1 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
3 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
4 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
6 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
7 Use FMAMODE_NOVF512 mode iterator.
8 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
9 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
10 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
11 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
13 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
14 Use VF_128_256 mode iterator.
15 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
18 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
20 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
21 static chain if needed.
23 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
26 * lra-constraints.c (index_part_to_reg): New.
27 (process_address): Use it.
29 2014-03-27 Jeff Law <law@redhat.com>
30 Jakub Jelinek <jakub@redhat.com>
33 * expr.c (do_tablejump): Use simplify_gen_binary rather than
34 gen_rtx_{PLUS,MULT} to build up the address expression.
36 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
37 creating non-canonical RTL.
39 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
42 * ipa-inline.c (want_inline_small_function_p): Short circuit large
43 functions; reorganize to make cheap checks first.
44 (inline_small_functions): Do not estimate growth when dumping;
46 * ipa-inline.h (inline_summary): Add min_size.
47 (growth_likely_positive): New function.
48 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
49 (set_cond_stmt_execution_predicate): Cleanup.
50 (estimate_edge_size_and_time): Compute min_size.
51 (estimate_calls_size_and_time): Likewise.
52 (estimate_node_size_and_time): Likewise.
53 (inline_update_overall_summary): Update min_size.
54 (do_estimate_edge_time): Likewise.
55 (do_estimate_edge_size): Update.
56 (do_estimate_edge_hints): Update.
57 (growth_likely_positive): New function.
59 2014-03-28 Jakub Jelinek <jakub@redhat.com>
62 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
63 also if addr has VOIDmode.
65 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
67 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
68 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
70 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
71 instructions as well as AdvancedSIMD loads.
73 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
75 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
77 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
78 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
79 crypto_aese, crypto_aesmc. Move to types.md.
80 * config/arm/types.md (crypto_aes): Split into crypto_aese,
82 * config/arm/iterators.md (crypto_type): Likewise.
84 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
86 * cgraph.c: Include expr.h and tree-dfa.h.
87 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
90 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
93 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
94 regs from checking multi-reg pseudos.
96 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
98 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
100 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
102 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
103 if it would clobber the stack pointer, even temporarily.
105 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
107 * mode-switching.c: Make small adjustments to the top comment.
109 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
111 * config/rs6000/constraints.md (wD constraint): New constraint to
112 match the constant integer to get the top DImode/DFmode out of a
113 vector in a VSX register.
115 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
116 match the constant integer to get the top DImode/DFmode out of a
117 vector in a VSX register.
119 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
122 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
125 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
126 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
128 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
129 Optimize vec_extract of 64-bit values, where the value being
130 extracted is in the top word, where we can use scalar
131 instructions. Add direct move and store support. Combine the big
132 endian/little endian vector select load support into a single insn.
133 (vsx_extract_<mode>_internal1): Likewise.
134 (vsx_extract_<mode>_internal2): Likewise.
135 (vsx_extract_<mode>_load): Likewise.
136 (vsx_extract_<mode>_store): Likewise.
137 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
138 combined into vsx_extract_<mode>_load.
139 (vsx_extract_<mode>_one_le): Likewise.
141 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
142 define the top 64-bit vector element.
144 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
147 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
148 Document vec_vbpermq builtin.
151 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
152 enable use of xxsldwi and xxpermdi builtin functions.
153 (vec_xxpermdi): Likewise.
155 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
156 Document use of vec_xxsldwi and vec_xxpermdi builtins.
158 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
160 PR rtl-optimization/60650
161 * lra-asign.c (find_hard_regno_for, spill_for): Add parameter
163 (find_spills_for): New.
164 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
165 Spill all pseudos on the second iteration.
167 2014-03-27 Marek Polacek <polacek@redhat.com>
170 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
173 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
175 * config/s390/s390.c (s390_can_use_return_insn): Check for
176 call-saved FPRs on 31 bit.
178 2014-03-27 Jakub Jelinek <jakub@redhat.com>
181 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
182 if they need regimplification, just drop them instead of
183 calling gimple_regimplify_operands on them.
185 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
188 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
189 (aarch64_frame_pointer_required): Adjust logic.
190 (aarch64_can_eliminate): Adjust logic.
191 (aarch64_override_options_after_change): Adjust logic.
193 2014-03-27 Dehao Chen <dehao@google.com>
195 * ipa-inline.c (early_inliner): Update node's inline info.
197 2014-03-26 Dehao Chen <dehao@google.com>
199 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
200 compiler inserted conditional jumps for NAN float check.
202 2014-03-26 Jakub Jelinek <jakub@redhat.com>
204 * ubsan.h (ubsan_create_data): Change second argument's type
205 to const location_t *.
206 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
208 (ubsan_create_data): Change second argument to const location_t *PLOC.
209 Create Loc field whenever PLOC is non-NULL.
210 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
211 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
215 * real.c (real_to_integer2): Change type of low to UHWI.
217 2014-03-26 Tobias Burnus <burnus@net-b.de>
219 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
220 (CILK_SELF_SPECS): New define.
221 (driver_self_specs): Use it.
223 2014-03-26 Richard Biener <rguenther@suse.de>
225 * tree-pretty-print.c (percent_K_format): Implement special
226 case for LTO and its stripped down BLOCK tree.
228 2014-03-26 Jakub Jelinek <jakub@redhat.com>
231 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
233 * tree-vrp.c (simplify_internal_call_using_ranges): If only
234 one range is range_int_cst_p, but not both, at least optimize
235 addition/subtraction of 0 and multiplication by 0 or 1.
236 * gimple-fold.c (gimple_fold_call): Fold
237 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
238 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
239 INTEGER_CSTs, try to fold at least x * 0 and y - y.
241 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
243 PR rtl-optimization/60452
244 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
245 <case REG>: Return 1 for invalid offsets from the frame pointer.
247 2014-03-26 Marek Polacek <polacek@redhat.com>
250 * doc/extend.texi (C Extensions): Mention variable-length arrays in
253 2014-03-26 Marek Polacek <polacek@redhat.com>
256 * doc/extend.texi (Designated Inits): Describe what happens to omitted
259 2014-03-26 Marek Polacek <polacek@redhat.com>
262 * ira-color.c (update_conflict_hard_regno_costs): Perform the
263 multiplication in unsigned type.
265 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
267 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
269 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
271 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
273 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
276 * cif-code.def (UNREACHABLE) New code.
277 * ipa-inline.c (inline_small_functions): Skip edges to
278 __builtlin_unreachable.
279 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
280 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
281 predicate to __bulitin_unreachable.
282 (set_cond_stmt_execution_predicate): Fix issue when
283 invert_tree_comparison returns ERROR_MARK.
284 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
285 propagate to inline clones.
286 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
288 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
289 * cgraphclones.c (cgraph_clone_node): If call destination is already
290 ureachable, do not redirect it back.
291 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
294 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
296 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
297 Do not modify inline clones.
299 2014-03-25 Jakub Jelinek <jakub@redhat.com>
301 * config/i386/i386.md (general_sext_operand): New mode attr.
302 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
303 don't generate (sign_extend (const_int)).
304 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
305 operands[2]. Use We constraint instead of <i> and
306 <general_sext_operand> predicate instead of <general_operand>.
307 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
308 * config/i386/constraints.md (We): New constraint.
309 * config/i386/predicates.md (x86_64_sext_operand,
310 sext_operand): New predicates.
312 2014-03-25 Martin Jambor <mjambor@suse.cz>
315 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
316 inconsistent devirtualizations to __builtin_unreachable.
318 2014-03-25 Marek Polacek <polacek@redhat.com>
321 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
323 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
325 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
326 order of elements for big-endian.
328 2014-03-25 Richard Biener <rguenther@suse.de>
331 * gimplify-me.c (gimple_regimplify_operands): Update the
334 2014-03-25 Martin Jambor <mjambor@suse.cz>
337 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
338 (lto_output_varpool_node): Likewise.
339 (input_overwrite_node): Likewise.
340 (input_varpool_node): Likewise.
342 2014-03-25 Richard Biener <rguenther@suse.de>
344 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
347 2014-03-25 Jakub Jelinek <jakub@redhat.com>
349 * combine.c (simplify_compare_const): Add MODE argument.
350 Handle mode_width 0 as very large mode_width.
351 (try_combine, simplify_comparison): Adjust callers.
353 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
354 type to avoid signed integer overflow.
355 * explow.c (plus_constant): Likewise.
357 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
359 * doc/generic.texi: Correct typos.
361 2014-03-24 Tobias Burnus <burnus@net-b.de>
363 * doc/invoke.texi (-flto): Expand section about
364 using static libraries with LTO.
366 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
368 PR rtl-optimization/60501
369 * optabs.def (addptr3_optab): New optab.
370 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
371 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
372 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
374 * lra.c (emit_add3_insn): Use the addptr pattern if available.
376 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
378 2014-03-24 Ulrich Drepper <drepper@gmail.com>
380 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
383 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
384 (_mm256_undefined_ps): Define.
385 (_mm256_undefined_pd): Define.
386 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
387 (_mm_undefined_pd): Define.
388 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
389 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
390 (_mm512_undefined_ps): Define.
391 (_mm512_undefined_pd): Define.
392 Use _mm*_undefined_*.
393 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
395 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
397 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
398 (lshr_simd): DI mode added.
399 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
400 (aarch64_ushr_simddi): Likewise.
401 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
402 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
403 (vshrd_n_u64): Likewise.
405 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
407 * Makefile.in (s-macro_list): Depend on cc1.
409 2014-03-23 Teresa Johnson <tejohnson@google.com>
411 * ipa-utils.c (ipa_print_order): Use specified dump file.
413 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
415 PR rtl-optimization/60601
416 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
418 * gcc.c (eval_spec_function): Initialize save_growing_value.
420 2014-03-22 Jakub Jelinek <jakub@redhat.com>
423 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
424 code == MINUS_EXPR, never swap op0 with op1.
426 * toplev.c (init_local_tick): Avoid signed integer multiplication
428 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
429 shift by first operand's bitsize.
431 2014-03-21 Jakub Jelinek <jakub@redhat.com>
434 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
436 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
437 TARGET_ISA_64BIT_P(x).
439 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
441 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
442 pattern for vector nor instead of subtract from splat(-1).
443 (altivec_expand_vec_perm_const_le): Likewise.
445 2014-03-21 Richard Henderson <rth@twiddle.net>
448 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
449 related insns after epilogue_completed.
451 2014-03-21 Martin Jambor <mjambor@suse.cz>
454 * cgraph.h (symtab_node): New flag body_removed.
455 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
456 when removing bodies.
457 * symtab.c (dump_symtab_base): Dump body_removed flag.
458 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
459 had their bodies removed.
461 2014-03-21 Martin Jambor <mjambor@suse.cz>
464 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
467 2014-03-21 Richard Biener <rguenther@suse.de>
469 PR tree-optimization/60577
470 * tree-core.h (struct tree_base): Document nothrow_flag use
471 in VAR_DECL_NONALIASED.
472 * tree.h (VAR_DECL_NONALIASED): New.
473 (may_be_aliased): Adjust.
474 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
476 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
478 * expr.c (expand_expr_real_1): Remove outdated comment.
480 2014-03-20 Jakub Jelinek <jakub@redhat.com>
483 * ira.c (adjust_cleared_regs): Call copy_rtx on
484 *reg_equiv[REGNO (loc)].src_p before passing it to
485 simplify_replace_fn_rtx.
488 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
489 into CONST, put pic register as first operand of PLUS. Use
490 gen_const_mem for both 32-bit and 64-bit PIC got loads.
492 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
494 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
496 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
498 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
499 around for store forwarding issue in the FPU on the UT699.
500 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
501 loads and operations if -mfix-ut699 is specified.
502 (divtf3_hq): Tweak attribute.
503 (sqrttf2_hq): Likewise.
505 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
507 * calls.c (store_one_arg): Remove incorrect const qualification on the
508 type of the temporary.
509 * cfgexpand.c (expand_return): Likewise.
510 * expr.c (expand_constructor): Likewise.
511 (expand_expr_real_1): Likewise.
513 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
515 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
518 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
521 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
523 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
525 * config/arm/aarch-common-protos.h
526 (alu_cost_table): Fix spelling of "extend".
527 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
529 2014-03-19 Richard Biener <rguenther@suse.de>
532 * tree-core.h (tree_type_common): Re-order pointer members
533 to reduce recursion depth during GC walks.
535 2014-03-19 Marek Polacek <polacek@redhat.com>
538 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
541 2014-03-19 Richard Biener <rguenther@suse.de>
544 * lto-streamer-in.c (input_function): In WPA stage do not drop
547 2014-03-19 Jakub Jelinek <jakub@redhat.com>
549 PR tree-optimization/60559
550 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
551 with build_zero_cst assignment.
553 2014-03-18 Kai Tietz <ktietz@redhat.com>
555 PR rtl-optimization/56356
556 * sdbout.c (sdbout_parms): Verify that parms'
557 incoming argument is valid.
558 (sdbout_reg_parms): Likewise.
560 2014-03-18 Richard Henderson <rth@redhat.com>
563 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
564 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
565 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
567 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
569 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
570 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
571 Italicize plugin event names in description. Explain that
572 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
573 Remind that no GCC functions should be called after PLUGIN_FINISH.
574 Explain what pragmas with expansion are.
576 2014-03-18 Martin Liska <mliska@suse.cz>
578 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
579 gimple call statement is update.
580 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
581 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
583 2014-03-18 Jakub Jelinek <jakub@redhat.com>
586 * ubsan.c (ubsan_instrument_unreachable): Call
587 initialize_sanitizer_builtins.
588 (ubsan_pass): Likewise.
591 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
592 varpool_finalize_decl instead of rest_of_decl_compilation.
594 2014-03-18 Richard Biener <rguenther@suse.de>
596 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
597 by using bitmap_and_compl instead of bitmap_and_compl_into.
598 (df_rd_transfer_function): Likewise.
600 2014-03-18 Richard Biener <rguenther@suse.de>
602 * doc/lto.texi (fresolution): Fix typo.
604 2014-03-18 Richard Biener <rguenther@suse.de>
606 * doc/invoke.texi (flto): Update for changes in 4.9.
608 2014-03-18 Richard Biener <rguenther@suse.de>
610 * doc/loop.texi: Remove section on the removed lambda framework.
611 Update loop docs with recent changes in preserving loop structure.
613 2014-03-18 Richard Biener <rguenther@suse.de>
615 * doc/lto.texi (-fresolution): Document.
617 2014-03-18 Richard Biener <rguenther@suse.de>
619 * doc/contrib.texi: Adjust my name.
621 2014-03-18 Jakub Jelinek <jakub@redhat.com>
624 * internal-fn.c: Include diagnostic-core.h.
625 (expand_BUILTIN_EXPECT): New function.
626 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
627 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
628 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
629 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
631 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
632 Revert 3 argument __builtin_expect code.
633 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
634 * gimple-fold.c (gimple_fold_call): Likewise.
635 * tree.h (fold_builtin_expect): New prototype.
636 * builtins.c (build_builtin_expect_predicate): Add predictor
637 argument, if non-NULL, create 3 argument __builtin_expect.
638 (fold_builtin_expect): No longer static. Add ARG2 argument,
639 pass it through to build_builtin_expect_predicate.
640 (fold_builtin_2): Adjust caller.
641 (fold_builtin_3): Handle BUILT_IN_EXPECT.
642 * internal-fn.def (BUILTIN_EXPECT): New.
644 2014-03-18 Tobias Burnus <burnus@net-b.de>
647 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
648 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
649 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
651 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
654 * predict.c (combine_predictions_for_bb): Fix up formatting.
655 (expr_expected_value_1, expr_expected_value): Add predictor argument,
656 fill what it points to if non-NULL.
657 (tree_predict_by_opcode): Adjust caller, use the predictor.
658 * predict.def (PRED_COMPARE_AND_SWAP): Add.
660 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
662 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
663 proper constant for the store mode.
665 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
667 * symtab.c (change_decl_assembler_name): Fix transparent alias
670 2014-03-16 Renlin Li <Renlin.Li@arm.com>
672 * config/aarch64/aarch64.c: Correct the comments about the
673 aarch64 stack layout.
675 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
677 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
678 check for GF_OMP_FOR_KIND_FOR.
680 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
682 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
683 ymm and zmm register names.
685 2014-03-17 Jakub Jelinek <jakub@redhat.com>
688 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
689 note creation for the 2010-08-31 changes.
691 2014-03-17 Marek Polacek <polacek@redhat.com>
694 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
695 as -fno-tree-loop-vectorize.
696 (expand_omp_simd): Likewise.
698 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
700 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
701 (eligible_for_call_delay): New prototype.
702 * config/sparc/sparc.c (tls_call_delay): Rename into...
703 (eligible_for_call_delay): ...this. Return false if the instruction
704 cannot be put in the delay slot of a branch.
705 (eligible_for_restore_insn): Simplify.
706 (eligible_for_return_delay): Return false if the instruction cannot be
707 put in the delay slot of a branch and simplify.
708 (eligible_for_sibcall_delay): Return false if the instruction cannot be
709 put in the delay slot of a branch.
710 * config/sparc/sparc.md (fix_ut699): New attribute.
711 (tls_call_delay): Delete.
712 (in_call_delay): Reimplement.
713 (eligible_for_sibcall_delay): Rename into...
714 (in_sibcall_delay): ...this.
715 (eligible_for_return_delay): Rename into...
716 (in_return_delay): ...this.
717 (in_branch_delay): Reimplement.
718 (in_uncond_branch_delay): Delete.
719 (in_annul_branch_delay): Delete.
721 2014-03-14 Richard Henderson <rth@redhat.com>
724 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
725 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
726 (*floathi<X87MODEF>2_i387_with_temp): Remove.
727 (floathi splitters): Remove.
728 (float<SWI48x>xf2): New pattern.
729 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
730 code that tried to handle DImode for 32-bit, but which was excluded
731 by the pattern's condition. Drop allocation of stack temporary.
732 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
733 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
734 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
735 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
736 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
737 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
738 (*float<SWI48><MODEF>2_sse_interunit): Remove.
