| blob | ded35eb7e72ccdc14ea591c7e859ece18671e61a |
1 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
3 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
4 static chain if needed.
6 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
8 PR target/60697
9 * lra-constraints.c (index_part_to_reg): New.
10 (process_address): Use it.
12 2014-03-27 Jeff Law <law@redhat.com>
13 Jakub Jelinek <jakub@redhat.com>
15 PR target/60648
16 * expr.c (do_tablejump): Use simplify_gen_binary rather than
17 gen_rtx_{PLUS,MULT} to build up the address expression.
19 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
20 creating non-canonical RTL.
22 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
24 PR ipa/60243
25 * ipa-inline.c (want_inline_small_function_p): Short circuit large
26 functions; reorganize to make cheap checks first.
27 (inline_small_functions): Do not estimate growth when dumping;
28 it is expensive.
29 * ipa-inline.h (inline_summary): Add min_size.
30 (growth_likely_positive): New function.
31 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
32 (set_cond_stmt_execution_predicate): Cleanup.
33 (estimate_edge_size_and_time): Compute min_size.
34 (estimate_calls_size_and_time): Likewise.
35 (estimate_node_size_and_time): Likewise.
36 (inline_update_overall_summary): Update min_size.
37 (do_estimate_edge_time): Likewise.
38 (do_estimate_edge_size): Update.
39 (do_estimate_edge_hints): Update.
40 (growth_likely_positive): New function.
42 2014-03-28 Jakub Jelinek <jakub@redhat.com>
44 PR target/60693
45 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
46 also if addr has VOIDmode.
48 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
50 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
51 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
52 Declare extern.
53 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
54 instructions as well as AdvancedSIMD loads.
56 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
58 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
59 Use crypto_aese type.
60 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
61 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
62 crypto_aese, crypto_aesmc. Move to types.md.
63 * config/arm/types.md (crypto_aes): Split into crypto_aese,
64 crypto_aesmc.
65 * config/arm/iterators.md (crypto_type): Likewise.
67 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
69 * cgraph.c: Include expr.h and tree-dfa.h.
70 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn; remove LHS.
72 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
74 PR target/60675
75 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
76 regs from checking multi-reg pseudos.
78 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
80 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
82 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
84 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
85 if it would clobber the stack pointer, even temporarily.
87 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
89 * mode-switching.c: Make small adjustments to the top comment.
91 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
93 * config/rs6000/constraints.md (wD constraint): New constraint to
94 match the constant integer to get the top DImode/DFmode out of a
95 vector in a VSX register.
97 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
98 match the constant integer to get the top DImode/DFmode out of a
99 vector in a VSX register.
101 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
102 for ISA 2.07.
104 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
105 vbpermq builtins.
107 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
108 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
110 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
111 Optimize vec_extract of 64-bit values, where the value being
112 extracted is in the top word, where we can use scalar
113 instructions. Add direct move and store support. Combine the big
114 endian/little endian vector select load support into a single
115 insn.
116 (vsx_extract_<mode>_internal1): Likewise.
117 (vsx_extract_<mode>_internal2): Likewise.
118 (vsx_extract_<mode>_load): Likewise.
119 (vsx_extract_<mode>_store): Likewise.
120 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
121 combined into vsx_extract_<mode>_load.
122 (vsx_extract_<mode>_one_le): Likewise.
124 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
125 define the top 64-bit vector element.
127 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
128 constraint.
130 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
131 Document vec_vbpermq builtin.
133 PR target/60672
134 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
135 enable use of xxsldwi and xxpermdi builtin functions.
136 (vec_xxpermdi): Likewise.
138 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
139 Document use of vec_xxsldwi and vec_xxpermdi builtins.
141 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
143 PR rtl-optimization/60650
144 * lra-asign.c (find_hard_regno_for, spill_for): Add parameter
145 first_p. Use it.
146 (find_spills_for): New.
147 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
148 Spill all pseudos on the second iteration.
150 2014-03-27 Marek Polacek <polacek@redhat.com>
152 PR c/50347
153 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
154 types.
156 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
158 * config/s390/s390.c (s390_can_use_return_insn): Check for
159 call-saved FPRs on 31 bit.
161 2014-03-27 Jakub Jelinek <jakub@redhat.com>
163 PR middle-end/60682
164 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
165 if they need regimplification, just drop them instead of
166 calling gimple_regimplify_operands on them.
168 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
170 PR target/60580
171 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
172 (aarch64_frame_pointer_required): Adjust logic.
173 (aarch64_can_eliminate): Adjust logic.
174 (aarch64_override_options_after_change): Adjust logic.
176 2014-03-27 Dehao Chen <dehao@google.com>
178 * ipa-inline.c (early_inliner): Update node's inline info.
180 2014-03-26 Dehao Chen <dehao@google.com>
182 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
183 compiler inserted conditional jumps for NAN float check.
185 2014-03-26 Jakub Jelinek <jakub@redhat.com>
187 * ubsan.h (ubsan_create_data): Change second argument's type
188 to const location_t *.
189 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
190 _("<unknown>").
191 (ubsan_create_data): Change second argument to const location_t *PLOC.
192 Create Loc field whenever PLOC is non-NULL.
193 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
194 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
195 callers.
197 PR other/59545
198 * real.c (real_to_integer2): Change type of low to UHWI.
200 2014-03-26 Tobias Burnus <burnus@net-b.de>
202 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
203 (CILK_SELF_SPECS): New define.
204 (driver_self_specs): Use it.
206 2014-03-26 Richard Biener <rguenther@suse.de>
208 * tree-pretty-print.c (percent_K_format): Implement special
209 case for LTO and its stripped down BLOCK tree.
211 2014-03-26 Jakub Jelinek <jakub@redhat.com>
213 PR sanitizer/60636
214 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
216 * tree-vrp.c (simplify_internal_call_using_ranges): If only
217 one range is range_int_cst_p, but not both, at least optimize
218 addition/subtraction of 0 and multiplication by 0 or 1.
219 * gimple-fold.c (gimple_fold_call): Fold
220 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
221 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
222 INTEGER_CSTs, try to fold at least x * 0 and y - y.
224 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
226 PR rtl-optimization/60452
227 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
228 <case REG>: Return 1 for invalid offsets from the frame pointer.
230 2014-03-26 Marek Polacek <polacek@redhat.com>
232 PR c/37428
233 * doc/extend.texi (C Extensions): Mention variable-length arrays in
234 a structure/union.
236 2014-03-26 Marek Polacek <polacek@redhat.com>
238 PR c/39525
239 * doc/extend.texi (Designated Inits): Describe what happens to omitted
240 field members.
242 2014-03-26 Marek Polacek <polacek@redhat.com>
244 PR other/59545
245 * ira-color.c (update_conflict_hard_regno_costs): Perform the
246 multiplication in unsigned type.
248 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
250 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
252 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
254 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
256 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
258 PR ipa/60315
259 * cif-code.def (UNREACHABLE) New code.
260 * ipa-inline.c (inline_small_functions): Skip edges to __builtlin_unreachable.
261 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
262 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
263 predicate to __bulitin_unreachable.
264 (set_cond_stmt_execution_predicate): Fix issue when invert_tree_comparison
265 returns ERROR_MARK.
266 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
267 propagate to inline clones.
