1 2014-03-31 Ulrich Drepper <drepper@gmail.com>
3 * config/i386/avx512fintrin.h (__v32hi): Define type.
5 (_mm512_set1_epi8): Define.
6 (_mm512_set1_epi16): Define.
7 (_mm512_set4_epi32): Define.
8 (_mm512_set4_epi64): Define.
9 (_mm512_set4_pd): Define.
10 (_mm512_set4_ps): Define.
11 (_mm512_setr4_epi64): Define.
12 (_mm512_setr4_epi32): Define.
13 (_mm512_setr4_pd): Define.
14 (_mm512_setr4_ps): Define.
15 (_mm512_setzero_epi32): Define.
17 2014-03-31 Martin Jambor <mjambor@suse.cz>
20 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
21 callsite_arguments_match_p. Updated all callers. Also check types of
22 corresponding formal parameters and actual arguments.
23 (not_all_callers_have_enough_arguments_p) Renamed to
24 some_callers_have_mismatched_arguments_p.
26 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
28 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
30 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
33 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
36 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
38 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
39 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
41 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
42 Use FMAMODE_NOVF512 mode iterator.
43 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
44 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
45 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
46 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
48 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
49 Use VF_128_256 mode iterator.
50 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
53 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
55 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
56 static chain if needed.
58 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
61 * lra-constraints.c (index_part_to_reg): New.
62 (process_address): Use it.
64 2014-03-27 Jeff Law <law@redhat.com>
65 Jakub Jelinek <jakub@redhat.com>
68 * expr.c (do_tablejump): Use simplify_gen_binary rather than
69 gen_rtx_{PLUS,MULT} to build up the address expression.
71 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
72 creating non-canonical RTL.
74 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
77 * ipa-inline.c (want_inline_small_function_p): Short circuit large
78 functions; reorganize to make cheap checks first.
79 (inline_small_functions): Do not estimate growth when dumping;
81 * ipa-inline.h (inline_summary): Add min_size.
82 (growth_likely_positive): New function.
83 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
84 (set_cond_stmt_execution_predicate): Cleanup.
85 (estimate_edge_size_and_time): Compute min_size.
86 (estimate_calls_size_and_time): Likewise.
87 (estimate_node_size_and_time): Likewise.
88 (inline_update_overall_summary): Update min_size.
89 (do_estimate_edge_time): Likewise.
90 (do_estimate_edge_size): Update.
91 (do_estimate_edge_hints): Update.
92 (growth_likely_positive): New function.
94 2014-03-28 Jakub Jelinek <jakub@redhat.com>
97 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
98 also if addr has VOIDmode.
100 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
102 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
103 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
105 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
106 instructions as well as AdvancedSIMD loads.
108 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
110 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
111 Use crypto_aese type.
112 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
113 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
114 crypto_aese, crypto_aesmc. Move to types.md.
115 * config/arm/types.md (crypto_aes): Split into crypto_aese,
117 * config/arm/iterators.md (crypto_type): Likewise.
119 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
121 * cgraph.c: Include expr.h and tree-dfa.h.
122 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
125 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
128 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
129 regs from checking multi-reg pseudos.
131 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
133 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
135 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
137 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
138 if it would clobber the stack pointer, even temporarily.
140 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
142 * mode-switching.c: Make small adjustments to the top comment.
144 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
146 * config/rs6000/constraints.md (wD constraint): New constraint to
147 match the constant integer to get the top DImode/DFmode out of a
148 vector in a VSX register.
150 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
151 match the constant integer to get the top DImode/DFmode out of a
152 vector in a VSX register.
154 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
157 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
160 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
161 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
163 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
164 Optimize vec_extract of 64-bit values, where the value being
165 extracted is in the top word, where we can use scalar
166 instructions. Add direct move and store support. Combine the big
167 endian/little endian vector select load support into a single insn.
168 (vsx_extract_<mode>_internal1): Likewise.
169 (vsx_extract_<mode>_internal2): Likewise.
170 (vsx_extract_<mode>_load): Likewise.
171 (vsx_extract_<mode>_store): Likewise.
172 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
173 combined into vsx_extract_<mode>_load.
174 (vsx_extract_<mode>_one_le): Likewise.
176 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
177 define the top 64-bit vector element.
179 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
182 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
183 Document vec_vbpermq builtin.
186 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
187 enable use of xxsldwi and xxpermdi builtin functions.
188 (vec_xxpermdi): Likewise.
190 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
191 Document use of vec_xxsldwi and vec_xxpermdi builtins.
193 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
195 PR rtl-optimization/60650
196 * lra-asign.c (find_hard_regno_for, spill_for): Add parameter
198 (find_spills_for): New.
199 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
200 Spill all pseudos on the second iteration.
202 2014-03-27 Marek Polacek <polacek@redhat.com>
205 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
208 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
210 * config/s390/s390.c (s390_can_use_return_insn): Check for
211 call-saved FPRs on 31 bit.
213 2014-03-27 Jakub Jelinek <jakub@redhat.com>
216 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
217 if they need regimplification, just drop them instead of
218 calling gimple_regimplify_operands on them.
220 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
223 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
224 (aarch64_frame_pointer_required): Adjust logic.
225 (aarch64_can_eliminate): Adjust logic.
226 (aarch64_override_options_after_change): Adjust logic.
228 2014-03-27 Dehao Chen <dehao@google.com>
230 * ipa-inline.c (early_inliner): Update node's inline info.
232 2014-03-26 Dehao Chen <dehao@google.com>
234 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
235 compiler inserted conditional jumps for NAN float check.
237 2014-03-26 Jakub Jelinek <jakub@redhat.com>
239 * ubsan.h (ubsan_create_data): Change second argument's type
240 to const location_t *.
241 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
243 (ubsan_create_data): Change second argument to const location_t *PLOC.
244 Create Loc field whenever PLOC is non-NULL.
245 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
246 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
250 * real.c (real_to_integer2): Change type of low to UHWI.
252 2014-03-26 Tobias Burnus <burnus@net-b.de>
254 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
255 (CILK_SELF_SPECS): New define.
256 (driver_self_specs): Use it.
258 2014-03-26 Richard Biener <rguenther@suse.de>
260 * tree-pretty-print.c (percent_K_format): Implement special
261 case for LTO and its stripped down BLOCK tree.
263 2014-03-26 Jakub Jelinek <jakub@redhat.com>
266 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
268 * tree-vrp.c (simplify_internal_call_using_ranges): If only
269 one range is range_int_cst_p, but not both, at least optimize
270 addition/subtraction of 0 and multiplication by 0 or 1.
271 * gimple-fold.c (gimple_fold_call): Fold
272 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
273 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
274 INTEGER_CSTs, try to fold at least x * 0 and y - y.
276 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
278 PR rtl-optimization/60452
279 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
280 <case REG>: Return 1 for invalid offsets from the frame pointer.
282 2014-03-26 Marek Polacek <polacek@redhat.com>
285 * doc/extend.texi (C Extensions): Mention variable-length arrays in
288 2014-03-26 Marek Polacek <polacek@redhat.com>
291 * doc/extend.texi (Designated Inits): Describe what happens to omitted
294 2014-03-26 Marek Polacek <polacek@redhat.com>
297 * ira-color.c (update_conflict_hard_regno_costs): Perform the
298 multiplication in unsigned type.
300 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
302 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
304 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
306 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
308 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
311 * cif-code.def (UNREACHABLE) New code.
312 * ipa-inline.c (inline_small_functions): Skip edges to
313 __builtlin_unreachable.
314 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
315 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
316 predicate to __bulitin_unreachable.
317 (set_cond_stmt_execution_predicate): Fix issue when
318 invert_tree_comparison returns ERROR_MARK.
319 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
320 propagate to inline clones.
321 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
323 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
324 * cgraphclones.c (cgraph_clone_node): If call destination is already
325 ureachable, do not redirect it back.
326 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
329 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
331 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
332 Do not modify inline clones.
334 2014-03-25 Jakub Jelinek <jakub@redhat.com>
336 * config/i386/i386.md (general_sext_operand): New mode attr.
337 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
338 don't generate (sign_extend (const_int)).
339 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
340 operands[2]. Use We constraint instead of <i> and
341 <general_sext_operand> predicate instead of <general_operand>.
342 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
343 * config/i386/constraints.md (We): New constraint.
344 * config/i386/predicates.md (x86_64_sext_operand,
345 sext_operand): New predicates.
347 2014-03-25 Martin Jambor <mjambor@suse.cz>
350 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
351 inconsistent devirtualizations to __builtin_unreachable.
353 2014-03-25 Marek Polacek <polacek@redhat.com>
356 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
358 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
360 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
361 order of elements for big-endian.
363 2014-03-25 Richard Biener <rguenther@suse.de>
366 * gimplify-me.c (gimple_regimplify_operands): Update the
369 2014-03-25 Martin Jambor <mjambor@suse.cz>
372 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
373 (lto_output_varpool_node): Likewise.
374 (input_overwrite_node): Likewise.
375 (input_varpool_node): Likewise.
377 2014-03-25 Richard Biener <rguenther@suse.de>
379 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
382 2014-03-25 Jakub Jelinek <jakub@redhat.com>
384 * combine.c (simplify_compare_const): Add MODE argument.
385 Handle mode_width 0 as very large mode_width.
386 (try_combine, simplify_comparison): Adjust callers.
388 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
389 type to avoid signed integer overflow.
390 * explow.c (plus_constant): Likewise.
392 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
394 * doc/generic.texi: Correct typos.
396 2014-03-24 Tobias Burnus <burnus@net-b.de>
398 * doc/invoke.texi (-flto): Expand section about
399 using static libraries with LTO.
401 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
403 PR rtl-optimization/60501
404 * optabs.def (addptr3_optab): New optab.
405 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
406 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
407 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
409 * lra.c (emit_add3_insn): Use the addptr pattern if available.
411 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
413 2014-03-24 Ulrich Drepper <drepper@gmail.com>
415 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
418 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
419 (_mm256_undefined_ps): Define.
420 (_mm256_undefined_pd): Define.
421 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
422 (_mm_undefined_pd): Define.
423 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
424 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
425 (_mm512_undefined_ps): Define.
426 (_mm512_undefined_pd): Define.
427 Use _mm*_undefined_*.
428 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
430 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
432 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
433 (lshr_simd): DI mode added.
434 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
435 (aarch64_ushr_simddi): Likewise.
436 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
437 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
438 (vshrd_n_u64): Likewise.
440 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
442 * Makefile.in (s-macro_list): Depend on cc1.
444 2014-03-23 Teresa Johnson <tejohnson@google.com>
446 * ipa-utils.c (ipa_print_order): Use specified dump file.
448 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
450 PR rtl-optimization/60601
451 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
453 * gcc.c (eval_spec_function): Initialize save_growing_value.
455 2014-03-22 Jakub Jelinek <jakub@redhat.com>
458 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
459 code == MINUS_EXPR, never swap op0 with op1.
461 * toplev.c (init_local_tick): Avoid signed integer multiplication
463 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
464 shift by first operand's bitsize.
466 2014-03-21 Jakub Jelinek <jakub@redhat.com>
469 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
471 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
472 TARGET_ISA_64BIT_P(x).
474 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
476 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
477 pattern for vector nor instead of subtract from splat(-1).
478 (altivec_expand_vec_perm_const_le): Likewise.
480 2014-03-21 Richard Henderson <rth@twiddle.net>
483 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
484 related insns after epilogue_completed.
486 2014-03-21 Martin Jambor <mjambor@suse.cz>
489 * cgraph.h (symtab_node): New flag body_removed.
490 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
491 when removing bodies.
492 * symtab.c (dump_symtab_base): Dump body_removed flag.
493 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
494 had their bodies removed.
496 2014-03-21 Martin Jambor <mjambor@suse.cz>
499 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
502 2014-03-21 Richard Biener <rguenther@suse.de>
504 PR tree-optimization/60577
505 * tree-core.h (struct tree_base): Document nothrow_flag use
506 in VAR_DECL_NONALIASED.
507 * tree.h (VAR_DECL_NONALIASED): New.
508 (may_be_aliased): Adjust.
509 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
511 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
513 * expr.c (expand_expr_real_1): Remove outdated comment.
515 2014-03-20 Jakub Jelinek <jakub@redhat.com>
518 * ira.c (adjust_cleared_regs): Call copy_rtx on
519 *reg_equiv[REGNO (loc)].src_p before passing it to
520 simplify_replace_fn_rtx.
523 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
524 into CONST, put pic register as first operand of PLUS. Use
525 gen_const_mem for both 32-bit and 64-bit PIC got loads.
527 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
529 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
531 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
533 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
534 around for store forwarding issue in the FPU on the UT699.
535 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
536 loads and operations if -mfix-ut699 is specified.
537 (divtf3_hq): Tweak attribute.
538 (sqrttf2_hq): Likewise.
540 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
542 * calls.c (store_one_arg): Remove incorrect const qualification on the
543 type of the temporary.
544 * cfgexpand.c (expand_return): Likewise.
545 * expr.c (expand_constructor): Likewise.
546 (expand_expr_real_1): Likewise.
548 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
550 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
553 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
556 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
558 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
560 * config/arm/aarch-common-protos.h
561 (alu_cost_table): Fix spelling of "extend".
562 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
564 2014-03-19 Richard Biener <rguenther@suse.de>
567 * tree-core.h (tree_type_common): Re-order pointer members
568 to reduce recursion depth during GC walks.
570 2014-03-19 Marek Polacek <polacek@redhat.com>
573 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
576 2014-03-19 Richard Biener <rguenther@suse.de>
579 * lto-streamer-in.c (input_function): In WPA stage do not drop
582 2014-03-19 Jakub Jelinek <jakub@redhat.com>
584 PR tree-optimization/60559
585 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
586 with build_zero_cst assignment.
588 2014-03-18 Kai Tietz <ktietz@redhat.com>
590 PR rtl-optimization/56356
591 * sdbout.c (sdbout_parms): Verify that parms'
592 incoming argument is valid.
593 (sdbout_reg_parms): Likewise.
595 2014-03-18 Richard Henderson <rth@redhat.com>
598 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
599 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
600 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
602 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
604 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
605 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
606 Italicize plugin event names in description. Explain that
607 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
608 Remind that no GCC functions should be called after PLUGIN_FINISH.
609 Explain what pragmas with expansion are.
611 2014-03-18 Martin Liska <mliska@suse.cz>
613 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
614 gimple call statement is update.
615 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
616 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
618 2014-03-18 Jakub Jelinek <jakub@redhat.com>
621 * ubsan.c (ubsan_instrument_unreachable): Call
622 initialize_sanitizer_builtins.
623 (ubsan_pass): Likewise.
626 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
627 varpool_finalize_decl instead of rest_of_decl_compilation.
629 2014-03-18 Richard Biener <rguenther@suse.de>
631 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
632 by using bitmap_and_compl instead of bitmap_and_compl_into.
633 (df_rd_transfer_function): Likewise.
635 2014-03-18 Richard Biener <rguenther@suse.de>
637 * doc/lto.texi (fresolution): Fix typo.
639 2014-03-18 Richard Biener <rguenther@suse.de>
641 * doc/invoke.texi (flto): Update for changes in 4.9.
643 2014-03-18 Richard Biener <rguenther@suse.de>
645 * doc/loop.texi: Remove section on the removed lambda framework.
646 Update loop docs with recent changes in preserving loop structure.
648 2014-03-18 Richard Biener <rguenther@suse.de>
650 * doc/lto.texi (-fresolution): Document.
652 2014-03-18 Richard Biener <rguenther@suse.de>
654 * doc/contrib.texi: Adjust my name.
656 2014-03-18 Jakub Jelinek <jakub@redhat.com>
659 * internal-fn.c: Include diagnostic-core.h.
660 (expand_BUILTIN_EXPECT): New function.
661 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
662 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
663 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
664 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
666 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
667 Revert 3 argument __builtin_expect code.
668 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
669 * gimple-fold.c (gimple_fold_call): Likewise.
670 * tree.h (fold_builtin_expect): New prototype.
671 * builtins.c (build_builtin_expect_predicate): Add predictor
672 argument, if non-NULL, create 3 argument __builtin_expect.
673 (fold_builtin_expect): No longer static. Add ARG2 argument,
674 pass it through to build_builtin_expect_predicate.
675 (fold_builtin_2): Adjust caller.
676 (fold_builtin_3): Handle BUILT_IN_EXPECT.
677 * internal-fn.def (BUILTIN_EXPECT): New.
679 2014-03-18 Tobias Burnus <burnus@net-b.de>
682 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
683 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
684 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
686 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
689 * predict.c (combine_predictions_for_bb): Fix up formatting.
690 (expr_expected_value_1, expr_expected_value): Add predictor argument,
691 fill what it points to if non-NULL.
692 (tree_predict_by_opcode): Adjust caller, use the predictor.
693 * predict.def (PRED_COMPARE_AND_SWAP): Add.
695 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
697 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
698 proper constant for the store mode.
700 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
702 * symtab.c (change_decl_assembler_name): Fix transparent alias
705 2014-03-16 Renlin Li <Renlin.Li@arm.com>
707 * config/aarch64/aarch64.c: Correct the comments about the
708 aarch64 stack layout.
710 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
712 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
713 check for GF_OMP_FOR_KIND_FOR.
715 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
717 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
718 ymm and zmm register names.
720 2014-03-17 Jakub Jelinek <jakub@redhat.com>
723 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
724 note creation for the 2010-08-31 changes.
726 2014-03-17 Marek Polacek <polacek@redhat.com>
729 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
730 as -fno-tree-loop-vectorize.
731 (expand_omp_simd): Likewise.
733 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
735 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
736 (eligible_for_call_delay): New prototype.
737 * config/sparc/sparc.c (tls_call_delay): Rename into...
738 (eligible_for_call_delay): ...this. Return false if the instruction
739 cannot be put in the delay slot of a branch.
740 (eligible_for_restore_insn): Simplify.
741 (eligible_for_return_delay): Return false if the instruction cannot be
742 put in the delay slot of a branch and simplify.
743 (eligible_for_sibcall_delay): Return false if the instruction cannot be
744 put in the delay slot of a branch.
745 * config/sparc/sparc.md (fix_ut699): New attribute.
746 (tls_call_delay): Delete.