739 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
740 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
741 (*float<SWI48x><X87MODEF>2_i387): Remove.
742 (all float _with_temp splitters): Remove.
743 (*float<SWI48x><MODEF>2_i387): New pattern.
744 (*float<SWI48><MODEF>2_sse): New pattern.
745 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
746 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
748 2014-03-14 Jakub Jelinek <jakub@redhat.com>
749 Marek Polacek <polacek@redhat.com>
752 * common.opt (dump_base_name_prefixed): New Variable.
753 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
754 if x_dump_base_name_prefixed is already set, set it at the end.
756 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
758 PR rtl-optimization/60508
759 * lra-constraints.c (get_reload_reg): Add new parameter
761 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
762 Pass the new parameter values.
764 2014-03-14 Richard Biener <rguenther@suse.de>
766 * common.opt: Revert unintented changes from r205065.
769 2014-03-14 Richard Biener <rguenther@suse.de>
772 * cfghooks.c (split_block): Properly adjust all loops the
773 block was a latch of.
775 2014-03-14 Martin Jambor <mjambor@suse.cz>
778 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
781 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
784 * config/avr/avr.c (avr_set_current_function): Pass function name
785 through default_strip_name_encoding before sanity checking instead
786 of skipping the first char of the assembler name.
788 2014-03-13 Richard Henderson <rth@redhat.com>
791 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
792 (ix86_force_to_memory, ix86_free_from_memory): Remove.
793 * config/i386/i386-protos.h: Likewise.
794 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
795 in the expander instead of a splitter.
796 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
797 any possibility of requiring a memory.
798 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
799 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
800 (fp branch splitters): Update for ix86_split_fp_branch.
801 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
802 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
803 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
804 (*fop_<MODEF>_2_i387): Remove f/r alternative.
805 (*fop_<MODEF>_3_i387): Likewise.
806 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
807 (splitters for the fop_* register patterns): Remove.
808 (fscalexf4_i387): Rename from *fscalexf4_i387.
809 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
811 2014-03-13 Jakub Jelinek <jakub@redhat.com>
813 PR tree-optimization/59779
814 * tree-dfa.c (get_ref_base_and_extent): Use double_int
815 type for bitsize and maxsize instead of HOST_WIDE_INT.
817 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
819 PR rtl-optimization/57320
820 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
821 the CFG after thread_prologue_and_epilogue_insns.
823 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
825 PR rtl-optimization/57189
826 * lra-constraints.c (process_alt_operands): Disfavor spilling
829 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
831 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
833 2014-03-13 Jakub Jelinek <jakub@redhat.com>
835 PR tree-optimization/59025
837 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
838 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
840 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
843 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
844 calls of avr_out_plus_1.
846 2014-03-13 Bin Cheng <bin.cheng@arm.com>
848 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
849 BB's single pred and update the father loop's latch info later.
851 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
853 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
857 (VEC_base): Likewise.
858 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
859 registers, we need to swap double words in little endian mode.
861 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
862 to be a container mode for 128-bit integer operations added in ISA
863 2.07. Unlike TImode and PTImode, the preferred register set is
864 the Altivec/VMX registers for the 128-bit operations.
866 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
868 (rs6000_split_128bit_ok_p): Likewise.
870 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
871 macros for creating ISA 2.07 normal and overloaded builtin
872 functions with 3 arguments.
873 (BU_P8V_OVERLOAD_3): Likewise.
874 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
875 for use as overloaded functions.
876 (VPERM_1TI_UNS): Likewise.
877 (VSEL_1TI): Likewise.
878 (VSEL_1TI_UNS): Likewise.
879 (ST_INTERNAL_1ti): Likewise.
880 (LD_INTERNAL_1ti): Likewise.
881 (XXSEL_1TI): Likewise.
882 (XXSEL_1TI_UNS): Likewise.
883 (VPERM_1TI): Likewise.
884 (VPERM_1TI_UNS): Likewise.
885 (XXPERMDI_1TI): Likewise.
887 (LXVD2X_V1TI): Likewise.
888 (STXVD2X_V1TI): Likewise.
889 (VEC_INIT_V1TI): Likewise.
890 (VEC_SET_V1TI): Likewise.
891 (VEC_EXT_V1TI): Likewise.
892 (EQV_V1TI): Likewise.
893 (NAND_V1TI): Likewise.
894 (ORC_V1TI): Likewise.
895 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
896 added in ISA 2.07. Add both normal 'altivec' builtins, and the
900 (VADDEUQM): Likewise.
901 (VADDECUQ): Likewise.
902 (VSUBEUQM): Likewise.
903 (VSUBECUQ): Likewise.
905 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
906 __int128_t and __uint128_t types.
907 (__uint128_type): Likewise.
908 (altivec_categorize_keyword): Add support for vector __int128_t,
909 vector __uint128_t, vector __int128, and vector unsigned __int128
910 as a container type for TImode operations that need to be done in
911 VSX/Altivec registers.
912 (rs6000_macro_to_expand): Likewise.
913 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
914 to support 128-bit integer instructions vaddcuq, vadduqm,
915 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
916 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
918 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
919 for V1TImode, and set up preferences to use VSX/Altivec registers.
920 Setup VSX reload handlers.
921 (rs6000_debug_reg_global): Likewise.
922 (rs6000_init_hard_regno_mode_ok): Likewise.
923 (rs6000_preferred_simd_mode): Likewise.
924 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
925 (easy_altivec_constant): Likewise.
926 (output_vec_const_move): Likewise.
927 (rs6000_expand_vector_set): Convert V1TImode set and extract to
929 (rs6000_expand_vector_extract): Likewise.
930 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
932 (rs6000_const_vec): Add support for V1TImode.
933 (rs6000_emit_le_vsx_load): Swap double words when loading or
934 storing TImode/V1TImode.
935 (rs6000_emit_le_vsx_store): Likewise.
936 (rs6000_emit_le_vsx_move): Likewise.
937 (rs6000_emit_move): Add support for V1TImode.
938 (altivec_expand_ld_builtin): Likewise.
939 (altivec_expand_st_builtin): Likewise.
940 (altivec_expand_vec_init_builtin): Likewise.
941 (altivec_expand_builtin): Likewise.
942 (rs6000_init_builtins): Add support for V1TImode type. Add
943 support for ISA 2.07 128-bit integer builtins. Define type names
944 for the VSX/Altivec vector types.
945 (altivec_init_builtins): Add support for overloaded vector
946 functions with V1TImode type.
947 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
948 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
950 (rs6000_split_128bit_ok_p): Likewise.
951 (rs6000_handle_altivec_attribute): Create V1TImode from vector
952 __int128_t and vector __uint128_t.
954 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
962 (VS_scalar): Likewise.
963 (VS_double): Likewise.
964 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
966 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
967 we support the ISA 2.07 128-bit integer arithmetic instructions.
968 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
969 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
970 and TImode types for use with the builtin functions.
971 (V1TI_type_node): Likewise.
972 (unsigned_V1TI_type_node): Likewise.
973 (intTI_type_internal_node): Likewise.
974 (uintTI_type_internal_node): Likewise.
976 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
977 128-bit builtin functions.
978 (UNSPEC_VADDEUQM): Likewise.
979 (UNSPEC_VADDECUQ): Likewise.
980 (UNSPEC_VSUBCUQ): Likewise.
981 (UNSPEC_VSUBEUQM): Likewise.
982 (UNSPEC_VSUBECUQ): Likewise.
983 (VM): Add V1TImode to vector mode iterators.
986 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
987 (altivec_vaddcuq): Likewise.
988 (altivec_vsubuqm): Likewise.
989 (altivec_vsubcuq): Likewise.
990 (altivec_vaddeuqm): Likewise.
991 (altivec_vaddecuq): Likewise.
992 (altivec_vsubeuqm): Likewise.
993 (altivec_vsubecuq): Likewise.
995 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
997 (BOOL_128): Likewise.
998 (BOOL_REGS_OUTPUT): Likewise.
999 (BOOL_REGS_OP1): Likewise.
1000 (BOOL_REGS_OP2): Likewise.
1001 (BOOL_REGS_UNARY): Likewise.
1002 (BOOL_REGS_AND_CR0): Likewise.
1004 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1005 128-bit integer builtin support.
1006 (vec_vadduqm): Likewise.
1007 (vec_vaddecuq): Likewise.
1008 (vec_vaddeuqm): Likewise.
1009 (vec_vsubecuq): Likewise.
1010 (vec_vsubeuqm): Likewise.
1011 (vec_vsubcuq): Likewise.
1012 (vec_vsubuqm): Likewise.
1014 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1015 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1016 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1017 128-bit integer add/subtract to ISA 2.07.
1019 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1021 * config/arc/arc.c (arc_predicate_delay_insns):
1022 Fix third argument passed to conditionalize_nonjump.
1024 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1026 * config/aarch64/aarch64-builtins.c
1027 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1028 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1029 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1030 instead of __builtin_lfloor.
1031 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1033 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1035 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1036 (tree_ssa_ifcombine_bb_1): New function.
1037 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1038 is an empty forwarder block to then_bb or vice versa and then_bb
1039 and else_bb are effectively swapped.
1041 2014-03-12 Christian Bruel <christian.bruel@st.com>
1044 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1045 REG_CFA_DEF_CFA note.
1046 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1047 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1049 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1051 PR tree-optimization/60454
1052 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1054 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1056 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1057 Do not define target_cpu_default2 to generic.
1058 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1059 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1060 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1062 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1063 Marc Glisse <marc.glisse@inria.fr>
1065 PR tree-optimization/60502
1066 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1067 instead of build_low_bits_mask.
1069 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1072 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1073 if there are multiple uses, but op doesn't live on E edge.
1074 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1075 clobber stmts before __builtin_unreachable.
1077 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1079 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1080 hard_frame_pointer_rtx.
1081 * cse.c (cse_insn): Remove volatile check.
1082 * cselib.c (cselib_process_insn): Likewise.
1083 * dse.c (scan_insn): Likewise.
1085 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1087 * config/arc/arc.c (conditionalize_nonjump): New function,
1089 (arc_ifcvt): ... this.
1090 (arc_predicate_delay_insns): Use it.
1092 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1094 * config/arc/predicates.md (extend_operand): During/after reload,
1095 allow const_int_operand.
1096 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1097 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1098 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1100 (umulsi3_highpart_i): Likewise.
1102 2014-03-11 Richard Biener <rguenther@suse.de>
1104 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1105 Add asserts to guard possible wrong-code bugs.
1107 2014-03-11 Richard Biener <rguenther@suse.de>
1109 PR tree-optimization/60429
1110 PR tree-optimization/60485
1111 * tree-ssa-structalias.c (set_union_with_increment): Properly
1112 take into account all fields that overlap the shifted vars.
1113 (do_sd_constraint): Likewise.
1114 (do_ds_constraint): Likewise.
1115 (get_constraint_for_ptr_offset): Likewise.
1117 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1119 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1120 (nios2_compute_frame_layout):
1121 Add calculation of cfun->machine->fp_save_offset.
1122 (nios2_expand_prologue): Correct setting of frame pointer register
1124 (nios2_expand_epilogue): Update recovery of stack pointer from
1125 frame pointer accordingly.
1126 (nios2_initial_elimination_offset): Update calculation of offset
1127 for eliminating to HARD_FRAME_POINTER_REGNUM.
1129 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1132 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1133 cgraph_get_create_node on VAR_DECLs.
1135 2014-03-10 Richard Biener <rguenther@suse.de>
1138 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1140 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1142 * config/vms/vms.opt (vms_float_format): New variable.
1144 2014-03-08 Tobias Burnus <burnus@net-b.de>
1146 * doc/invoke.texi (-fcilkplus): Update implementation status.
1148 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1149 Richard Biener <rguenther@suse.de>
1151 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1152 consistently accross all TUs.
1153 (run_gcc): Enable -fshort-double automatically at link at link-time
1154 and disallow override.
1156 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1159 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1160 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1161 if they can't be used.
1163 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1165 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1166 for Solaris 11/x86 ld.
1167 * configure: Regenerate.
1169 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1171 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1172 (LIB_TLS_SPEC): Save as ld_tls_libs.
1173 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1174 (HAVE_AS_IX86_TLSLDM): New test.
1175 * configure, config.in: Regenerate.
1176 * config/i386/i386.c (legitimize_tls_address): Fall back to
1177 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1178 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1179 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1180 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1182 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1184 * common.opt (fira-loop-pressure): Mark as optimization.
1186 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1188 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1189 an OpenMP mappable type.
1191 2014-03-06 Matthias Klose <doko@ubuntu.com>
1193 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1194 MULTILIB_OSDIRNAMES is not defined.
1196 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1197 Meador Inge <meadori@codesourcery.com>
1200 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1201 (arm_legitimize_address): Call legitimize_tls_address for any
1202 arm_tls_referenced_p expression, handle constant addend. Call it
1203 before testing for !TARGET_ARM.
1204 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1206 2014-03-06 Richard Biener <rguenther@suse.de>
1212 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1214 * tree-streamer.c (record_common_node): Assert we don't record
1215 nodes with type double.
1216 (preload_common_node): Skip type double, complex double and double
1217 pointer since it is now frontend dependent due to fshort-double option.
1219 2014-03-06 Richard Biener <rguenther@suse.de>
1221 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1222 or -fno-lto is specified and the linker has full plugin support.
1223 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1224 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1225 * lto-wrapper.c (merge_and_complain): Merge compile-time
1226 optimization levels.
1227 (run_gcc): And pass it through to the link options.
1229 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1233 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1235 * cselib.c (remove_useless_values): Skip to avoid quadratic
1236 behavior if the condition moved from...
1237 (cselib_process_insn): ... here holds.
1239 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1242 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1243 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1246 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1247 (TM_H): Add x86-tune.def.
1249 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1251 * config/aarch64/aarch64.c (generic_tunings):
1252 Use cortexa57_extra_costs.
1254 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1257 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1258 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1259 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1262 2014-03-04 Heiher <r@hev.cc>
1264 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1265 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1267 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1269 * config/i386/predicates.md (const2356_operand): Change to ...
1270 (const2367_operand): ... this.
1271 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1273 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1274 (*avx512pf_scatterpf<mode>sf): Ditto.
1275 (avx512pf_scatterpf<mode>df): Ditto.
1276 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1277 (*avx512pf_scatterpf<mode>df): Ditto.
1278 * config/i386/i386.c (ix86_expand_builtin): Update
1279 incorrect hint operand error message.
1281 2014-03-04 Richard Biener <rguenther@suse.de>
1283 * lto-section-in.c (lto_get_section_data): Fix const cast.
1285 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1287 * tree-streamer.c (record_common_node): Assert we don't record
1288 nodes with type double.
1289 (preload_common_node): Skip type double, complex double and double
1290 pointer since it is now frontend dependent due to fshort-double option.
1292 2014-03-04 Richard Biener <rguenther@suse.de>
1295 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1296 (lto_input_toplevel_asms): Likewise.
1297 * lto-section-in.c (lto_get_section_data): Instead do it here
1300 2014-03-04 Richard Biener <rguenther@suse.de>
1302 PR tree-optimization/60382
1303 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1304 dead PHIs a reduction.
1306 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1308 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1310 (_mm_prefetch): Move out of GCC target("sse") pragma.
1311 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1312 GCC target("prfchw") pragma.
1313 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1315 * config/i386/i386.c (ix86_option_override_internal): Enable
1316 -mprfchw with -mprefetchwt1.
1318 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1320 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1323 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1325 * opts.h (CL_PCH_IGNORE): Define.
1326 * targhooks.c (option_affects_pch_p):
1327 Return false for options that have CL_PCH_IGNORE set.
1328 * opt-functions.awk: Process PchIgnore.
1329 * doc/options.texi: Document PchIgnore.
1331 * config/arc/arc.opt (misize): Add PchIgnore property.
1333 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1335 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1336 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1337 constraint on constants to permit them being loaded into
1338 GENERAL_REGS or BASE_REGS.
1340 2014-03-03 Nick Clifton <nickc@redhat.com>
1342 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1343 anti-cacnonical alternatives.
1344 (negandhi3_real): New pattern.
1345 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1347 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1349 * config/avr/avr-mcus.def: Remove atxmega16x1.
1350 * config/avr/avr-tables.opt: Regenerate.
1351 * config/avr/t-multilib: Regenerate.
1352 * doc/avr-mmcu.texi: Regenerate.
1354 2014-03-03 Tobias Grosser <tobias@grosser.es>
1355 Mircea Namolaru <mircea.namolaru@inria.fr>
1357 PR tree-optimization/58028
1358 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1361 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1363 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1364 not handled by recognizers.
1366 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1369 * function.c (expand_function_end): Don't emit
1370 clobber_return_register sequence if clobber_after is a BARRIER.
1371 * cfgexpand.c (construct_exit_block): Append instructions before
1372 return_label to prev_bb.
1374 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1376 * config/rs6000/constraints.md: Document reserved use of "wc".
1378 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1381 * ipa.c (function_and_variable_visibility): When dissolving comdat
1382 group, also set all symbols to local.
1384 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1389 2013-12-14 Jan Hubicka <jh@suse.cz>
1391 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1393 2014-03-02 Jon Beniston <jon@beniston.com>
1399 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1400 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1401 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1402 (simple_return, *simple_return): New patterns
1403 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1404 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1406 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1408 * dwarf2out.c (gen_subprogram_die): Tidy.
1410 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1413 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1414 (*mov_t_msb_neg_negc): ... this new insn.
1416 2014-02-28 Jason Merrill <jason@redhat.com>
1419 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1422 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1425 * dwarf2out.c (decltype_auto_die): New static.
1426 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1427 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1428 (is_cxx_auto): Likewise.
1430 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1432 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1433 we are not using general regs only.
1435 2014-02-28 Richard Biener <rguenther@suse.de>
1438 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1439 previous fix and only allow to remove trivial pre-headers
1440 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1441 (remove_forwarder_block): Properly update the latch of a loop.
1443 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1446 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1447 (cselib_preserved_hash_table): New.
1448 (preserve_constants_and_equivs): Move preserved vals to it.
1449 (cselib_find_slot): Look it up first.