268 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
269 to unreachable.
270 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
271 * cgraphclones.c (cgraph_clone_node): If call destination is already
272 ureachable, do not redirect it back.
273 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
274 unreachable.
276 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
278 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
279 Do not modify inline clones.
281 2014-03-25 Jakub Jelinek <jakub@redhat.com>
283 * config/i386/i386.md (general_sext_operand): New mode attr.
284 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
285 don't generate (sign_extend (const_int)).
286 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
287 operands[2]. Use We constraint instead of <i> and <general_sext_operand>
288 predicate instead of <general_operand>.
289 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
290 * config/i386/constraints.md (We): New constraint.
291 * config/i386/predicates.md (x86_64_sext_operand,
292 sext_operand): New predicates.
294 2014-03-25 Martin Jambor <mjambor@suse.cz>
296 PR ipa/60600
297 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
298 inconsistent devirtualizations to __builtin_unreachable.
300 2014-03-25 Marek Polacek <polacek@redhat.com>
302 PR c/35449
303 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
305 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
307 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
308 order of elements for big-endian.
310 2014-03-25 Richard Biener <rguenther@suse.de>
312 PR middle-end/60635
313 * gimplify-me.c (gimple_regimplify_operands): Update the
314 re-gimplifed stmt.
316 2014-03-25 Martin Jambor <mjambor@suse.cz>
318 PR ipa/59176
319 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
320 (lto_output_varpool_node): Likewise.
321 (input_overwrite_node): Likewise.
322 (input_varpool_node): Likewise.
324 2014-03-25 Richard Biener <rguenther@suse.de>
326 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like
327 OPT_fpie.
328 (run_gcc): Likewise.
330 2014-03-25 Jakub Jelinek <jakub@redhat.com>
332 * combine.c (simplify_compare_const): Add MODE argument.
333 Handle mode_width 0 as very large mode_width.
334 (try_combine, simplify_comparison): Adjust callers.
336 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
337 type to avoid signed integer overflow.
338 * explow.c (plus_constant): Likewise.
340 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
342 * doc/generic.texi: Correct typos.
344 2014-03-24 Tobias Burnus <burnus@net-b.de>
346 * doc/invoke.texi (-flto): Expand section about
347 using static libraries with LTO.
349 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
351 PR rtl-optimization/60501
352 * optabs.def (addptr3_optab): New optab.
353 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
354 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
355 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
357 * lra.c (emit_add3_insn): Use the addptr pattern if available.
359 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
361 2014-03-24 Ulrich Drepper <drepper@gmail.com>
363 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
364 _mm512_set1_pd.
366 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
367 (_mm256_undefined_ps): Define.
368 (_mm256_undefined_pd): Define.
369 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
370 (_mm_undefined_pd): Define.
371 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
372 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
373 (_mm512_undefined_ps): Define.
374 (_mm512_undefined_pd): Define.
375 Use _mm*_undefined_*.
376 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
378 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
380 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
381 (lshr_simd): DI mode added.
382 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
383 (aarch64_ushr_simddi): Likewise.
384 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
385 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
386 (vshrd_n_u64): Likewise.
388 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
390 * Makefile.in (s-macro_list): Depend on cc1.
392 2014-03-23 Teresa Johnson <tejohnson@google.com>
394 * ipa-utils.c (ipa_print_order): Use specified dump file.
396 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
398 PR rtl-optimization/60601
399 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
401 * gcc.c (eval_spec_function): Initialize save_growing_value.
403 2014-03-22 Jakub Jelinek <jakub@redhat.com>
405 PR sanitizer/60613
406 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
407 code == MINUS_EXPR, never swap op0 with op1.
409 * toplev.c (init_local_tick): Avoid signed integer multiplication
410 overflow.
411 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
412 shift by first operand's bitsize.
414 2014-03-21 Jakub Jelinek <jakub@redhat.com>
416 PR target/60610
417 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
418 redefine to 1 or 0.
419 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
420 TARGET_ISA_64BIT_P(x).
422 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
424 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
425 pattern for vector nor instead of subtract from splat(-1).
426 (altivec_expand_vec_perm_const_le): Likewise.
428 2014-03-21 Richard Henderson <rth@twiddle.net>
430 PR target/60598
431 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
432 related insns after epilogue_completed.
434 2014-03-21 Martin Jambor <mjambor@suse.cz>
436 PR ipa/59176
437 * cgraph.h (symtab_node): New flag body_removed.
438 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
439 when removing bodies.
440 * symtab.c (dump_symtab_base): Dump body_removed flag.
441 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
442 had their bodies removed.
444 2014-03-21 Martin Jambor <mjambor@suse.cz>
446 PR ipa/60419
447 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
448 in the border.
450 2014-03-21 Richard Biener <rguenther@suse.de>
452 PR tree-optimization/60577
453 * tree-core.h (struct tree_base): Document nothrow_flag use
454 in VAR_DECL_NONALIASED.
455 * tree.h (VAR_DECL_NONALIASED): New.
456 (may_be_aliased): Adjust.
457 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
459 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
461 * expr.c (expand_expr_real_1): Remove outdated comment.
463 2014-03-20 Jakub Jelinek <jakub@redhat.com>
465 PR middle-end/60597
466 * ira.c (adjust_cleared_regs): Call copy_rtx on
467 *reg_equiv[REGNO (loc)].src_p before passing it to
468 simplify_replace_fn_rtx.
470 PR target/60568
471 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
472 into CONST, put pic register as first operand of PLUS. Use
473 gen_const_mem for both 32-bit and 64-bit PIC got loads.
475 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
477 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
479 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
481 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
482 around for store forwarding issue in the FPU on the UT699.
483 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
484 loads and operations if -mfix-ut699 is specified.
485 (divtf3_hq): Tweak attribute.
486 (sqrttf2_hq): Likewise.
488 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
490 * calls.c (store_one_arg): Remove incorrect const qualification on the
491 type of the temporary.
492 * cfgexpand.c (expand_return): Likewise.
493 * expr.c (expand_constructor): Likewise.
494 (expand_expr_real_1): Likewise.
496 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
498 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
499 of parts.
501 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
503 PR target/60039
504 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
506 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
508 * config/arm/aarch-common-protos.h
509 (alu_cost_table): Fix spelling of "extend".
510 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
512 2014-03-19 Richard Biener <rguenther@suse.de>
514 PR middle-end/60553
515 * tree-core.h (tree_type_common): Re-order pointer members
516 to reduce recursion depth during GC walks.
518 2014-03-19 Marek Polacek <polacek@redhat.com>
520 PR sanitizer/60569
521 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
522 before accessing it.
524 2014-03-19 Richard Biener <rguenther@suse.de>
526 PR lto/59543
527 * lto-streamer-in.c (input_function): In WPA stage do not drop
528 debug stmts.
530 2014-03-19 Jakub Jelinek <jakub@redhat.com>
532 PR tree-optimization/60559
533 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
534 with build_zero_cst assignment.
536 2014-03-18 Kai Tietz <ktietz@redhat.com>
538 PR rtl-optimization/56356
539 * sdbout.c (sdbout_parms): Verify that parms'
540 incoming argument is valid.
541 (sdbout_reg_parms): Likewise.