747 (in_call_delay): Reimplement.
748 (eligible_for_sibcall_delay): Rename into...
749 (in_sibcall_delay): ...this.
750 (eligible_for_return_delay): Rename into...
751 (in_return_delay): ...this.
752 (in_branch_delay): Reimplement.
753 (in_uncond_branch_delay): Delete.
754 (in_annul_branch_delay): Delete.
756 2014-03-14 Richard Henderson <rth@redhat.com>
759 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
760 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
761 (*floathi<X87MODEF>2_i387_with_temp): Remove.
762 (floathi splitters): Remove.
763 (float<SWI48x>xf2): New pattern.
764 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
765 code that tried to handle DImode for 32-bit, but which was excluded
766 by the pattern's condition. Drop allocation of stack temporary.
767 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
768 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
769 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
770 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
771 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
772 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
773 (*float<SWI48><MODEF>2_sse_interunit): Remove.
774 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
775 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
776 (*float<SWI48x><X87MODEF>2_i387): Remove.
777 (all float _with_temp splitters): Remove.
778 (*float<SWI48x><MODEF>2_i387): New pattern.
779 (*float<SWI48><MODEF>2_sse): New pattern.
780 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
781 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
783 2014-03-14 Jakub Jelinek <jakub@redhat.com>
784 Marek Polacek <polacek@redhat.com>
787 * common.opt (dump_base_name_prefixed): New Variable.
788 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
789 if x_dump_base_name_prefixed is already set, set it at the end.
791 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
793 PR rtl-optimization/60508
794 * lra-constraints.c (get_reload_reg): Add new parameter
796 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
797 Pass the new parameter values.
799 2014-03-14 Richard Biener <rguenther@suse.de>
801 * common.opt: Revert unintented changes from r205065.
804 2014-03-14 Richard Biener <rguenther@suse.de>
807 * cfghooks.c (split_block): Properly adjust all loops the
808 block was a latch of.
810 2014-03-14 Martin Jambor <mjambor@suse.cz>
813 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
816 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
819 * config/avr/avr.c (avr_set_current_function): Pass function name
820 through default_strip_name_encoding before sanity checking instead
821 of skipping the first char of the assembler name.
823 2014-03-13 Richard Henderson <rth@redhat.com>
826 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
827 (ix86_force_to_memory, ix86_free_from_memory): Remove.
828 * config/i386/i386-protos.h: Likewise.
829 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
830 in the expander instead of a splitter.
831 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
832 any possibility of requiring a memory.
833 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
834 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
835 (fp branch splitters): Update for ix86_split_fp_branch.
836 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
837 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
838 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
839 (*fop_<MODEF>_2_i387): Remove f/r alternative.
840 (*fop_<MODEF>_3_i387): Likewise.
841 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
842 (splitters for the fop_* register patterns): Remove.
843 (fscalexf4_i387): Rename from *fscalexf4_i387.
844 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
846 2014-03-13 Jakub Jelinek <jakub@redhat.com>
848 PR tree-optimization/59779
849 * tree-dfa.c (get_ref_base_and_extent): Use double_int
850 type for bitsize and maxsize instead of HOST_WIDE_INT.
852 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
854 PR rtl-optimization/57320
855 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
856 the CFG after thread_prologue_and_epilogue_insns.
858 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
860 PR rtl-optimization/57189
861 * lra-constraints.c (process_alt_operands): Disfavor spilling
864 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
866 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
868 2014-03-13 Jakub Jelinek <jakub@redhat.com>
870 PR tree-optimization/59025
872 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
873 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
875 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
878 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
879 calls of avr_out_plus_1.
881 2014-03-13 Bin Cheng <bin.cheng@arm.com>
883 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
884 BB's single pred and update the father loop's latch info later.
886 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
888 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
892 (VEC_base): Likewise.
893 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
894 registers, we need to swap double words in little endian mode.
896 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
897 to be a container mode for 128-bit integer operations added in ISA
898 2.07. Unlike TImode and PTImode, the preferred register set is
899 the Altivec/VMX registers for the 128-bit operations.
901 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
903 (rs6000_split_128bit_ok_p): Likewise.
905 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
906 macros for creating ISA 2.07 normal and overloaded builtin
907 functions with 3 arguments.
908 (BU_P8V_OVERLOAD_3): Likewise.
909 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
910 for use as overloaded functions.
911 (VPERM_1TI_UNS): Likewise.
912 (VSEL_1TI): Likewise.
913 (VSEL_1TI_UNS): Likewise.
914 (ST_INTERNAL_1ti): Likewise.
915 (LD_INTERNAL_1ti): Likewise.
916 (XXSEL_1TI): Likewise.
917 (XXSEL_1TI_UNS): Likewise.
918 (VPERM_1TI): Likewise.
919 (VPERM_1TI_UNS): Likewise.
920 (XXPERMDI_1TI): Likewise.
922 (LXVD2X_V1TI): Likewise.
923 (STXVD2X_V1TI): Likewise.
924 (VEC_INIT_V1TI): Likewise.
925 (VEC_SET_V1TI): Likewise.
926 (VEC_EXT_V1TI): Likewise.
927 (EQV_V1TI): Likewise.
928 (NAND_V1TI): Likewise.
929 (ORC_V1TI): Likewise.
930 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
931 added in ISA 2.07. Add both normal 'altivec' builtins, and the
935 (VADDEUQM): Likewise.
936 (VADDECUQ): Likewise.
937 (VSUBEUQM): Likewise.
938 (VSUBECUQ): Likewise.
940 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
941 __int128_t and __uint128_t types.
942 (__uint128_type): Likewise.
943 (altivec_categorize_keyword): Add support for vector __int128_t,
944 vector __uint128_t, vector __int128, and vector unsigned __int128
945 as a container type for TImode operations that need to be done in
946 VSX/Altivec registers.
947 (rs6000_macro_to_expand): Likewise.
948 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
949 to support 128-bit integer instructions vaddcuq, vadduqm,
950 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
951 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
953 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
954 for V1TImode, and set up preferences to use VSX/Altivec registers.
955 Setup VSX reload handlers.
956 (rs6000_debug_reg_global): Likewise.
957 (rs6000_init_hard_regno_mode_ok): Likewise.
958 (rs6000_preferred_simd_mode): Likewise.
959 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
960 (easy_altivec_constant): Likewise.
961 (output_vec_const_move): Likewise.
962 (rs6000_expand_vector_set): Convert V1TImode set and extract to
964 (rs6000_expand_vector_extract): Likewise.
965 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
967 (rs6000_const_vec): Add support for V1TImode.
968 (rs6000_emit_le_vsx_load): Swap double words when loading or
969 storing TImode/V1TImode.
970 (rs6000_emit_le_vsx_store): Likewise.
971 (rs6000_emit_le_vsx_move): Likewise.
972 (rs6000_emit_move): Add support for V1TImode.
973 (altivec_expand_ld_builtin): Likewise.
974 (altivec_expand_st_builtin): Likewise.
975 (altivec_expand_vec_init_builtin): Likewise.
976 (altivec_expand_builtin): Likewise.
977 (rs6000_init_builtins): Add support for V1TImode type. Add
978 support for ISA 2.07 128-bit integer builtins. Define type names
979 for the VSX/Altivec vector types.
980 (altivec_init_builtins): Add support for overloaded vector
981 functions with V1TImode type.
982 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
983 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
985 (rs6000_split_128bit_ok_p): Likewise.
986 (rs6000_handle_altivec_attribute): Create V1TImode from vector
987 __int128_t and vector __uint128_t.
989 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
997 (VS_scalar): Likewise.
998 (VS_double): Likewise.
999 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1001 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1002 we support the ISA 2.07 128-bit integer arithmetic instructions.
1003 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1004 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1005 and TImode types for use with the builtin functions.
1006 (V1TI_type_node): Likewise.
1007 (unsigned_V1TI_type_node): Likewise.
1008 (intTI_type_internal_node): Likewise.
1009 (uintTI_type_internal_node): Likewise.
1011 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1012 128-bit builtin functions.
1013 (UNSPEC_VADDEUQM): Likewise.
1014 (UNSPEC_VADDECUQ): Likewise.
1015 (UNSPEC_VSUBCUQ): Likewise.
1016 (UNSPEC_VSUBEUQM): Likewise.
1017 (UNSPEC_VSUBECUQ): Likewise.
1018 (VM): Add V1TImode to vector mode iterators.
1020 (VI_unit): Likewise.
1021 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1022 (altivec_vaddcuq): Likewise.
1023 (altivec_vsubuqm): Likewise.
1024 (altivec_vsubcuq): Likewise.
1025 (altivec_vaddeuqm): Likewise.
1026 (altivec_vaddecuq): Likewise.
1027 (altivec_vsubeuqm): Likewise.
1028 (altivec_vsubecuq): Likewise.
1030 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1032 (BOOL_128): Likewise.
1033 (BOOL_REGS_OUTPUT): Likewise.
1034 (BOOL_REGS_OP1): Likewise.
1035 (BOOL_REGS_OP2): Likewise.
1036 (BOOL_REGS_UNARY): Likewise.
1037 (BOOL_REGS_AND_CR0): Likewise.
1039 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1040 128-bit integer builtin support.
1041 (vec_vadduqm): Likewise.
1042 (vec_vaddecuq): Likewise.
1043 (vec_vaddeuqm): Likewise.
1044 (vec_vsubecuq): Likewise.
1045 (vec_vsubeuqm): Likewise.
1046 (vec_vsubcuq): Likewise.
1047 (vec_vsubuqm): Likewise.
1049 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1050 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1051 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1052 128-bit integer add/subtract to ISA 2.07.
1054 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1056 * config/arc/arc.c (arc_predicate_delay_insns):
1057 Fix third argument passed to conditionalize_nonjump.
1059 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1061 * config/aarch64/aarch64-builtins.c
1062 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1063 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1064 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1065 instead of __builtin_lfloor.
1066 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1068 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1070 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1071 (tree_ssa_ifcombine_bb_1): New function.
1072 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1073 is an empty forwarder block to then_bb or vice versa and then_bb
1074 and else_bb are effectively swapped.
1076 2014-03-12 Christian Bruel <christian.bruel@st.com>
1079 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1080 REG_CFA_DEF_CFA note.
1081 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1082 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1084 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1086 PR tree-optimization/60454
1087 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1089 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1091 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1092 Do not define target_cpu_default2 to generic.
1093 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1094 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1095 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1097 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1098 Marc Glisse <marc.glisse@inria.fr>
1100 PR tree-optimization/60502
1101 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1102 instead of build_low_bits_mask.
1104 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1107 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1108 if there are multiple uses, but op doesn't live on E edge.
1109 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1110 clobber stmts before __builtin_unreachable.
1112 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1114 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1115 hard_frame_pointer_rtx.
1116 * cse.c (cse_insn): Remove volatile check.
1117 * cselib.c (cselib_process_insn): Likewise.
1118 * dse.c (scan_insn): Likewise.
1120 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1122 * config/arc/arc.c (conditionalize_nonjump): New function,
1124 (arc_ifcvt): ... this.
1125 (arc_predicate_delay_insns): Use it.
1127 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1129 * config/arc/predicates.md (extend_operand): During/after reload,
1130 allow const_int_operand.
1131 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1132 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1133 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1135 (umulsi3_highpart_i): Likewise.
1137 2014-03-11 Richard Biener <rguenther@suse.de>
1139 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1140 Add asserts to guard possible wrong-code bugs.
1142 2014-03-11 Richard Biener <rguenther@suse.de>
1144 PR tree-optimization/60429
1145 PR tree-optimization/60485
1146 * tree-ssa-structalias.c (set_union_with_increment): Properly
1147 take into account all fields that overlap the shifted vars.
1148 (do_sd_constraint): Likewise.
1149 (do_ds_constraint): Likewise.
1150 (get_constraint_for_ptr_offset): Likewise.
1152 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1154 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1155 (nios2_compute_frame_layout):
1156 Add calculation of cfun->machine->fp_save_offset.
1157 (nios2_expand_prologue): Correct setting of frame pointer register
1159 (nios2_expand_epilogue): Update recovery of stack pointer from
1160 frame pointer accordingly.
1161 (nios2_initial_elimination_offset): Update calculation of offset
1162 for eliminating to HARD_FRAME_POINTER_REGNUM.
1164 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1167 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1168 cgraph_get_create_node on VAR_DECLs.
1170 2014-03-10 Richard Biener <rguenther@suse.de>
1173 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1175 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1177 * config/vms/vms.opt (vms_float_format): New variable.
1179 2014-03-08 Tobias Burnus <burnus@net-b.de>
1181 * doc/invoke.texi (-fcilkplus): Update implementation status.
1183 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1184 Richard Biener <rguenther@suse.de>
1186 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1187 consistently accross all TUs.
1188 (run_gcc): Enable -fshort-double automatically at link at link-time
1189 and disallow override.
1191 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1194 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1195 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1196 if they can't be used.
1198 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1200 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1201 for Solaris 11/x86 ld.
1202 * configure: Regenerate.
1204 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1206 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1207 (LIB_TLS_SPEC): Save as ld_tls_libs.
1208 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1209 (HAVE_AS_IX86_TLSLDM): New test.
1210 * configure, config.in: Regenerate.
1211 * config/i386/i386.c (legitimize_tls_address): Fall back to
1212 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1213 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1214 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1215 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1217 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1219 * common.opt (fira-loop-pressure): Mark as optimization.
1221 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1223 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1224 an OpenMP mappable type.
1226 2014-03-06 Matthias Klose <doko@ubuntu.com>
1228 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1229 MULTILIB_OSDIRNAMES is not defined.
1231 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1232 Meador Inge <meadori@codesourcery.com>
1235 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1236 (arm_legitimize_address): Call legitimize_tls_address for any
1237 arm_tls_referenced_p expression, handle constant addend. Call it
1238 before testing for !TARGET_ARM.
1239 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1241 2014-03-06 Richard Biener <rguenther@suse.de>
1247 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1249 * tree-streamer.c (record_common_node): Assert we don't record
1250 nodes with type double.
1251 (preload_common_node): Skip type double, complex double and double
1252 pointer since it is now frontend dependent due to fshort-double option.
1254 2014-03-06 Richard Biener <rguenther@suse.de>
1256 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1257 or -fno-lto is specified and the linker has full plugin support.
1258 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1259 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1260 * lto-wrapper.c (merge_and_complain): Merge compile-time
1261 optimization levels.
1262 (run_gcc): And pass it through to the link options.
1264 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1268 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1270 * cselib.c (remove_useless_values): Skip to avoid quadratic
1271 behavior if the condition moved from...
1272 (cselib_process_insn): ... here holds.
1274 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1277 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1278 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1281 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1282 (TM_H): Add x86-tune.def.
1284 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1286 * config/aarch64/aarch64.c (generic_tunings):
1287 Use cortexa57_extra_costs.
1289 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1292 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1293 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1294 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1297 2014-03-04 Heiher <r@hev.cc>
1299 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1300 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1302 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1304 * config/i386/predicates.md (const2356_operand): Change to ...
1305 (const2367_operand): ... this.
1306 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1308 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1309 (*avx512pf_scatterpf<mode>sf): Ditto.
1310 (avx512pf_scatterpf<mode>df): Ditto.
1311 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1312 (*avx512pf_scatterpf<mode>df): Ditto.
1313 * config/i386/i386.c (ix86_expand_builtin): Update
1314 incorrect hint operand error message.
1316 2014-03-04 Richard Biener <rguenther@suse.de>
1318 * lto-section-in.c (lto_get_section_data): Fix const cast.
1320 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1322 * tree-streamer.c (record_common_node): Assert we don't record
1323 nodes with type double.
1324 (preload_common_node): Skip type double, complex double and double
1325 pointer since it is now frontend dependent due to fshort-double option.
1327 2014-03-04 Richard Biener <rguenther@suse.de>
1330 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1331 (lto_input_toplevel_asms): Likewise.
1332 * lto-section-in.c (lto_get_section_data): Instead do it here
1335 2014-03-04 Richard Biener <rguenther@suse.de>
1337 PR tree-optimization/60382
1338 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1339 dead PHIs a reduction.
1341 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1343 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1345 (_mm_prefetch): Move out of GCC target("sse") pragma.
1346 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1347 GCC target("prfchw") pragma.
1348 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1350 * config/i386/i386.c (ix86_option_override_internal): Enable
1351 -mprfchw with -mprefetchwt1.
1353 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1355 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1358 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1360 * opts.h (CL_PCH_IGNORE): Define.
1361 * targhooks.c (option_affects_pch_p):
1362 Return false for options that have CL_PCH_IGNORE set.
1363 * opt-functions.awk: Process PchIgnore.
1364 * doc/options.texi: Document PchIgnore.
1366 * config/arc/arc.opt (misize): Add PchIgnore property.
1368 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1370 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1371 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1372 constraint on constants to permit them being loaded into
1373 GENERAL_REGS or BASE_REGS.
1375 2014-03-03 Nick Clifton <nickc@redhat.com>
1377 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1378 anti-cacnonical alternatives.
1379 (negandhi3_real): New pattern.
1380 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1382 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1384 * config/avr/avr-mcus.def: Remove atxmega16x1.
1385 * config/avr/avr-tables.opt: Regenerate.
1386 * config/avr/t-multilib: Regenerate.
1387 * doc/avr-mmcu.texi: Regenerate.
1389 2014-03-03 Tobias Grosser <tobias@grosser.es>
1390 Mircea Namolaru <mircea.namolaru@inria.fr>
1392 PR tree-optimization/58028
1393 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1396 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1398 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1399 not handled by recognizers.
1401 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1404 * function.c (expand_function_end): Don't emit
1405 clobber_return_register sequence if clobber_after is a BARRIER.
1406 * cfgexpand.c (construct_exit_block): Append instructions before
1407 return_label to prev_bb.
1409 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1411 * config/rs6000/constraints.md: Document reserved use of "wc".
1413 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1416 * ipa.c (function_and_variable_visibility): When dissolving comdat
1417 group, also set all symbols to local.
1419 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1424 2013-12-14 Jan Hubicka <jh@suse.cz>
1426 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1428 2014-03-02 Jon Beniston <jon@beniston.com>
1434 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1435 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1436 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1437 (simple_return, *simple_return): New patterns
1438 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1439 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1441 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1443 * dwarf2out.c (gen_subprogram_die): Tidy.