1450 (cselib_init): Initialize it.
1451 (cselib_finish): Release it.
1452 (dump_cselib_table): Dump it.
1454 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1457 * cselib.c (remove_useless_values): Skip to avoid quadratic
1458 behavior if the condition moved from...
1459 (cselib_process_insn): ... here holds.
1461 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1464 * var-tracking.c (vt_initialize): Apply the same condition to
1465 preserve the CFA base value.
1467 2014-02-28 Joey Ye <joey.ye@arm.com>
1470 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1471 if reload in progress or completed.
1473 2014-02-28 Tobias Burnus <burnus@net-b.de>
1476 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1479 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1481 * doc/tm.texi.in (Condition Code Status): Update documention for
1482 relative locations of cc0-setter and cc0-user.
1484 2014-02-27 Jeff Law <law@redhat.com>
1486 PR rtl-optimization/52714
1487 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1488 into two independent simple sets, if I3 is a jump, ensure the
1489 pattern we place into I3 is a (set (pc) ...).
1491 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1492 Jeff Law <law@redhat.com>
1494 PR rtl-optimization/49847
1495 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1496 are in different blocks.
1497 * doc/tm.texi (Condition Code Status): Update documention for
1498 relative locations of cc0-setter and cc0-user.
1500 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1503 * lra.c (lra_emit_add): Check SUBREG too.
1505 2014-02-27 Andreas Schwab <schwab@suse.de>
1507 * config/m68k/m68k.c (m68k_option_override): Disable
1508 -flive-range-shrinkage for classic m68k.
1509 (m68k_override_options_after_change): Likewise.
1511 2014-02-27 Marek Polacek <polacek@redhat.com>
1514 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1515 -Wmaybe-uninitialized.
1517 2014-02-27 Alan Modra <amodra@gmail.com>
1520 * reload1.c (emit_input_reload_insns): When reload_override_in,
1521 set old to rl->in_reg when rl->in_reg is a subreg.
1523 2014-02-26 Richard Biener <rguenther@suse.de>
1526 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1528 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1530 * common/config/i386/predicates.md (const1256_operand): Remove.
1531 (const2356_operand): New.
1532 (const_1_to_2_operand): Remove.
1533 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1534 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1535 (*avx512pf_gatherpf<mode>sf): Ditto.
1536 (avx512pf_gatherpf<mode>df): Ditto.
1537 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1538 (*avx512pf_gatherpf<mode>df): Ditto.
1539 (avx512pf_scatterpf<mode>sf): Ditto.
1540 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1541 (*avx512pf_scatterpf<mode>sf): Ditto.
1542 (avx512pf_scatterpf<mode>df): Ditto.
1543 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1544 (*avx512pf_scatterpf<mode>df): Ditto.
1545 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1547 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1549 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1550 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1551 (_mm512_mask_testn_epi64_mask): Move to ...
1552 * config/i386/avx512cdintrin.h: Here.
1553 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1554 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1555 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1556 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1557 TARGET_AVX512F from TARGET_AVX512CD.
1559 2014-02-26 Richard Biener <rguenther@suse.de>
1562 * ipa.c (walk_polymorphic_call_targets): Properly guard
1563 call to inline_update_overall_summary.
1565 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1568 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1569 and latches only if requested. Fix latch if it is removed.
1570 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1571 LOOPS_HAVE_PREHEADERS.
1573 2014-02-25 Andrew Pinski <apinski@cavium.com>
1575 * builtins.c (expand_builtin_thread_pointer): Create a new target
1576 when the target is NULL.
1578 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1580 PR rtl-optimization/60317
1581 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1582 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1583 * lra-assigns.c: Include params.h.
1584 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1585 other reload pseudos considerations.
1587 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1589 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1590 to use canonical form for nor<mode>3.
1592 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1595 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1598 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1600 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1601 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1602 (ix86_handle_option): Handle OPT_mprefetchwt1.
1603 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1604 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1606 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1607 OPTION_MASK_ISA_PREFETCHWT1.
1608 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1609 (PTA_PREFETCHWT1): New.
1610 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1611 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1612 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1613 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1614 (*prefetch_avx512pf_<mode>_: Change into ...
1615 (*prefetch_prefetchwt1_<mode>: This.
1616 * config/i386/i386.opt (mprefetchwt1): New.
1617 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1618 (_mm_prefetch): Handle intent to write.
1619 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1621 2014-02-25 Richard Biener <rguenther@suse.de>
1624 * emit-rtl.c (mem_attrs_htab): Remove.
1625 (mem_attrs_htab_hash): Likewise.
1626 (mem_attrs_htab_eq): Likewise.
1627 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1628 (init_emit_once): Do not allocate mem_attrs_htab.
1630 2014-02-25 Richard Biener <rguenther@suse.de>
1633 * lto-opts.c (lto_write_options): Output non-explicit conservative
1634 -fwrapv, -fno-trapv and -fno-strict-overflow.
1635 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1636 (run_gcc): And pass them through.
1638 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1640 * sel-sched.c (calculate_new_fences): New parameter ptime.
1641 Calculate it as a maximum over all fence cycles.
1642 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1643 Print the final schedule timing when sched_verbose.
1645 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1647 PR rtl-optimization/60292
1648 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1649 bit fot the fence instruction.
1651 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1653 * calls.h: Fix typo in comment.
1655 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1657 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1658 adjusting offsetable addresses.
1660 2014-02-24 Guozhi Wei <carrot@google.com>
1662 * sparseset.h (sparseset_pop): Fix the wrong index.
1664 2014-02-24 Walter Lee <walt@tilera.com>
1666 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1667 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1669 * common/config/tilegx/tilegx-common.c
1670 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1671 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1673 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1674 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1675 (tilegx_gimplify_va_arg_expr): Handle big endian.
1676 (tilegx_expand_unaligned_load): Ditto.
1677 (tilegx_expand_unaligned_store): Ditto.
1678 (TARGET_RETURN_IN_MSB): New.
1679 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1680 (TARGET_ENDIAN_DEFAULT): New.
1681 (TARGET_BIG_ENDIAN): Handle big endian.
1682 (BYTES_BIG_ENDIAN): Ditto.
1683 (WORDS_BIG_ENDIAN): Ditto.
1684 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1687 * config/tilegx/tilegx.md (extv): Handle big endian.
1689 (insn_st<n>): Ditto.
1690 (insn_st<n>_add<bitsuffix>): Ditto.
1691 (insn_stnt<n>): Ditto.
1692 (insn_stnt<n>_add<bitsuffix>):Ditto.
1693 (vec_interleave_highv8qi): Handle big endian.
1694 (vec_interleave_highv8qi_be): New.
1695 (vec_interleave_highv8qi_le): New.
1696 (insn_v1int_h): Handle big endian.
1697 (vec_interleave_lowv8qi): Handle big endian.
1698 (vec_interleave_lowv8qi_be): New.
1699 (vec_interleave_lowv8qi_le): New.
1700 (insn_v1int_l): Handle big endian.
1701 (vec_interleave_highv4hi): Handle big endian.
1702 (vec_interleave_highv4hi_be): New.
1703 (vec_interleave_highv4hi_le): New.
1704 (insn_v2int_h): Handle big endian.
1705 (vec_interleave_lowv4hi): Handle big endian.
1706 (vec_interleave_lowv4hi_be): New.
1707 (vec_interleave_lowv4hi_le): New.
1708 (insn_v2int_l): Handle big endian.
1709 (vec_interleave_highv2si): Handle big endian.
1710 (vec_interleave_highv2si_be): New.
1711 (vec_interleave_highv2si_le): New.
1712 (insn_v4int_h): Handle big endian.
1713 (vec_interleave_lowv2si): Handle big endian.
1714 (vec_interleave_lowv2si_be): New.
1715 (vec_interleave_lowv2si_le): New.
1716 (insn_v4int_l): Handle big endian.
1717 * config/tilegx/tilegx.opt (mbig-endian): New option.
1718 (mlittle-endian): New option.
1719 * doc/install.texi: Document tilegxbe-linux.
1720 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1722 2014-02-24 Martin Jambor <mjambor@suse.cz>
1725 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1726 there are no parameter descriptors.
1728 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1730 PR rtl-optimization/60268
1731 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1732 initialization to ...
1733 (sched_rgn_init): ... here.
1734 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1736 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1738 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1741 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1743 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1746 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1748 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1749 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1751 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1753 * config/microblaze/predicates.md: Add cmp_op predicate.
1754 * config/microblaze/microblaze.md: Add branch_compare instruction
1755 which uses cmp_op predicate and emits cmp insn before branch.
1756 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1757 to microblaze_expand_conditional_branch and consolidate logic.
1758 (microblaze_expand_conditional_branch): emit branch_compare
1759 insn instead of handling cmp op separate from branch insn.
1761 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1763 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1766 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1768 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1769 define_insn with define_expand and new define_insn
1770 *altivec_lve<VI_char>x_internal.
1771 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1772 and new define_insn *altivec_stve<VI_char>x_internal.
1773 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1775 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1777 (altivec_expand_stvex_be): New function.
1779 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1781 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1782 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1783 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1784 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1786 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1789 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1790 instead of emit_move_insn.
1792 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1794 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1796 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1797 gen_altivec_vsumsws.
1799 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1801 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1802 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1803 (altivec_lvxl_<mode>): New define_expand incorporating
1804 -maltivec=be semantics where needed.
1805 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1806 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1807 semantics where needed.
1808 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1809 (altivec_stvx_<mode>): New define_expand incorporating
1810 -maltivec=be semantics where needed.
1811 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1812 VM2 iterator instead of V4SI.
1813 (altivec_stvxl_<mode>): New define_expand incorporating
1814 -maltivec=be semantics where needed.
1815 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1816 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1817 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1818 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1819 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1820 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1821 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1822 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1823 ALTIVEC_BUILTIN_STVXL.
1824 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1825 (altivec_expand_stvx_be): Likewise.
1826 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1827 (altivec_expand_lvx_be): Likewise.
1828 (altivec_expand_stvx_be): Likewise.
1829 (altivec_expand_builtin): Add cases for
1830 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1831 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1832 (altivec_init_builtins): Add definitions for
1833 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1834 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1836 2014-02-21 Catherine Moore <clm@codesourcery.com>
1838 * doc/invoke.texi (mvirt, mno-virt): Document.
1839 * config/mips/mips.opt (mvirt): New option.
1840 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1842 2014-02-21 Richard Biener <rguenther@suse.de>
1844 PR tree-optimization/60276
1845 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1846 (STMT_VINFO_MIN_NEG_DIST): New macro.
1847 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1848 STMT_VINFO_MIN_NEG_DIST.
1849 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1850 made for negative dependence distances still hold.
1852 2014-02-21 Richard Biener <rguenther@suse.de>
1855 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1856 DECL_INITIAL for globals not in the current function context.
1858 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1860 PR tree-optimization/56490
1861 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1862 * tree-ssa-uninit.c: Include params.h.
1863 (compute_control_dep_chain): Add num_calls argument, return false
1864 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1865 num_calls to recursive call.
1866 (find_predicates): Change dep_chain into normal array,
1867 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1868 variable and adjust compute_control_dep_chain caller.
1869 (find_def_preds): Likewise.
1871 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1873 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1874 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1876 2014-02-21 Nick Clifton <nickc@redhat.com>
1878 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1879 (pushhi1): Likewise.
1880 (popqi1): Add mode to pre_dec.
1883 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1885 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1886 mode for mask of V8SFmode permutation.
1888 2014-02-20 Richard Henderson <rth@redhat.com>
1891 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1892 a new pseudo for OLDVAL.
1894 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1897 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1898 gen_reg_rtx if d->testing_p.
1899 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1900 if d->testing_p and we will certainly return true.
1901 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1904 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1906 * emit-rtl.c (gen_reg_rtx): Assert that
1907 crtl->emit.regno_pointer_align_length is non-zero.
1909 2014-02-20 Richard Henderson <rth@redhat.com>
1912 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1913 on failure the store back into EXPECT.
1915 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1916 Sandra Loosemore <sandra@codesourcery.com>
1918 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1919 * config/nios2/nios2.c (nios2_function_profiler): Add
1920 -fPIC (flag_pic == 2) support.
1921 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1922 (nios2_large_offset_p): New function.
1923 (nios2_unspec_reloc_p): Move up position, update to use
1924 nios2_large_offset_p.
1925 (nios2_unspec_address): Remove function.
1926 (nios2_unspec_offset): New function.
1927 (nios2_large_got_address): New function.
1928 (nios2_got_address): Add large offset support.
1929 (nios2_legitimize_tls_address): Update usage of removed and new
1931 (nios2_symbol_binds_local_p): New function.
1932 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1933 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1934 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1935 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1936 processing for (const (unspec ...)).
1937 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1939 2014-02-20 Richard Biener <rguenther@suse.de>
1941 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1942 doing the substitution.
1943 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1945 2014-02-20 Martin Jambor <mjambor@suse.cz>
1948 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1949 info when checking whether lattices are bottom.
1951 2014-02-20 Richard Biener <rguenther@suse.de>
1954 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1957 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1960 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1961 parameter specifying the scaling.
1962 (inline_call): Update.
1963 (want_inline_recursively): Guard division by zero.
1964 (recursive_inlining): Update.
1965 * ipa-inline.h (clone_inlined_nodes): Update.
1967 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1970 * config/i386/i386.c (classify_argument): Pass structures of size
1971 64 bytes or less in register.
1973 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1974 Kirill Yukhin <kirill.yukhin@intel.com>
1976 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1977 (_mm_rcp28_round_ss): Ditto.
1978 (_mm_rsqrt28_round_sd): Ditto.
1979 (_mm_rsqrt28_round_ss): Ditto.
1980 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1981 (_mm_rcp14_round_ss): Ditto.
1982 (_mm_rsqrt14_round_sd): Ditto.
1983 (_mm_rsqrt14_round_ss): Ditto.
1984 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1985 the first input operand, get rid of match_dup.
1986 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1988 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1990 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1991 operand as the first input operand, set type attribute.
1992 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1994 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1995 operand as the first input operand, set type attribute.
1997 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1999 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2002 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2005 * config/i386/i386.c (construct_container): Remove TFmode check
2006 for X86_64_INTEGER_CLASS.
2008 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2011 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2012 only when -Wpsabi is enabled.
2014 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2017 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2018 passing arrays in registers are the same as for structs, so remove the
2019 special case for them.
2021 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2023 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2024 destination type, extract only the valid bits if the source type is not
2025 integral and has a different mode.
2027 2014-02-19 Richard Biener <rguenther@suse.de>
2030 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2033 2014-02-19 Richard Biener <rguenther@suse.de>
2036 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2037 (ipa_modify_call_arguments): Emit an argument load explicitely and
2038 preserve virtual SSA form there and for the replacement call.
2039 Do not update SSA form nor free dominance info.
2041 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2043 * ipa.c (function_and_variable_visibility): Also clear WEAK
2044 flag when disolving COMDAT_GROUP.
2046 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2048 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2049 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2051 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2052 do more sanity checks.
2053 (detect_type_change): Return true when giving up early.
2054 (compute_complex_assign_jump_func): Fix type parameter of
2055 ipa_set_ancestor_jf.
2056 (compute_complex_ancestor_jump_func): Likewise.
2057 (update_jump_functions_after_inlining): Fix updating of
2059 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2061 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2063 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2064 inline clones when edge disappears.
2066 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2069 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2070 Split 64-bit moves into 2 patterns. Do not allow the use of
2071 direct move for TDmode in little endian, since the decimal value
2072 has little endian bytes within a word, but the 64-bit pieces are
2073 ordered in a big endian fashion, and normal subreg's of TDmode are
2075 (mov<mode>_64bit_dm): Likewise.
2076 (movtd_64bit_nodm): Likewise.
2078 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2080 PR tree-optimization/60174
2081 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2082 statement of an SSA_NAME that occurs in an abnormal PHI node.
2084 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2087 * final.c (SEEN_BB): Remove.
2088 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2089 (final_scan_insn): Don't force_source_line on second
2090 NOTE_INSN_BASIC_BLOCK.
2092 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2095 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2096 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2097 (type_natural_mode): Warn ABI change when %zmm register is not
2098 available for AVX512F vector value passing.
2100 2014-02-18 Kai Tietz <ktietz@redhat.com>
2103 * config/i386/i386.c (ix86_expand_prologue): Use value in
2104 rax register as displacement when restoring %r10 or %rax.
2105 Fix wrong offset when restoring both registers.
2107 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2109 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2110 assertion with conditional return.
2112 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2113 Uros Bizjak <ubizjak@gmail.com>
2116 * config/i386/driver-i386.c (host_detect_local_cpu): If
2117 YMM state is not saved by the OS, also clear has_f16c. Move
2118 CPUID 0x80000001 handling before YMM state saving checking.
2120 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2122 PR rtl-optimization/58960
2123 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2124 factored out from ...
2125 (sched_init): ... here.
2126 (free_global_sched_pressure_data): New, factored out from ...
2127 (sched_finish): ... here.
2128 * sched-int.h (free_global_sched_pressure_data): Declare.
2129 * sched-rgn.c (nr_regions_initial): New static global.
2130 (haifa_find_rgns): Initialize it.
2131 (schedule_region): Disable sched-pressure for the newly
2134 2014-02-17 Richard Biener <rguenther@suse.de>
2136 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2137 release SSA defs of pattern stmts.
2139 2014-02-17 Richard Biener <rguenther@suse.de>
2141 * tree-inline.c (expand_call_inline): Release the virtual
2142 operand defined by the call we are about to inline.
2144 2014-02-17 Richard Biener <rguenther@suse.de>
2146 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2148 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2149 Ilya Tocar <ilya.tocar@intel.com>
2151 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2152 arguments order in builtin.
2153 (_mm512_permutexvar_epi64): Ditto.
2154 (_mm512_mask_permutexvar_epi64): Ditto
2155 (_mm512_maskz_permutexvar_epi32): Ditto
2156 (_mm512_permutexvar_epi32): Ditto
2157 (_mm512_mask_permutexvar_epi32): Ditto
2159 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2161 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2162 (p8_vmrgow): Likewise.
