitted in any medium without royalty provided the copyright
4564 notice and this notice are preserved.