1445 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1448 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1449 (*mov_t_msb_neg_negc): ... this new insn.
1451 2014-02-28 Jason Merrill <jason@redhat.com>
1454 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1457 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1460 * dwarf2out.c (decltype_auto_die): New static.
1461 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1462 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1463 (is_cxx_auto): Likewise.
1465 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1467 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1468 we are not using general regs only.
1470 2014-02-28 Richard Biener <rguenther@suse.de>
1473 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1474 previous fix and only allow to remove trivial pre-headers
1475 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1476 (remove_forwarder_block): Properly update the latch of a loop.
1478 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1481 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1482 (cselib_preserved_hash_table): New.
1483 (preserve_constants_and_equivs): Move preserved vals to it.
1484 (cselib_find_slot): Look it up first.
1485 (cselib_init): Initialize it.
1486 (cselib_finish): Release it.
1487 (dump_cselib_table): Dump it.
1489 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1492 * cselib.c (remove_useless_values): Skip to avoid quadratic
1493 behavior if the condition moved from...
1494 (cselib_process_insn): ... here holds.
1496 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1499 * var-tracking.c (vt_initialize): Apply the same condition to
1500 preserve the CFA base value.
1502 2014-02-28 Joey Ye <joey.ye@arm.com>
1505 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1506 if reload in progress or completed.
1508 2014-02-28 Tobias Burnus <burnus@net-b.de>
1511 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1514 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1516 * doc/tm.texi.in (Condition Code Status): Update documention for
1517 relative locations of cc0-setter and cc0-user.
1519 2014-02-27 Jeff Law <law@redhat.com>
1521 PR rtl-optimization/52714
1522 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1523 into two independent simple sets, if I3 is a jump, ensure the
1524 pattern we place into I3 is a (set (pc) ...).
1526 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1527 Jeff Law <law@redhat.com>
1529 PR rtl-optimization/49847
1530 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1531 are in different blocks.
1532 * doc/tm.texi (Condition Code Status): Update documention for
1533 relative locations of cc0-setter and cc0-user.
1535 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1538 * lra.c (lra_emit_add): Check SUBREG too.
1540 2014-02-27 Andreas Schwab <schwab@suse.de>
1542 * config/m68k/m68k.c (m68k_option_override): Disable
1543 -flive-range-shrinkage for classic m68k.
1544 (m68k_override_options_after_change): Likewise.
1546 2014-02-27 Marek Polacek <polacek@redhat.com>
1549 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1550 -Wmaybe-uninitialized.
1552 2014-02-27 Alan Modra <amodra@gmail.com>
1555 * reload1.c (emit_input_reload_insns): When reload_override_in,
1556 set old to rl->in_reg when rl->in_reg is a subreg.
1558 2014-02-26 Richard Biener <rguenther@suse.de>
1561 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1563 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1565 * common/config/i386/predicates.md (const1256_operand): Remove.
1566 (const2356_operand): New.
1567 (const_1_to_2_operand): Remove.
1568 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1569 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1570 (*avx512pf_gatherpf<mode>sf): Ditto.
1571 (avx512pf_gatherpf<mode>df): Ditto.
1572 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1573 (*avx512pf_gatherpf<mode>df): Ditto.
1574 (avx512pf_scatterpf<mode>sf): Ditto.
1575 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1576 (*avx512pf_scatterpf<mode>sf): Ditto.
1577 (avx512pf_scatterpf<mode>df): Ditto.
1578 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1579 (*avx512pf_scatterpf<mode>df): Ditto.
1580 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1582 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1584 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1585 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1586 (_mm512_mask_testn_epi64_mask): Move to ...
1587 * config/i386/avx512cdintrin.h: Here.
1588 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1589 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1590 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1591 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1592 TARGET_AVX512F from TARGET_AVX512CD.
1594 2014-02-26 Richard Biener <rguenther@suse.de>
1597 * ipa.c (walk_polymorphic_call_targets): Properly guard
1598 call to inline_update_overall_summary.
1600 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1603 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1604 and latches only if requested. Fix latch if it is removed.
1605 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1606 LOOPS_HAVE_PREHEADERS.
1608 2014-02-25 Andrew Pinski <apinski@cavium.com>
1610 * builtins.c (expand_builtin_thread_pointer): Create a new target
1611 when the target is NULL.
1613 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1615 PR rtl-optimization/60317
1616 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1617 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1618 * lra-assigns.c: Include params.h.
1619 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1620 other reload pseudos considerations.
1622 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1624 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1625 to use canonical form for nor<mode>3.
1627 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1630 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1633 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1635 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1636 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1637 (ix86_handle_option): Handle OPT_mprefetchwt1.
1638 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1639 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1641 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1642 OPTION_MASK_ISA_PREFETCHWT1.
1643 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1644 (PTA_PREFETCHWT1): New.
1645 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1646 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1647 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1648 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1649 (*prefetch_avx512pf_<mode>_: Change into ...
1650 (*prefetch_prefetchwt1_<mode>: This.
1651 * config/i386/i386.opt (mprefetchwt1): New.
1652 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1653 (_mm_prefetch): Handle intent to write.
1654 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1656 2014-02-25 Richard Biener <rguenther@suse.de>
1659 * emit-rtl.c (mem_attrs_htab): Remove.
1660 (mem_attrs_htab_hash): Likewise.
1661 (mem_attrs_htab_eq): Likewise.
1662 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1663 (init_emit_once): Do not allocate mem_attrs_htab.
1665 2014-02-25 Richard Biener <rguenther@suse.de>
1668 * lto-opts.c (lto_write_options): Output non-explicit conservative
1669 -fwrapv, -fno-trapv and -fno-strict-overflow.
1670 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1671 (run_gcc): And pass them through.
1673 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1675 * sel-sched.c (calculate_new_fences): New parameter ptime.
1676 Calculate it as a maximum over all fence cycles.
1677 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1678 Print the final schedule timing when sched_verbose.
1680 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1682 PR rtl-optimization/60292
1683 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1684 bit fot the fence instruction.
1686 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1688 * calls.h: Fix typo in comment.
1690 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1692 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1693 adjusting offsetable addresses.
1695 2014-02-24 Guozhi Wei <carrot@google.com>
1697 * sparseset.h (sparseset_pop): Fix the wrong index.
1699 2014-02-24 Walter Lee <walt@tilera.com>
1701 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1702 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1704 * common/config/tilegx/tilegx-common.c
1705 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1706 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1708 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1709 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1710 (tilegx_gimplify_va_arg_expr): Handle big endian.
1711 (tilegx_expand_unaligned_load): Ditto.
1712 (tilegx_expand_unaligned_store): Ditto.
1713 (TARGET_RETURN_IN_MSB): New.
1714 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1715 (TARGET_ENDIAN_DEFAULT): New.
1716 (TARGET_BIG_ENDIAN): Handle big endian.
1717 (BYTES_BIG_ENDIAN): Ditto.
1718 (WORDS_BIG_ENDIAN): Ditto.
1719 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1722 * config/tilegx/tilegx.md (extv): Handle big endian.
1724 (insn_st<n>): Ditto.
1725 (insn_st<n>_add<bitsuffix>): Ditto.
1726 (insn_stnt<n>): Ditto.
1727 (insn_stnt<n>_add<bitsuffix>):Ditto.
1728 (vec_interleave_highv8qi): Handle big endian.
1729 (vec_interleave_highv8qi_be): New.
1730 (vec_interleave_highv8qi_le): New.
1731 (insn_v1int_h): Handle big endian.
1732 (vec_interleave_lowv8qi): Handle big endian.
1733 (vec_interleave_lowv8qi_be): New.
1734 (vec_interleave_lowv8qi_le): New.
1735 (insn_v1int_l): Handle big endian.
1736 (vec_interleave_highv4hi): Handle big endian.
1737 (vec_interleave_highv4hi_be): New.
1738 (vec_interleave_highv4hi_le): New.
1739 (insn_v2int_h): Handle big endian.
1740 (vec_interleave_lowv4hi): Handle big endian.
1741 (vec_interleave_lowv4hi_be): New.
1742 (vec_interleave_lowv4hi_le): New.
1743 (insn_v2int_l): Handle big endian.
1744 (vec_interleave_highv2si): Handle big endian.
1745 (vec_interleave_highv2si_be): New.
1746 (vec_interleave_highv2si_le): New.
1747 (insn_v4int_h): Handle big endian.
1748 (vec_interleave_lowv2si): Handle big endian.
1749 (vec_interleave_lowv2si_be): New.
1750 (vec_interleave_lowv2si_le): New.
1751 (insn_v4int_l): Handle big endian.
1752 * config/tilegx/tilegx.opt (mbig-endian): New option.
1753 (mlittle-endian): New option.
1754 * doc/install.texi: Document tilegxbe-linux.
1755 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1757 2014-02-24 Martin Jambor <mjambor@suse.cz>
1760 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1761 there are no parameter descriptors.
1763 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1765 PR rtl-optimization/60268
1766 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1767 initialization to ...
1768 (sched_rgn_init): ... here.
1769 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1771 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1773 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1776 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1778 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1781 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1783 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1784 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1786 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1788 * config/microblaze/predicates.md: Add cmp_op predicate.
1789 * config/microblaze/microblaze.md: Add branch_compare instruction
1790 which uses cmp_op predicate and emits cmp insn before branch.
1791 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1792 to microblaze_expand_conditional_branch and consolidate logic.
1793 (microblaze_expand_conditional_branch): emit branch_compare
1794 insn instead of handling cmp op separate from branch insn.
1796 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1798 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1801 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1803 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1804 define_insn with define_expand and new define_insn
1805 *altivec_lve<VI_char>x_internal.
1806 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1807 and new define_insn *altivec_stve<VI_char>x_internal.
1808 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1810 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1812 (altivec_expand_stvex_be): New function.
1814 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1816 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1817 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1818 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1819 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1821 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1824 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1825 instead of emit_move_insn.
1827 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1829 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1831 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1832 gen_altivec_vsumsws.
1834 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1836 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1837 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1838 (altivec_lvxl_<mode>): New define_expand incorporating
1839 -maltivec=be semantics where needed.
1840 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1841 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1842 semantics where needed.
1843 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1844 (altivec_stvx_<mode>): New define_expand incorporating
1845 -maltivec=be semantics where needed.
1846 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1847 VM2 iterator instead of V4SI.
1848 (altivec_stvxl_<mode>): New define_expand incorporating
1849 -maltivec=be semantics where needed.
1850 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1851 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1852 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1853 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1854 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1855 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1856 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1857 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1858 ALTIVEC_BUILTIN_STVXL.
1859 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1860 (altivec_expand_stvx_be): Likewise.
1861 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1862 (altivec_expand_lvx_be): Likewise.
1863 (altivec_expand_stvx_be): Likewise.
1864 (altivec_expand_builtin): Add cases for
1865 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1866 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1867 (altivec_init_builtins): Add definitions for
1868 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1869 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1871 2014-02-21 Catherine Moore <clm@codesourcery.com>
1873 * doc/invoke.texi (mvirt, mno-virt): Document.
1874 * config/mips/mips.opt (mvirt): New option.
1875 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1877 2014-02-21 Richard Biener <rguenther@suse.de>
1879 PR tree-optimization/60276
1880 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1881 (STMT_VINFO_MIN_NEG_DIST): New macro.
1882 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1883 STMT_VINFO_MIN_NEG_DIST.
1884 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1885 made for negative dependence distances still hold.
1887 2014-02-21 Richard Biener <rguenther@suse.de>
1890 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1891 DECL_INITIAL for globals not in the current function context.
1893 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1895 PR tree-optimization/56490
1896 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1897 * tree-ssa-uninit.c: Include params.h.
1898 (compute_control_dep_chain): Add num_calls argument, return false
1899 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1900 num_calls to recursive call.
1901 (find_predicates): Change dep_chain into normal array,
1902 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1903 variable and adjust compute_control_dep_chain caller.
1904 (find_def_preds): Likewise.
1906 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1908 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1909 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1911 2014-02-21 Nick Clifton <nickc@redhat.com>
1913 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1914 (pushhi1): Likewise.
1915 (popqi1): Add mode to pre_dec.
1918 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1920 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1921 mode for mask of V8SFmode permutation.
1923 2014-02-20 Richard Henderson <rth@redhat.com>
1926 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1927 a new pseudo for OLDVAL.
1929 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1932 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1933 gen_reg_rtx if d->testing_p.
1934 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1935 if d->testing_p and we will certainly return true.
1936 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1939 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1941 * emit-rtl.c (gen_reg_rtx): Assert that
1942 crtl->emit.regno_pointer_align_length is non-zero.
1944 2014-02-20 Richard Henderson <rth@redhat.com>
1947 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1948 on failure the store back into EXPECT.
1950 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1951 Sandra Loosemore <sandra@codesourcery.com>
1953 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1954 * config/nios2/nios2.c (nios2_function_profiler): Add
1955 -fPIC (flag_pic == 2) support.
1956 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1957 (nios2_large_offset_p): New function.
1958 (nios2_unspec_reloc_p): Move up position, update to use
1959 nios2_large_offset_p.
1960 (nios2_unspec_address): Remove function.
1961 (nios2_unspec_offset): New function.
1962 (nios2_large_got_address): New function.
1963 (nios2_got_address): Add large offset support.
1964 (nios2_legitimize_tls_address): Update usage of removed and new
1966 (nios2_symbol_binds_local_p): New function.
1967 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1968 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1969 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1970 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1971 processing for (const (unspec ...)).
1972 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1974 2014-02-20 Richard Biener <rguenther@suse.de>
1976 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1977 doing the substitution.
1978 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1980 2014-02-20 Martin Jambor <mjambor@suse.cz>
1983 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1984 info when checking whether lattices are bottom.
1986 2014-02-20 Richard Biener <rguenther@suse.de>
1989 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1992 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1995 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1996 parameter specifying the scaling.
1997 (inline_call): Update.
1998 (want_inline_recursively): Guard division by zero.
1999 (recursive_inlining): Update.
2000 * ipa-inline.h (clone_inlined_nodes): Update.
2002 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2005 * config/i386/i386.c (classify_argument): Pass structures of size
2006 64 bytes or less in register.
2008 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2009 Kirill Yukhin <kirill.yukhin@intel.com>
2011 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2012 (_mm_rcp28_round_ss): Ditto.
2013 (_mm_rsqrt28_round_sd): Ditto.
2014 (_mm_rsqrt28_round_ss): Ditto.
2015 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2016 (_mm_rcp14_round_ss): Ditto.
2017 (_mm_rsqrt14_round_sd): Ditto.
2018 (_mm_rsqrt14_round_ss): Ditto.
2019 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2020 the first input operand, get rid of match_dup.
2021 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2023 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2025 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2026 operand as the first input operand, set type attribute.
2027 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2029 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2030 operand as the first input operand, set type attribute.
2032 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2034 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2037 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2040 * config/i386/i386.c (construct_container): Remove TFmode check
2041 for X86_64_INTEGER_CLASS.
2043 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2046 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2047 only when -Wpsabi is enabled.
2049 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2052 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2053 passing arrays in registers are the same as for structs, so remove the
2054 special case for them.
2056 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2058 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2059 destination type, extract only the valid bits if the source type is not
2060 integral and has a different mode.
2062 2014-02-19 Richard Biener <rguenther@suse.de>
2065 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2068 2014-02-19 Richard Biener <rguenther@suse.de>
2071 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2072 (ipa_modify_call_arguments): Emit an argument load explicitely and
2073 preserve virtual SSA form there and for the replacement call.
2074 Do not update SSA form nor free dominance info.
2076 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2078 * ipa.c (function_and_variable_visibility): Also clear WEAK
2079 flag when disolving COMDAT_GROUP.
2081 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2083 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2084 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2086 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2087 do more sanity checks.
2088 (detect_type_change): Return true when giving up early.
2089 (compute_complex_assign_jump_func): Fix type parameter of
2090 ipa_set_ancestor_jf.
2091 (compute_complex_ancestor_jump_func): Likewise.
2092 (update_jump_functions_after_inlining): Fix updating of
2094 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2096 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2098 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2099 inline clones when edge disappears.
2101 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2104 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2105 Split 64-bit moves into 2 patterns. Do not allow the use of
2106 direct move for TDmode in little endian, since the decimal value
2107 has little endian bytes within a word, but the 64-bit pieces are
2108 ordered in a big endian fashion, and normal subreg's of TDmode are
2110 (mov<mode>_64bit_dm): Likewise.
2111 (movtd_64bit_nodm): Likewise.
2113 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2115 PR tree-optimization/60174
2116 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2117 statement of an SSA_NAME that occurs in an abnormal PHI node.
2119 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2122 * final.c (SEEN_BB): Remove.
2123 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2124 (final_scan_insn): Don't force_source_line on second
2125 NOTE_INSN_BASIC_BLOCK.
2127 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2130 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2131 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2132 (type_natural_mode): Warn ABI change when %zmm register is not
2133 available for AVX512F vector value passing.
2135 2014-02-18 Kai Tietz <ktietz@redhat.com>
2138 * config/i386/i386.c (ix86_expand_prologue): Use value in
2139 rax register as displacement when restoring %r10 or %rax.
2140 Fix wrong offset when restoring both registers.
2142 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2144 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2145 assertion with conditional return.
2147 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2148 Uros Bizjak <ubizjak@gmail.com>
2151 * config/i386/driver-i386.c (host_detect_local_cpu): If
2152 YMM state is not saved by the OS, also clear has_f16c. Move
2153 CPUID 0x80000001 handling before YMM state saving checking.
2155 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2157 PR rtl-optimization/58960
2158 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2159 factored out from ...
2160 (sched_init): ... here.
2161 (free_global_sched_pressure_data): New, factored out from ...
2162 (sched_finish): ... here.
2163 * sched-int.h (free_global_sched_pressure_data): Declare.
2164 * sched-rgn.c (nr_regions_initial): New static global.
2165 (haifa_find_rgns): Initialize it.
2166 (schedule_region): Disable sched-pressure for the newly
2169 2014-02-17 Richard Biener <rguenther@suse.de>
2171 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2172 release SSA defs of pattern stmts.