2164 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2166 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2169 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2172 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2173 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2174 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2175 using direct move instructions on ISA 2.07. Also adjust
2176 instruction length for 64-bit.
2177 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2178 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2180 2014-02-15 Alan Modra <amodra@gmail.com>
2184 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2185 find_replacement on parts of insn rtl that might be reloaded.
2187 2014-02-15 Richard Biener <rguenther@suse.de>
2189 PR tree-optimization/60183
2190 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2191 (tree_ssa_phiprop): Calculate and free post-dominators.
2193 2014-02-14 Jeff Law <law@redhat.com>
2195 PR rtl-optimization/60131
2196 * ree.c (get_extended_src_reg): New function.
2197 (combine_reaching_defs): Use it rather than assuming location of REG.
2198 (find_and_remove_re): Verify first operand of extension is
2199 a REG before adding the insns to the copy list.
2201 2014-02-14 Roland McGrath <mcgrathr@google.com>
2203 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2204 * configure: Regenerated.
2205 * config.in: Regenerated.
2206 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2207 instead of ASM_SHORT.
2209 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2210 Richard Earnshaw <rearnsha@arm.com>
2212 PR rtl-optimization/59535
2213 * lra-constraints.c (process_alt_operands): Encourage alternative
2214 when unassigned pseudo class is superset of the alternative class.
2215 (inherit_reload_reg): Don't inherit when optimizing for code size.
2216 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2217 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2218 modes not less than 4 for Thumb1.
2220 2014-02-14 Kyle McMartin <kyle@redhat.com>
2223 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2225 2014-02-14 Richard Biener <rguenther@suse.de>
2227 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2228 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2229 Do not strip INDIRECT_REFs.
2231 2014-02-14 Richard Biener <rguenther@suse.de>
2234 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2235 DECL_FUNCTION_SPECIFIC_TARGET.
2236 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2237 * tree-streamer-out.c (pack_ts_target_option): Remove.
2238 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2239 (write_ts_function_decl_tree_pointers): Do not stream
2240 DECL_FUNCTION_SPECIFIC_TARGET.
2241 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2242 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2243 (lto_input_ts_function_decl_tree_pointers): Do not stream
2244 DECL_FUNCTION_SPECIFIC_TARGET.
2246 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2248 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2249 (get_initial_def_for_induction, vectorizable_induction): Ignore
2250 debug stmts when looking for exit_phi.
2251 (vectorizable_live_operation): Fix up condition.
2253 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2255 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2256 nreverse() because it changes the content of original tree list.
2258 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2260 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2261 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2263 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2265 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2266 GNU coding standards.
2268 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2271 * dwarf2out.c (gen_subprogram_die): Don't call
2272 add_calling_convention_attribute if subr_die is old_die.
2274 2014-02-13 Sharad Singhai <singhai@google.com>
2276 * doc/optinfo.texi: Fix order of nodes.
2278 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2280 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2281 operands[2], not operands[3].
2283 2014-02-13 Richard Biener <rguenther@suse.de>
2286 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2287 mention the possibility of an in-tree build.
2288 (CLooG): Update recommended version to 0.18.1, mention the
2289 possibility of an in-tree build and clarify that the ISL
2290 bundled with CLooG does not work.
2292 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2295 * expr.c (compress_float_constant): If x is a hard register,
2296 extend into a pseudo and then move to x.
2298 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2300 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2301 caused by bad second argument to warning_at() with -mhotpatch and
2302 nested functions (e.g. with gfortran).
2304 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2306 * opts.c (option_name): Remove "enabled by default" rider.
2308 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2310 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2312 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2313 Uros Bizjak <ubizjak@gmail.com>
2316 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2317 * configure: Regenerated.
2319 2014-02-12 Richard Biener <rguenther@suse.de>
2321 * vec.c (vec_prefix::calculate_allocation): Move as
2322 inline variant to vec.h.
2323 (vec_prefix::calculate_allocation_1): New out-of-line version.
2324 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2325 (vec_prefix::m_has_auto_buf): Rename to ...
2326 (vec_prefix::m_using_auto_storage): ... this.
2327 (vec_prefix::calculate_allocation): Inline the easy cases
2328 and dispatch to calculate_allocation_1 which doesn't need the
2330 (va_heap::reserve): Use gcc_checking_assert.
2331 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2332 m_using_auto_storage.
2333 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2335 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2336 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2337 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2339 2014-02-12 Richard Biener <rguenther@suse.de>
2341 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2342 when we found a dependence.
2344 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2346 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2348 (maybe_fold_stmt): ... into this new function.
2349 * omp-low.c (lower_omp): Update comment.
2351 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2354 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2357 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2359 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2360 identifiers in comments.
2361 (cortexa53_extra_costs): Likewise.
2362 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2363 (cortexa7_extra_costs): Likewise.
2364 (cortexa12_extra_costs): Likewise.
2365 (cortexa15_extra_costs): Likewise.
2366 (v7m_extra_costs): Likewise.
2368 2014-02-12 Richard Biener <rguenther@suse.de>
2371 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2372 of posix_memalign being successful.
2373 (lower_stmt): Restrict lowering of posix_memalign to when
2374 -ftree-bit-ccp is enabled.
2376 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2378 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2380 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2382 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2384 PR rtl-optimization/60116
2385 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2386 other_insn once the combination has been validated.
2388 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2391 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2393 * ipa-devirt.c: Include demangle.h
2394 (odr_violation_reported): New static variable.
2395 (add_type_duplicate): Update odr_violations.
2396 (maybe_record_node): Add completep parameter; update it.
2397 (record_target_from_binfo): Add COMPLETEP parameter;
2398 update it as needed.
2399 (possible_polymorphic_call_targets_1): Likewise.
2400 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2401 rename FINAL to COMPLETE.
2402 (record_targets_from_bases): Sanity check we found the binfo;
2403 fix COMPLETEP updating.
2404 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2405 parameter, fix computing of COMPLETEP.
2406 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2407 at LTO time do demangling.
2408 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2409 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2411 (gimple_get_virt_method_for_binfo): Likewise.
2412 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2413 gimple_get_virt_method_for_vtable): Update prototypes.
2415 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2418 * genautomata.c (add_presence_absence): Fix typo with
2419 {final_}presence_list.
2421 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2424 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2425 for VSX/Altivec vectors that land in GPR registers.
2427 2014-02-11 Richard Henderson <rth@redhat.com>
2428 Jakub Jelinek <jakub@redhat.com>
2431 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2432 around drhs if type conversion to lacc->type is not useless.
2434 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2436 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2438 (cortex-a57.cortex-a53): Likewise.
2439 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2441 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2443 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2446 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2448 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2449 add_options_for_arm_vfp3.
2451 2014-02-11 Jeff Law <law@redhat.com>
2454 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2455 object with an undesirable mode.
2457 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2460 * config/i386/sol2-9.h: New file.
2461 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2462 *-*-solaris2.9*): Use it.
2464 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2466 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2467 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2469 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2471 * config/microblaze/microblaze.c: Extend mcpu version format
2473 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2475 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2477 2014-02-10 Richard Henderson <rth@redhat.com>
2480 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2481 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2482 ms-abi vs -mno-accumulate-outgoing-args.
2483 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2484 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2487 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2490 * cfgexpand.c (expand_asm_operands): Attach source location to
2491 ASM_INPUT rtx objects.
2492 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2494 2014-02-10 Nick Clifton <nickc@redhat.com>
2496 * config/mn10300/mn10300.c (popcount): New function.
2497 (mn10300_expand_prologue): Include saved registers in stack usage
2500 2014-02-10 Jeff Law <law@redhat.com>
2503 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2504 when changing the SET_DEST of a prior insn to avoid an input reload.
2506 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2508 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2509 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2510 -mcall-openbsd, or -mcall-linux.
2511 (CC1_ENDIAN_BIG_SPEC): Remove.
2512 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2513 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2514 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2515 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2516 and %cc1_endian_default.
2517 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2519 2014-02-10 Richard Biener <rguenther@suse.de>
2521 PR tree-optimization/60115
2522 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2523 MEM_REF handling. Properly verify that the accesses are not
2524 out of the objects bound.
2526 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2528 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2531 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2533 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2534 proper constants and fix formatting.
2535 (possible_polymorphic_call_targets): Fix formatting.
2537 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2538 Ilya Tocar <ilya.tocar@intel.com>
2540 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2541 (_mm512_loadu_epi32): Renamed into...
2542 (_mm512_loadu_si512): This.
2543 (_mm512_storeu_epi32): Renamed into...
2544 (_mm512_storeu_si512): This.
2545 (_mm512_maskz_ceil_ps): Removed.
2546 (_mm512_maskz_ceil_pd): Ditto.
2547 (_mm512_maskz_floor_ps): Ditto.
2548 (_mm512_maskz_floor_pd): Ditto.
2549 (_mm512_floor_round_ps): Ditto.
2550 (_mm512_floor_round_pd): Ditto.
2551 (_mm512_ceil_round_ps): Ditto.
2552 (_mm512_ceil_round_pd): Ditto.
2553 (_mm512_mask_floor_round_ps): Ditto.
2554 (_mm512_mask_floor_round_pd): Ditto.
2555 (_mm512_mask_ceil_round_ps): Ditto.
2556 (_mm512_mask_ceil_round_pd): Ditto.
2557 (_mm512_maskz_floor_round_ps): Ditto.
2558 (_mm512_maskz_floor_round_pd): Ditto.
2559 (_mm512_maskz_ceil_round_ps): Ditto.
2560 (_mm512_maskz_ceil_round_pd): Ditto.
2561 (_mm512_expand_pd): Ditto.
2562 (_mm512_expand_ps): Ditto.
2563 * config/i386/i386.c (ix86_builtins): Remove
2564 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2565 (bdesc_args): Ditto.
2566 * config/i386/predicates.md (const1256_operand): New.
2567 (const_1_to_2_operand): Ditto.
2568 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2569 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2570 (*avx512pf_gatherpf<mode>sf): Ditto.
2571 (avx512pf_gatherpf<mode>df): Ditto.
2572 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2573 (*avx512pf_gatherpf<mode>df): Ditto.
2574 (avx512pf_scatterpf<mode>sf): Ditto.
2575 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2576 (*avx512pf_scatterpf<mode>sf): Ditto.
2577 (avx512pf_scatterpf<mode>df): Ditto.
2578 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2579 (*avx512pf_scatterpf<mode>df): Ditto.
2580 (avx512f_expand<mode>): Removed.
2581 (<shift_insn><mode>3<mask_name>): Change predicate type.
2583 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2585 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2586 not at the end of datarefs vector use ordered_remove to avoid
2587 reordering datarefs vector.
2590 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2591 mark local addressable non-static vars as GOVD_PRIVATE
2592 instead of GOVD_LOCAL.
2593 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2594 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2598 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2599 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2600 assume_aligned or alloc_align attributes.
2601 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2602 arguments. Handle also assume_aligned and alloc_align attributes.
2603 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2604 calls to functions with assume_aligned or alloc_align attributes.
2605 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2607 2014-02-08 Terry Guo <terry.guo@arm.com>
2609 * doc/invoke.texi: Document ARM -march=armv7e-m.
2611 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2613 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2614 flag on __cilkrts_rethrow builtin.
2617 * ipa-cp.c (determine_versionability): Fail at -O0
2618 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2619 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2622 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2625 * tree-inline.c (copy_forbidden): Fail for
2626 __attribute__((optimize (0))) functions.
2628 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2630 * varpool.c: Include pointer-set.h.
2631 (varpool_remove_unreferenced_decls): Variables in other partitions
2632 will not be output; be however careful to not lose information
2635 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2637 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2638 lookup in the vtable constructor.
2640 2014-02-07 Jeff Law <law@redhat.com>
2643 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2644 define_insn_and_split.
2646 * ipa-inline.c (inline_small_functions): Fix typos.
2648 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2650 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2651 (s390_can_use_return_insn): Declare.
2652 * config/s390/s390.h (EPILOGUE_USES): Define.
2653 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2655 (s390_chunkify_start): Handle return JUMP_LABELs.
2656 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2657 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2658 (s390_can_use_return_insn): New functions.
2659 (s390_fix_long_loop_prediction): Handle conditional returns.
2660 (TARGET_SET_UP_BY_PROLOGUE): Define.
2661 * config/s390/s390.md (ANY_RETURN): New code iterator.
2662 (*creturn, *csimple_return, return, simple_return): New patterns.
2664 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2666 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2667 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2668 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2669 REG_CFA_RESTORE list when deciding not to restore a register.
2671 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2673 * config/s390/s390.c: Include tree-pass.h and context.h.
2674 (s390_early_mach): New function, split out from...
2675 (s390_emit_prologue): ...here.
2676 (pass_data_s390_early_mach): New pass structure.
2677 (pass_s390_early_mach): New class.
2678 (s390_option_override): Create and register early_mach pass.
2679 Move to end of file.
2681 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2683 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2684 to match for the exit block.
2686 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2688 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2689 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2690 Reject misaligned operands.
2692 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2694 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2696 2014-02-07 Richard Biener <rguenther@suse.de>
2699 * gimple-low.c (lower_builtin_posix_memalign): New function.
2700 (lower_stmt): Call it to lower posix_memalign in a way
2701 to make alignment info accessible.
2703 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2706 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2707 __builtin_setjmp_receiver.
2709 2014-02-07 Richard Biener <rguenther@suse.de>
2712 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2713 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2714 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2715 Handle BUILT_IN_POSIX_MEMALIGN.
2716 (find_func_clobbers): Likewise.
2717 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2718 (call_may_clobber_ref_p_1): Likewise.
2720 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2723 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2726 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2729 * lto-cgraph.c (lto_output_node): Use
2730 symtab_get_symbol_partitioning_class.
2731 (lto_output_varpool_node): likewise.
2732 (symtab_get_symbol_partitioning_class): Move here from
2734 * cgraph.h (symbol_partitioning_class): Likewise.
2735 (symtab_get_symbol_partitioning_class): Declare.
2737 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2739 * ggc.h (ggc_internal_cleared_alloc): New macro.
2740 * vec.h (vec_safe_copy): Handle memory stats.
2741 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2742 * target-globals.c (save_target_globals): Likewise.
2744 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2747 * expr.c (emit_move_resolve_push): Export; be bit more selective
2748 on when to clear alias set.
2749 * expr.h (emit_move_resolve_push): Declare.
2750 * function.h (struct function): Add tail_call_marked.
2751 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2752 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2753 * config/i386/i386.md (TImode move expander): De not call
2755 (FP push expanders): Preserve memory attributes.
2756 * config/i386/sse.md (push<mode>1): Remove.
2757 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2758 (ix86_expand_push): Remove.
2759 * config/i386/mmx.md (push<mode>1): Remove.
2761 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2763 PR rtl-optimization/60030
2764 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2765 lopart with paradoxical subreg before shifting it up by hprec.
2767 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2769 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2770 Remove extra newline at end of file.
2771 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2772 (arm_issue_rate): Handle cortexa57.
2773 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2774 (cortex-a57.cortex-a53): Likewise.
2776 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2779 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2780 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2782 (arm_expand_prologue): Adjust all callers.
2783 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2784 info, registers also at the lowest numbered registers side. Use
2785 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2789 * var-tracking.c (adjust_mems): Before adding a SET to
2790 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2792 2014-02-06 Alan Modra <amodra@gmail.com>
2795 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2796 change SDmode to DDmode when lra_in_progress.
2798 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2801 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2802 free_data_ref on the dr first, and before goto again also set dr
2803 to the next dr. For simd_lane_access, free old datarefs[i] before
2804 overwriting it. For get_vectype_for_scalar_type failure, don't
2805 free_data_ref if simd_lane_access.
2807 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2810 * tree.h (opts_for_fn): New inline function.
2811 (opt_for_fn): Define.
2812 * config/i386/i386.c (ix86_function_regparm): Use
2813 opt_for_fn (decl, optimize) instead of optimize.
2815 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2817 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2818 for SYMBOL_REF in large memory model.
2820 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2822 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2824 (cortex-a57): Likewise.
2825 (cortex-a57.cortex-a53): Likewise.
2827 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2828 Kugan Vivekanandarajah <kuganv@linaro.org>
2830 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2832 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2834 2014-02-06 Richard Biener <rguenther@suse.de>
2836 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2837 set_loop_copy and initialize_original_copy_tables.
2839 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2841 * config/aarch64/aarch64-simd.md
2842 (aarch64_ashr_simddi): Change QI to SI.
2844 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2845 Jakub Jelinek <jakub@redhat.com>
2848 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2851 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2853 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2854 CODE_FOR_altivec_vpku[hw]um to
2855 CODE_FOR_altivec_vpku[hw]um_direct.
2856 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2857 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2858 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2859 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2861 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2863 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2864 generation for -maltivec=be.
2865 (altivec_vsumsws): Simplify redundant test.
2867 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2869 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2870 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2871 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2872 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2873 gen_altivec_vpkuwum.
2874 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2876 (altivec_vpks<VI_char>ss): Likewise.
2877 (altivec_vpks<VI_char>us): Likewise.
2878 (altivec_vpku<VI_char>us): Likewise.
2879 (altivec_vpku<VI_char>um): Likewise.
2880 (altivec_vpku<VI_char>um_direct): New (copy of
2881 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2883 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2884 target is little endian and -maltivec=be is not specified.
2885 (*altivec_vupkhs<VU_char>_direct): New (copy of
2886 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2887 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2888 target is little endian and -maltivec=be is not specified.
2889 (*altivec_vupkls<VU_char>_direct): New (copy of
2890 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2891 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2892 little endian and -maltivec=be is not specified.
2893 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2894 little endian and -maltivec=be is not specified.
2896 2014-02-05 Richard Henderson <rth@redhat.com>
2899 * combine-stack-adj.c: Revert r206943.
2900 * sched-int.h (struct deps_desc): Add last_args_size.
2901 * sched-deps.c (init_deps): Initialize it.
2902 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2903 contain REG_ARGS_SIZE notes.
2905 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2907 * lto-cgraph.c (asm_nodes_output): Make global.