543 2014-03-18 Richard Henderson <rth@redhat.com>
545 PR target/60562
546 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
547 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
548 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
550 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
552 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
553 * doc/plugins.texi (Plugin callbacks): Mention
554 PLUGIN_INCLUDE_FILE.
555 Italicize plugin event names in description. Explain that
556 PLUGIN_PRAGMAS has no sense for lto1. Explain
557 PLUGIN_INCLUDE_FILE.
558 Remind that no GCC functions should be called after
559 PLUGIN_FINISH.
560 Explain what pragmas with expansion are.
562 2014-03-18 Martin Liska <mliska@suse.cz>
564 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
565 gimple call statement is update.
566 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
567 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
569 2014-03-18 Jakub Jelinek <jakub@redhat.com>
571 PR sanitizer/60557
572 * ubsan.c (ubsan_instrument_unreachable): Call
573 initialize_sanitizer_builtins.
574 (ubsan_pass): Likewise.
576 PR sanitizer/60535
577 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
578 varpool_finalize_decl instead of rest_of_decl_compilation.
580 2014-03-18 Richard Biener <rguenther@suse.de>
582 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
583 by using bitmap_and_compl instead of bitmap_and_compl_into.
584 (df_rd_transfer_function): Likewise.
586 2014-03-18 Richard Biener <rguenther@suse.de>
588 * doc/lto.texi (fresolution): Fix typo.
590 2014-03-18 Richard Biener <rguenther@suse.de>
592 * doc/invoke.texi (flto): Update for changes in 4.9.
594 2014-03-18 Richard Biener <rguenther@suse.de>
596 * doc/loop.texi: Remove section on the removed lambda framework.
597 Update loop docs with recent changes in preserving loop structure.
599 2014-03-18 Richard Biener <rguenther@suse.de>
601 * doc/lto.texi (-fresolution): Document.
603 2014-03-18 Richard Biener <rguenther@suse.de>
605 * doc/contrib.texi: Adjust my name.
607 2014-03-18 Jakub Jelinek <jakub@redhat.com>
609 PR ipa/58721
610 * internal-fn.c: Include diagnostic-core.h.
611 (expand_BUILTIN_EXPECT): New function.
612 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
613 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
614 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
615 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
616 IFN_BUILTIN_EXPECT.
617 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
618 Revert 3 argument __builtin_expect code.
619 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
620 * gimple-fold.c (gimple_fold_call): Likewise.
621 * tree.h (fold_builtin_expect): New prototype.
622 * builtins.c (build_builtin_expect_predicate): Add predictor
623 argument, if non-NULL, create 3 argument __builtin_expect.
624 (fold_builtin_expect): No longer static. Add ARG2 argument,
625 pass it through to build_builtin_expect_predicate.
626 (fold_builtin_2): Adjust caller.
627 (fold_builtin_3): Handle BUILT_IN_EXPECT.
628 * internal-fn.def (BUILTIN_EXPECT): New.
630 2014-03-18 Tobias Burnus <burnus@net-b.de>
632 PR ipa/58721
633 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
634 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
635 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
637 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
639 PR ipa/58721
640 * predict.c (combine_predictions_for_bb): Fix up formatting.
641 (expr_expected_value_1, expr_expected_value): Add predictor argument,
642 fill what it points to if non-NULL.
643 (tree_predict_by_opcode): Adjust caller, use the predictor.
644 * predict.def (PRED_COMPARE_AND_SWAP): Add.
646 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
648 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
649 proper constant for the store mode.
651 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
653 * symtab.c (change_decl_assembler_name): Fix transparent alias
654 chain construction.
656 2014-03-16 Renlin Li <Renlin.Li@arm.com>
658 * config/aarch64/aarch64.c: Correct the comments about the
659 aarch64 stack layout.
661 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
663 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
664 check for GF_OMP_FOR_KIND_FOR.
666 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
668 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
669 ymm and zmm register names.
671 2014-03-17 Jakub Jelinek <jakub@redhat.com>
673 PR target/60516
674 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
675 note creation for the 2010-08-31 changes.
677 2014-03-17 Marek Polacek <polacek@redhat.com>
679 PR middle-end/60534
680 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
681 as -fno-tree-loop-vectorize.
682 (expand_omp_simd): Likewise.
684 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
686 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
687 (eligible_for_call_delay): New prototype.
688 * config/sparc/sparc.c (tls_call_delay): Rename into...
689 (eligible_for_call_delay): ...this. Return false if the instruction
690 cannot be put in the delay slot of a branch.
691 (eligible_for_restore_insn): Simplify.
692 (eligible_for_return_delay): Return false if the instruction cannot be
693 put in the delay slot of a branch and simplify.
694 (eligible_for_sibcall_delay): Return false if the instruction cannot be
695 put in the delay slot of a branch.
696 * config/sparc/sparc.md (fix_ut699): New attribute.
697 (tls_call_delay): Delete.
698 (in_call_delay): Reimplement.
699 (eligible_for_sibcall_delay): Rename into...
700 (in_sibcall_delay): ...this.
701 (eligible_for_return_delay): Rename into...
702 (in_return_delay): ...this.
703 (in_branch_delay): Reimplement.
704 (in_uncond_branch_delay): Delete.
705 (in_annul_branch_delay): Delete.
707 2014-03-14 Richard Henderson <rth@redhat.com>
709 PR target/60525
710 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
711 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
712 (*floathi<X87MODEF>2_i387_with_temp): Remove.
713 (floathi splitters): Remove.
714 (float<SWI48x>xf2): New pattern.
715 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
716 code that tried to handle DImode for 32-bit, but which was excluded
717 by the pattern's condition. Drop allocation of stack temporary.
718 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
719 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
720 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
721 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
722 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
723 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
724 (*float<SWI48><MODEF>2_sse_interunit): Remove.
725 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
726 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
727 (*float<SWI48x><X87MODEF>2_i387): Remove.
728 (all float _with_temp splitters): Remove.
729 (*float<SWI48x><MODEF>2_i387): New pattern.
730 (*float<SWI48><MODEF>2_sse): New pattern.
731 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
732 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
734 2014-03-14 Jakub Jelinek <jakub@redhat.com>
735 Marek Polacek <polacek@redhat.com>
737 PR middle-end/60484
738 * common.opt (dump_base_name_prefixed): New Variable.
739 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
740 if x_dump_base_name_prefixed is already set, set it at the end.
742 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
744 PR rtl-optimization/60508
745 * lra-constraints.c (get_reload_reg): Add new parameter
746 in_subreg_p.
747 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
748 Pass the new parameter values.
750 2014-03-14 Richard Biener <rguenther@suse.de>
752 * common.opt: Revert unintented changes from r205065.
753 * opts.c: Likewise.
755 2014-03-14 Richard Biener <rguenther@suse.de>
757 PR middle-end/60518
758 * cfghooks.c (split_block): Properly adjust all loops the
759 block was a latch of.
761 2014-03-14 Martin Jambor <mjambor@suse.cz>
763 PR lto/60461
764 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
765 and simplify it.
767 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
769 PR target/59396
770 * config/avr/avr.c (avr_set_current_function): Pass function name
771 through default_strip_name_encoding before sanity checking instead
772 of skipping the first char of the assembler name.
774 2014-03-13 Richard Henderson <rth@redhat.com>
776 PR debug/60438
777 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
778 (ix86_force_to_memory, ix86_free_from_memory): Remove.