2174 2014-02-17 Richard Biener <rguenther@suse.de>
2176 * tree-inline.c (expand_call_inline): Release the virtual
2177 operand defined by the call we are about to inline.
2179 2014-02-17 Richard Biener <rguenther@suse.de>
2181 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2183 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2184 Ilya Tocar <ilya.tocar@intel.com>
2186 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2187 arguments order in builtin.
2188 (_mm512_permutexvar_epi64): Ditto.
2189 (_mm512_mask_permutexvar_epi64): Ditto
2190 (_mm512_maskz_permutexvar_epi32): Ditto
2191 (_mm512_permutexvar_epi32): Ditto
2192 (_mm512_mask_permutexvar_epi32): Ditto
2194 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2196 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2197 (p8_vmrgow): Likewise.
2199 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2201 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2204 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2207 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2208 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2209 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2210 using direct move instructions on ISA 2.07. Also adjust
2211 instruction length for 64-bit.
2212 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2213 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2215 2014-02-15 Alan Modra <amodra@gmail.com>
2219 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2220 find_replacement on parts of insn rtl that might be reloaded.
2222 2014-02-15 Richard Biener <rguenther@suse.de>
2224 PR tree-optimization/60183
2225 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2226 (tree_ssa_phiprop): Calculate and free post-dominators.
2228 2014-02-14 Jeff Law <law@redhat.com>
2230 PR rtl-optimization/60131
2231 * ree.c (get_extended_src_reg): New function.
2232 (combine_reaching_defs): Use it rather than assuming location of REG.
2233 (find_and_remove_re): Verify first operand of extension is
2234 a REG before adding the insns to the copy list.
2236 2014-02-14 Roland McGrath <mcgrathr@google.com>
2238 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2239 * configure: Regenerated.
2240 * config.in: Regenerated.
2241 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2242 instead of ASM_SHORT.
2244 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2245 Richard Earnshaw <rearnsha@arm.com>
2247 PR rtl-optimization/59535
2248 * lra-constraints.c (process_alt_operands): Encourage alternative
2249 when unassigned pseudo class is superset of the alternative class.
2250 (inherit_reload_reg): Don't inherit when optimizing for code size.
2251 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2252 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2253 modes not less than 4 for Thumb1.
2255 2014-02-14 Kyle McMartin <kyle@redhat.com>
2258 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2260 2014-02-14 Richard Biener <rguenther@suse.de>
2262 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2263 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2264 Do not strip INDIRECT_REFs.
2266 2014-02-14 Richard Biener <rguenther@suse.de>
2269 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2270 DECL_FUNCTION_SPECIFIC_TARGET.
2271 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2272 * tree-streamer-out.c (pack_ts_target_option): Remove.
2273 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2274 (write_ts_function_decl_tree_pointers): Do not stream
2275 DECL_FUNCTION_SPECIFIC_TARGET.
2276 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2277 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2278 (lto_input_ts_function_decl_tree_pointers): Do not stream
2279 DECL_FUNCTION_SPECIFIC_TARGET.
2281 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2283 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2284 (get_initial_def_for_induction, vectorizable_induction): Ignore
2285 debug stmts when looking for exit_phi.
2286 (vectorizable_live_operation): Fix up condition.
2288 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2290 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2291 nreverse() because it changes the content of original tree list.
2293 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2295 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2296 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2298 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2300 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2301 GNU coding standards.
2303 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2306 * dwarf2out.c (gen_subprogram_die): Don't call
2307 add_calling_convention_attribute if subr_die is old_die.
2309 2014-02-13 Sharad Singhai <singhai@google.com>
2311 * doc/optinfo.texi: Fix order of nodes.
2313 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2315 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2316 operands[2], not operands[3].
2318 2014-02-13 Richard Biener <rguenther@suse.de>
2321 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2322 mention the possibility of an in-tree build.
2323 (CLooG): Update recommended version to 0.18.1, mention the
2324 possibility of an in-tree build and clarify that the ISL
2325 bundled with CLooG does not work.
2327 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2330 * expr.c (compress_float_constant): If x is a hard register,
2331 extend into a pseudo and then move to x.
2333 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2335 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2336 caused by bad second argument to warning_at() with -mhotpatch and
2337 nested functions (e.g. with gfortran).
2339 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2341 * opts.c (option_name): Remove "enabled by default" rider.
2343 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2345 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2347 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2348 Uros Bizjak <ubizjak@gmail.com>
2351 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2352 * configure: Regenerated.
2354 2014-02-12 Richard Biener <rguenther@suse.de>
2356 * vec.c (vec_prefix::calculate_allocation): Move as
2357 inline variant to vec.h.
2358 (vec_prefix::calculate_allocation_1): New out-of-line version.
2359 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2360 (vec_prefix::m_has_auto_buf): Rename to ...
2361 (vec_prefix::m_using_auto_storage): ... this.
2362 (vec_prefix::calculate_allocation): Inline the easy cases
2363 and dispatch to calculate_allocation_1 which doesn't need the
2365 (va_heap::reserve): Use gcc_checking_assert.
2366 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2367 m_using_auto_storage.
2368 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2370 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2371 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2372 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2374 2014-02-12 Richard Biener <rguenther@suse.de>
2376 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2377 when we found a dependence.
2379 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2381 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2383 (maybe_fold_stmt): ... into this new function.
2384 * omp-low.c (lower_omp): Update comment.
2386 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2389 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2392 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2394 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2395 identifiers in comments.
2396 (cortexa53_extra_costs): Likewise.
2397 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2398 (cortexa7_extra_costs): Likewise.
2399 (cortexa12_extra_costs): Likewise.
2400 (cortexa15_extra_costs): Likewise.
2401 (v7m_extra_costs): Likewise.
2403 2014-02-12 Richard Biener <rguenther@suse.de>
2406 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2407 of posix_memalign being successful.
2408 (lower_stmt): Restrict lowering of posix_memalign to when
2409 -ftree-bit-ccp is enabled.
2411 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2413 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2415 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2417 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2419 PR rtl-optimization/60116
2420 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2421 other_insn once the combination has been validated.
2423 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2426 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2428 * ipa-devirt.c: Include demangle.h
2429 (odr_violation_reported): New static variable.
2430 (add_type_duplicate): Update odr_violations.
2431 (maybe_record_node): Add completep parameter; update it.
2432 (record_target_from_binfo): Add COMPLETEP parameter;
2433 update it as needed.
2434 (possible_polymorphic_call_targets_1): Likewise.
2435 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2436 rename FINAL to COMPLETE.
2437 (record_targets_from_bases): Sanity check we found the binfo;
2438 fix COMPLETEP updating.
2439 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2440 parameter, fix computing of COMPLETEP.
2441 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2442 at LTO time do demangling.
2443 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2444 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2446 (gimple_get_virt_method_for_binfo): Likewise.
2447 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2448 gimple_get_virt_method_for_vtable): Update prototypes.
2450 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2453 * genautomata.c (add_presence_absence): Fix typo with
2454 {final_}presence_list.
2456 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2459 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2460 for VSX/Altivec vectors that land in GPR registers.
2462 2014-02-11 Richard Henderson <rth@redhat.com>
2463 Jakub Jelinek <jakub@redhat.com>
2466 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2467 around drhs if type conversion to lacc->type is not useless.
2469 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2471 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2473 (cortex-a57.cortex-a53): Likewise.
2474 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2476 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2478 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2481 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2483 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2484 add_options_for_arm_vfp3.
2486 2014-02-11 Jeff Law <law@redhat.com>
2489 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2490 object with an undesirable mode.
2492 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2495 * config/i386/sol2-9.h: New file.
2496 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2497 *-*-solaris2.9*): Use it.
2499 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2501 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2502 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2504 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2506 * config/microblaze/microblaze.c: Extend mcpu version format
2508 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2510 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2512 2014-02-10 Richard Henderson <rth@redhat.com>
2515 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2516 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2517 ms-abi vs -mno-accumulate-outgoing-args.
2518 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2519 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2522 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2525 * cfgexpand.c (expand_asm_operands): Attach source location to
2526 ASM_INPUT rtx objects.
2527 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2529 2014-02-10 Nick Clifton <nickc@redhat.com>
2531 * config/mn10300/mn10300.c (popcount): New function.
2532 (mn10300_expand_prologue): Include saved registers in stack usage
2535 2014-02-10 Jeff Law <law@redhat.com>
2538 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2539 when changing the SET_DEST of a prior insn to avoid an input reload.
2541 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2543 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2544 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2545 -mcall-openbsd, or -mcall-linux.
2546 (CC1_ENDIAN_BIG_SPEC): Remove.
2547 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2548 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2549 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2550 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2551 and %cc1_endian_default.
2552 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2554 2014-02-10 Richard Biener <rguenther@suse.de>
2556 PR tree-optimization/60115
2557 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2558 MEM_REF handling. Properly verify that the accesses are not
2559 out of the objects bound.
2561 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2563 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2566 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2568 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2569 proper constants and fix formatting.
2570 (possible_polymorphic_call_targets): Fix formatting.
2572 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2573 Ilya Tocar <ilya.tocar@intel.com>
2575 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2576 (_mm512_loadu_epi32): Renamed into...
2577 (_mm512_loadu_si512): This.
2578 (_mm512_storeu_epi32): Renamed into...
2579 (_mm512_storeu_si512): This.
2580 (_mm512_maskz_ceil_ps): Removed.
2581 (_mm512_maskz_ceil_pd): Ditto.
2582 (_mm512_maskz_floor_ps): Ditto.
2583 (_mm512_maskz_floor_pd): Ditto.
2584 (_mm512_floor_round_ps): Ditto.
2585 (_mm512_floor_round_pd): Ditto.
2586 (_mm512_ceil_round_ps): Ditto.
2587 (_mm512_ceil_round_pd): Ditto.
2588 (_mm512_mask_floor_round_ps): Ditto.
2589 (_mm512_mask_floor_round_pd): Ditto.
2590 (_mm512_mask_ceil_round_ps): Ditto.
2591 (_mm512_mask_ceil_round_pd): Ditto.
2592 (_mm512_maskz_floor_round_ps): Ditto.
2593 (_mm512_maskz_floor_round_pd): Ditto.
2594 (_mm512_maskz_ceil_round_ps): Ditto.
2595 (_mm512_maskz_ceil_round_pd): Ditto.
2596 (_mm512_expand_pd): Ditto.
2597 (_mm512_expand_ps): Ditto.
2598 * config/i386/i386.c (ix86_builtins): Remove
2599 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2600 (bdesc_args): Ditto.
2601 * config/i386/predicates.md (const1256_operand): New.
2602 (const_1_to_2_operand): Ditto.
2603 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2604 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2605 (*avx512pf_gatherpf<mode>sf): Ditto.
2606 (avx512pf_gatherpf<mode>df): Ditto.
2607 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2608 (*avx512pf_gatherpf<mode>df): Ditto.
2609 (avx512pf_scatterpf<mode>sf): Ditto.
2610 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2611 (*avx512pf_scatterpf<mode>sf): Ditto.
2612 (avx512pf_scatterpf<mode>df): Ditto.
2613 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2614 (*avx512pf_scatterpf<mode>df): Ditto.
2615 (avx512f_expand<mode>): Removed.
2616 (<shift_insn><mode>3<mask_name>): Change predicate type.
2618 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2620 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2621 not at the end of datarefs vector use ordered_remove to avoid
2622 reordering datarefs vector.
2625 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2626 mark local addressable non-static vars as GOVD_PRIVATE
2627 instead of GOVD_LOCAL.
2628 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2629 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2633 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2634 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2635 assume_aligned or alloc_align attributes.
2636 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2637 arguments. Handle also assume_aligned and alloc_align attributes.
2638 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2639 calls to functions with assume_aligned or alloc_align attributes.
2640 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2642 2014-02-08 Terry Guo <terry.guo@arm.com>
2644 * doc/invoke.texi: Document ARM -march=armv7e-m.
2646 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2648 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2649 flag on __cilkrts_rethrow builtin.
2652 * ipa-cp.c (determine_versionability): Fail at -O0
2653 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2654 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2657 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2660 * tree-inline.c (copy_forbidden): Fail for
2661 __attribute__((optimize (0))) functions.
2663 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2665 * varpool.c: Include pointer-set.h.
2666 (varpool_remove_unreferenced_decls): Variables in other partitions
2667 will not be output; be however careful to not lose information
2670 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2672 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2673 lookup in the vtable constructor.
2675 2014-02-07 Jeff Law <law@redhat.com>
2678 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2679 define_insn_and_split.
2681 * ipa-inline.c (inline_small_functions): Fix typos.
2683 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2685 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2686 (s390_can_use_return_insn): Declare.
2687 * config/s390/s390.h (EPILOGUE_USES): Define.
2688 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2690 (s390_chunkify_start): Handle return JUMP_LABELs.
2691 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2692 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2693 (s390_can_use_return_insn): New functions.
2694 (s390_fix_long_loop_prediction): Handle conditional returns.
2695 (TARGET_SET_UP_BY_PROLOGUE): Define.
2696 * config/s390/s390.md (ANY_RETURN): New code iterator.
2697 (*creturn, *csimple_return, return, simple_return): New patterns.
2699 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2701 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2702 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2703 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2704 REG_CFA_RESTORE list when deciding not to restore a register.
2706 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2708 * config/s390/s390.c: Include tree-pass.h and context.h.
2709 (s390_early_mach): New function, split out from...
2710 (s390_emit_prologue): ...here.
2711 (pass_data_s390_early_mach): New pass structure.
2712 (pass_s390_early_mach): New class.
2713 (s390_option_override): Create and register early_mach pass.
2714 Move to end of file.
2716 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2718 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2719 to match for the exit block.
2721 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2723 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2724 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2725 Reject misaligned operands.
2727 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2729 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2731 2014-02-07 Richard Biener <rguenther@suse.de>
2734 * gimple-low.c (lower_builtin_posix_memalign): New function.
2735 (lower_stmt): Call it to lower posix_memalign in a way
2736 to make alignment info accessible.
2738 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2741 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2742 __builtin_setjmp_receiver.
2744 2014-02-07 Richard Biener <rguenther@suse.de>
2747 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2748 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2749 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2750 Handle BUILT_IN_POSIX_MEMALIGN.
2751 (find_func_clobbers): Likewise.
2752 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2753 (call_may_clobber_ref_p_1): Likewise.
2755 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2758 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2761 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2764 * lto-cgraph.c (lto_output_node): Use
2765 symtab_get_symbol_partitioning_class.
2766 (lto_output_varpool_node): likewise.
2767 (symtab_get_symbol_partitioning_class): Move here from
2769 * cgraph.h (symbol_partitioning_class): Likewise.
2770 (symtab_get_symbol_partitioning_class): Declare.
2772 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2774 * ggc.h (ggc_internal_cleared_alloc): New macro.
2775 * vec.h (vec_safe_copy): Handle memory stats.
2776 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2777 * target-globals.c (save_target_globals): Likewise.
2779 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2782 * expr.c (emit_move_resolve_push): Export; be bit more selective
2783 on when to clear alias set.
2784 * expr.h (emit_move_resolve_push): Declare.
2785 * function.h (struct function): Add tail_call_marked.
2786 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2787 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2788 * config/i386/i386.md (TImode move expander): De not call
2790 (FP push expanders): Preserve memory attributes.
2791 * config/i386/sse.md (push<mode>1): Remove.
2792 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2793 (ix86_expand_push): Remove.
2794 * config/i386/mmx.md (push<mode>1): Remove.
2796 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2798 PR rtl-optimization/60030
2799 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2800 lopart with paradoxical subreg before shifting it up by hprec.
2802 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2804 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2805 Remove extra newline at end of file.
2806 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2807 (arm_issue_rate): Handle cortexa57.
2808 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2809 (cortex-a57.cortex-a53): Likewise.
2811 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2814 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2815 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2817 (arm_expand_prologue): Adjust all callers.
2818 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2819 info, registers also at the lowest numbered registers side. Use
2820 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2824 * var-tracking.c (adjust_mems): Before adding a SET to
2825 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2827 2014-02-06 Alan Modra <amodra@gmail.com>
2830 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2831 change SDmode to DDmode when lra_in_progress.
2833 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2836 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2837 free_data_ref on the dr first, and before goto again also set dr
2838 to the next dr. For simd_lane_access, free old datarefs[i] before
2839 overwriting it. For get_vectype_for_scalar_type failure, don't
2840 free_data_ref if simd_lane_access.
2842 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2845 * tree.h (opts_for_fn): New inline function.
2846 (opt_for_fn): Define.
2847 * config/i386/i386.c (ix86_function_regparm): Use
2848 opt_for_fn (decl, optimize) instead of optimize.
2850 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2852 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2853 for SYMBOL_REF in large memory model.
2855 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2857 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2859 (cortex-a57): Likewise.
2860 (cortex-a57.cortex-a53): Likewise.
2862 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2863 Kugan Vivekanandarajah <kuganv@linaro.org>
2865 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2867 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2869 2014-02-06 Richard Biener <rguenther@suse.de>
2871 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2872 set_loop_copy and initialize_original_copy_tables.
2874 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2876 * config/aarch64/aarch64-simd.md
2877 (aarch64_ashr_simddi): Change QI to SI.
2879 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2880 Jakub Jelinek <jakub@redhat.com>
2883 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2886 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2888 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2889 CODE_FOR_altivec_vpku[hw]um to
2890 CODE_FOR_altivec_vpku[hw]um_direct.
2891 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2892 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2893 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2894 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2896 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2898 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2899 generation for -maltivec=be.
2900 (altivec_vsumsws): Simplify redundant test.
2902 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2904 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2905 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2906 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2907 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2908 gen_altivec_vpkuwum.
2909 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2911 (altivec_vpks<VI_char>ss): Likewise.
2912 (altivec_vpks<VI_char>us): Likewise.
2913 (altivec_vpku<VI_char>us): Likewise.
2914 (altivec_vpku<VI_char>um): Likewise.
2915 (altivec_vpku<VI_char>um_direct): New (copy of
2916 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2918 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2919 target is little endian and -maltivec=be is not specified.
2920 (*altivec_vupkhs<VU_char>_direct): New (copy of
2921 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2922 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2923 target is little endian and -maltivec=be is not specified.