2908 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2909 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2910 (driver_handle_option): Handle OPT_fwpa.
2912 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2915 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2916 a comment typo and formatting issue. If odr_hash hasn't been
2917 created, return vNULL and set *completep to false.
2920 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2921 bb with no successors.
2923 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2926 * doc/invoke.texi (-march): Clarify documentation for ARM.
2930 2014-02-05 Richard Biener <rguenther@suse.de>
2932 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2933 when not vectorizing because of too many alias checks.
2934 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2935 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2937 2014-02-05 Nick Clifton <nickc@redhat.com>
2939 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2940 accept extended registers in any mode when compiling for the MN10300.
2942 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2944 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2945 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2946 sanitization attributes.
2947 (can_inline_edge_p): Likewise.
2948 (sanitize_attrs_match_for_inline_p): New function.
2950 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2952 * ipa-prop.c (detect_type_change): Shor circuit testing of
2953 type changes on THIS pointer.
2955 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2958 * config/pa/pa.c (legitimize_tls_address): Return original address
2959 if not passed a SYMBOL_REF rtx.
2960 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2962 (pa_emit_move_sequence): Simplify TLS source operands.
2963 (pa_legitimate_constant_p): Reject all TLS constants.
2964 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2965 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2967 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2969 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2970 groups when we know they are controlled by LTO.
2971 * varasm.c (default_binds_local_p_1): If object is in other partition,
2972 it will be resolved locally.
2974 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2976 * config/host-linux.c (linux_gt_pch_use_address): Don't
2977 use SSIZE_MAX because it is not always defined.
2979 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2982 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2983 threshold for pseudo splitting.
2984 (update_ebb_live_info): Process call argument hard registers and
2985 hard registers from insn definition too.
2986 (max_small_class_regs_num): New constant.
2987 (inherit_in_ebb): Update live hard regs through EBBs. Update
2988 reloads_num only for small register classes. Don't split for
2991 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2994 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2997 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2999 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3002 2014-02-04 Marek Polacek <polacek@redhat.com>
3004 * gdbinit.in (pel): Define.
3006 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3008 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3011 2014-02-04 Richard Biener <rguenther@suse.de>
3014 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3015 in function context local.
3016 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3017 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3018 similar to LTO_imported_decl_ref.
3020 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3022 PR tree-optimization/60002
3023 * cgraphclones.c (build_function_decl_skip_args): Clear
3026 PR tree-optimization/60023
3027 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3028 false to gsi_replace.
3029 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3030 has been in some EH region and vec_stmt could throw, add
3031 vec_stmt into the same EH region.
3032 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3033 has no lhs, ignore it.
3034 * internal-fn.c (expand_MASK_LOAD): Likewise.
3037 * tree-inline.c (copy_forbidden): Fail for
3038 __attribute__((optimize (0))) functions.
3041 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3042 is set, copy one less argument.
3043 (expand_simd_clones): Don't subtract clone_info->inbranch
3044 from simd_clone_struct_alloc argument.
3046 PR rtl-optimization/57915
3047 * recog.c (simplify_while_replacing): If all unary/binary/relational
3048 operation arguments are constant, attempt to simplify those.
3051 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3052 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3054 2014-02-04 Richard Biener <rguenther@suse.de>
3056 PR tree-optimization/60012
3057 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3058 TBAA disambiguation to all DDRs.
3060 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3063 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3064 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3066 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3069 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3071 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3073 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3074 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3076 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3079 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3080 to figure out targets of polymorphic calls with known decl.
3081 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3082 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3083 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3084 (get_polymorphic_call_info): ... here.
3085 (get_polymorphic_call_info_from_invariant): New function.
3087 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3089 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3090 lookup via vtable pointer; check for type consistency
3091 and turn inconsitent facts into UNREACHABLE.
3092 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3093 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3094 type inconsistent querries; return UNREACHABLE instead.
3096 2014-02-03 Richard Henderson <rth@twiddle.net>
3099 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3100 already processed this node.
3101 (normalize_one_pred_1): Pass along mark_set.
3102 (normalize_one_pred): Create and destroy a pointer_set_t.
3103 (normalize_one_pred_chain): Likewise.
3105 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3107 PR gcov-profile/58602
3108 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3110 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3113 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3114 -fno-devirtualize; try to devirtualize by the knowledge of
3115 virtual table pointer given by aggregate propagation.
3116 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3117 (ipa_print_node_jump_functions): Dump also offset that
3118 is relevant for polymorphic calls.
3119 (determine_known_aggregate_parts): Add arg_type parameter; use it
3120 instead of determining the type from pointer type.
3121 (ipa_compute_jump_functions_for_edge): Update call of
3122 determine_known_aggregate_parts.
3123 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3124 (gimple_get_virt_method_for_binfo): ... here; simplify using
3125 vtable_pointer_value_to_vtable.
3126 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3127 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3128 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3129 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3131 (vtable_pointer_value_to_binfo): ... here.
3132 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3134 2014-02-03 Teresa Johnson <tejohnson@google.com>
3136 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3137 redef of outer loop index variable.
3139 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3143 * doc/extend.texi (Function Attributes): Typo.
3145 2014-02-03 Cong Hou <congh@google.com>
3147 PR tree-optimization/60000
3148 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3149 if the vectorized statement is a store. A store statement can only
3150 appear at the end of pattern statements.
3152 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3154 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3155 (ix86_option_override_internal): Default long double to 64-bit for
3156 32-bit Bionic and to 128-bit for 64-bit Bionic.
3158 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3159 TARGET_LONG_DOUBLE_128 is true.
3160 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3162 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3163 (mlong-double-64): Negate -mlong-double-128.
3164 (mlong-double-128): New option.
3166 * config/i386/i386-c.c (ix86_target_macros): Define
3167 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3169 * doc/invoke.texi: Document -mlong-double-128.
3171 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3173 PR rtl-optimization/60024
3174 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3176 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3178 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3180 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3182 PR rtl-optimization/57662
3183 * sel-sched.c (code_motion_path_driver): Do not mark already not
3184 existing blocks in the visiting bitmap.
3186 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3188 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3189 on the insn being emitted.
3191 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3192 Will Deacon <will.deacon@arm.com>
3194 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3196 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3198 * config/arm/arm-tables.opt: Regenerate.
3200 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3202 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3203 for vector types other than V16QImode.
3204 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3205 define_expand, and call altivec_expand_vec_perm_le when producing
3206 code with little endian element order.
3207 (*altivec_vperm_<mode>_internal): New insn having previous
3208 behavior of altivec_vperm_<mode>.
3209 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3210 altivec_expand_vec_perm_le when producing code with little endian
3212 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3213 behavior of altivec_vperm_<mode>_uns.
3215 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3217 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3218 (altivec_vsumsws): Add handling for -maltivec=be with a little
3220 (altivec_vsumsws_direct): New.
3221 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3222 gen_altivec_vsumsws.
3224 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3226 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3227 vtable_pointer_value_to_binfo): New functions.
3228 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3229 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3231 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3233 * config/nios2/nios2.md (load_got_register): Initialize GOT
3234 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3235 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3237 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3239 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3240 preserverd by passthrough, do not propagate the type.
3242 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3244 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3245 (mips_atomic_assign_expand_fenv): New function.
3246 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3248 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3250 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3251 (__builtin_mips_set_fcsr): Likewise.
3252 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3253 MIPS_USI_FTYPE_VOID.
3254 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3255 (mips16_expand_set_fcsr): Likewise.
3256 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3257 (mips16_set_fcsr_stub): Likewise.
3258 (mips16_get_fcsr_one_only_stub): New class.
3259 (mips16_set_fcsr_one_only_stub): Likewise.
3260 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3261 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3262 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3263 (hard_float): New availability predicate.
3264 (mips_builtins): Add get_fcsr and set_fcsr.
3265 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3266 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3267 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3268 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3269 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3272 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3274 * config/mips/mips.c (mips_one_only_stub): New class.
3275 (mips_need_mips16_rdhwr_p): Replace with...
3276 (mips16_rdhwr_stub): ...this new variable.
3277 (mips16_stub_call_address): New function.
3278 (mips16_rdhwr_one_only_stub): New class.
3279 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3280 (mips_output_mips16_rdhwr): Delete.
3281 (mips_finish_stub): New function.
3282 (mips_code_end): Use it to handle rdhwr stubs.
3284 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3287 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3288 when calculating size of integer atomic types.
3290 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3292 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3294 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3296 PR tree-optimization/60003
3297 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3298 * profile.c (branch_prob): Use gimple_call_builtin_p
3299 to check for BUILT_IN_SETJMP_RECEIVER.
3300 * tree-inline.c (copy_bb): Call notice_special_calls.
3302 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3305 * lra-constraints.c (process_alt_operands): Update reload_sum only
3308 2014-01-31 Richard Henderson <rth@redhat.com>
3311 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3312 until after else_eh is processed.
3314 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3316 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3317 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3318 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3319 in smmintrin.h, remove them.
3320 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3321 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3322 * config/i386/i386.md (ROUND_SAE): Fix value.
3323 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3324 (const48_operand): New.
3325 * config/i386/subst.md (round), (round_expand): Use
3326 const_4_or_8_to_11_operand.
3327 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3329 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3331 * config/i386/constraints.md (Yk): Swap meaning with k.
3332 * config/i386/i386.md (movhi_internal): Change Yk to k.
3333 (movqi_internal): Ditto.
3334 (*k<logic><mode>): Ditto.
3337 (kandn<mode>): Ditto.
3338 (*<code>hi_1): Ditto.
3339 (*<code>qi_1): Ditto.
3340 (kxnor<mode>): Ditto.
3341 (kortestzhi): Ditto.
3342 (kortestchi): Ditto.
3344 (*one_cmplhi2_1): Ditto.
3345 (*one_cmplqi2_1): Ditto.
3346 * config/i386/sse.md (): Change k to Yk.
3347 (avx512f_load<mode>_mask): Ditto.
3348 (avx512f_blendm<mode>): Ditto.
3349 (avx512f_store<mode>_mask): Ditto.
3350 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3351 (avx512f_storedqu<mode>_mask): Ditto.
3352 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3354 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3355 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3356 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3357 (avx512f_maskcmp<mode>3): Ditto.
3358 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3359 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3360 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3361 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3362 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3363 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3364 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3365 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3366 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3367 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3368 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3369 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3370 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3371 (vec_extract_lo_<mode>_maskm): Ditto.
3372 (vec_extract_hi_<mode>_maskm): Ditto.
3373 (avx512f_vternlog<mode>_mask): Ditto.
3374 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3375 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3376 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3377 (avx512f_<code>v8div16qi2_mask): Ditto.
3378 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3379 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3380 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3381 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3382 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3383 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3384 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3385 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3386 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3387 (avx512cd_maskb_vec_dupv8di): Ditto.
3388 (avx512cd_maskw_vec_dupv16si): Ditto.
3389 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3390 (avx512f_vpermi2var<mode>3_mask): Ditto.
3391 (avx512f_vpermi2var<mode>3_mask): Ditto.
3392 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3393 (*avx512f_gathersi<mode>): Ditto.
3394 (*avx512f_gathersi<mode>_2): Ditto.
3395 (*avx512f_gatherdi<mode>): Ditto.
3396 (*avx512f_gatherdi<mode>_2): Ditto.
3397 (*avx512f_scattersi<mode>): Ditto.
3398 (*avx512f_scatterdi<mode>): Ditto.
3399 (avx512f_compress<mode>_mask): Ditto.
3400 (avx512f_compressstore<mode>_mask): Ditto.
3401 (avx512f_expand<mode>_mask): Ditto.
3402 * config/i386/subst.md (mask): Change k to Yk.
3403 (mask_scalar_merge): Ditto.
3406 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3408 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3410 2014-01-31 Richard Biener <rguenther@suse.de>
3413 * builtins.c (fold_builtin_memory_op): Make sure to not
3414 use a floating-point mode or a boolean or enumeral type for
3417 2014-01-30 DJ Delorie <dj@redhat.com>
3419 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3420 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3421 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3422 whenever main() has an epilogue.
3424 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3426 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3427 unused variable "field".
3428 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3429 (vsx_mergeh_<mode>): Likewise.
3430 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3431 (altivec_vmrghh): Likewise.
3432 (altivec_vmrghw): Likewise.
3433 (altivec_vmrglb): Likewise.
3434 (altivec_vmrglh): Likewise.
3435 (altivec_vmrglw): Likewise.
3436 (altivec_vspltb): Add missing uses.
3437 (altivec_vsplth): Likewise.
3438 (altivec_vspltw): Likewise.
3439 (altivec_vspltsf): Likewise.
3441 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3444 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3445 frame related instructions.
3447 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3449 PR rtl-optimization/59959
3450 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3451 any reload of register whose subreg is invalid.
3453 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3455 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3456 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3457 Add missing return type - void.
3459 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3461 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3462 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3463 remove element index adjustment for endian (now handled in vsx.md
3465 (altivec_expand_vec_perm_const): Use
3466 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3467 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3468 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3469 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3470 define_expand and a new define_insn *altivec_vspltb_internal;
3471 adjust for -maltivec=be on a little endian target.
3472 (altivec_vspltb_direct): New.
3473 (altivec_vsplth): Divide into a define_expand and a new
3474 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3475 little endian target.
3476 (altivec_vsplth_direct): New.
3477 (altivec_vspltw): Divide into a define_expand and a new
3478 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3479 little endian target.
3480 (altivec_vspltw_direct): New.
3481 (altivec_vspltsf): Divide into a define_expand and a new
3482 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3483 a little endian target.
3485 2014-01-30 Richard Biener <rguenther@suse.de>
3487 PR tree-optimization/59993
3488 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3489 can propagate form the earlier stmt and avoid the transform
3490 when the intermediate result is needed.
3492 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3494 * README.Portability: Fix typo.
3496 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3498 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3499 comparison_operator with ordered_comparison_operator.
3501 2014-01-30 Nick Clifton <nickc@redhat.com>
3503 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3504 Rename to mn10300_store_multiple_regs.
3505 * config/mn10300/mn10300.c: Likewise.
3506 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3507 store_multiple_regs.
3508 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3509 Call mn10300_store_multiple_regs.
3511 2014-01-30 Nick Clifton <nickc@redhat.com>
3512 DJ Delorie <dj@redhat.com>
3514 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3515 %fp 2 to keep registers after it properly word-aligned.
3516 (rl78_alloc_physical_registers_umul): Handle the case where both
3517 input operands are the same.
3519 2014-01-30 Richard Biener <rguenther@suse.de>
3521 PR tree-optimization/59903
3522 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3525 2014-01-30 Jason Merrill <jason@redhat.com>
3528 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3531 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3533 2014-01-30 Richard Biener <rguenther@suse.de>
3535 PR tree-optimization/59951
3536 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3538 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3541 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3542 SFmode to DFmode case.
3544 2014-01-29 DJ Delorie <dj@redhat.com>
3546 * config/msp430/msp430.opt (-minrt): New.
3547 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3549 (ENDFILE_SPEC): Likewise.
3551 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3553 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3554 (estimate_function_body_sizes): Use it.
3556 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3559 * dwarf2out.c (is_cxx_auto): New.
3560 (is_base_type): Use it.
3561 (gen_type_die_with_usage): Likewise.
3563 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3565 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3566 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3567 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3568 -maltivec=be with LE targets.
3569 (vsx_mergeh_<mode>): Likewise.
3570 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3571 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3572 (altivec_vmrghb): Replace with define_expand and new
3573 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3574 (altivec_vmrghb_direct): New define_insn.
3575 (altivec_vmrghh): Replace with define_expand and new
3576 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3577 (altivec_vmrghh_direct): New define_insn.
3578 (altivec_vmrghw): Replace with define_expand and new
3579 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3580 (altivec_vmrghw_direct): New define_insn.
3581 (*altivec_vmrghsf): Adjust for endianness.
3582 (altivec_vmrglb): Replace with define_expand and new
3583 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3584 (altivec_vmrglb_direct): New define_insn.
3585 (altivec_vmrglh): Replace with define_expand and new
3586 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3587 (altivec_vmrglh_direct): New define_insn.
3588 (altivec_vmrglw): Replace with define_expand and new
3589 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3590 (altivec_vmrglw_direct): New define_insn.
3591 (*altivec_vmrglsf): Adjust for endianness.
3592 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3593 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3594 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3595 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3596 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3597 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3598 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3599 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3601 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3603 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3604 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3607 2014-01-29 Richard Biener <rguenther@suse.de>
3609 PR tree-optimization/58742
3610 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3611 associate_pointerplus_align.
3612 (associate_pointerplus_diff): New function.
3613 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3614 and associate_pointerplus_diff.
3616 2014-01-29 Richard Biener <rguenther@suse.de>
3618 * lto-streamer.h (LTO_major_version): Bump to 3.
3619 (LTO_minor_version): Reset to 0.
3621 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3623 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3624 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3625 (arm_file_start): Generate correct asm header for armv7ve.
3626 * config/arm/bpabi.h: Add multilib support for armv7ve.
3627 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3628 and cortex-a15 to armv7ve.
3629 * config/arm/t-aprofile: Add multilib support for armv7ve.
3630 * doc/invoke.texi: Document -march=armv7ve.
3632 2014-01-29 Richard Biener <rguenther@suse.de>
3634 PR tree-optimization/58742
3635 * tree-ssa-forwprop.c (associate_plusminus): Return true
3636 if we changed sth, defer EH cleanup to ...
3637 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3638 (simplify_mult): New function.
3640 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3643 PR tree-optimization/59920
3644 * tree.c (build_common_builtin_nodes): Remove
3645 __builtin_setjmp_dispatcher initialization.
3646 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3647 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3648 instead of gsi_after_labels + manually skipping debug stmts.
3649 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3650 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3651 * tree-inline.c (copy_edges_for_bb): Remove
3652 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3653 argument. Ignore computed_goto_p stmts. Don't call
3654 make_abnormal_goto_edges. If a call might need abnormal edges
3655 for non-local gotos, see if it already has an edge to
3656 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3657 with true argument, don't do anything then, otherwise add
3658 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3659 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3661 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3662 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3663 (lower_stmt): Don't set data->calls_builtin_setjmp.