779 * config/i386/i386-protos.h: Likewise.
780 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
781 in the expander instead of a splitter.
782 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
783 any possibility of requiring a memory.
784 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
785 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
786 (fp branch splitters): Update for ix86_split_fp_branch.
787 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
788 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
789 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
790 (*fop_<MODEF>_2_i387): Remove f/r alternative.
791 (*fop_<MODEF>_3_i387): Likewise.
792 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
793 (splitters for the fop_* register patterns): Remove.
794 (fscalexf4_i387): Rename from *fscalexf4_i387.
795 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
797 2014-03-13 Jakub Jelinek <jakub@redhat.com>
799 PR tree-optimization/59779
800 * tree-dfa.c (get_ref_base_and_extent): Use double_int
801 type for bitsize and maxsize instead of HOST_WIDE_INT.
803 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
805 PR rtl-optimization/57320
806 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
807 the CFG after thread_prologue_and_epilogue_insns.
809 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
811 PR rtl-optimization/57189
812 * lra-constraints.c (process_alt_operands): Disfavor spilling
813 vector pseudos.
815 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
817 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
818 messages.
820 2014-03-13 Jakub Jelinek <jakub@redhat.com>
822 PR tree-optimization/59025
823 PR middle-end/60418
824 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
825 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
827 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
829 PR target/60486
830 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
831 calls of avr_out_plus_1.
833 2014-03-13 Bin Cheng <bin.cheng@arm.com>
835 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
836 BB's single pred and update the father loop's latch info later.
838 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
840 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
841 (VEC_M): Likewise.
842 (VEC_N): Likewise.
843 (VEC_R): Likewise.
844 (VEC_base): Likewise.
845 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
846 registers, we need to swap double words in little endian mode.
848 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
849 to be a container mode for 128-bit integer operations added in ISA
850 2.07. Unlike TImode and PTImode, the preferred register set is
851 the Altivec/VMX registers for the 128-bit operations.
853 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
854 declarations.
855 (rs6000_split_128bit_ok_p): Likewise.
857 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
858 macros for creating ISA 2.07 normal and overloaded builtin
859 functions with 3 arguments.
860 (BU_P8V_OVERLOAD_3): Likewise.
861 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
862 for use as overloaded functions.
863 (VPERM_1TI_UNS): Likewise.
864 (VSEL_1TI): Likewise.
865 (VSEL_1TI_UNS): Likewise.
866 (ST_INTERNAL_1ti): Likewise.
867 (LD_INTERNAL_1ti): Likewise.
868 (XXSEL_1TI): Likewise.
869 (XXSEL_1TI_UNS): Likewise.
870 (VPERM_1TI): Likewise.
871 (VPERM_1TI_UNS): Likewise.
872 (XXPERMDI_1TI): Likewise.
873 (SET_1TI): Likewise.
874 (LXVD2X_V1TI): Likewise.
875 (STXVD2X_V1TI): Likewise.
876 (VEC_INIT_V1TI): Likewise.
877 (VEC_SET_V1TI): Likewise.
878 (VEC_EXT_V1TI): Likewise.
879 (EQV_V1TI): Likewise.
880 (NAND_V1TI): Likewise.
881 (ORC_V1TI): Likewise.
882 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
883 added in ISA 2.07. Add both normal 'altivec' builtins, and the
884 overloaded builtin.
885 (VADDUQM): Likewise.
886 (VSUBCUQ): Likewise.
887 (VADDEUQM): Likewise.
888 (VADDECUQ): Likewise.
889 (VSUBEUQM): Likewise.
890 (VSUBECUQ): Likewise.
892 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
893 __int128_t and __uint128_t types.
894 (__uint128_type): Likewise.
895 (altivec_categorize_keyword): Add support for vector __int128_t,
896 vector __uint128_t, vector __int128, and vector unsigned __int128
897 as a container type for TImode operations that need to be done in
898 VSX/Altivec registers.
899 (rs6000_macro_to_expand): Likewise.
900 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
901 to support 128-bit integer instructions vaddcuq, vadduqm,
902 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
903 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
905 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
906 for V1TImode, and set up preferences to use VSX/Altivec registers.
907 Setup VSX reload handlers.
908 (rs6000_debug_reg_global): Likewise.
909 (rs6000_init_hard_regno_mode_ok): Likewise.
910 (rs6000_preferred_simd_mode): Likewise.
911 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
912 (easy_altivec_constant): Likewise.
913 (output_vec_const_move): Likewise.
914 (rs6000_expand_vector_set): Convert V1TImode set and extract to
915 simple move.
916 (rs6000_expand_vector_extract): Likewise.
917 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
918 addressing.
919 (rs6000_const_vec): Add support for V1TImode.
920 (rs6000_emit_le_vsx_load): Swap double words when loading or
921 storing TImode/V1TImode.
922 (rs6000_emit_le_vsx_store): Likewise.
923 (rs6000_emit_le_vsx_move): Likewise.
924 (rs6000_emit_move): Add support for V1TImode.
925 (altivec_expand_ld_builtin): Likewise.
926 (altivec_expand_st_builtin): Likewise.
927 (altivec_expand_vec_init_builtin): Likewise.
928 (altivec_expand_builtin): Likewise.
929 (rs6000_init_builtins): Add support for V1TImode type. Add
930 support for ISA 2.07 128-bit integer builtins. Define type names
931 for the VSX/Altivec vector types.
932 (altivec_init_builtins): Add support for overloaded vector
933 functions with V1TImode type.
934 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
935 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
936 external function.
937 (rs6000_split_128bit_ok_p): Likewise.
938 (rs6000_handle_altivec_attribute): Create V1TImode from vector
939 __int128_t and vector __uint128_t.
941 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
942 and mode attributes.
943 (VSX_M): Likewise.
944 (VSX_M2): Likewise.
945 (VSm): Likewise.
946 (VSs): Likewise.
947 (VSr): Likewise.
948 (VSv): Likewise.
949 (VS_scalar): Likewise.
950 (VS_double): Likewise.
951 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
953 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
954 we support the ISA 2.07 128-bit integer arithmetic instructions.
955 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
956 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
957 and TImode types for use with the builtin functions.
958 (V1TI_type_node): Likewise.
959 (unsigned_V1TI_type_node): Likewise.
960 (intTI_type_internal_node): Likewise.
961 (uintTI_type_internal_node): Likewise.
963 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
964 128-bit builtin functions.
965 (UNSPEC_VADDEUQM): Likewise.
966 (UNSPEC_VADDECUQ): Likewise.
967 (UNSPEC_VSUBCUQ): Likewise.
968 (UNSPEC_VSUBEUQM): Likewise.
969 (UNSPEC_VSUBECUQ): Likewise.
970 (VM): Add V1TImode to vector mode iterators.
971 (VM2): Likewise.
972 (VI_unit): Likewise.
973 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
974 (altivec_vaddcuq): Likewise.
975 (altivec_vsubuqm): Likewise.
976 (altivec_vsubcuq): Likewise.
977 (altivec_vaddeuqm): Likewise.
978 (altivec_vaddecuq): Likewise.
979 (altivec_vsubeuqm): Likewise.
980 (altivec_vsubecuq): Likewise.
982 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
983 mode iterators.