2924 (*altivec_vupkls<VU_char>_direct): New (copy of
2925 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2926 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2927 little endian and -maltivec=be is not specified.
2928 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2929 little endian and -maltivec=be is not specified.
2931 2014-02-05 Richard Henderson <rth@redhat.com>
2934 * combine-stack-adj.c: Revert r206943.
2935 * sched-int.h (struct deps_desc): Add last_args_size.
2936 * sched-deps.c (init_deps): Initialize it.
2937 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2938 contain REG_ARGS_SIZE notes.
2940 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2942 * lto-cgraph.c (asm_nodes_output): Make global.
2943 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2944 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2945 (driver_handle_option): Handle OPT_fwpa.
2947 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2950 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2951 a comment typo and formatting issue. If odr_hash hasn't been
2952 created, return vNULL and set *completep to false.
2955 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2956 bb with no successors.
2958 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2961 * doc/invoke.texi (-march): Clarify documentation for ARM.
2965 2014-02-05 Richard Biener <rguenther@suse.de>
2967 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2968 when not vectorizing because of too many alias checks.
2969 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2970 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2972 2014-02-05 Nick Clifton <nickc@redhat.com>
2974 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2975 accept extended registers in any mode when compiling for the MN10300.
2977 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2979 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2980 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2981 sanitization attributes.
2982 (can_inline_edge_p): Likewise.
2983 (sanitize_attrs_match_for_inline_p): New function.
2985 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2987 * ipa-prop.c (detect_type_change): Shor circuit testing of
2988 type changes on THIS pointer.
2990 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2993 * config/pa/pa.c (legitimize_tls_address): Return original address
2994 if not passed a SYMBOL_REF rtx.
2995 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2997 (pa_emit_move_sequence): Simplify TLS source operands.
2998 (pa_legitimate_constant_p): Reject all TLS constants.
2999 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3000 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3002 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3004 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3005 groups when we know they are controlled by LTO.
3006 * varasm.c (default_binds_local_p_1): If object is in other partition,
3007 it will be resolved locally.
3009 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3011 * config/host-linux.c (linux_gt_pch_use_address): Don't
3012 use SSIZE_MAX because it is not always defined.
3014 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3017 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3018 threshold for pseudo splitting.
3019 (update_ebb_live_info): Process call argument hard registers and
3020 hard registers from insn definition too.
3021 (max_small_class_regs_num): New constant.
3022 (inherit_in_ebb): Update live hard regs through EBBs. Update
3023 reloads_num only for small register classes. Don't split for
3026 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3029 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3032 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3034 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3037 2014-02-04 Marek Polacek <polacek@redhat.com>
3039 * gdbinit.in (pel): Define.
3041 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3043 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3046 2014-02-04 Richard Biener <rguenther@suse.de>
3049 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3050 in function context local.
3051 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3052 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3053 similar to LTO_imported_decl_ref.
3055 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3057 PR tree-optimization/60002
3058 * cgraphclones.c (build_function_decl_skip_args): Clear
3061 PR tree-optimization/60023
3062 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3063 false to gsi_replace.
3064 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3065 has been in some EH region and vec_stmt could throw, add
3066 vec_stmt into the same EH region.
3067 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3068 has no lhs, ignore it.
3069 * internal-fn.c (expand_MASK_LOAD): Likewise.
3072 * tree-inline.c (copy_forbidden): Fail for
3073 __attribute__((optimize (0))) functions.
3076 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3077 is set, copy one less argument.
3078 (expand_simd_clones): Don't subtract clone_info->inbranch
3079 from simd_clone_struct_alloc argument.
3081 PR rtl-optimization/57915
3082 * recog.c (simplify_while_replacing): If all unary/binary/relational
3083 operation arguments are constant, attempt to simplify those.
3086 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3087 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3089 2014-02-04 Richard Biener <rguenther@suse.de>
3091 PR tree-optimization/60012
3092 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3093 TBAA disambiguation to all DDRs.
3095 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3098 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3099 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3101 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3104 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3106 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3108 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3109 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3111 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3114 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3115 to figure out targets of polymorphic calls with known decl.
3116 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3117 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3118 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3119 (get_polymorphic_call_info): ... here.
3120 (get_polymorphic_call_info_from_invariant): New function.
3122 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3124 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3125 lookup via vtable pointer; check for type consistency
3126 and turn inconsitent facts into UNREACHABLE.
3127 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3128 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3129 type inconsistent querries; return UNREACHABLE instead.
3131 2014-02-03 Richard Henderson <rth@twiddle.net>
3134 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3135 already processed this node.
3136 (normalize_one_pred_1): Pass along mark_set.
3137 (normalize_one_pred): Create and destroy a pointer_set_t.
3138 (normalize_one_pred_chain): Likewise.
3140 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3142 PR gcov-profile/58602
3143 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3145 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3148 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3149 -fno-devirtualize; try to devirtualize by the knowledge of
3150 virtual table pointer given by aggregate propagation.
3151 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3152 (ipa_print_node_jump_functions): Dump also offset that
3153 is relevant for polymorphic calls.
3154 (determine_known_aggregate_parts): Add arg_type parameter; use it
3155 instead of determining the type from pointer type.
3156 (ipa_compute_jump_functions_for_edge): Update call of
3157 determine_known_aggregate_parts.
3158 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3159 (gimple_get_virt_method_for_binfo): ... here; simplify using
3160 vtable_pointer_value_to_vtable.
3161 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3162 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3163 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3164 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3166 (vtable_pointer_value_to_binfo): ... here.
3167 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3169 2014-02-03 Teresa Johnson <tejohnson@google.com>
3171 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3172 redef of outer loop index variable.
3174 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3178 * doc/extend.texi (Function Attributes): Typo.
3180 2014-02-03 Cong Hou <congh@google.com>
3182 PR tree-optimization/60000
3183 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3184 if the vectorized statement is a store. A store statement can only
3185 appear at the end of pattern statements.
3187 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3189 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3190 (ix86_option_override_internal): Default long double to 64-bit for
3191 32-bit Bionic and to 128-bit for 64-bit Bionic.
3193 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3194 TARGET_LONG_DOUBLE_128 is true.
3195 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3197 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3198 (mlong-double-64): Negate -mlong-double-128.
3199 (mlong-double-128): New option.
3201 * config/i386/i386-c.c (ix86_target_macros): Define
3202 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3204 * doc/invoke.texi: Document -mlong-double-128.
3206 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3208 PR rtl-optimization/60024
3209 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3211 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3213 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3215 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3217 PR rtl-optimization/57662
3218 * sel-sched.c (code_motion_path_driver): Do not mark already not
3219 existing blocks in the visiting bitmap.
3221 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3223 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3224 on the insn being emitted.
3226 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3227 Will Deacon <will.deacon@arm.com>
3229 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3231 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3233 * config/arm/arm-tables.opt: Regenerate.
3235 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3237 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3238 for vector types other than V16QImode.
3239 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3240 define_expand, and call altivec_expand_vec_perm_le when producing
3241 code with little endian element order.
3242 (*altivec_vperm_<mode>_internal): New insn having previous
3243 behavior of altivec_vperm_<mode>.
3244 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3245 altivec_expand_vec_perm_le when producing code with little endian
3247 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3248 behavior of altivec_vperm_<mode>_uns.
3250 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3252 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3253 (altivec_vsumsws): Add handling for -maltivec=be with a little
3255 (altivec_vsumsws_direct): New.
3256 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3257 gen_altivec_vsumsws.
3259 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3261 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3262 vtable_pointer_value_to_binfo): New functions.
3263 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3264 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3266 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3268 * config/nios2/nios2.md (load_got_register): Initialize GOT
3269 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3270 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3272 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3274 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3275 preserverd by passthrough, do not propagate the type.
3277 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3279 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3280 (mips_atomic_assign_expand_fenv): New function.
3281 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3283 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3285 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3286 (__builtin_mips_set_fcsr): Likewise.
3287 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3288 MIPS_USI_FTYPE_VOID.
3289 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3290 (mips16_expand_set_fcsr): Likewise.
3291 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3292 (mips16_set_fcsr_stub): Likewise.
3293 (mips16_get_fcsr_one_only_stub): New class.
3294 (mips16_set_fcsr_one_only_stub): Likewise.
3295 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3296 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3297 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3298 (hard_float): New availability predicate.
3299 (mips_builtins): Add get_fcsr and set_fcsr.
3300 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3301 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3302 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3303 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3304 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3307 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3309 * config/mips/mips.c (mips_one_only_stub): New class.
3310 (mips_need_mips16_rdhwr_p): Replace with...
3311 (mips16_rdhwr_stub): ...this new variable.
3312 (mips16_stub_call_address): New function.
3313 (mips16_rdhwr_one_only_stub): New class.
3314 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3315 (mips_output_mips16_rdhwr): Delete.
3316 (mips_finish_stub): New function.
3317 (mips_code_end): Use it to handle rdhwr stubs.
3319 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3322 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3323 when calculating size of integer atomic types.
3325 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3327 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3329 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3331 PR tree-optimization/60003
3332 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3333 * profile.c (branch_prob): Use gimple_call_builtin_p
3334 to check for BUILT_IN_SETJMP_RECEIVER.
3335 * tree-inline.c (copy_bb): Call notice_special_calls.
3337 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3340 * lra-constraints.c (process_alt_operands): Update reload_sum only
3343 2014-01-31 Richard Henderson <rth@redhat.com>
3346 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3347 until after else_eh is processed.
3349 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3351 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3352 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3353 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3354 in smmintrin.h, remove them.
3355 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3356 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3357 * config/i386/i386.md (ROUND_SAE): Fix value.
3358 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3359 (const48_operand): New.
3360 * config/i386/subst.md (round), (round_expand): Use
3361 const_4_or_8_to_11_operand.
3362 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3364 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3366 * config/i386/constraints.md (Yk): Swap meaning with k.
3367 * config/i386/i386.md (movhi_internal): Change Yk to k.
3368 (movqi_internal): Ditto.
3369 (*k<logic><mode>): Ditto.
3372 (kandn<mode>): Ditto.
3373 (*<code>hi_1): Ditto.
3374 (*<code>qi_1): Ditto.
3375 (kxnor<mode>): Ditto.
3376 (kortestzhi): Ditto.
3377 (kortestchi): Ditto.
3379 (*one_cmplhi2_1): Ditto.
3380 (*one_cmplqi2_1): Ditto.
3381 * config/i386/sse.md (): Change k to Yk.
3382 (avx512f_load<mode>_mask): Ditto.
3383 (avx512f_blendm<mode>): Ditto.
3384 (avx512f_store<mode>_mask): Ditto.
3385 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3386 (avx512f_storedqu<mode>_mask): Ditto.
3387 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3389 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3390 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3391 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3392 (avx512f_maskcmp<mode>3): Ditto.
3393 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3394 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3395 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3396 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3397 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3398 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3399 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3400 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3401 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3402 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3403 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3404 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3405 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3406 (vec_extract_lo_<mode>_maskm): Ditto.
3407 (vec_extract_hi_<mode>_maskm): Ditto.
3408 (avx512f_vternlog<mode>_mask): Ditto.
3409 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3410 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3411 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3412 (avx512f_<code>v8div16qi2_mask): Ditto.
3413 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3414 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3415 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3416 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3417 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3418 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3419 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3420 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3421 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3422 (avx512cd_maskb_vec_dupv8di): Ditto.
3423 (avx512cd_maskw_vec_dupv16si): Ditto.
3424 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3425 (avx512f_vpermi2var<mode>3_mask): Ditto.
3426 (avx512f_vpermi2var<mode>3_mask): Ditto.
3427 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3428 (*avx512f_gathersi<mode>): Ditto.
3429 (*avx512f_gathersi<mode>_2): Ditto.
3430 (*avx512f_gatherdi<mode>): Ditto.
3431 (*avx512f_gatherdi<mode>_2): Ditto.
3432 (*avx512f_scattersi<mode>): Ditto.
3433 (*avx512f_scatterdi<mode>): Ditto.
3434 (avx512f_compress<mode>_mask): Ditto.
3435 (avx512f_compressstore<mode>_mask): Ditto.
3436 (avx512f_expand<mode>_mask): Ditto.
3437 * config/i386/subst.md (mask): Change k to Yk.
3438 (mask_scalar_merge): Ditto.
3441 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3443 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3445 2014-01-31 Richard Biener <rguenther@suse.de>
3448 * builtins.c (fold_builtin_memory_op): Make sure to not
3449 use a floating-point mode or a boolean or enumeral type for
3452 2014-01-30 DJ Delorie <dj@redhat.com>
3454 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3455 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3456 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3457 whenever main() has an epilogue.
3459 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3461 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3462 unused variable "field".
3463 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3464 (vsx_mergeh_<mode>): Likewise.
3465 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3466 (altivec_vmrghh): Likewise.
3467 (altivec_vmrghw): Likewise.
3468 (altivec_vmrglb): Likewise.
3469 (altivec_vmrglh): Likewise.
3470 (altivec_vmrglw): Likewise.
3471 (altivec_vspltb): Add missing uses.
3472 (altivec_vsplth): Likewise.
3473 (altivec_vspltw): Likewise.
3474 (altivec_vspltsf): Likewise.
3476 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3479 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3480 frame related instructions.
3482 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3484 PR rtl-optimization/59959
3485 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3486 any reload of register whose subreg is invalid.
3488 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3490 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3491 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3492 Add missing return type - void.
3494 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3496 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3497 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3498 remove element index adjustment for endian (now handled in vsx.md
3500 (altivec_expand_vec_perm_const): Use
3501 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3502 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3503 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3504 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3505 define_expand and a new define_insn *altivec_vspltb_internal;
3506 adjust for -maltivec=be on a little endian target.
3507 (altivec_vspltb_direct): New.
3508 (altivec_vsplth): Divide into a define_expand and a new
3509 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3510 little endian target.
3511 (altivec_vsplth_direct): New.
3512 (altivec_vspltw): Divide into a define_expand and a new
3513 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3514 little endian target.
3515 (altivec_vspltw_direct): New.
3516 (altivec_vspltsf): Divide into a define_expand and a new
3517 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3518 a little endian target.
3520 2014-01-30 Richard Biener <rguenther@suse.de>
3522 PR tree-optimization/59993
3523 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3524 can propagate form the earlier stmt and avoid the transform
3525 when the intermediate result is needed.
3527 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3529 * README.Portability: Fix typo.
3531 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3533 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3534 comparison_operator with ordered_comparison_operator.
3536 2014-01-30 Nick Clifton <nickc@redhat.com>
3538 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3539 Rename to mn10300_store_multiple_regs.
3540 * config/mn10300/mn10300.c: Likewise.
3541 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3542 store_multiple_regs.
3543 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3544 Call mn10300_store_multiple_regs.
3546 2014-01-30 Nick Clifton <nickc@redhat.com>
3547 DJ Delorie <dj@redhat.com>
3549 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3550 %fp 2 to keep registers after it properly word-aligned.
3551 (rl78_alloc_physical_registers_umul): Handle the case where both
3552 input operands are the same.
3554 2014-01-30 Richard Biener <rguenther@suse.de>
3556 PR tree-optimization/59903
3557 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3560 2014-01-30 Jason Merrill <jason@redhat.com>
3563 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3566 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3568 2014-01-30 Richard Biener <rguenther@suse.de>
3570 PR tree-optimization/59951
3571 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3573 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3576 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3577 SFmode to DFmode case.
3579 2014-01-29 DJ Delorie <dj@redhat.com>
3581 * config/msp430/msp430.opt (-minrt): New.
3582 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3584 (ENDFILE_SPEC): Likewise.
3586 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3588 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3589 (estimate_function_body_sizes): Use it.
3591 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3594 * dwarf2out.c (is_cxx_auto): New.
3595 (is_base_type): Use it.
3596 (gen_type_die_with_usage): Likewise.
3598 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3600 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3601 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3602 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3603 -maltivec=be with LE targets.
3604 (vsx_mergeh_<mode>): Likewise.
3605 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3606 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3607 (altivec_vmrghb): Replace with define_expand and new
3608 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3609 (altivec_vmrghb_direct): New define_insn.
3610 (altivec_vmrghh): Replace with define_expand and new
3611 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3612 (altivec_vmrghh_direct): New define_insn.
3613 (altivec_vmrghw): Replace with define_expand and new
3614 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3615 (altivec_vmrghw_direct): New define_insn.
3616 (*altivec_vmrghsf): Adjust for endianness.
3617 (altivec_vmrglb): Replace with define_expand and new
3618 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3619 (altivec_vmrglb_direct): New define_insn.
3620 (altivec_vmrglh): Replace with define_expand and new
3621 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3622 (altivec_vmrglh_direct): New define_insn.
3623 (altivec_vmrglw): Replace with define_expand and new
3624 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3625 (altivec_vmrglw_direct): New define_insn.
3626 (*altivec_vmrglsf): Adjust for endianness.
3627 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3628 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3629 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3630 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3631 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3632 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3633 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3634 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3636 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3638 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3639 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3642 2014-01-29 Richard Biener <rguenther@suse.de>
3644 PR tree-optimization/58742
3645 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3646 associate_pointerplus_align.
3647 (associate_pointerplus_diff): New function.
3648 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3649 and associate_pointerplus_diff.
3651 2014-01-29 Richard Biener <rguenther@suse.de>
3653 * lto-streamer.h (LTO_major_version): Bump to 3.
3654 (LTO_minor_version): Reset to 0.
3656 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3658 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3659 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3660 (arm_file_start): Generate correct asm header for armv7ve.
3661 * config/arm/bpabi.h: Add multilib support for armv7ve.
3662 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3663 and cortex-a15 to armv7ve.
3664 * config/arm/t-aprofile: Add multilib support for armv7ve.
3665 * doc/invoke.texi: Document -march=armv7ve.
3667 2014-01-29 Richard Biener <rguenther@suse.de>
3669 PR tree-optimization/58742
3670 * tree-ssa-forwprop.c (associate_plusminus): Return true
3671 if we changed sth, defer EH cleanup to ...
3672 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3673 (simplify_mult): New function.
3675 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3678 PR tree-optimization/59920
3679 * tree.c (build_common_builtin_nodes): Remove
3680 __builtin_setjmp_dispatcher initialization.
3681 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3682 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3683 instead of gsi_after_labels + manually skipping debug stmts.