3664 (lower_builtin_setjmp): Adjust comment.
3665 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3666 * tree-cfg.c (found_computed_goto): Remove.
3667 (factor_computed_gotos): Remove.
3668 (make_goto_expr_edges): Return bool, true for computed gotos.
3669 Don't call make_abnormal_goto_edges.
3670 (build_gimple_cfg): Don't set found_computed_goto, don't call
3671 factor_computed_gotos.
3672 (computed_goto_p): No longer static.
3673 (make_blocks): Don't set found_computed_goto.
3674 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3675 (make_edges): If make_goto_expr_edges returns true, push bb
3676 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3677 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3678 vector. Record mapping between bbs and OpenMP regions if there
3679 are any, adjust make_gimple_omp_edges caller. Call
3680 handle_abnormal_edges.
3681 (make_abnormal_goto_edges): Remove.
3682 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3683 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3684 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3685 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3686 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3687 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3688 filling *region also set *region_idx to (*region)->entry->index.
3691 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3692 For REGs set ORIGINAL_REGNO.
3694 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3696 * doc/md.texi: Mention that a target shouldn't implement
3697 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3700 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3702 PR tree-optimization/59594
3703 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3704 a copy of the datarefs vector rather than the vector itself.
3706 2014-01-28 Jason Merrill <jason@redhat.com>
3709 * dwarf2out.c (auto_die): New static.
3710 (gen_type_die_with_usage): Handle C++1y 'auto'.
3711 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3714 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3717 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3718 (SPEC_X32): Likewise.
3719 (SPEC_64): Likewise.
3720 * config/i386/i386.c (ix86_option_override_internal): Turn off
3721 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3723 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3724 * config/i386/i386.h (TARGET_16BIT): New macro.
3725 (TARGET_16BIT_P): Likewise.
3726 * config/i386/i386.opt: Add m16.
3727 * doc/invoke.texi: Document -m16.
3729 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3731 PR preprocessor/59935
3732 * input.c (location_get_source_line): Bail out on when line number
3733 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3735 2014-01-28 Richard Biener <rguenther@suse.de>
3737 PR tree-optimization/58742
3738 * tree-ssa-forwprop.c (associate_plusminus): Handle
3739 pointer subtraction of the form (T)(P + A) - (T)P.
3741 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3743 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3746 2014-01-28 Richard Biener <rguenther@suse.de>
3749 2014-01-28 Richard Biener <rguenther@suse.de>
3751 PR rtl-optimization/45364
3752 PR rtl-optimization/59890
3753 * var-tracking.c (local_get_addr_clear_given_value): Handle
3754 already cleared slot.
3755 (val_reset): Handle not allocated local_get_addr_cache.
3756 (vt_find_locations): Use post-order on the inverted CFG.
3758 2014-01-28 Richard Biener <rguenther@suse.de>
3760 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3762 2014-01-28 Richard Biener <rguenther@suse.de>
3764 PR rtl-optimization/45364
3765 PR rtl-optimization/59890
3766 * var-tracking.c (local_get_addr_clear_given_value): Handle
3767 already cleared slot.
3768 (val_reset): Handle not allocated local_get_addr_cache.
3769 (vt_find_locations): Use post-order on the inverted CFG.
3771 2014-01-28 Alan Modra <amodra@gmail.com>
3773 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3774 * configure.ac <recursive call for build != host>: Define
3775 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3776 and LD_FOR_BUILD too.
3777 * configure: Regenerate.
3779 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3781 * config/i386/i386.c (get_builtin_code_for_version): Separate
3782 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3783 Broadwell from Haswell.
3785 2014-01-27 Steve Ellcey <sellcey@mips.com>
3787 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3788 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3789 * config/mips/mips.c (mips_option_override): Change setting
3791 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3792 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3793 Change from Mask to Var.
3795 2014-01-27 Jeff Law <law@redhat.com>
3797 * ipa-inline.c (inline_small_functions): Fix typo.
3799 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3801 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3802 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3803 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3804 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3805 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3806 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3807 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3808 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3809 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3810 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3811 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3812 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3813 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3814 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3815 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3816 (_mm512_storeu_epi64): Ditto.
3817 (_mm512_cmpge_epi32_mask): Ditto.
3818 (_mm512_cmpge_epu32_mask): Ditto.
3819 (_mm512_cmpge_epi64_mask): Ditto.
3820 (_mm512_cmpge_epu64_mask): Ditto.
3821 (_mm512_cmple_epi32_mask): Ditto.
3822 (_mm512_cmple_epu32_mask): Ditto.
3823 (_mm512_cmple_epi64_mask): Ditto.
3824 (_mm512_cmple_epu64_mask): Ditto.
3825 (_mm512_cmplt_epi32_mask): Ditto.
3826 (_mm512_cmplt_epu32_mask): Ditto.
3827 (_mm512_cmplt_epi64_mask): Ditto.
3828 (_mm512_cmplt_epu64_mask): Ditto.
3829 (_mm512_cmpneq_epi32_mask): Ditto.
3830 (_mm512_cmpneq_epu32_mask): Ditto.
3831 (_mm512_cmpneq_epi64_mask): Ditto.
3832 (_mm512_cmpneq_epu64_mask): Ditto.
3833 (_mm512_expand_pd): Ditto.
3834 (_mm512_expand_ps): Ditto.
3835 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3836 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3837 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3838 * config/i386/i386.c (ix86_builtins): Add
3839 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3840 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3841 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3842 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3843 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3844 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3845 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3846 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3847 IX86_BUILTIN_PMOVUSQW512_MEM.
3848 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3849 __builtin_ia32_pmovsqd512mem_mask,
3850 __builtin_ia32_pmovqd512mem_mask,
3851 __builtin_ia32_pmovusqw512mem_mask,
3852 __builtin_ia32_pmovsqw512mem_mask,
3853 __builtin_ia32_pmovqw512mem_mask,
3854 __builtin_ia32_pmovusdw512mem_mask,
3855 __builtin_ia32_pmovsdw512mem_mask,
3856 __builtin_ia32_pmovdw512mem_mask,
3857 __builtin_ia32_pmovqb512mem_mask,
3858 __builtin_ia32_pmovusqb512mem_mask,
3859 __builtin_ia32_pmovsqb512mem_mask,
3860 __builtin_ia32_pmovusdb512mem_mask,
3861 __builtin_ia32_pmovsdb512mem_mask,
3862 __builtin_ia32_pmovdb512mem_mask.
3863 (bdesc_args): Add __builtin_ia32_expanddf512,
3864 __builtin_ia32_expandsf512.
3865 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3866 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3867 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3868 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3869 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3870 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3871 (avx512f_<code>v8div16qi2_mask_store): This.
3872 (avx512f_expand<mode>): New.
3874 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3876 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3878 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3879 (_mm512_prefetch_i32scatter_pd): Ditto.
3880 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3881 (_mm512_prefetch_i64scatter_pd): Ditto.
3882 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3883 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3884 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3885 (_mm512_prefetch_i32scatter_ps): Ditto.
3886 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3887 (_mm512_prefetch_i64scatter_ps): Ditto.
3888 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3889 * config/i386/i386-builtin-types.def: Define
3890 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3891 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3892 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3893 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3894 IX86_BUILTIN_SCATTERPFQPD.
3895 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3896 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3897 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3898 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3899 __builtin_ia32_scatterpfqps.
3900 (ix86_expand_builtin): Expand new built-ins.
3901 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3902 fix memory access data type.
3903 (*avx512pf_gatherpf<mode>_mask): Ditto.
3904 (*avx512pf_gatherpf<mode>): Ditto.
3905 (avx512pf_scatterpf<mode>): Ditto.
3906 (*avx512pf_scatterpf<mode>_mask): Ditto.
3907 (*avx512pf_scatterpf<mode>): Ditto.
3908 (GATHER_SCATTER_SF_MEM_MODE): New.
3909 (avx512pf_gatherpf<mode>df): Ditto.
3910 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3911 (*avx512pf_scatterpf<mode>df): Ditto.
3913 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3916 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3917 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3920 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3922 * common/config/arm/arm-common.c
3923 (arm_rewrite_mcpu): Handle multiple names.
3925 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3927 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3929 * gimple-builder.h (create_gimple_tmp): Delete.
3931 2014-01-27 Christian Bruel <christian.bruel@st.com>
3933 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3936 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3938 * config/pa/pa.md (call): Generate indirect long calls to non-local
3939 functions when outputing 32-bit code.
3940 (call_value): Likewise except for special call to buggy powf function.
3942 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3943 portable runtime and PIC indirect calls.
3944 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3945 and PIC call sequences. Use ldo instead of blr to set return register
3946 in PIC call sequence.
3948 2014-01-25 Walter Lee <walt@tilera.com>
3950 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3951 avoid clobbering a live register.
3953 2014-01-25 Walter Lee <walt@tilera.com>
3955 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3956 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3957 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3958 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3960 2014-01-25 Walter Lee <walt@tilera.com>
3962 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3963 arguments on even registers.
3964 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3966 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3967 (BIGGEST_ALIGNMENT): Ditto.
3968 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3970 2014-01-25 Walter Lee <walt@tilera.com>
3972 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3973 insns before bundling.
3974 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3976 2014-01-25 Walter Lee <walt@tilera.com>
3978 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3979 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3980 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3982 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3984 * config/mips/constraints.md (kl): Delete.
3985 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3986 define expands, using...
3987 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3988 instructions for MIPS16.
3989 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3990 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3992 2014-01-25 Walter Lee <walt@tilera.com>
3994 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3998 2014-01-25 Walter Lee <walt@tilera.com>
4000 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4001 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4003 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4005 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4008 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4010 * print-rtl.c (in_call_function_usage): New var.
4011 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4012 EXPR_LIST mode as mode and not as reg note name.
4015 * cfgloopmanip.c (copy_loop_info): If
4016 loop->warned_aggressive_loop_optimizations, make sure
4017 the flag is set in target loop too.
4019 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4021 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4023 * builtins.def: Likewise.
4024 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4025 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4026 * ira.c (ira_setup_eliminable_regset): Likewise.
4027 * omp-low.c (gate_expand_omp): Likewise.
4028 (execute_lower_omp): Likewise.
4029 (diagnose_sb_0): Likewise.
4030 (gate_diagnose_omp_blocks): Likewise.
4031 (simd_clone_clauses_extract): Likewise.
4034 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4036 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4037 correction for little endian...
4038 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4041 2014-01-24 Jeff Law <law@redhat.com>
4043 PR tree-optimization/59919
4044 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4045 for non-returning calls.
4047 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4049 * common/config/aarch64/aarch64-common.c
4050 (aarch64_rewrite_mcpu): Handle multiple names.
4051 * config/aarch64/aarch64.h
4052 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4054 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4056 * input.c (add_file_to_cache_tab): Handle the case where fopen
4059 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4062 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4063 from push operand if code of push isn't PRE_DEC.
4065 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4068 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4069 -mquad-memory-atomic. Update -mquad-memory documentation to say
4070 it is only used for non-atomic loads/stores.
4072 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4073 -mquad-memory or -mquad-memory-atomic switches.
4075 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4076 -mquad-memory-atomic to ISA 2.07 support.
4078 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4079 to separate support of normal quad word memory operations (ldq, stq)
4080 from the atomic quad word memory operations.
4082 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4083 support to separate non-atomic quad word operations from atomic
4084 quad word operations. Disable non-atomic quad word operations in
4085 little endian mode so that we don't have to swap words after the
4086 load and before the store.
4087 (quad_load_store_p): Add comment about atomic quad word support.
4088 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4089 options printed with -mdebug=reg.
4091 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4092 -mquad-memory-atomic as the test for whether we have quad word
4093 atomic instructions.
4094 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4095 or -mp8-vector are used, allow byte/half-word atomic operations.
4097 * config/rs6000/sync.md (load_lockedti): Insure that the address
4098 is a proper indexed or indirect address for the lqarx instruction.
4099 On little endian systems, swap the hi/lo registers after the lqarx
4101 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4102 insure the address is valid for the lqarx instruction.
4103 (store_conditionalti): Insure that the address is a proper indexed
4104 or indirect address for the stqcrx. instruction. On little endian
4105 systems, swap the hi/lo registers before doing the stqcrx.
4107 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4108 insure the address is valid for the stqcrx. instruction.
4110 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4111 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4112 type of quad memory support is available.
4114 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4117 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4118 there is a danger of looping.
4120 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4122 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4123 force flag_ira_loop_pressure if set via command line.
4125 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4127 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4128 (ashr_simd): New builtin handling DI mode.
4129 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4130 (aarch64_sshr_simddi): New match pattern.
4131 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4132 (vshrd_n_s64): Likewise.
4133 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4135 2014-01-23 Nick Clifton <nickc@redhat.com>
4137 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4138 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4139 favour of mcu specific scripts.
4140 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4143 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4144 Alex Velenko <Alex.Velenko@arm.com>
4146 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4147 (vaddv_s16): Likewise.
4148 (vaddv_s32): Likewise.
4149 (vaddv_u8): Likewise.
4150 (vaddv_u16): Likewise.
4151 (vaddv_u32): Likewise.
4152 (vaddvq_s8): Likewise.
4153 (vaddvq_s16): Likewise.
4154 (vaddvq_s32): Likewise.
4155 (vaddvq_s64): Likewise.
4156 (vaddvq_u8): Likewise.
4157 (vaddvq_u16): Likewise.
4158 (vaddvq_u32): Likewise.
4159 (vaddvq_u64): Likewise.
4160 (vaddv_f32): Likewise.
4161 (vaddvq_f32): Likewise.
4162 (vaddvq_f64): Likewise.
4163 (vmaxv_f32): Likewise.
4164 (vmaxv_s8): Likewise.
4165 (vmaxv_s16): Likewise.
4166 (vmaxv_s32): Likewise.
4167 (vmaxv_u8): Likewise.
4168 (vmaxv_u16): Likewise.
4169 (vmaxv_u32): Likewise.
4170 (vmaxvq_f32): Likewise.
4171 (vmaxvq_f64): Likewise.
4172 (vmaxvq_s8): Likewise.
4173 (vmaxvq_s16): Likewise.
4174 (vmaxvq_s32): Likewise.
4175 (vmaxvq_u8): Likewise.
4176 (vmaxvq_u16): Likewise.
4177 (vmaxvq_u32): Likewise.
4178 (vmaxnmv_f32): Likewise.
4179 (vmaxnmvq_f32): Likewise.
4180 (vmaxnmvq_f64): Likewise.
4181 (vminv_f32): Likewise.
4182 (vminv_s8): Likewise.
4183 (vminv_s16): Likewise.
4184 (vminv_s32): Likewise.
4185 (vminv_u8): Likewise.
4186 (vminv_u16): Likewise.
4187 (vminv_u32): Likewise.
4188 (vminvq_f32): Likewise.
4189 (vminvq_f64): Likewise.
4190 (vminvq_s8): Likewise.
4191 (vminvq_s16): Likewise.
4192 (vminvq_s32): Likewise.
4193 (vminvq_u8): Likewise.
4194 (vminvq_u16): Likewise.
4195 (vminvq_u32): Likewise.
4196 (vminnmv_f32): Likewise.
4197 (vminnmvq_f32): Likewise.
4198 (vminnmvq_f64): Likewise.
4200 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4202 * config/aarch64/aarch64-simd.md
4203 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4204 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4205 (*aarch64_mul3_elt<mode>): Likewise.
4206 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4207 (*aarch64_mul3_elt_to_64v2df): Likewise.
4208 (*aarch64_mla_elt<mode>): Likewise.
4209 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4210 (*aarch64_mls_elt<mode>): Likewise.
4211 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4212 (*aarch64_fma4_elt<mode>): Likewise.
4213 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4214 (*aarch64_fma4_elt_to_64v2df): Likewise.
4215 (*aarch64_fnma4_elt<mode>): Likewise.
4216 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4217 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4218 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4219 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4220 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4221 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4222 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4223 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4224 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4226 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4228 * config/aarch64/aarch64-simd.md
4229 (aarch64_be_checked_get_lane<mode>): New define_expand.
4230 * config/aarch64/aarch64-simd-builtins.def
4231 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4232 New builtin definition.
4233 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4234 Use new safe be builtin.
4236 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4238 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4240 (aarch64_be_st1<mode>): Likewise.
4241 (aarch_ld1<VALL:mode>): Define_expand modified.
4242 (aarch_st1<VALL:mode>): Likewise.
4243 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4244 (UNSPEC_ST1): Likewise.
4246 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4248 * config/microblaze/microblaze.md: Add trap insn and attribute
4250 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4252 PR preprocessor/58580
4253 * input.h (location_get_source_line): Take an additional line_size
4255 (void diagnostics_file_cache_fini): Declare new function.
4256 * input.c (struct fcache): New type.
4257 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4258 New static constants.
4259 (diagnostic_file_cache_init, total_lines_num)
4260 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4261 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4262 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4263 (get_next_line, read_next_line, goto_next_line, read_line_num):
4264 New static function definitions.
4265 (diagnostic_file_cache_fini): New function.
4266 (location_get_source_line): Take an additional output line_len
4267 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4269 * diagnostic.c (diagnostic_finish): Call
4270 diagnostic_file_cache_fini.
4271 (adjust_line): Take an additional input parameter for the length
4272 of the line, rather than calculating it with strlen.
4273 (diagnostic_show_locus): Adjust the use of
4274 location_get_source_line and adjust_line with respect to their new
4275 signature. While displaying a line now, do not stop at the first
4276 null byte. Rather, display the zero byte as a space and keep
4277 going until we reach the size of the line.
4278 * Makefile.in: Add vec.o to OBJS-libcommon
4280 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4281 Ilya Tocar <ilya.tocar@intel.com>
4283 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4284 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4285 (__builtin_ia32_kmov16): Ditto.
4286 * config/i386/i386.md (UNSPEC_KMOV): New.
4289 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4291 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4292 (_mm512_storeu_si512): Ditto.
4294 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4297 * rtl.h (get_referenced_operands): Declare.
4298 * recog.c (get_referenced_operands): New function.
4299 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4300 operands have been referenced when recording LO_SUM references.