984 (BOOL_128): Likewise.
985 (BOOL_REGS_OUTPUT): Likewise.
986 (BOOL_REGS_OP1): Likewise.
987 (BOOL_REGS_OP2): Likewise.
988 (BOOL_REGS_UNARY): Likewise.
989 (BOOL_REGS_AND_CR0): Likewise.
991 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
992 128-bit integer builtin support.
993 (vec_vadduqm): Likewise.
994 (vec_vaddecuq): Likewise.
995 (vec_vaddeuqm): Likewise.
996 (vec_vsubecuq): Likewise.
997 (vec_vsubeuqm): Likewise.
998 (vec_vsubcuq): Likewise.
999 (vec_vsubuqm): Likewise.
1001 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1002 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1003 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1004 128-bit integer add/subtract to ISA 2.07.
1006 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1008 * config/arc/arc.c (arc_predicate_delay_insns):
1009 Fix third argument passed to conditionalize_nonjump.
1011 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1013 * config/aarch64/aarch64-builtins.c
1014 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1015 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1016 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1017 instead of __builtin_lfloor.
1018 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1020 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1022 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1023 (tree_ssa_ifcombine_bb_1): New function.
1024 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1025 is an empty forwarder block to then_bb or vice versa and then_bb
1026 and else_bb are effectively swapped.
1028 2014-03-12 Christian Bruel <christian.bruel@st.com>
1030 PR target/60264
1031 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1032 REG_CFA_DEF_CFA note.
1033 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1034 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1036 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1038 PR tree-optimization/60454
1039 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1041 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1043 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1044 Do not define target_cpu_default2 to generic.
1045 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1046 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1047 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1049 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1050 Marc Glisse <marc.glisse@inria.fr>
1052 PR tree-optimization/60502
1053 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1054 instead of build_low_bits_mask.
1056 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1058 PR middle-end/60482
1059 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1060 if there are multiple uses, but op doesn't live on E edge.
1061 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1062 clobber stmts before __builtin_unreachable.
1064 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1066 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1067 hard_frame_pointer_rtx.
1068 * cse.c (cse_insn): Remove volatile check.
1069 * cselib.c (cselib_process_insn): Likewise.
1070 * dse.c (scan_insn): Likewise.
1072 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1074 * config/arc/arc.c (conditionalize_nonjump): New function,
1075 broken out of ...
1076 (arc_ifcvt): ... this.
1077 (arc_predicate_delay_insns): Use it.
1079 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1081 * config/arc/predicates.md (extend_operand): During/after reload,
1082 allow const_int_operand.
1083 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1084 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1085 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1086 to "i".
1087 (umulsi3_highpart_i): Likewise.
1089 2014-03-11 Richard Biener <rguenther@suse.de>
1091 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1092 Add asserts to guard possible wrong-code bugs.
1094 2014-03-11 Richard Biener <rguenther@suse.de>
1096 PR tree-optimization/60429
1097 PR tree-optimization/60485
1098 * tree-ssa-structalias.c (set_union_with_increment): Properly
1099 take into account all fields that overlap the shifted vars.
1100 (do_sd_constraint): Likewise.
1101 (do_ds_constraint): Likewise.
1102 (get_constraint_for_ptr_offset): Likewise.
1104 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1106 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1107 (nios2_compute_frame_layout):
1108 Add calculation of cfun->machine->fp_save_offset.
1109 (nios2_expand_prologue): Correct setting of frame pointer register
1110 in prologue.
1111 (nios2_expand_epilogue): Update recovery of stack pointer from
1112 frame pointer accordingly.
1113 (nios2_initial_elimination_offset): Update calculation of offset
1114 for eliminating to HARD_FRAME_POINTER_REGNUM.
1116 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1118 PR ipa/60457
1119 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1120 cgraph_get_create_node on VAR_DECLs.
1122 2014-03-10 Richard Biener <rguenther@suse.de>
1124 PR middle-end/60474
1125 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1127 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1129 * config/vms/vms.opt (vms_float_format): New variable.
1131 2014-03-08 Tobias Burnus <burnus@net-b.de>
1133 * doc/invoke.texi (-fcilkplus): Update implementation status.
1135 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1136 Richard Biener <rguenther@suse.de>
1138 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1139 consistently accross all TUs.
1140 (run_gcc): Enable -fshort-double automatically at link at link-time
1141 and disallow override.
1143 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1145 PR target/58271
1146 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1147 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1148 if they can't be used.
1150 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1152 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1153 for Solaris 11/x86 ld.
1154 * configure: Regenerate.
1156 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1158 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1159 (LIB_TLS_SPEC): Save as ld_tls_libs.
1160 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1161 (HAVE_AS_IX86_TLSLDM): New test.
1162 * configure, config.in: Regenerate.
1163 * config/i386/i386.c (legitimize_tls_address): Fall back to
1164 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1165 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1166 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1167 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1169 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1171 * common.opt (fira-loop-pressure): Mark as optimization.
1173 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1175 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1176 an OpenMP mappable type.
1178 2014-03-06 Matthias Klose <doko@ubuntu.com>
1180 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1181 MULTILIB_OSDIRNAMES is not defined.
1183 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1184 Meador Inge <meadori@codesourcery.com>
1186 PR target/58595
1187 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1188 (arm_legitimize_address): Call legitimize_tls_address for any
1189 arm_tls_referenced_p expression, handle constant addend. Call it
1190 before testing for !TARGET_ARM.
1191 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1193 2014-03-06 Richard Biener <rguenther@suse.de>
1195 PR middle-end/60445
1196 PR lto/60424
1197 PR lto/60427
1198 Revert
1199 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1201 * tree-streamer.c (record_common_node): Assert we don't record
1202 nodes with type double.
1203 (preload_common_node): Skip type double, complex double and double
1204 pointer since it is now frontend dependent due to fshort-double option.
1206 2014-03-06 Richard Biener <rguenther@suse.de>
1208 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1209 or -fno-lto is specified and the linker has full plugin support.
1210 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1211 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1212 * lto-wrapper.c (merge_and_complain): Merge compile-time
1213 optimization levels.
1214 (run_gcc): And pass it through to the link options.
1216 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1218 PR debug/60381
1219 Revert:
1220 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1221 PR debug/59992
1222 * cselib.c (remove_useless_values): Skip to avoid quadratic
1223 behavior if the condition moved from...
1224 (cselib_process_insn): ... here holds.
1226 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1228 PR plugins/59335
1229 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1230 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1232 PR plugins/59335
1233 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1234 (TM_H): Add x86-tune.def.
1236 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1238 * config/aarch64/aarch64.c (generic_tunings):
1239 Use cortexa57_extra_costs.
1241 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1243 PR lto/60404
1244 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1245 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1246 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1247 cost for in_lto_p.
1249 2014-03-04 Heiher <r@hev.cc>
1251 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1252 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1254 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1256 * config/i386/predicates.md (const2356_operand): Change to ...
1257 (const2367_operand): ... this.
1258 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1259 const2367_operand.
1260 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1261 (*avx512pf_scatterpf<mode>sf): Ditto.
1262 (avx512pf_scatterpf<mode>df): Ditto.
1263 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1264 (*avx512pf_scatterpf<mode>df): Ditto.
1265 * config/i386/i386.c (ix86_expand_builtin): Update
1266 incorrect hint operand error message.