3684 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3685 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3686 * tree-inline.c (copy_edges_for_bb): Remove
3687 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3688 argument. Ignore computed_goto_p stmts. Don't call
3689 make_abnormal_goto_edges. If a call might need abnormal edges
3690 for non-local gotos, see if it already has an edge to
3691 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3692 with true argument, don't do anything then, otherwise add
3693 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3694 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3696 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3697 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3698 (lower_stmt): Don't set data->calls_builtin_setjmp.
3699 (lower_builtin_setjmp): Adjust comment.
3700 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3701 * tree-cfg.c (found_computed_goto): Remove.
3702 (factor_computed_gotos): Remove.
3703 (make_goto_expr_edges): Return bool, true for computed gotos.
3704 Don't call make_abnormal_goto_edges.
3705 (build_gimple_cfg): Don't set found_computed_goto, don't call
3706 factor_computed_gotos.
3707 (computed_goto_p): No longer static.
3708 (make_blocks): Don't set found_computed_goto.
3709 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3710 (make_edges): If make_goto_expr_edges returns true, push bb
3711 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3712 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3713 vector. Record mapping between bbs and OpenMP regions if there
3714 are any, adjust make_gimple_omp_edges caller. Call
3715 handle_abnormal_edges.
3716 (make_abnormal_goto_edges): Remove.
3717 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3718 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3719 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3720 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3721 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3722 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3723 filling *region also set *region_idx to (*region)->entry->index.
3726 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3727 For REGs set ORIGINAL_REGNO.
3729 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3731 * doc/md.texi: Mention that a target shouldn't implement
3732 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3735 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3737 PR tree-optimization/59594
3738 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3739 a copy of the datarefs vector rather than the vector itself.
3741 2014-01-28 Jason Merrill <jason@redhat.com>
3744 * dwarf2out.c (auto_die): New static.
3745 (gen_type_die_with_usage): Handle C++1y 'auto'.
3746 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3749 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3752 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3753 (SPEC_X32): Likewise.
3754 (SPEC_64): Likewise.
3755 * config/i386/i386.c (ix86_option_override_internal): Turn off
3756 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3758 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3759 * config/i386/i386.h (TARGET_16BIT): New macro.
3760 (TARGET_16BIT_P): Likewise.
3761 * config/i386/i386.opt: Add m16.
3762 * doc/invoke.texi: Document -m16.
3764 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3766 PR preprocessor/59935
3767 * input.c (location_get_source_line): Bail out on when line number
3768 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3770 2014-01-28 Richard Biener <rguenther@suse.de>
3772 PR tree-optimization/58742
3773 * tree-ssa-forwprop.c (associate_plusminus): Handle
3774 pointer subtraction of the form (T)(P + A) - (T)P.
3776 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3778 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3781 2014-01-28 Richard Biener <rguenther@suse.de>
3784 2014-01-28 Richard Biener <rguenther@suse.de>
3786 PR rtl-optimization/45364
3787 PR rtl-optimization/59890
3788 * var-tracking.c (local_get_addr_clear_given_value): Handle
3789 already cleared slot.
3790 (val_reset): Handle not allocated local_get_addr_cache.
3791 (vt_find_locations): Use post-order on the inverted CFG.
3793 2014-01-28 Richard Biener <rguenther@suse.de>
3795 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3797 2014-01-28 Richard Biener <rguenther@suse.de>
3799 PR rtl-optimization/45364
3800 PR rtl-optimization/59890
3801 * var-tracking.c (local_get_addr_clear_given_value): Handle
3802 already cleared slot.
3803 (val_reset): Handle not allocated local_get_addr_cache.
3804 (vt_find_locations): Use post-order on the inverted CFG.
3806 2014-01-28 Alan Modra <amodra@gmail.com>
3808 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3809 * configure.ac <recursive call for build != host>: Define
3810 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3811 and LD_FOR_BUILD too.
3812 * configure: Regenerate.
3814 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3816 * config/i386/i386.c (get_builtin_code_for_version): Separate
3817 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3818 Broadwell from Haswell.
3820 2014-01-27 Steve Ellcey <sellcey@mips.com>
3822 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3823 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3824 * config/mips/mips.c (mips_option_override): Change setting
3826 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3827 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3828 Change from Mask to Var.
3830 2014-01-27 Jeff Law <law@redhat.com>
3832 * ipa-inline.c (inline_small_functions): Fix typo.
3834 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3836 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3837 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3838 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3839 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3840 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3841 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3842 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3843 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3844 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3845 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3846 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3847 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3848 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3849 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3850 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3851 (_mm512_storeu_epi64): Ditto.
3852 (_mm512_cmpge_epi32_mask): Ditto.
3853 (_mm512_cmpge_epu32_mask): Ditto.
3854 (_mm512_cmpge_epi64_mask): Ditto.
3855 (_mm512_cmpge_epu64_mask): Ditto.
3856 (_mm512_cmple_epi32_mask): Ditto.
3857 (_mm512_cmple_epu32_mask): Ditto.
3858 (_mm512_cmple_epi64_mask): Ditto.
3859 (_mm512_cmple_epu64_mask): Ditto.
3860 (_mm512_cmplt_epi32_mask): Ditto.
3861 (_mm512_cmplt_epu32_mask): Ditto.
3862 (_mm512_cmplt_epi64_mask): Ditto.
3863 (_mm512_cmplt_epu64_mask): Ditto.
3864 (_mm512_cmpneq_epi32_mask): Ditto.
3865 (_mm512_cmpneq_epu32_mask): Ditto.
3866 (_mm512_cmpneq_epi64_mask): Ditto.
3867 (_mm512_cmpneq_epu64_mask): Ditto.
3868 (_mm512_expand_pd): Ditto.
3869 (_mm512_expand_ps): Ditto.
3870 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3871 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3872 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3873 * config/i386/i386.c (ix86_builtins): Add
3874 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3875 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3876 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3877 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3878 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3879 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3880 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3881 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3882 IX86_BUILTIN_PMOVUSQW512_MEM.
3883 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3884 __builtin_ia32_pmovsqd512mem_mask,
3885 __builtin_ia32_pmovqd512mem_mask,
3886 __builtin_ia32_pmovusqw512mem_mask,
3887 __builtin_ia32_pmovsqw512mem_mask,
3888 __builtin_ia32_pmovqw512mem_mask,
3889 __builtin_ia32_pmovusdw512mem_mask,
3890 __builtin_ia32_pmovsdw512mem_mask,
3891 __builtin_ia32_pmovdw512mem_mask,
3892 __builtin_ia32_pmovqb512mem_mask,
3893 __builtin_ia32_pmovusqb512mem_mask,
3894 __builtin_ia32_pmovsqb512mem_mask,
3895 __builtin_ia32_pmovusdb512mem_mask,
3896 __builtin_ia32_pmovsdb512mem_mask,
3897 __builtin_ia32_pmovdb512mem_mask.
3898 (bdesc_args): Add __builtin_ia32_expanddf512,
3899 __builtin_ia32_expandsf512.
3900 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3901 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3902 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3903 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3904 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3905 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3906 (avx512f_<code>v8div16qi2_mask_store): This.
3907 (avx512f_expand<mode>): New.
3909 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3911 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3913 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3914 (_mm512_prefetch_i32scatter_pd): Ditto.
3915 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3916 (_mm512_prefetch_i64scatter_pd): Ditto.
3917 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3918 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3919 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3920 (_mm512_prefetch_i32scatter_ps): Ditto.
3921 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3922 (_mm512_prefetch_i64scatter_ps): Ditto.
3923 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3924 * config/i386/i386-builtin-types.def: Define
3925 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3926 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3927 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3928 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3929 IX86_BUILTIN_SCATTERPFQPD.
3930 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3931 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3932 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3933 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3934 __builtin_ia32_scatterpfqps.
3935 (ix86_expand_builtin): Expand new built-ins.
3936 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3937 fix memory access data type.
3938 (*avx512pf_gatherpf<mode>_mask): Ditto.
3939 (*avx512pf_gatherpf<mode>): Ditto.
3940 (avx512pf_scatterpf<mode>): Ditto.
3941 (*avx512pf_scatterpf<mode>_mask): Ditto.
3942 (*avx512pf_scatterpf<mode>): Ditto.
3943 (GATHER_SCATTER_SF_MEM_MODE): New.
3944 (avx512pf_gatherpf<mode>df): Ditto.
3945 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3946 (*avx512pf_scatterpf<mode>df): Ditto.
3948 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3951 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3952 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3955 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3957 * common/config/arm/arm-common.c
3958 (arm_rewrite_mcpu): Handle multiple names.
3960 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3962 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3964 * gimple-builder.h (create_gimple_tmp): Delete.
3966 2014-01-27 Christian Bruel <christian.bruel@st.com>
3968 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3971 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3973 * config/pa/pa.md (call): Generate indirect long calls to non-local
3974 functions when outputing 32-bit code.
3975 (call_value): Likewise except for special call to buggy powf function.
3977 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3978 portable runtime and PIC indirect calls.
3979 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3980 and PIC call sequences. Use ldo instead of blr to set return register
3981 in PIC call sequence.
3983 2014-01-25 Walter Lee <walt@tilera.com>
3985 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3986 avoid clobbering a live register.
3988 2014-01-25 Walter Lee <walt@tilera.com>
3990 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3991 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3992 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3993 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3995 2014-01-25 Walter Lee <walt@tilera.com>
3997 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3998 arguments on even registers.
3999 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4001 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4002 (BIGGEST_ALIGNMENT): Ditto.
4003 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4005 2014-01-25 Walter Lee <walt@tilera.com>
4007 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4008 insns before bundling.
4009 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4011 2014-01-25 Walter Lee <walt@tilera.com>
4013 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4014 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4015 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4017 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4019 * config/mips/constraints.md (kl): Delete.
4020 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4021 define expands, using...
4022 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4023 instructions for MIPS16.
4024 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4025 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4027 2014-01-25 Walter Lee <walt@tilera.com>
4029 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4033 2014-01-25 Walter Lee <walt@tilera.com>
4035 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4036 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4038 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4040 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4043 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4045 * print-rtl.c (in_call_function_usage): New var.
4046 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4047 EXPR_LIST mode as mode and not as reg note name.
4050 * cfgloopmanip.c (copy_loop_info): If
4051 loop->warned_aggressive_loop_optimizations, make sure
4052 the flag is set in target loop too.
4054 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4056 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4058 * builtins.def: Likewise.
4059 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4060 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4061 * ira.c (ira_setup_eliminable_regset): Likewise.
4062 * omp-low.c (gate_expand_omp): Likewise.
4063 (execute_lower_omp): Likewise.
4064 (diagnose_sb_0): Likewise.
4065 (gate_diagnose_omp_blocks): Likewise.
4066 (simd_clone_clauses_extract): Likewise.
4069 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4071 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4072 correction for little endian...
4073 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4076 2014-01-24 Jeff Law <law@redhat.com>
4078 PR tree-optimization/59919
4079 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4080 for non-returning calls.
4082 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4084 * common/config/aarch64/aarch64-common.c
4085 (aarch64_rewrite_mcpu): Handle multiple names.
4086 * config/aarch64/aarch64.h
4087 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4089 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4091 * input.c (add_file_to_cache_tab): Handle the case where fopen
4094 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4097 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4098 from push operand if code of push isn't PRE_DEC.
4100 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4103 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4104 -mquad-memory-atomic. Update -mquad-memory documentation to say
4105 it is only used for non-atomic loads/stores.
4107 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4108 -mquad-memory or -mquad-memory-atomic switches.
4110 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4111 -mquad-memory-atomic to ISA 2.07 support.
4113 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4114 to separate support of normal quad word memory operations (ldq, stq)
4115 from the atomic quad word memory operations.
4117 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4118 support to separate non-atomic quad word operations from atomic
4119 quad word operations. Disable non-atomic quad word operations in
4120 little endian mode so that we don't have to swap words after the
4121 load and before the store.
4122 (quad_load_store_p): Add comment about atomic quad word support.
4123 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4124 options printed with -mdebug=reg.
4126 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4127 -mquad-memory-atomic as the test for whether we have quad word
4128 atomic instructions.
4129 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4130 or -mp8-vector are used, allow byte/half-word atomic operations.
4132 * config/rs6000/sync.md (load_lockedti): Insure that the address
4133 is a proper indexed or indirect address for the lqarx instruction.
4134 On little endian systems, swap the hi/lo registers after the lqarx
4136 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4137 insure the address is valid for the lqarx instruction.
4138 (store_conditionalti): Insure that the address is a proper indexed
4139 or indirect address for the stqcrx. instruction. On little endian
4140 systems, swap the hi/lo registers before doing the stqcrx.
4142 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4143 insure the address is valid for the stqcrx. instruction.
4145 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4146 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4147 type of quad memory support is available.
4149 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4152 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4153 there is a danger of looping.
4155 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4157 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4158 force flag_ira_loop_pressure if set via command line.
4160 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4162 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4163 (ashr_simd): New builtin handling DI mode.
4164 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4165 (aarch64_sshr_simddi): New match pattern.
4166 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4167 (vshrd_n_s64): Likewise.
4168 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4170 2014-01-23 Nick Clifton <nickc@redhat.com>
4172 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4173 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4174 favour of mcu specific scripts.
4175 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4178 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4179 Alex Velenko <Alex.Velenko@arm.com>
4181 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4182 (vaddv_s16): Likewise.
4183 (vaddv_s32): Likewise.
4184 (vaddv_u8): Likewise.
4185 (vaddv_u16): Likewise.
4186 (vaddv_u32): Likewise.
4187 (vaddvq_s8): Likewise.
4188 (vaddvq_s16): Likewise.
4189 (vaddvq_s32): Likewise.
4190 (vaddvq_s64): Likewise.
4191 (vaddvq_u8): Likewise.
4192 (vaddvq_u16): Likewise.
4193 (vaddvq_u32): Likewise.
4194 (vaddvq_u64): Likewise.
4195 (vaddv_f32): Likewise.
4196 (vaddvq_f32): Likewise.
4197 (vaddvq_f64): Likewise.
4198 (vmaxv_f32): Likewise.
4199 (vmaxv_s8): Likewise.
4200 (vmaxv_s16): Likewise.
4201 (vmaxv_s32): Likewise.
4202 (vmaxv_u8): Likewise.
4203 (vmaxv_u16): Likewise.
4204 (vmaxv_u32): Likewise.
4205 (vmaxvq_f32): Likewise.
4206 (vmaxvq_f64): Likewise.
4207 (vmaxvq_s8): Likewise.
4208 (vmaxvq_s16): Likewise.
4209 (vmaxvq_s32): Likewise.
4210 (vmaxvq_u8): Likewise.
4211 (vmaxvq_u16): Likewise.
4212 (vmaxvq_u32): Likewise.
4213 (vmaxnmv_f32): Likewise.
4214 (vmaxnmvq_f32): Likewise.
4215 (vmaxnmvq_f64): Likewise.
4216 (vminv_f32): Likewise.
4217 (vminv_s8): Likewise.
4218 (vminv_s16): Likewise.
4219 (vminv_s32): Likewise.
4220 (vminv_u8): Likewise.
4221 (vminv_u16): Likewise.
4222 (vminv_u32): Likewise.
4223 (vminvq_f32): Likewise.
4224 (vminvq_f64): Likewise.
4225 (vminvq_s8): Likewise.
4226 (vminvq_s16): Likewise.
4227 (vminvq_s32): Likewise.
4228 (vminvq_u8): Likewise.
4229 (vminvq_u16): Likewise.
4230 (vminvq_u32): Likewise.
4231 (vminnmv_f32): Likewise.
4232 (vminnmvq_f32): Likewise.
4233 (vminnmvq_f64): Likewise.
4235 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4237 * config/aarch64/aarch64-simd.md
4238 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4239 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4240 (*aarch64_mul3_elt<mode>): Likewise.
4241 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4242 (*aarch64_mul3_elt_to_64v2df): Likewise.
4243 (*aarch64_mla_elt<mode>): Likewise.
4244 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4245 (*aarch64_mls_elt<mode>): Likewise.
4246 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4247 (*aarch64_fma4_elt<mode>): Likewise.
4248 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4249 (*aarch64_fma4_elt_to_64v2df): Likewise.
4250 (*aarch64_fnma4_elt<mode>): Likewise.
4251 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4252 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4253 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4254 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4255 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4256 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4257 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4258 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4259 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4261 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4263 * config/aarch64/aarch64-simd.md
4264 (aarch64_be_checked_get_lane<mode>): New define_expand.
4265 * config/aarch64/aarch64-simd-builtins.def
4266 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4267 New builtin definition.
4268 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4269 Use new safe be builtin.
4271 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4273 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4275 (aarch64_be_st1<mode>): Likewise.
4276 (aarch_ld1<VALL:mode>): Define_expand modified.
4277 (aarch_st1<VALL:mode>): Likewise.
4278 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4279 (UNSPEC_ST1): Likewise.
4281 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4283 * config/microblaze/microblaze.md: Add trap insn and attribute
4285 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4287 PR preprocessor/58580
4288 * input.h (location_get_source_line): Take an additional line_size
4290 (void diagnostics_file_cache_fini): Declare new function.
4291 * input.c (struct fcache): New type.
4292 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4293 New static constants.
4294 (diagnostic_file_cache_init, total_lines_num)
4295 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4296 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4297 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4298 (get_next_line, read_next_line, goto_next_line, read_line_num):
4299 New static function definitions.
4300 (diagnostic_file_cache_fini): New function.
4301 (location_get_source_line): Take an additional output line_len
4302 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4304 * diagnostic.c (diagnostic_finish): Call
4305 diagnostic_file_cache_fini.
4306 (adjust_line): Take an additional input parameter for the length
4307 of the line, rather than calculating it with strlen.
4308 (diagnostic_show_locus): Adjust the use of
4309 location_get_source_line and adjust_line with respect to their new
4310 signature. While displaying a line now, do not stop at the first
4311 null byte. Rather, display the zero byte as a space and keep
4312 going until we reach the size of the line.
4313 * Makefile.in: Add vec.o to OBJS-libcommon
4315 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4316 Ilya Tocar <ilya.tocar@intel.com>
4318 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4319 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4320 (__builtin_ia32_kmov16): Ditto.