4302 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4304 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4306 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4308 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4309 Enable for generic and recent AMD targets.
4311 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4313 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4314 ARG_SIZE note when adjustment was eliminated.
4316 2014-01-22 Jeff Law <law@redhat.com>
4318 PR tree-optimization/59597
4319 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4320 in file. Accept new argument REGISTERING and use it to modify
4321 dump output appropriately.
4322 (register_jump_thread): Corresponding changes.
4323 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4324 thread paths involving joiner blocks. Add code to dump cancelled
4325 jump threading paths.
4327 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4329 PR rtl-optimization/59477
4330 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4331 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4333 2014-01-22 Tom Tromey <tromey@redhat.com>
4335 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4337 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4339 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4341 PR rtl-optimization/59896
4342 * lra-constraints.c (process_alt_operands): Check unused note for
4343 matched operands of insn with no output reloads.
4345 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4347 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4348 (mips_move_from_gpr_cost): Likewise.
4350 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4352 PR rtl-optimization/59858
4353 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4354 ira_class_hard_regs_num.
4355 (process_alt_operands): Increase reject for dying matched operand.
4357 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4360 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4361 smaller than size, perform several stores or loads and stores
4362 at dst + count - size to store or copy all of size bytes, rather
4363 than just last modesize bytes.
4365 2014-01-20 DJ Delorie <dj@redhat.com>
4367 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4368 that CLOBBERs are REGs before propogating their values.
4370 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4373 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4374 (cgraph_inline_failed_type): New function.
4375 * cgraph.h (DEFCIFCODE): Add type.
4376 (cgraph_inline_failed_type_t): New enum.
4377 (cgraph_inline_failed_type): New prototype.
4378 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4379 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4380 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4381 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4382 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4383 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4384 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4385 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4386 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4387 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4388 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4389 OPTIMIZATION_MISMATCH.
4390 * tree-inline.c (expand_call_inline): Emit errors during
4391 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4393 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4396 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4397 mode attribute in insn output.
4399 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4401 * output.h (output_constant): Delete.
4402 * varasm.c (output_constant): Make private.
4404 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4406 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4408 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4411 * tree.h (fold_builtin_strcat): New prototype.
4412 * builtins.c (fold_builtin_strcat): No longer static. Add len
4413 argument, if non-NULL, don't call c_strlen. Optimize
4414 directly into __builtin_memcpy instead of __builtin_strcpy.
4415 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4416 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4418 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4420 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4421 for SImode_address_operand operands, having only a REG argument.
4423 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4425 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4426 loader name using mbig-endian.
4427 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4429 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4431 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4435 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4437 * config/aarch64/aarch64-protos.h
4438 (aarch64_cannot_change_mode_class_ptr): Declare.
4439 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4440 aarch64_cannot_change_mode_class_ptr): New.
4441 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4442 backend hook aarch64_cannot_change_mode_class.
4444 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4446 * common/config/aarch64/aarch64-common.c
4447 (aarch64_handle_option): Don't handle any option order logic here.
4448 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4449 selected_cpu, warn on architecture version mismatch.
4450 (aarch64_override_options): Fix parsing order for option strings.
4452 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4453 Iain Sandoe <iain@codesourcery.com>
4456 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4457 warning. Amend comment to reflect current functionality.
4459 2014-01-20 Richard Biener <rguenther@suse.de>
4462 * builtins.c (fold_builtin_strcat): Remove case better handled
4463 by tree-ssa-strlen.c.
4465 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4467 * config/aarch64/aarch64.opt
4468 (mcpu, march, mtune): Make case-insensitive.
4470 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4473 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4474 if operands[1] is a REG or ZERO_EXTEND of a REG.
4476 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4478 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4480 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4482 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4483 long non-pic millicode calls.
4485 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4487 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4489 2014-01-19 Kito Cheng <kito@0xlab.org>
4491 * builtins.c (expand_movstr): Check movstr expand done or fail.
4493 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4494 H.J. Lu <hongjiu.lu@intel.com>
4497 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4498 to DImode for zero-extended addresses.
4500 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4502 PR rtl-optimization/57763
4503 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4504 on the new indirect jump_insn and increment LABEL_NUSES (label).
4506 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4510 * config.gcc (x86_archs): New variable.
4511 (x86_64_archs): Likewise.
4512 (x86_cpus): Likewise.
4513 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4514 --with-arch/--with-cpu= options.
4515 Support --with-arch=/--with-cpu={nehalem,westmere,
4516 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4518 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4520 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4521 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4523 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4525 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4527 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4530 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4531 clear cpp_get_options (parse_in)->warn_unused_macros for
4532 ix86_target_macros_internal with cpp_define.
4534 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4536 * jump.c (delete_related_insns): Keep (use (insn))s.
4537 * reorg.c (redundant_insn): Check for barriers too.
4539 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4541 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4543 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4545 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4546 call to $$dyncall when TARGET_LONG_CALLS is true.
4548 2014-01-17 Jeff Law <law@redhat.com>
4550 * ree.c (combine_set_extension): Temporarily disable test for
4551 changing number of hard registers.
4553 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4556 * ipa-inline-analysis.c (inline_free_summary):
4557 Do not free summary of aliases.
4559 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4562 * gimplify.c (gimplify_expr): Use create_tmp_var
4563 instead of create_tmp_var_raw. If cond doesn't have
4564 integral type, don't add the IFN_ANNOTATE builtin at all.
4566 2014-01-17 Martin Jambor <mjambor@suse.cz>
4569 * ipa-cp.c (prev_edge_clone): New variable.
4570 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4571 Also resize prev_edge_clone vector.
4572 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4573 (ipcp_edge_removal_hook): New function.
4574 (ipcp_driver): Register ipcp_edge_removal_hook.
4576 2014-01-17 Andrew Pinski <apinski@cavium.com>
4577 Steve Ellcey <sellcey@mips.com>
4580 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4583 2014-01-17 Jeff Law <law@redhat.com>
4586 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4587 so that pass_ccp runs first.
4589 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4591 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4592 (ix86_adjust_cost): Use !TARGET_XXX.
4593 (do_reorder_for_imul): Likewise.
4594 (swap_top_of_ready_list): Likewise.
4595 (ix86_sched_reorder): Likewise.
4597 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4599 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4600 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4601 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4602 (intel_memset): New. Duplicate slm_memset.
4603 (intel_cost): New. Duplicate slm_cost.
4604 (m_INTEL): New macro.
4605 (processor_target_table): Add "intel".
4606 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4607 with PROCESSOR_INTEL for "intel".
4608 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4609 PROCESSOR_SILVERMONT.
4610 (ix86_issue_rate): Likewise.
4611 (ix86_adjust_cost): Likewise.
4612 (ia32_multipass_dfa_lookahead): Likewise.
4613 (swap_top_of_ready_list): Likewise.
4614 (ix86_sched_reorder): Likewise.
4615 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4616 instead of TARGET_OPT_AGU.
4617 * config/i386/i386.h (TARGET_INTEL): New.
4618 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4619 (processor_type): Add PROCESSOR_INTEL.
4620 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4621 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4623 2014-01-17 Marek Polacek <polacek@redhat.com>
4626 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4629 2014-01-17 Richard Biener <rguenther@suse.de>
4631 PR tree-optimization/46590
4632 * opts.c (default_options_table): Add entries for
4633 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4634 all enabled at -O1 but not for -Og.
4635 * common.opt (fbranch-count-reg): Remove Init(1).
4636 (fmove-loop-invariants): Likewise.
4637 (ftree-pta): Likewise.
4639 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4641 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4642 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4643 decls to at least the GCC 4.8 used alignments.
4646 * tree-nested.c (convert_nonlocal_reference_stmt,
4647 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4648 of GIMPLE_BIND stmts, adjust associated decls.
4650 2014-01-17 Richard Biener <rguenther@suse.de>
4652 PR tree-optimization/46590
4653 * vec.h (vec<>::bseach): New member function implementing
4654 binary search according to C89 bsearch.
4655 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4656 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4657 bitmap pointer again. Make accesses_in_loop a flat array.
4658 (mem_ref_obstack): New global.
4659 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4660 (mark_ref_stored): Likewise.
4661 (ref_indep_loop_p_2): Likewise.
4662 (set_ref_stored_in_loop): New helper function.
4663 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4664 (memref_free): Adjust.
4665 (record_mem_ref_loc): Simplify.
4666 (gather_mem_refs_stmt): Adjust.
4667 (sort_locs_in_loop_postorder_cmp): New function.
4668 (analyze_memory_references): Sort accesses_in_loop after
4669 loop postorder number.
4670 (find_ref_loc_in_loop_cmp): New function.
4671 (for_all_locs_in_loop): Find relevant cluster of locs in
4672 accesses_in_loop and iterate without recursion.
4673 (execute_sm): Avoid uninit warning.
4674 (struct ref_always_accessed): Simplify.
4675 (ref_always_accessed::operator ()): Likewise.
4676 (ref_always_accessed_p): Likewise.
4677 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4678 loop postorder numbers here.
4679 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4682 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4685 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4686 on decls for which assemble_alias has been called.
4688 2014-01-17 Nick Clifton <nickc@redhat.com>
4690 * config/msp430/msp430.opt: (mcpu): New option.
4691 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4692 (msp430_option_override): Parse target_cpu. If the MCU name
4693 matches a generic string, clear target_mcu.
4694 (msp430_attr): Allow numeric interrupt values up to 63.
4695 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4696 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4698 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4700 * config/msp430/msp430.md (popm): Use %J rather than %I.
4701 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4702 (addhi_cy_i): Use immediate_operand for operand 2.
4703 * doc/invoke.texi: Document -mcpu option.
4705 2014-01-17 Richard Biener <rguenther@suse.de>
4707 PR rtl-optimization/38518
4708 * df.h (df_analyze_loop): Declare.
4709 * df-core.c: Include cfgloop.h.
4710 (df_analyze_1): Split out main part of df_analyze.
4711 (df_analyze): Adjust.
4712 (loop_inverted_post_order_compute): New function.
4713 (loop_post_order_compute): Likewise.
4714 (df_analyze_loop): New function avoiding whole-function
4716 * loop-invariant.c (find_defs): Use df_analyze_loop.
4717 (find_invariants): Adjust.
4718 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4720 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4722 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4723 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4725 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4727 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4728 traversal when removing references.
4730 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4733 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4735 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4738 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4739 pushing a reload for an autoinc when we had previously reloaded an
4740 inner part of the address.
4742 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4744 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4746 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4747 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4748 when not giving up or versioning for alias only because of
4750 (vect_analyze_data_ref_dependences): Set to true.
4751 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4753 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4754 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4758 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4761 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4762 operand 0 predicate for gathers, use a new pseudo as subtarget.
4764 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4767 * lra-constraints.c (process_alt_operands): Add printing debug info.
4768 Check absence of input/output reloads for matched operands too.
4770 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4772 PR rtl-optimization/59835
4773 * ira.c (ira_init_register_move_cost): Increase cost for
4776 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4778 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4780 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4783 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4784 non-register objects. Use gen_(high/low)part more consistently.
4787 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4790 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4791 endian support, remove tests for WORDS_BIG_ENDIAN.
4792 (p8_mfvsrd_3_<mode>): Likewise.
4793 (reload_gpr_from_vsx<mode>): Likewise.
4794 (reload_gpr_from_vsxsf): Likewise.
4795 (p8_mfvsrd_4_disf): Likewise.
4797 2014-01-16 Richard Biener <rguenther@suse.de>
4799 PR rtl-optimization/46590
4800 * lcm.c (compute_antinout_edge): Use postorder iteration.
4801 (compute_laterin): Use inverted postorder iteration.
4803 2014-01-16 Nick Clifton <nickc@redhat.com>
4806 * varasm.c (output_constant): Return the number of bytes actually
4808 (output_constructor_array_range): Update the field size with the
4809 number of bytes emitted by output_constant.
4810 (output_constructor_regular_field): Likewise. Also do not
4811 complain if the total number of bytes emitted is now greater
4812 than the expected fieldpos.
4813 * output.h (output_constant): Update prototype and descriptive comment.
4815 2014-01-16 Marek Polacek <polacek@redhat.com>
4818 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4819 it is error_mark_node.
4821 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4823 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4824 VALID_AVX256_REG_OR_OI_MODE.
4826 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4828 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4829 current procedure should be profiled.
4831 2014-01-15 Andrew Pinski <apinski@cavium.com>
4833 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4834 of moving from/to the STACK_REG register class.
4836 2014-01-15 Richard Henderson <rth@redhat.com>
4839 * reginfo.c (global_regs_decl): Globalize.
4840 * rtl.h (global_regs_decl): Declare.
4841 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4842 reserved via global_regs.
4844 2014-01-15 Teresa Johnson <tejohnson@google.com>
4846 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4848 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4850 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4851 and vmulosh rather than call gen_vec_widen_smult_*.
4852 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4853 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4854 (vec_widen_smult_even_v16qi): Likewise.
4855 (vec_widen_umult_even_v8hi): Likewise.
4856 (vec_widen_smult_even_v8hi): Likewise.
4857 (vec_widen_umult_odd_v16qi): Likewise.
4858 (vec_widen_smult_odd_v16qi): Likewise.
4859 (vec_widen_umult_odd_v8hi): Likewise.
4860 (vec_widen_smult_odd_v8hi): Likewise.
4861 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4862 vmuloub rather than call gen_vec_widen_umult_*.
4863 (vec_widen_umult_lo_v16qi): Likewise.
4864 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4865 vmulosb rather than call gen_vec_widen_smult_*.
4866 (vec_widen_smult_lo_v16qi): Likewise.
4867 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4868 rather than call gen_vec_widen_umult_*.
4869 (vec_widen_umult_lo_v8hi): Likewise.
4870 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4871 rather than call gen_vec_widen_smult_*.
4872 (vec_widen_smult_lo_v8hi): Likewise.
4874 2014-01-15 Jeff Law <law@redhat.com>
4876 PR tree-optimization/59747
4877 * ree.c (find_and_remove_re): Properly handle case where a second
4878 eliminated extension requires widening a copy created for elimination
4879 of a prior extension.
4880 (combine_set_extension): Ensure that the number of hard regs needed
4881 for a destination register does not change when we widen it.
4883 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4885 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4886 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4887 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4888 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4889 (avr-*-rtems*): Likewise.
4890 (bfin*-rtems*): Likewise.
4891 (moxie-*-rtems*): Likewise.
4892 (h8300-*-rtems*): Likewise.
4893 (i[34567]86-*-rtems*): Likewise.
4894 (lm32-*-rtems*): Likewise.
4895 (m32r-*-rtems*): Likewise.
4896 (m68k-*-rtems*): Likewise.
4897 (microblaze*-*-rtems*): Likewise.
4898 (mips*-*-rtems*): Likewise.
4899 (powerpc-*-rtems*): Likewise.
4900 (sh-*-rtems*): Likewise.
4901 (sparc-*-rtems*): Likewise.
4902 (sparc64-*-rtems*): Likewise.
4903 (v850-*-rtems*): Likewise.
4904 (m32c-*-rtems*): Likewise.
4906 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4908 PR rtl-optimization/59511
4909 * ira.c (ira_init_register_move_cost): Use memory costs for some
4910 cases of register move cost calculations.
4911 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4912 instead of BB frequency.
4913 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4914 * lra-assigns.c (find_hard_regno_for): Ditto.
4916 2014-01-15 Richard Biener <rguenther@suse.de>
4918 PR tree-optimization/59822
4919 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4920 (vectorizable_load): Use it to hoist defs of uses of invariant
4921 loads out of the loop.
4923 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4924 Kugan Vivekanandarajah <kuganv@linaro.org>
4927 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4930 2014-01-15 Richard Biener <rguenther@suse.de>
4932 PR rtl-optimization/59802
4933 * lcm.c (compute_available): Use inverted postorder to seed
4934 the initial worklist.
4936 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4939 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4940 ADDR_REGS for invalid symrefs in non-PIC code.
4942 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4945 * builtins.c (determine_block_size): Initialize *probable_max_size
4946 even if len_rtx is CONST_INT.
4948 2014-01-14 Andrew Pinski <apinski@cavium.com>
4950 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4951 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4952 (cortexa53_tunings): Likewise.
4953 (aarch64_sched_issue_rate): New function.
4954 (TARGET_SCHED_ISSUE_RATE): Define.
4956 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4958 * ira-costs.c (find_costs_and_classes): Add missed
4959 ira_init_register_move_cost_if_necessary.
4961 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4964 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4966 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4969 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4970 to indicate if type is used for function return value. Warn ABI
4971 change if the vector mode isn't available for function return value.
4972 (ix86_function_arg_advance): Pass false to type_natural_mode.
4973 (ix86_function_arg): Likewise.
4974 (ix86_gimplify_va_arg): Likewise.
4975 (function_arg_32): Don't warn ABI change.
4976 (ix86_function_value): Pass true to type_natural_mode.
4977 (ix86_return_in_memory): Likewise.
4978 (ix86_struct_value_rtx): Removed.
4979 (TARGET_STRUCT_VALUE_RTX): Likewise.
4981 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4983 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4984 converting a conditional jump into a conditional return.
4986 2014-01-14 Richard Biener <rguenther@suse.de>
4988 PR tree-optimization/58921
4989 PR tree-optimization/59006
4990 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4991 hoisting invariant stmts.
4992 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4993 invariant loads on the preheader edge if possible.
4995 2014-01-14 Joey Ye <joey.ye@arm.com>
4997 * doc/plugin.texi (Building GCC plugins): Update to C++.
4999 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5001 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5002 (_mm_rcp28_round_ss): Ditto.
5003 (_mm_rsqrt28_round_sd): Ditto.
5004 (_mm_rsqrt28_round_ss): Ditto.
5005 (_mm_rcp28_sd): Ditto.
5006 (_mm_rcp28_ss): Ditto.
5007 (_mm_rsqrt28_sd): Ditto.
5008 (_mm_rsqrt28_ss): Ditto.
5009 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5010 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5011 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5012 (IX86_BUILTIN_RCP28SD): Ditto.