1268 2014-03-04 Richard Biener <rguenther@suse.de>
1270 * lto-section-in.c (lto_get_section_data): Fix const cast.
1272 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1274 * tree-streamer.c (record_common_node): Assert we don't record
1275 nodes with type double.
1276 (preload_common_node): Skip type double, complex double and double
1277 pointer since it is now frontend dependent due to fshort-double option.
1279 2014-03-04 Richard Biener <rguenther@suse.de>
1281 PR lto/60405
1282 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1283 (lto_input_toplevel_asms): Likewise.
1284 * lto-section-in.c (lto_get_section_data): Instead do it here
1285 for every section.
1287 2014-03-04 Richard Biener <rguenther@suse.de>
1289 PR tree-optimization/60382
1290 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1291 dead PHIs a reduction.
1293 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1295 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1296 hint value.
1297 (_mm_prefetch): Move out of GCC target("sse") pragma.
1298 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1299 GCC target("prfchw") pragma.
1300 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1301 for locality <= 2.
1302 * config/i386/i386.c (ix86_option_override_internal): Enable
1303 -mprfchw with -mprefetchwt1.
1305 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1307 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1308 Mark as varying.
1310 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1312 * opts.h (CL_PCH_IGNORE): Define.
1313 * targhooks.c (option_affects_pch_p):
1314 Return false for options that have CL_PCH_IGNORE set.
1315 * opt-functions.awk: Process PchIgnore.
1316 * doc/options.texi: Document PchIgnore.
1318 * config/arc/arc.opt (misize): Add PchIgnore property.
1320 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1322 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1323 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1324 constraint on constants to permit them being loaded into
1325 GENERAL_REGS or BASE_REGS.
1327 2014-03-03 Nick Clifton <nickc@redhat.com>
1329 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1330 anti-cacnonical alternatives.
1331 (negandhi3_real): New pattern.
1332 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1334 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1336 * config/avr/avr-mcus.def: Remove atxmega16x1.
1337 * config/avr/avr-tables.opt: Regenerate.
1338 * config/avr/t-multilib: Regenerate.
1339 * doc/avr-mmcu.texi: Regenerate.
1341 2014-03-03 Tobias Grosser <tobias@grosser.es>
1342 Mircea Namolaru <mircea.namolaru@inria.fr>
1344 PR tree-optimization/58028
1345 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1346 scalar dimensions.
1348 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1350 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1351 not handled by recognizers.
1353 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1355 PR middle-end/60175
1356 * function.c (expand_function_end): Don't emit
1357 clobber_return_register sequence if clobber_after is a BARRIER.
1358 * cfgexpand.c (construct_exit_block): Append instructions before
1359 return_label to prev_bb.
1361 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1363 * config/rs6000/constraints.md: Document reserved use of "wc".
1365 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1367 PR ipa/60150
1368 * ipa.c (function_and_variable_visibility): When dissolving comdat
1369 group, also set all symbols to local.
1371 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1373 PR ipa/60306
1375 Revert:
1376 2013-12-14 Jan Hubicka <jh@suse.cz>
1377 PR middle-end/58477
1378 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1380 2014-03-02 Jon Beniston <jon@beniston.com>
1382 PR bootstrap/48230
1383 PR bootstrap/50927
1384 PR bootstrap/52466
1385 PR target/46898
1386 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1387 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1388 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1389 (simple_return, *simple_return): New patterns
1390 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1391 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1393 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1395 * dwarf2out.c (gen_subprogram_die): Tidy.
1397 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1399 PR target/60071
1400 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1401 (*mov_t_msb_neg_negc): ... this new insn.
1403 2014-02-28 Jason Merrill <jason@redhat.com>
1405 PR c++/58678
1406 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1407 function.
1409 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1411 PR c++/60314
1412 * dwarf2out.c (decltype_auto_die): New static.
1413 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1414 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1415 (is_cxx_auto): Likewise.
1417 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1419 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1420 we are not using general regs only.
1422 2014-02-28 Richard Biener <rguenther@suse.de>
1424 PR target/60280
1425 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1426 previous fix and only allow to remove trivial pre-headers
1427 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1428 (remove_forwarder_block): Properly update the latch of a loop.
1430 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1432 PR debug/59992
1433 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1434 (cselib_preserved_hash_table): New.
1435 (preserve_constants_and_equivs): Move preserved vals to it.
1436 (cselib_find_slot): Look it up first.
1437 (cselib_init): Initialize it.
1438 (cselib_finish): Release it.
1439 (dump_cselib_table): Dump it.
1441 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1443 PR debug/59992
1444 * cselib.c (remove_useless_values): Skip to avoid quadratic
1445 behavior if the condition moved from...
1446 (cselib_process_insn): ... here holds.
1448 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1450 PR debug/57232
1451 * var-tracking.c (vt_initialize): Apply the same condition to
1452 preserve the CFA base value.
1454 2014-02-28 Joey Ye <joey.ye@arm.com>
1456 PR target/PR60169
1457 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1458 if reload in progress or completed.
1460 2014-02-28 Tobias Burnus <burnus@net-b.de>
1462 PR middle-end/60147
1463 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1464 NAMELIST_DECL.
1466 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1468 * doc/tm.texi.in (Condition Code Status): Update documention for
1469 relative locations of cc0-setter and cc0-user.
1471 2014-02-27 Jeff Law <law@redhat.com>
1473 PR rtl-optimization/52714
1474 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1475 into two independent simple sets, if I3 is a jump, ensure the
1476 pattern we place into I3 is a (set (pc) ...).
1478 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1479 Jeff Law <law@redhat.com>
1481 PR rtl-optimization/49847
1482 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1483 are in different blocks.
1484 * doc/tm.texi (Condition Code Status): Update documention for
1485 relative locations of cc0-setter and cc0-user.
1487 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1489 PR target/59222
1490 * lra.c (lra_emit_add): Check SUBREG too.
1492 2014-02-27 Andreas Schwab <schwab@suse.de>
1494 * config/m68k/m68k.c (m68k_option_override): Disable
1495 -flive-range-shrinkage for classic m68k.
1496 (m68k_override_options_after_change): Likewise.
1498 2014-02-27 Marek Polacek <polacek@redhat.com>
1500 PR middle-end/59223
1501 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1502 -Wmaybe-uninitialized.
1504 2014-02-27 Alan Modra <amodra@gmail.com>
1506 PR target/57936
1507 * reload1.c (emit_input_reload_insns): When reload_override_in,
1508 set old to rl->in_reg when rl->in_reg is a subreg.
1510 2014-02-26 Richard Biener <rguenther@suse.de>
1512 PR bootstrap/60343
1513 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1515 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1517 * common/config/i386/predicates.md (const1256_operand): Remove.
1518 (const2356_operand): New.
1519 (const_1_to_2_operand): Remove.
1520 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1521 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1522 (*avx512pf_gatherpf<mode>sf): Ditto.
1523 (avx512pf_gatherpf<mode>df): Ditto.
1524 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1525 (*avx512pf_gatherpf<mode>df): Ditto.
1526 (avx512pf_scatterpf<mode>sf): Ditto.
1527 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1528 (*avx512pf_scatterpf<mode>sf): Ditto.
1529 (avx512pf_scatterpf<mode>df): Ditto.