4321 * config/i386/i386.md (UNSPEC_KMOV): New.
4324 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4326 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4327 (_mm512_storeu_si512): Ditto.
4329 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4332 * rtl.h (get_referenced_operands): Declare.
4333 * recog.c (get_referenced_operands): New function.
4334 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4335 operands have been referenced when recording LO_SUM references.
4337 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4339 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4341 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4343 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4344 Enable for generic and recent AMD targets.
4346 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4348 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4349 ARG_SIZE note when adjustment was eliminated.
4351 2014-01-22 Jeff Law <law@redhat.com>
4353 PR tree-optimization/59597
4354 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4355 in file. Accept new argument REGISTERING and use it to modify
4356 dump output appropriately.
4357 (register_jump_thread): Corresponding changes.
4358 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4359 thread paths involving joiner blocks. Add code to dump cancelled
4360 jump threading paths.
4362 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4364 PR rtl-optimization/59477
4365 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4366 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4368 2014-01-22 Tom Tromey <tromey@redhat.com>
4370 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4372 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4374 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4376 PR rtl-optimization/59896
4377 * lra-constraints.c (process_alt_operands): Check unused note for
4378 matched operands of insn with no output reloads.
4380 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4382 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4383 (mips_move_from_gpr_cost): Likewise.
4385 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4387 PR rtl-optimization/59858
4388 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4389 ira_class_hard_regs_num.
4390 (process_alt_operands): Increase reject for dying matched operand.
4392 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4395 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4396 smaller than size, perform several stores or loads and stores
4397 at dst + count - size to store or copy all of size bytes, rather
4398 than just last modesize bytes.
4400 2014-01-20 DJ Delorie <dj@redhat.com>
4402 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4403 that CLOBBERs are REGs before propogating their values.
4405 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4408 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4409 (cgraph_inline_failed_type): New function.
4410 * cgraph.h (DEFCIFCODE): Add type.
4411 (cgraph_inline_failed_type_t): New enum.
4412 (cgraph_inline_failed_type): New prototype.
4413 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4414 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4415 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4416 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4417 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4418 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4419 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4420 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4421 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4422 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4423 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4424 OPTIMIZATION_MISMATCH.
4425 * tree-inline.c (expand_call_inline): Emit errors during
4426 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4428 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4431 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4432 mode attribute in insn output.
4434 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4436 * output.h (output_constant): Delete.
4437 * varasm.c (output_constant): Make private.
4439 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4441 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4443 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4446 * tree.h (fold_builtin_strcat): New prototype.
4447 * builtins.c (fold_builtin_strcat): No longer static. Add len
4448 argument, if non-NULL, don't call c_strlen. Optimize
4449 directly into __builtin_memcpy instead of __builtin_strcpy.
4450 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4451 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4453 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4455 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4456 for SImode_address_operand operands, having only a REG argument.
4458 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4460 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4461 loader name using mbig-endian.
4462 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4464 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4466 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4470 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4472 * config/aarch64/aarch64-protos.h
4473 (aarch64_cannot_change_mode_class_ptr): Declare.
4474 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4475 aarch64_cannot_change_mode_class_ptr): New.
4476 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4477 backend hook aarch64_cannot_change_mode_class.
4479 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4481 * common/config/aarch64/aarch64-common.c
4482 (aarch64_handle_option): Don't handle any option order logic here.
4483 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4484 selected_cpu, warn on architecture version mismatch.
4485 (aarch64_override_options): Fix parsing order for option strings.
4487 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4488 Iain Sandoe <iain@codesourcery.com>
4491 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4492 warning. Amend comment to reflect current functionality.
4494 2014-01-20 Richard Biener <rguenther@suse.de>
4497 * builtins.c (fold_builtin_strcat): Remove case better handled
4498 by tree-ssa-strlen.c.
4500 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4502 * config/aarch64/aarch64.opt
4503 (mcpu, march, mtune): Make case-insensitive.
4505 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4508 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4509 if operands[1] is a REG or ZERO_EXTEND of a REG.
4511 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4513 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4515 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4517 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4518 long non-pic millicode calls.
4520 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4522 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4524 2014-01-19 Kito Cheng <kito@0xlab.org>
4526 * builtins.c (expand_movstr): Check movstr expand done or fail.
4528 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4529 H.J. Lu <hongjiu.lu@intel.com>
4532 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4533 to DImode for zero-extended addresses.
4535 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4537 PR rtl-optimization/57763
4538 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4539 on the new indirect jump_insn and increment LABEL_NUSES (label).
4541 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4545 * config.gcc (x86_archs): New variable.
4546 (x86_64_archs): Likewise.
4547 (x86_cpus): Likewise.
4548 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4549 --with-arch/--with-cpu= options.
4550 Support --with-arch=/--with-cpu={nehalem,westmere,
4551 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4553 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4555 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4556 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4558 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4560 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4562 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4565 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4566 clear cpp_get_options (parse_in)->warn_unused_macros for
4567 ix86_target_macros_internal with cpp_define.
4569 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4571 * jump.c (delete_related_insns): Keep (use (insn))s.
4572 * reorg.c (redundant_insn): Check for barriers too.
4574 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4576 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4578 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4580 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4581 call to $$dyncall when TARGET_LONG_CALLS is true.
4583 2014-01-17 Jeff Law <law@redhat.com>
4585 * ree.c (combine_set_extension): Temporarily disable test for
4586 changing number of hard registers.
4588 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4591 * ipa-inline-analysis.c (inline_free_summary):
4592 Do not free summary of aliases.
4594 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4597 * gimplify.c (gimplify_expr): Use create_tmp_var
4598 instead of create_tmp_var_raw. If cond doesn't have
4599 integral type, don't add the IFN_ANNOTATE builtin at all.
4601 2014-01-17 Martin Jambor <mjambor@suse.cz>
4604 * ipa-cp.c (prev_edge_clone): New variable.
4605 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4606 Also resize prev_edge_clone vector.
4607 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4608 (ipcp_edge_removal_hook): New function.
4609 (ipcp_driver): Register ipcp_edge_removal_hook.
4611 2014-01-17 Andrew Pinski <apinski@cavium.com>
4612 Steve Ellcey <sellcey@mips.com>
4615 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4618 2014-01-17 Jeff Law <law@redhat.com>
4621 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4622 so that pass_ccp runs first.
4624 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4626 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4627 (ix86_adjust_cost): Use !TARGET_XXX.
4628 (do_reorder_for_imul): Likewise.
4629 (swap_top_of_ready_list): Likewise.
4630 (ix86_sched_reorder): Likewise.
4632 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4634 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4635 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4636 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4637 (intel_memset): New. Duplicate slm_memset.
4638 (intel_cost): New. Duplicate slm_cost.
4639 (m_INTEL): New macro.
4640 (processor_target_table): Add "intel".
4641 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4642 with PROCESSOR_INTEL for "intel".
4643 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4644 PROCESSOR_SILVERMONT.
4645 (ix86_issue_rate): Likewise.
4646 (ix86_adjust_cost): Likewise.
4647 (ia32_multipass_dfa_lookahead): Likewise.
4648 (swap_top_of_ready_list): Likewise.
4649 (ix86_sched_reorder): Likewise.
4650 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4651 instead of TARGET_OPT_AGU.
4652 * config/i386/i386.h (TARGET_INTEL): New.
4653 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4654 (processor_type): Add PROCESSOR_INTEL.
4655 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4656 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4658 2014-01-17 Marek Polacek <polacek@redhat.com>
4661 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4664 2014-01-17 Richard Biener <rguenther@suse.de>
4666 PR tree-optimization/46590
4667 * opts.c (default_options_table): Add entries for
4668 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4669 all enabled at -O1 but not for -Og.
4670 * common.opt (fbranch-count-reg): Remove Init(1).
4671 (fmove-loop-invariants): Likewise.
4672 (ftree-pta): Likewise.
4674 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4676 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4677 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4678 decls to at least the GCC 4.8 used alignments.
4681 * tree-nested.c (convert_nonlocal_reference_stmt,
4682 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4683 of GIMPLE_BIND stmts, adjust associated decls.
4685 2014-01-17 Richard Biener <rguenther@suse.de>
4687 PR tree-optimization/46590
4688 * vec.h (vec<>::bseach): New member function implementing
4689 binary search according to C89 bsearch.
4690 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4691 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4692 bitmap pointer again. Make accesses_in_loop a flat array.
4693 (mem_ref_obstack): New global.
4694 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4695 (mark_ref_stored): Likewise.
4696 (ref_indep_loop_p_2): Likewise.
4697 (set_ref_stored_in_loop): New helper function.
4698 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4699 (memref_free): Adjust.
4700 (record_mem_ref_loc): Simplify.
4701 (gather_mem_refs_stmt): Adjust.
4702 (sort_locs_in_loop_postorder_cmp): New function.
4703 (analyze_memory_references): Sort accesses_in_loop after
4704 loop postorder number.
4705 (find_ref_loc_in_loop_cmp): New function.
4706 (for_all_locs_in_loop): Find relevant cluster of locs in
4707 accesses_in_loop and iterate without recursion.
4708 (execute_sm): Avoid uninit warning.
4709 (struct ref_always_accessed): Simplify.
4710 (ref_always_accessed::operator ()): Likewise.
4711 (ref_always_accessed_p): Likewise.
4712 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4713 loop postorder numbers here.
4714 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4717 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4720 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4721 on decls for which assemble_alias has been called.
4723 2014-01-17 Nick Clifton <nickc@redhat.com>
4725 * config/msp430/msp430.opt: (mcpu): New option.
4726 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4727 (msp430_option_override): Parse target_cpu. If the MCU name
4728 matches a generic string, clear target_mcu.
4729 (msp430_attr): Allow numeric interrupt values up to 63.
4730 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4731 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4733 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4735 * config/msp430/msp430.md (popm): Use %J rather than %I.
4736 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4737 (addhi_cy_i): Use immediate_operand for operand 2.
4738 * doc/invoke.texi: Document -mcpu option.
4740 2014-01-17 Richard Biener <rguenther@suse.de>
4742 PR rtl-optimization/38518
4743 * df.h (df_analyze_loop): Declare.
4744 * df-core.c: Include cfgloop.h.
4745 (df_analyze_1): Split out main part of df_analyze.
4746 (df_analyze): Adjust.
4747 (loop_inverted_post_order_compute): New function.
4748 (loop_post_order_compute): Likewise.
4749 (df_analyze_loop): New function avoiding whole-function
4751 * loop-invariant.c (find_defs): Use df_analyze_loop.
4752 (find_invariants): Adjust.
4753 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4755 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4757 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4758 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4760 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4762 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4763 traversal when removing references.
4765 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4768 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4770 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4773 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4774 pushing a reload for an autoinc when we had previously reloaded an
4775 inner part of the address.
4777 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4779 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4781 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4782 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4783 when not giving up or versioning for alias only because of
4785 (vect_analyze_data_ref_dependences): Set to true.
4786 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4788 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4789 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4793 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4796 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4797 operand 0 predicate for gathers, use a new pseudo as subtarget.
4799 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4802 * lra-constraints.c (process_alt_operands): Add printing debug info.
4803 Check absence of input/output reloads for matched operands too.
4805 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4807 PR rtl-optimization/59835
4808 * ira.c (ira_init_register_move_cost): Increase cost for
4811 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4813 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4815 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4818 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4819 non-register objects. Use gen_(high/low)part more consistently.
4822 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4825 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4826 endian support, remove tests for WORDS_BIG_ENDIAN.
4827 (p8_mfvsrd_3_<mode>): Likewise.
4828 (reload_gpr_from_vsx<mode>): Likewise.
4829 (reload_gpr_from_vsxsf): Likewise.
4830 (p8_mfvsrd_4_disf): Likewise.
4832 2014-01-16 Richard Biener <rguenther@suse.de>
4834 PR rtl-optimization/46590
4835 * lcm.c (compute_antinout_edge): Use postorder iteration.
4836 (compute_laterin): Use inverted postorder iteration.
4838 2014-01-16 Nick Clifton <nickc@redhat.com>
4841 * varasm.c (output_constant): Return the number of bytes actually
4843 (output_constructor_array_range): Update the field size with the
4844 number of bytes emitted by output_constant.
4845 (output_constructor_regular_field): Likewise. Also do not
4846 complain if the total number of bytes emitted is now greater
4847 than the expected fieldpos.
4848 * output.h (output_constant): Update prototype and descriptive comment.
4850 2014-01-16 Marek Polacek <polacek@redhat.com>
4853 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4854 it is error_mark_node.
4856 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4858 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4859 VALID_AVX256_REG_OR_OI_MODE.
4861 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4863 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4864 current procedure should be profiled.
4866 2014-01-15 Andrew Pinski <apinski@cavium.com>
4868 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4869 of moving from/to the STACK_REG register class.
4871 2014-01-15 Richard Henderson <rth@redhat.com>
4874 * reginfo.c (global_regs_decl): Globalize.
4875 * rtl.h (global_regs_decl): Declare.
4876 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4877 reserved via global_regs.
4879 2014-01-15 Teresa Johnson <tejohnson@google.com>
4881 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4883 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4885 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4886 and vmulosh rather than call gen_vec_widen_smult_*.
4887 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4888 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4889 (vec_widen_smult_even_v16qi): Likewise.
4890 (vec_widen_umult_even_v8hi): Likewise.
4891 (vec_widen_smult_even_v8hi): Likewise.
4892 (vec_widen_umult_odd_v16qi): Likewise.
4893 (vec_widen_smult_odd_v16qi): Likewise.
4894 (vec_widen_umult_odd_v8hi): Likewise.
4895 (vec_widen_smult_odd_v8hi): Likewise.
4896 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4897 vmuloub rather than call gen_vec_widen_umult_*.
4898 (vec_widen_umult_lo_v16qi): Likewise.
4899 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4900 vmulosb rather than call gen_vec_widen_smult_*.
4901 (vec_widen_smult_lo_v16qi): Likewise.
4902 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4903 rather than call gen_vec_widen_umult_*.
4904 (vec_widen_umult_lo_v8hi): Likewise.
4905 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4906 rather than call gen_vec_widen_smult_*.
4907 (vec_widen_smult_lo_v8hi): Likewise.
4909 2014-01-15 Jeff Law <law@redhat.com>
4911 PR tree-optimization/59747
4912 * ree.c (find_and_remove_re): Properly handle case where a second
4913 eliminated extension requires widening a copy created for elimination
4914 of a prior extension.
4915 (combine_set_extension): Ensure that the number of hard regs needed
4916 for a destination register does not change when we widen it.
4918 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4920 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4921 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4922 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4923 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4924 (avr-*-rtems*): Likewise.
4925 (bfin*-rtems*): Likewise.
4926 (moxie-*-rtems*): Likewise.
4927 (h8300-*-rtems*): Likewise.
4928 (i[34567]86-*-rtems*): Likewise.
4929 (lm32-*-rtems*): Likewise.
4930 (m32r-*-rtems*): Likewise.
4931 (m68k-*-rtems*): Likewise.
4932 (microblaze*-*-rtems*): Likewise.
4933 (mips*-*-rtems*): Likewise.
4934 (powerpc-*-rtems*): Likewise.
4935 (sh-*-rtems*): Likewise.
4936 (sparc-*-rtems*): Likewise.
4937 (sparc64-*-rtems*): Likewise.
4938 (v850-*-rtems*): Likewise.
4939 (m32c-*-rtems*): Likewise.
4941 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4943 PR rtl-optimization/59511
4944 * ira.c (ira_init_register_move_cost): Use memory costs for some
4945 cases of register move cost calculations.
4946 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4947 instead of BB frequency.
4948 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4949 * lra-assigns.c (find_hard_regno_for): Ditto.
4951 2014-01-15 Richard Biener <rguenther@suse.de>
4953 PR tree-optimization/59822
4954 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4955 (vectorizable_load): Use it to hoist defs of uses of invariant
4956 loads out of the loop.
4958 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4959 Kugan Vivekanandarajah <kuganv@linaro.org>
4962 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4965 2014-01-15 Richard Biener <rguenther@suse.de>
4967 PR rtl-optimization/59802
4968 * lcm.c (compute_available): Use inverted postorder to seed
4969 the initial worklist.
4971 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4974 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4975 ADDR_REGS for invalid symrefs in non-PIC code.
4977 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4980 * builtins.c (determine_block_size): Initialize *probable_max_size
4981 even if len_rtx is CONST_INT.
4983 2014-01-14 Andrew Pinski <apinski@cavium.com>
4985 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4986 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4987 (cortexa53_tunings): Likewise.
4988 (aarch64_sched_issue_rate): New function.
4989 (TARGET_SCHED_ISSUE_RATE): Define.
4991 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4993 * ira-costs.c (find_costs_and_classes): Add missed
4994 ira_init_register_move_cost_if_necessary.
4996 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4999 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5001 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5004 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5005 to indicate if type is used for function return value. Warn ABI
5006 change if the vector mode isn't available for function return value.
5007 (ix86_function_arg_advance): Pass false to type_natural_mode.
5008 (ix86_function_arg): Likewise.
5009 (ix86_gimplify_va_arg): Likewise.
5010 (function_arg_32): Don't warn ABI change.
5011 (ix86_function_value): Pass true to type_natural_mode.
5012 (ix86_return_in_memory): Likewise.
5013 (ix86_struct_value_rtx): Removed.
5014 (TARGET_STRUCT_VALUE_RTX): Likewise.
5016 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5018 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5019 converting a conditional jump into a conditional return.
5021 2014-01-14 Richard Biener <rguenther@suse.de>
5023 PR tree-optimization/58921
5024 PR tree-optimization/59006
5025 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5026 hoisting invariant stmts.
5027 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5028 invariant loads on the preheader edge if possible.
5030 2014-01-14 Joey Ye <joey.ye@arm.com>
5032 * doc/plugin.texi (Building GCC plugins): Update to C++.
5034 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5036 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5037 (_mm_rcp28_round_ss): Ditto.
5038 (_mm_rsqrt28_round_sd): Ditto.
5039 (_mm_rsqrt28_round_ss): Ditto.
5040 (_mm_rcp28_sd): Ditto.
5041 (_mm_rcp28_ss): Ditto.
5042 (_mm_rsqrt28_sd): Ditto.
5043 (_mm_rsqrt28_ss): Ditto.