5013 (IX86_BUILTIN_RCP28SS): Ditto.
5014 (IX86_BUILTIN_RSQRT28SD): Ditto.
5015 (IX86_BUILTIN_RSQRT28SS): Ditto.
5016 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5017 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5018 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5019 (ix86_expand_special_args_builtin): Expand new FTYPE.
5020 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5021 (srcp14<mode>): Make insn unary.
5022 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5023 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5024 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5025 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5026 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5027 Fix rounding: make it SAE only.
5028 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5030 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5032 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5033 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5034 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5035 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5036 (round_saeonly_mask_scalar_operand4): Ditto.
5037 (round_saeonly_mask_scalar_op3): Ditto.
5038 (round_saeonly_mask_scalar_op4): Ditto.
5040 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5042 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5043 Implement -maltivec=be for vec_insert and vec_extract.
5045 2014-01-10 DJ Delorie <dj@redhat.com>
5047 * config/msp430/msp430.md (call_internal): Don't allow memory
5048 references with SP as the base register.
5049 (call_value_internal): Likewise.
5050 * config/msp430/constraints.md (Yc): New. For memory references
5051 that don't use SP as a base register.
5053 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5054 "an integer without a # prefix"
5055 * config/msp430/msp430.md (epilogue_helper): Use it.
5057 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5060 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5061 AVX512F gather builtins.
5062 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5063 on gather decls with INTEGER_TYPE masktype.
5064 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5065 directly into the builtin rather than hoisting it before loop.
5067 PR tree-optimization/59387
5068 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5069 (scev_const_prop): If folded_casts and type has undefined overflow,
5070 use force_gimple_operand instead of force_gimple_operand_gsi and
5071 for each added stmt if it is assign with
5072 arith_code_with_undefined_signed_overflow, call
5073 rewrite_to_defined_overflow.
5074 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5075 gimple-fold.h instead.
5076 (arith_code_with_undefined_signed_overflow,
5077 rewrite_to_defined_overflow): Moved to ...
5078 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5079 rewrite_to_defined_overflow): ... here. No longer static.
5080 Include gimplify-me.h.
5081 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5082 rewrite_to_defined_overflow): New prototypes.
5084 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5086 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5088 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5090 * builtins.c (get_object_alignment_2): Minor tweak.
5091 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5093 2014-01-13 Christian Bruel <christian.bruel@st.com>
5095 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5096 optimized non constant lengths.
5098 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5101 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5102 load as __atomic_load_N if possible.
5104 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5106 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5108 (rs6000_expand_builtin): Adjust call.
5110 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5113 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5114 * config/rs6000/rs6000.c: Include target-globals.h.
5115 (rs6000_set_current_function): Instead of doing target_reinit
5116 unconditionally, use save_target_globals_default_opts and
5117 restore_target_globals.
5119 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5121 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5122 (rs6000_expand_builtin): Handle mffs and mtfsf.
5123 (rs6000_init_builtins): Define mffs and mtfsf.
5124 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5125 (rs6000_mffs): New pattern.
5126 (rs6000_mtfsf): New pattern.
5128 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5130 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5131 Start narrowing with START. Apply candidate-use pair
5132 and check overall cost in narrowing.
5133 (iv_ca_prune): Pass new argument.
5135 2014-01-10 Jeff Law <law@redhat.com>
5138 * ree.c (combine_reaching_defs): Ensure the defining statement
5139 occurs before the extension when optimizing extensions with
5140 different source and destination hard registers.
5142 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5145 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5146 vtables into the type inheritance graph.
5148 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5150 PR rtl-optimization/59754
5151 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5152 modes in the REGNO != REGNO case.
5154 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5156 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5158 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5160 PR tree-optimization/59745
5161 * tree-predcom.c (tree_predictive_commoning_loop): Call
5162 free_affine_expand_cache if giving up because components is NULL.
5164 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5165 GC in payload of target_globals struct instead of allocating them on
5166 the heap and the larger structs separately using GC.
5167 * target-globals.h (struct target_globals): Make regs, hard_regs,
5168 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5169 of GTY((skip)) and change type to void *.
5170 (reset_target_globals): Cast loads from those fields to corresponding
5173 2014-01-10 Steve Ellcey <sellcey@mips.com>
5176 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5177 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5178 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5180 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5183 * aarch64-modes.def (CC_Zmode): New flags mode.
5184 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5185 represents an equality.
5186 (aarch64_get_condition_code): Handle CC_Zmode.
5187 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5189 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5191 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5192 extraction in good case.
5194 2014-01-10 Richard Biener <rguenther@suse.de>
5196 PR tree-optimization/59374
5197 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5198 checking after SLP discovery. Mark stmts not participating
5199 in any SLP instance properly.
5201 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5203 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5204 when handling a SET rtx.
5206 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5208 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5209 (cortex-a57): Likewise.
5210 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5212 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5214 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5215 non-iwmmxt builtins.
5217 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5221 * ipa-devirt.c record_target_from_binfo): Take as argument
5222 stack of binfos and lookup matching one for virtual inheritance.
5223 (possible_polymorphic_call_targets_1): Update.
5225 2014-01-10 Huacai Chen <chenhc@lemote.com>
5227 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5228 kernel strings for Loongson-2E/2F/3A.
5230 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5233 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5234 is_gimple_call before calling gimple_call_internal_p.
5236 2014-01-09 Steve Ellcey <sellcey@mips.com>
5238 * Makefile.in (TREE_FLOW_H): Remove.
5239 (TREE_SSA_H): Add file names from tree-flow.h.
5240 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5241 * tree.h: Remove tree-flow.h reference.
5242 * hash-table.h: Remove tree-flow.h reference.
5243 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5244 reference with tree-ssa-loop.h.
5246 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5248 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5249 default element-order behavior for -maltivec.
5250 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5251 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5252 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5253 when targeting big endian, at least for now.
5254 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5256 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5259 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5260 var satisfies use_register_for_decl, just take into account type
5261 alignment, rather than decl alignment.
5263 PR tree-optimization/59622
5264 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5265 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5266 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5267 Don't devirtualize for inplace at all. For targets.length () == 1,
5268 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5270 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5272 * config/i386/i386.md (cpu): Remove the unused btver1.
5274 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5276 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5278 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5281 * tree-core.h (struct target_globals): New forward declaration.
5282 (struct tree_target_option): Add globals field.
5283 * tree.h (TREE_TARGET_GLOBALS): Define.
5284 (prepare_target_option_nodes_for_pch): New prototype.
5285 * target-globals.h (struct target_globals): Define even if
5287 * tree.c (prepare_target_option_node_for_pch,
5288 prepare_target_option_nodes_for_pch): New functions.
5289 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5290 * config/i386/i386.c: Include target-globals.h.
5291 (ix86_set_current_function): Instead of doing target_reinit
5292 unconditionally, use save_target_globals_default_opts and
5293 restore_target_globals.
5295 2014-01-09 Richard Biener <rguenther@suse.de>
5297 PR tree-optimization/59715
5298 * tree-cfg.h (split_critical_edges): Declare.
5299 * tree-cfg.c (split_critical_edges): Export.
5300 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5302 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5304 * cfgexpand.c (expand_stack_vars): Optionally disable
5305 asan stack protection.
5306 (expand_used_vars): Likewise.
5307 (partition_stack_vars): Likewise.
5308 * asan.c (asan_emit_stack_protection): Optionally disable
5309 after return stack usage.
5310 (instrument_derefs): Optionally disable memory access instrumentation.
5311 (instrument_builtin_call): Likewise.
5312 (instrument_strlen_call): Likewise.
5313 (asan_protect_global): Optionally disable global variables protection.
5314 * doc/invoke.texi: Added doc for new options.
5315 * params.def: Added new options.
5316 * params.h: Likewise.
5318 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5320 PR rtl-optimization/59724
5321 * ifcvt.c (cond_exec_process_if_block): Don't call
5322 flow_find_head_matching_sequence with 0 longest_match.
5323 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5324 non-active insns if !stop_after.
5325 (try_head_merge_bb): Revert 2014-01-07 changes.
5327 2014-01-08 Jeff Law <law@redhat.com>
5329 * ree.c (get_sub_rtx): New function, extracted from...
5330 (merge_def_and_ext): Here.
5331 (combine_reaching_defs): Use get_sub_rtx.
5333 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5335 * cgraph.h (varpool_variable_node): Do not choke on null node.
5337 2014-01-08 Catherine Moore <clm@codesourcery.com>
5339 * config/mips/mips.md (simple_return): Attempt to use JRC
5341 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5343 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5345 PR rtl-optimization/59137
5346 * reorg.c (steal_delay_list_from_target): Call update_block for
5348 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5350 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5352 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5353 two duplicate entries.
5355 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5358 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5360 * config/mips/mips.c (mips_truncated_op_cost): New function.
5361 (mips_rtx_costs): Adjust test for BADDU.
5362 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5364 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5366 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5367 (*baddu_si): ...this new pattern.
5369 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5372 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5374 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5377 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5379 (expand_expr, expand_normal): Adjust.
5380 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5381 inner_reference_p. Use inner_reference_p to expand inner references.
5382 (store_expr): Adjust.
5383 * cfgexpand.c (expand_call_stmt): Adjust.
5385 2014-01-08 Rong Xu <xur@google.com>
5387 * gcov-io.c (gcov_var): Move from gcov-io.h.
5388 (gcov_position): Ditto.
5389 (gcov_is_error): Ditto.
5390 (gcov_rewrite): Ditto.
5391 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5392 only part to libgcc/libgcov.h.
5394 2014-01-08 Marek Polacek <polacek@redhat.com>
5397 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5399 2014-01-08 Marek Polacek <polacek@redhat.com>
5402 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5404 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5406 PR rtl-optimization/59649
5407 * stor-layout.c (get_mode_bounds): For BImode return
5408 0 and STORE_FLAG_VALUE.
5410 2014-01-08 Richard Biener <rguenther@suse.de>
5413 * gimple.h (is_gimple_builtin_call): Remove.
5414 (gimple_builtin_call_types_compatible_p): New.
5415 (gimple_call_builtin_p): New overload.
5416 * gimple.c (is_gimple_builtin_call): Remove.
5417 (validate_call): Rename to ...
5418 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5420 (validate_type): New static function.
5421 (gimple_call_builtin_p): New overload and adjust.
5422 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5423 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5424 (gimple_fold_stmt_to_constant_1): Likewise.
5425 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5427 2014-01-08 Richard Biener <rguenther@suse.de>
5430 * gimplify.c (gimplify_expr): Gimplify register-register type
5431 VIEW_CONVERT_EXPRs to separate stmts.
5433 2014-01-07 Jeff Law <law@redhat.com>
5436 * ree.c (combine_set_extension): Handle case where source
5437 and destination registers in an extension insn are different.
5438 (combine_reaching_defs): Allow source and destination registers
5439 in extension to be different under limited circumstances.
5440 (add_removable_extension): Remove restriction that the
5441 source and destination registers in the extension are the same.
5442 (find_and_remove_re): Emit a copy from the extension's
5443 destination to its source after the defining insn if
5444 the source and destination registers are different.
5447 * ifcvt.c (merge_if_block): If we are merging a block with more than
5448 one successor with a block with no successors, remove any BARRIER
5449 after the second block.
5451 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5453 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5455 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5458 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5459 for 14-bit register offsets when INT14_OK_STRICT is false.
5461 2014-01-07 Roland Stigge <stigge@antcom.de>
5462 Michael Meissner <meissner@linux.vnet.ibm.com>
5465 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5466 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5468 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5470 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5473 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5475 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5476 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5477 rtx is const0_rtx or not.
5479 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5482 * target-globals.c (save_target_globals): Remove this_fn_optab
5484 * toplev.c: Include optabs.h.
5485 (target_reinit): Temporarily restore the global options if another
5486 set of options are in force.
5488 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5490 PR rtl-optimization/58668
5491 * cfgcleanup.c (flow_find_cross_jump): Don't count
5492 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5493 to determine what is counted.
5494 (flow_find_head_matching_sequence): Use active_insn_p to determine
5496 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5498 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5499 determine what is counted.
5501 PR tree-optimization/59643
5502 * tree-predcom.c (split_data_refs_to_components): If one dr is
5503 read and one write, determine_offset fails and the write isn't
5504 in the bad component, just put the read into the bad component.
5506 2014-01-07 Mike Stump <mikestump@comcast.net>
5507 Jakub Jelinek <jakub@redhat.com>
5510 * tree-core.h (struct tree_optimization_option): Change optabs
5511 type from unsigned char * to void *.
5512 * optabs.c (init_tree_optimization_optabs): Adjust
5513 TREE_OPTIMIZATION_OPTABS initialization.
5515 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5518 * config/i386/i386.h (struct machine_function): Add
5519 no_drap_save_restore field.
5520 * config/i386/i386.c (ix86_save_reg): Use
5521 !cfun->machine->no_drap_save_restore instead of
5522 crtl->stack_realign_needed.
5523 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5524 this function clears frame_pointer_needed. Set
5525 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5526 and DRAP reg is needed.
5528 2014-01-06 Marek Polacek <polacek@redhat.com>
5531 * doc/implement-c.texi: Mention that other integer types are
5532 permitted as bit-field types in strictly conforming mode.
5534 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5536 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5539 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5541 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5543 2014-01-06 Martin Jambor <mjambor@suse.cz>
5546 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5548 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5550 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5554 * var-tracking.c (add_stores): Preserve the value of the source even if
5555 we don't record the store.
5557 2014-01-06 Terry Guo <terry.guo@arm.com>
5559 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5561 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5564 * config/darwin.c (darwin_function_section): Adjust return values to
5565 correspond to optimisation changes made in r206070.
5567 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5569 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5570 from prefetch_block tune setting.
5571 (nocona_cost): Correct size of prefetch block to 64.
5573 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5575 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5576 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5577 used to save the static chain register in the computation of the offset
5578 from which the FP registers need to be restored.
5580 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5582 PR tree-optimization/59519
5583 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5584 ICE if get_current_def (current_new_name) is already non-NULL, as long
5585 as it is a phi result of some other phi in *new_exit_bb that has
5588 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5589 or vmovdqu* for misaligned_operand.
5590 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5591 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5592 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5593 aligned_mem for AVX512F masked aligned load and store builtins and for
5596 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5598 PR tree-optimization/59651
5599 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5600 Address range for negative step should be added by TYPE_SIZE_UNIT.
5602 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5604 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5605 overlapping registers also for registers other than the stack pointer.
5607 2014-01-03 Marek Polacek <polacek@redhat.com>
5610 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5613 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5616 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5619 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5620 (push splitter): Use <P:MODE_SIZE> instead of
5621 GET_MODE_SIZE (<P:MODE>mode).
5622 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5623 (mov -1, reg peephole2): Likewise.
5624 * config/i386/sse.md (*mov<mode>_internal,
5625 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5626 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5627 *<code><mode>3, *andnot<mode>3<mask_name>,
5628 <mask_codefor><code><mode>3<mask_name>): Likewise.
5629 * config/i386/subst.md (mask_mode512bit_condition,
5630 sd_mask_mode512bit_condition): Likewise.
5632 2014-01-02 Xinliang David Li <davidxl@google.com>
5634 PR tree-optimization/59303
5635 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5636 (dump_predicates): Better output format.
5637 (pred_equal_p): New function.
5638 (is_neq_relop_p): Ditto.
5639 (is_neq_zero_form_p): Ditto.
5640 (pred_expr_equal_p): Ditto.
5641 (pred_neg_p): Ditto.
5642 (simplify_pred): Ditto.
5643 (simplify_preds_2): Ditto.
5644 (simplify_preds_3): Ditto.
5645 (simplify_preds_4): Ditto.
5646 (simplify_preds): Ditto.
5648 (push_to_worklist): Ditto.
5649 (get_pred_info_from_cmp): Ditto.
5650 (is_degenerated_phi): Ditto.
5651 (normalize_one_pred_1): Ditto.
5652 (normalize_one_pred): Ditto.
5653 (normalize_one_pred_chain): Ditto.
5654 (normalize_preds): Ditto.
5655 (normalize_cond_1): Remove function.
5656 (normalize_cond): Ditto.
5657 (is_gcond_subset_of): Ditto.
5658 (is_subset_of_any): Ditto.
5659 (is_or_set_subset_of): Ditto.
5660 (is_and_set_subset_of): Ditto.
5661 (is_norm_cond_subset_of): Ditto.
5662 (pred_chain_length_cmp): Ditto.
5663 (convert_control_dep_chain_into_preds): Type change.
5664 (find_predicates): Ditto.
5665 (find_def_preds): Ditto.
5666 (destroy_predicates_vecs): Ditto.
5667 (find_matching_predicates_in_rest_chains): Ditto.
5668 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5669 (is_pred_expr_subset): Ditto.
5670 (is_pred_chain_subset_of): Ditto.
5671 (is_included_in): Ditto.
5672 (is_superset_of): Ditto.
5674 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5676 Update copyright years.
5678 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5680 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5681 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5682 config/arc/arc.md, config/arc/arc.opt,
5683 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5684 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5685 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5686 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5687 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5688 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5689 vtable-verify.c, vtable-verify.h: Use the standard form for the
5692 2014-01-02 Tobias Burnus <burnus@net-b.de>
5694 * gcc.c (process_command): Update copyright notice dates.
5695 * gcov-dump.c: Ditto.
5697 * doc/cpp.texi: Bump @copying's copyright year.
5698 * doc/cppinternals.texi: Ditto.
5699 * doc/gcc.texi: Ditto.
5700 * doc/gccint.texi: Ditto.
5701 * doc/gcov.texi: Ditto.
5702 * doc/install.texi: Ditto.
5703 * doc/invoke.texi: Ditto.
5705 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5707 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5709 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5711 * config/i386/sse.md (*mov<mode>_internal): Guard
5712 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5714 PR rtl-optimization/59647
5715 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5716 new_rtx into UNSIGNED_FLOAT rtxes.
5718 Copyright (C) 2014 Free Software Foundation, Inc.
5720 Copying and distribution of this file, with or without modification,
5721 are permitted in any medium without royalty provided the copyright
5722 notice and this notice are preserved.