1530 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1531 (*avx512pf_scatterpf<mode>df): Ditto.
1532 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1534 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1536 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1537 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1538 (_mm512_mask_testn_epi64_mask): Move to ...
1539 * config/i386/avx512cdintrin.h: Here.
1540 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1541 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1542 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1543 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1544 TARGET_AVX512F from TARGET_AVX512CD.
1546 2014-02-26 Richard Biener <rguenther@suse.de>
1548 PR ipa/60327
1549 * ipa.c (walk_polymorphic_call_targets): Properly guard
1550 call to inline_update_overall_summary.
1552 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1554 PR target/60280
1555 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1556 and latches only if requested. Fix latch if it is removed.
1557 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1558 LOOPS_HAVE_PREHEADERS.
1560 2014-02-25 Andrew Pinski <apinski@cavium.com>
1562 * builtins.c (expand_builtin_thread_pointer): Create a new target
1563 when the target is NULL.
1565 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1567 PR rtl-optimization/60317
1568 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1569 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1570 * lra-assigns.c: Include params.h.
1571 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1572 other reload pseudos considerations.
1574 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1576 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1577 to use canonical form for nor<mode>3.
1579 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1581 PR target/55426
1582 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1583 conversions.
1585 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1587 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1588 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1589 (ix86_handle_option): Handle OPT_mprefetchwt1.
1590 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1591 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1592 PREFETCHWT1 CPUID.
1593 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1594 OPTION_MASK_ISA_PREFETCHWT1.
1595 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1596 (PTA_PREFETCHWT1): New.
1597 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1598 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1599 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1600 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1601 (*prefetch_avx512pf_<mode>_: Change into ...
1602 (*prefetch_prefetchwt1_<mode>: This.
1603 * config/i386/i386.opt (mprefetchwt1): New.
1604 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1605 (_mm_prefetch): Handle intent to write.
1606 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1608 2014-02-25 Richard Biener <rguenther@suse.de>
1610 PR middle-end/60291
1611 * emit-rtl.c (mem_attrs_htab): Remove.
1612 (mem_attrs_htab_hash): Likewise.
1613 (mem_attrs_htab_eq): Likewise.
1614 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1615 (init_emit_once): Do not allocate mem_attrs_htab.
1617 2014-02-25 Richard Biener <rguenther@suse.de>
1619 PR lto/60319
1620 * lto-opts.c (lto_write_options): Output non-explicit conservative
1621 -fwrapv, -fno-trapv and -fno-strict-overflow.
1622 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1623 (run_gcc): And pass them through.
1625 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1627 * sel-sched.c (calculate_new_fences): New parameter ptime.
1628 Calculate it as a maximum over all fence cycles.
1629 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1630 Print the final schedule timing when sched_verbose.
1632 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1634 PR rtl-optimization/60292
1635 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1636 bit fot the fence instruction.
1638 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1640 * calls.h: Fix typo in comment.
1642 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1644 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1645 adjusting offsetable addresses.
1647 2014-02-24 Guozhi Wei <carrot@google.com>
1649 * sparseset.h (sparseset_pop): Fix the wrong index.
1651 2014-02-24 Walter Lee <walt@tilera.com>
1653 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1654 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1655 triplet.
1656 * common/config/tilegx/tilegx-common.c
1657 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1658 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1659 (LINK_SPEC): Ditto.
1660 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1661 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1662 (tilegx_gimplify_va_arg_expr): Handle big endian.
1663 (tilegx_expand_unaligned_load): Ditto.
1664 (tilegx_expand_unaligned_store): Ditto.
1665 (TARGET_RETURN_IN_MSB): New.
1666 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1667 (TARGET_ENDIAN_DEFAULT): New.
1668 (TARGET_BIG_ENDIAN): Handle big endian.
1669 (BYTES_BIG_ENDIAN): Ditto.
1670 (WORDS_BIG_ENDIAN): Ditto.
1671 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1672 (ENDIAN_SPEC): New.
1673 (EXTRA_SPECS): New.
1674 * config/tilegx/tilegx.md (extv): Handle big endian.
1675 (extzv): Ditto.
1676 (insn_st<n>): Ditto.
1677 (insn_st<n>_add<bitsuffix>): Ditto.
1678 (insn_stnt<n>): Ditto.
1679 (insn_stnt<n>_add<bitsuffix>):Ditto.
1680 (vec_interleave_highv8qi): Handle big endian.
1681 (vec_interleave_highv8qi_be): New.
1682 (vec_interleave_highv8qi_le): New.
1683 (insn_v1int_h): Handle big endian.
1684 (vec_interleave_lowv8qi): Handle big endian.
1685 (vec_interleave_lowv8qi_be): New.
1686 (vec_interleave_lowv8qi_le): New.
1687 (insn_v1int_l): Handle big endian.
1688 (vec_interleave_highv4hi): Handle big endian.
1689 (vec_interleave_highv4hi_be): New.
1690 (vec_interleave_highv4hi_le): New.
1691 (insn_v2int_h): Handle big endian.
1692 (vec_interleave_lowv4hi): Handle big endian.
1693 (vec_interleave_lowv4hi_be): New.
1694 (vec_interleave_lowv4hi_le): New.
1695 (insn_v2int_l): Handle big endian.
1696 (vec_interleave_highv2si): Handle big endian.
1697 (vec_interleave_highv2si_be): New.
1698 (vec_interleave_highv2si_le): New.
1699 (insn_v4int_h): Handle big endian.
1700 (vec_interleave_lowv2si): Handle big endian.
1701 (vec_interleave_lowv2si_be): New.
1702 (vec_interleave_lowv2si_le): New.
1703 (insn_v4int_l): Handle big endian.
1704 * config/tilegx/tilegx.opt (mbig-endian): New option.
1705 (mlittle-endian): New option.
1706 * doc/install.texi: Document tilegxbe-linux.
1707 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1709 2014-02-24 Martin Jambor <mjambor@suse.cz>
1711 PR ipa/60266
1712 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1713 there are no parameter descriptors.
1715 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1717 PR rtl-optimization/60268
1718 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1719 initialization to ...
1720 (sched_rgn_init): ... here.
1721 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1723 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1725 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1726 names.
1728 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1730 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1731 definition.
1733 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1735 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1736 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1738 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1740 * config/microblaze/predicates.md: Add cmp_op predicate.
1741 * config/microblaze/microblaze.md: Add branch_compare instruction
1742 which uses cmp_op predicate and emits cmp insn before branch.
1743 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1744 to microblaze_expand_conditional_branch and consolidate logic.
1745 (microblaze_expand_conditional_branch): emit branch_compare
1746 insn instead of handling cmp op separate from branch insn.
1748 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1750 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1751 to permit subregs.
1753 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1755 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1756 define_insn with define_expand and new define_insn
1757 *altivec_lve<VI_char>x_internal.
1758 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1759 and new define_insn *altivec_stve<VI_char>x_internal.
1760 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1761 prototype.
1762 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1763 lve*x built-ins.
1764 (altivec_expand_stvex_be): New function.
1766 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1768 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1769 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1770 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1771 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1773 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1775 PR target/60298
1776 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1777 instead of emit_move_insn.
1779 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1781 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1782 vspltw with vsldoi.
1783 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1784 gen_altivec_vsumsws.