5044 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5045 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5046 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5047 (IX86_BUILTIN_RCP28SD): Ditto.
5048 (IX86_BUILTIN_RCP28SS): Ditto.
5049 (IX86_BUILTIN_RSQRT28SD): Ditto.
5050 (IX86_BUILTIN_RSQRT28SS): Ditto.
5051 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5052 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5053 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5054 (ix86_expand_special_args_builtin): Expand new FTYPE.
5055 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5056 (srcp14<mode>): Make insn unary.
5057 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5058 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5059 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5060 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5061 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5062 Fix rounding: make it SAE only.
5063 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5065 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5067 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5068 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5069 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5070 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5071 (round_saeonly_mask_scalar_operand4): Ditto.
5072 (round_saeonly_mask_scalar_op3): Ditto.
5073 (round_saeonly_mask_scalar_op4): Ditto.
5075 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5077 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5078 Implement -maltivec=be for vec_insert and vec_extract.
5080 2014-01-10 DJ Delorie <dj@redhat.com>
5082 * config/msp430/msp430.md (call_internal): Don't allow memory
5083 references with SP as the base register.
5084 (call_value_internal): Likewise.
5085 * config/msp430/constraints.md (Yc): New. For memory references
5086 that don't use SP as a base register.
5088 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5089 "an integer without a # prefix"
5090 * config/msp430/msp430.md (epilogue_helper): Use it.
5092 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5095 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5096 AVX512F gather builtins.
5097 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5098 on gather decls with INTEGER_TYPE masktype.
5099 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5100 directly into the builtin rather than hoisting it before loop.
5102 PR tree-optimization/59387
5103 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5104 (scev_const_prop): If folded_casts and type has undefined overflow,
5105 use force_gimple_operand instead of force_gimple_operand_gsi and
5106 for each added stmt if it is assign with
5107 arith_code_with_undefined_signed_overflow, call
5108 rewrite_to_defined_overflow.
5109 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5110 gimple-fold.h instead.
5111 (arith_code_with_undefined_signed_overflow,
5112 rewrite_to_defined_overflow): Moved to ...
5113 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5114 rewrite_to_defined_overflow): ... here. No longer static.
5115 Include gimplify-me.h.
5116 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5117 rewrite_to_defined_overflow): New prototypes.
5119 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5121 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5123 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5125 * builtins.c (get_object_alignment_2): Minor tweak.
5126 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5128 2014-01-13 Christian Bruel <christian.bruel@st.com>
5130 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5131 optimized non constant lengths.
5133 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5136 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5137 load as __atomic_load_N if possible.
5139 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5141 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5143 (rs6000_expand_builtin): Adjust call.
5145 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5148 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5149 * config/rs6000/rs6000.c: Include target-globals.h.
5150 (rs6000_set_current_function): Instead of doing target_reinit
5151 unconditionally, use save_target_globals_default_opts and
5152 restore_target_globals.
5154 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5156 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5157 (rs6000_expand_builtin): Handle mffs and mtfsf.
5158 (rs6000_init_builtins): Define mffs and mtfsf.
5159 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5160 (rs6000_mffs): New pattern.
5161 (rs6000_mtfsf): New pattern.
5163 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5165 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5166 Start narrowing with START. Apply candidate-use pair
5167 and check overall cost in narrowing.
5168 (iv_ca_prune): Pass new argument.
5170 2014-01-10 Jeff Law <law@redhat.com>
5173 * ree.c (combine_reaching_defs): Ensure the defining statement
5174 occurs before the extension when optimizing extensions with
5175 different source and destination hard registers.
5177 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5180 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5181 vtables into the type inheritance graph.
5183 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5185 PR rtl-optimization/59754
5186 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5187 modes in the REGNO != REGNO case.
5189 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5191 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5193 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5195 PR tree-optimization/59745
5196 * tree-predcom.c (tree_predictive_commoning_loop): Call
5197 free_affine_expand_cache if giving up because components is NULL.
5199 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5200 GC in payload of target_globals struct instead of allocating them on
5201 the heap and the larger structs separately using GC.
5202 * target-globals.h (struct target_globals): Make regs, hard_regs,
5203 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5204 of GTY((skip)) and change type to void *.
5205 (reset_target_globals): Cast loads from those fields to corresponding
5208 2014-01-10 Steve Ellcey <sellcey@mips.com>
5211 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5212 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5213 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5215 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5218 * aarch64-modes.def (CC_Zmode): New flags mode.
5219 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5220 represents an equality.
5221 (aarch64_get_condition_code): Handle CC_Zmode.
5222 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5224 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5226 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5227 extraction in good case.
5229 2014-01-10 Richard Biener <rguenther@suse.de>
5231 PR tree-optimization/59374
5232 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5233 checking after SLP discovery. Mark stmts not participating
5234 in any SLP instance properly.
5236 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5238 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5239 when handling a SET rtx.
5241 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5243 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5244 (cortex-a57): Likewise.
5245 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5247 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5249 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5250 non-iwmmxt builtins.
5252 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5256 * ipa-devirt.c record_target_from_binfo): Take as argument
5257 stack of binfos and lookup matching one for virtual inheritance.
5258 (possible_polymorphic_call_targets_1): Update.
5260 2014-01-10 Huacai Chen <chenhc@lemote.com>
5262 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5263 kernel strings for Loongson-2E/2F/3A.
5265 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5268 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5269 is_gimple_call before calling gimple_call_internal_p.
5271 2014-01-09 Steve Ellcey <sellcey@mips.com>
5273 * Makefile.in (TREE_FLOW_H): Remove.
5274 (TREE_SSA_H): Add file names from tree-flow.h.
5275 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5276 * tree.h: Remove tree-flow.h reference.
5277 * hash-table.h: Remove tree-flow.h reference.
5278 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5279 reference with tree-ssa-loop.h.
5281 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5283 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5284 default element-order behavior for -maltivec.
5285 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5286 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5287 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5288 when targeting big endian, at least for now.
5289 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5291 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5294 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5295 var satisfies use_register_for_decl, just take into account type
5296 alignment, rather than decl alignment.
5298 PR tree-optimization/59622
5299 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5300 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5301 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5302 Don't devirtualize for inplace at all. For targets.length () == 1,
5303 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5305 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5307 * config/i386/i386.md (cpu): Remove the unused btver1.
5309 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5311 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5313 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5316 * tree-core.h (struct target_globals): New forward declaration.
5317 (struct tree_target_option): Add globals field.
5318 * tree.h (TREE_TARGET_GLOBALS): Define.
5319 (prepare_target_option_nodes_for_pch): New prototype.
5320 * target-globals.h (struct target_globals): Define even if
5322 * tree.c (prepare_target_option_node_for_pch,
5323 prepare_target_option_nodes_for_pch): New functions.
5324 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5325 * config/i386/i386.c: Include target-globals.h.
5326 (ix86_set_current_function): Instead of doing target_reinit
5327 unconditionally, use save_target_globals_default_opts and
5328 restore_target_globals.
5330 2014-01-09 Richard Biener <rguenther@suse.de>
5332 PR tree-optimization/59715
5333 * tree-cfg.h (split_critical_edges): Declare.
5334 * tree-cfg.c (split_critical_edges): Export.
5335 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5337 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5339 * cfgexpand.c (expand_stack_vars): Optionally disable
5340 asan stack protection.
5341 (expand_used_vars): Likewise.
5342 (partition_stack_vars): Likewise.
5343 * asan.c (asan_emit_stack_protection): Optionally disable
5344 after return stack usage.
5345 (instrument_derefs): Optionally disable memory access instrumentation.
5346 (instrument_builtin_call): Likewise.
5347 (instrument_strlen_call): Likewise.
5348 (asan_protect_global): Optionally disable global variables protection.
5349 * doc/invoke.texi: Added doc for new options.
5350 * params.def: Added new options.
5351 * params.h: Likewise.
5353 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5355 PR rtl-optimization/59724
5356 * ifcvt.c (cond_exec_process_if_block): Don't call
5357 flow_find_head_matching_sequence with 0 longest_match.
5358 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5359 non-active insns if !stop_after.
5360 (try_head_merge_bb): Revert 2014-01-07 changes.
5362 2014-01-08 Jeff Law <law@redhat.com>
5364 * ree.c (get_sub_rtx): New function, extracted from...
5365 (merge_def_and_ext): Here.
5366 (combine_reaching_defs): Use get_sub_rtx.
5368 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5370 * cgraph.h (varpool_variable_node): Do not choke on null node.
5372 2014-01-08 Catherine Moore <clm@codesourcery.com>
5374 * config/mips/mips.md (simple_return): Attempt to use JRC
5376 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5378 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5380 PR rtl-optimization/59137
5381 * reorg.c (steal_delay_list_from_target): Call update_block for
5383 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5385 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5387 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5388 two duplicate entries.
5390 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5393 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5395 * config/mips/mips.c (mips_truncated_op_cost): New function.
5396 (mips_rtx_costs): Adjust test for BADDU.
5397 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5399 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5401 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5402 (*baddu_si): ...this new pattern.
5404 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5407 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5409 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5412 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5414 (expand_expr, expand_normal): Adjust.
5415 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5416 inner_reference_p. Use inner_reference_p to expand inner references.
5417 (store_expr): Adjust.
5418 * cfgexpand.c (expand_call_stmt): Adjust.
5420 2014-01-08 Rong Xu <xur@google.com>
5422 * gcov-io.c (gcov_var): Move from gcov-io.h.
5423 (gcov_position): Ditto.
5424 (gcov_is_error): Ditto.
5425 (gcov_rewrite): Ditto.
5426 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5427 only part to libgcc/libgcov.h.
5429 2014-01-08 Marek Polacek <polacek@redhat.com>
5432 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5434 2014-01-08 Marek Polacek <polacek@redhat.com>
5437 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5439 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5441 PR rtl-optimization/59649
5442 * stor-layout.c (get_mode_bounds): For BImode return
5443 0 and STORE_FLAG_VALUE.
5445 2014-01-08 Richard Biener <rguenther@suse.de>
5448 * gimple.h (is_gimple_builtin_call): Remove.
5449 (gimple_builtin_call_types_compatible_p): New.
5450 (gimple_call_builtin_p): New overload.
5451 * gimple.c (is_gimple_builtin_call): Remove.
5452 (validate_call): Rename to ...
5453 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5455 (validate_type): New static function.
5456 (gimple_call_builtin_p): New overload and adjust.
5457 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5458 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5459 (gimple_fold_stmt_to_constant_1): Likewise.
5460 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5462 2014-01-08 Richard Biener <rguenther@suse.de>
5465 * gimplify.c (gimplify_expr): Gimplify register-register type
5466 VIEW_CONVERT_EXPRs to separate stmts.
5468 2014-01-07 Jeff Law <law@redhat.com>
5471 * ree.c (combine_set_extension): Handle case where source
5472 and destination registers in an extension insn are different.
5473 (combine_reaching_defs): Allow source and destination registers
5474 in extension to be different under limited circumstances.
5475 (add_removable_extension): Remove restriction that the
5476 source and destination registers in the extension are the same.
5477 (find_and_remove_re): Emit a copy from the extension's
5478 destination to its source after the defining insn if
5479 the source and destination registers are different.
5482 * ifcvt.c (merge_if_block): If we are merging a block with more than
5483 one successor with a block with no successors, remove any BARRIER
5484 after the second block.
5486 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5488 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5490 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5493 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5494 for 14-bit register offsets when INT14_OK_STRICT is false.
5496 2014-01-07 Roland Stigge <stigge@antcom.de>
5497 Michael Meissner <meissner@linux.vnet.ibm.com>
5500 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5501 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5503 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5505 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5508 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5510 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5511 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5512 rtx is const0_rtx or not.
5514 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5517 * target-globals.c (save_target_globals): Remove this_fn_optab
5519 * toplev.c: Include optabs.h.
5520 (target_reinit): Temporarily restore the global options if another
5521 set of options are in force.
5523 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5525 PR rtl-optimization/58668
5526 * cfgcleanup.c (flow_find_cross_jump): Don't count
5527 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5528 to determine what is counted.
5529 (flow_find_head_matching_sequence): Use active_insn_p to determine
5531 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5533 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5534 determine what is counted.
5536 PR tree-optimization/59643
5537 * tree-predcom.c (split_data_refs_to_components): If one dr is
5538 read and one write, determine_offset fails and the write isn't
5539 in the bad component, just put the read into the bad component.
5541 2014-01-07 Mike Stump <mikestump@comcast.net>
5542 Jakub Jelinek <jakub@redhat.com>
5545 * tree-core.h (struct tree_optimization_option): Change optabs
5546 type from unsigned char * to void *.
5547 * optabs.c (init_tree_optimization_optabs): Adjust
5548 TREE_OPTIMIZATION_OPTABS initialization.
5550 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5553 * config/i386/i386.h (struct machine_function): Add
5554 no_drap_save_restore field.
5555 * config/i386/i386.c (ix86_save_reg): Use
5556 !cfun->machine->no_drap_save_restore instead of
5557 crtl->stack_realign_needed.
5558 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5559 this function clears frame_pointer_needed. Set
5560 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5561 and DRAP reg is needed.
5563 2014-01-06 Marek Polacek <polacek@redhat.com>
5566 * doc/implement-c.texi: Mention that other integer types are
5567 permitted as bit-field types in strictly conforming mode.
5569 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5571 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5574 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5576 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5578 2014-01-06 Martin Jambor <mjambor@suse.cz>
5581 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5583 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5585 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5589 * var-tracking.c (add_stores): Preserve the value of the source even if
5590 we don't record the store.
5592 2014-01-06 Terry Guo <terry.guo@arm.com>
5594 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5596 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5599 * config/darwin.c (darwin_function_section): Adjust return values to
5600 correspond to optimisation changes made in r206070.
5602 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5604 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5605 from prefetch_block tune setting.
5606 (nocona_cost): Correct size of prefetch block to 64.
5608 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5610 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5611 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5612 used to save the static chain register in the computation of the offset
5613 from which the FP registers need to be restored.
5615 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5617 PR tree-optimization/59519
5618 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5619 ICE if get_current_def (current_new_name) is already non-NULL, as long
5620 as it is a phi result of some other phi in *new_exit_bb that has
5623 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5624 or vmovdqu* for misaligned_operand.
5625 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5626 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5627 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5628 aligned_mem for AVX512F masked aligned load and store builtins and for
5631 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5633 PR tree-optimization/59651
5634 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5635 Address range for negative step should be added by TYPE_SIZE_UNIT.
5637 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5639 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5640 overlapping registers also for registers other than the stack pointer.
5642 2014-01-03 Marek Polacek <polacek@redhat.com>
5645 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5648 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5651 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5654 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5655 (push splitter): Use <P:MODE_SIZE> instead of
5656 GET_MODE_SIZE (<P:MODE>mode).
5657 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5658 (mov -1, reg peephole2): Likewise.
5659 * config/i386/sse.md (*mov<mode>_internal,
5660 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5661 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5662 *<code><mode>3, *andnot<mode>3<mask_name>,
5663 <mask_codefor><code><mode>3<mask_name>): Likewise.
5664 * config/i386/subst.md (mask_mode512bit_condition,
5665 sd_mask_mode512bit_condition): Likewise.
5667 2014-01-02 Xinliang David Li <davidxl@google.com>
5669 PR tree-optimization/59303
5670 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5671 (dump_predicates): Better output format.
5672 (pred_equal_p): New function.
5673 (is_neq_relop_p): Ditto.
5674 (is_neq_zero_form_p): Ditto.
5675 (pred_expr_equal_p): Ditto.
5676 (pred_neg_p): Ditto.
5677 (simplify_pred): Ditto.
5678 (simplify_preds_2): Ditto.
5679 (simplify_preds_3): Ditto.
5680 (simplify_preds_4): Ditto.
5681 (simplify_preds): Ditto.
5683 (push_to_worklist): Ditto.
5684 (get_pred_info_from_cmp): Ditto.
5685 (is_degenerated_phi): Ditto.
5686 (normalize_one_pred_1): Ditto.
5687 (normalize_one_pred): Ditto.
5688 (normalize_one_pred_chain): Ditto.
5689 (normalize_preds): Ditto.
5690 (normalize_cond_1): Remove function.
5691 (normalize_cond): Ditto.
5692 (is_gcond_subset_of): Ditto.
5693 (is_subset_of_any): Ditto.
5694 (is_or_set_subset_of): Ditto.
5695 (is_and_set_subset_of): Ditto.
5696 (is_norm_cond_subset_of): Ditto.
5697 (pred_chain_length_cmp): Ditto.
5698 (convert_control_dep_chain_into_preds): Type change.
5699 (find_predicates): Ditto.
5700 (find_def_preds): Ditto.
5701 (destroy_predicates_vecs): Ditto.
5702 (find_matching_predicates_in_rest_chains): Ditto.
5703 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5704 (is_pred_expr_subset): Ditto.
5705 (is_pred_chain_subset_of): Ditto.
5706 (is_included_in): Ditto.
5707 (is_superset_of): Ditto.
5709 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5711 Update copyright years.
5713 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5715 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5716 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5717 config/arc/arc.md, config/arc/arc.opt,
5718 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5719 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5720 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5721 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5722 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5723 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5724 vtable-verify.c, vtable-verify.h: Use the standard form for the
5727 2014-01-02 Tobias Burnus <burnus@net-b.de>
5729 * gcc.c (process_command): Update copyright notice dates.
5730 * gcov-dump.c: Ditto.
5732 * doc/cpp.texi: Bump @copying's copyright year.
5733 * doc/cppinternals.texi: Ditto.
5734 * doc/gcc.texi: Ditto.
5735 * doc/gccint.texi: Ditto.
5736 * doc/gcov.texi: Ditto.
5737 * doc/install.texi: Ditto.
5738 * doc/invoke.texi: Ditto.
5740 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5742 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5744 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5746 * config/i386/sse.md (*mov<mode>_internal): Guard
5747 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5749 PR rtl-optimization/59647
5750 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5751 new_rtx into UNSIGNED_FLOAT rtxes.
5753 Copyright (C) 2014 Free Software Foundation, Inc.
5755 Copying and distribution of this file, with or without modification,
5756 are permitted in any medium without royalty provided the copyright
5757 notice and this notice are preserved.