1786 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1788 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1789 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1790 (altivec_lvxl_<mode>): New define_expand incorporating
1791 -maltivec=be semantics where needed.
1792 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1793 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1794 semantics where needed.
1795 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1796 (altivec_stvx_<mode>): New define_expand incorporating
1797 -maltivec=be semantics where needed.
1798 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1799 VM2 iterator instead of V4SI.
1800 (altivec_stvxl_<mode>): New define_expand incorporating
1801 -maltivec=be semantics where needed.
1802 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1803 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1804 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1805 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1806 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1807 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1808 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1809 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1810 ALTIVEC_BUILTIN_STVXL.
1811 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1812 (altivec_expand_stvx_be): Likewise.
1813 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1814 (altivec_expand_lvx_be): Likewise.
1815 (altivec_expand_stvx_be): Likewise.
1816 (altivec_expand_builtin): Add cases for
1817 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1818 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1819 (altivec_init_builtins): Add definitions for
1820 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1821 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1823 2014-02-21 Catherine Moore <clm@codesourcery.com>
1825 * doc/invoke.texi (mvirt, mno-virt): Document.
1826 * config/mips/mips.opt (mvirt): New option.
1827 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1829 2014-02-21 Richard Biener <rguenther@suse.de>
1831 PR tree-optimization/60276
1832 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1833 (STMT_VINFO_MIN_NEG_DIST): New macro.
1834 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1835 STMT_VINFO_MIN_NEG_DIST.
1836 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1837 made for negative dependence distances still hold.
1839 2014-02-21 Richard Biener <rguenther@suse.de>
1841 PR middle-end/60291
1842 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1843 DECL_INITIAL for globals not in the current function context.
1845 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1847 PR tree-optimization/56490
1848 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1849 * tree-ssa-uninit.c: Include params.h.
1850 (compute_control_dep_chain): Add num_calls argument, return false
1851 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1852 num_calls to recursive call.
1853 (find_predicates): Change dep_chain into normal array,
1854 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1855 variable and adjust compute_control_dep_chain caller.
1856 (find_def_preds): Likewise.
1858 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1860 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1861 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1863 2014-02-21 Nick Clifton <nickc@redhat.com>
1865 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1866 (pushhi1): Likewise.
1867 (popqi1): Add mode to pre_dec.
1868 (pophi1): Likewise.
1870 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1872 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1873 mode for mask of V8SFmode permutation.
1875 2014-02-20 Richard Henderson <rth@redhat.com>
1877 PR c++/60272
1878 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1879 a new pseudo for OLDVAL.
1881 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1883 PR target/57896
1884 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1885 gen_reg_rtx if d->testing_p.
1886 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1887 if d->testing_p and we will certainly return true.
1888 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1889 if d->testing_p.
1891 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1893 * emit-rtl.c (gen_reg_rtx): Assert that
1894 crtl->emit.regno_pointer_align_length is non-zero.
1896 2014-02-20 Richard Henderson <rth@redhat.com>
1898 PR c++/60272
1899 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1900 on failure the store back into EXPECT.
1902 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1903 Sandra Loosemore <sandra@codesourcery.com>
1905 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1906 * config/nios2/nios2.c (nios2_function_profiler): Add
1907 -fPIC (flag_pic == 2) support.
1908 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1909 (nios2_large_offset_p): New function.
1910 (nios2_unspec_reloc_p): Move up position, update to use
1911 nios2_large_offset_p.
1912 (nios2_unspec_address): Remove function.
1913 (nios2_unspec_offset): New function.
1914 (nios2_large_got_address): New function.
1915 (nios2_got_address): Add large offset support.
1916 (nios2_legitimize_tls_address): Update usage of removed and new
1917 functions.
1918 (nios2_symbol_binds_local_p): New function.
1919 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1920 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1921 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1922 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1923 processing for (const (unspec ...)).
1924 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1926 2014-02-20 Richard Biener <rguenther@suse.de>
1928 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1929 doing the substitution.
1930 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1932 2014-02-20 Martin Jambor <mjambor@suse.cz>
1934 PR ipa/55260
1935 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1936 info when checking whether lattices are bottom.
1938 2014-02-20 Richard Biener <rguenther@suse.de>
1940 PR middle-end/60221
1941 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1942 regions at -O0.
1944 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1946 PR ipa/58555
1947 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1948 parameter specifying the scaling.
1949 (inline_call): Update.
1950 (want_inline_recursively): Guard division by zero.
1951 (recursive_inlining): Update.
1952 * ipa-inline.h (clone_inlined_nodes): Update.
1954 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1956 PR target/60204
1957 * config/i386/i386.c (classify_argument): Pass structures of size
1958 64 bytes or less in register.
1960 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1961 Kirill Yukhin <kirill.yukhin@intel.com>
1963 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1964 (_mm_rcp28_round_ss): Ditto.
1965 (_mm_rsqrt28_round_sd): Ditto.
1966 (_mm_rsqrt28_round_ss): Ditto.
1967 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1968 (_mm_rcp14_round_ss): Ditto.
1969 (_mm_rsqrt14_round_sd): Ditto.
1970 (_mm_rsqrt14_round_ss): Ditto.
1971 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1972 the first input operand, get rid of match_dup.
1973 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1974 attribute to sse.
1975 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1976 Ditto.
1977 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1978 operand as the first input operand, set type attribute.
1979 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1980 Set type attribute.
1981 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1982 operand as the first input operand, set type attribute.
1984 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1986 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1987 bit of zero.
1989 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1991 PR target/60207
1992 * config/i386/i386.c (construct_container): Remove TFmode check
1993 for X86_64_INTEGER_CLASS.
1995 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1997 PR target/59794
1998 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1999 only when -Wpsabi is enabled.
2001 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2003 PR target/59799
2004 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2005 passing arrays in registers are the same as for structs, so remove the
2006 special case for them.
2008 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2010 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2011 destination type, extract only the valid bits if the source type is not
2012 integral and has a different mode.
2014 2014-02-19 Richard Biener <rguenther@suse.de>
2016 PR ipa/60243
2017 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2018 for all calls.
2020 2014-02-19 Richard Biener <rguenther@suse.de>
2022 PR ipa/60243
2023 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2024 (ipa_modify_call_arguments): Emit an argument load explicitely and
2025 preserve virtual SSA form there and for the replacement call.
2026 Do not update SSA form nor free dominance info.
2028 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2030 * ipa.c (function_and_variable_visibility): Also clear WEAK
2031 flag when disolving COMDAT_GROUP.
2033 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2035 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2036 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2037 not devirtualizing.
2038 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2039 do more sanity checks.
2040 (detect_type_change): Return true when giving up early.
2041 (compute_complex_assign_jump_func): Fix type parameter of
2042 ipa_set_ancestor_jf.
2043 (compute_complex_ancestor_jump_func): Likewise.
2044 (update_jump_functions_after_inlining): Fix updating of
2045 ancestor function.
2046 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2048 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2050 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2051 inline clones when edge disappears.
2053 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2055 PR target/60203
2056 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2057 Split 64-bit moves into 2 patterns. Do not allow the use of
2058 direct move for TDmode in little endian, since the decimal value
2059 has little endian bytes within a word, but the 64-bit pieces are
2060 ordered in a big endian fashion, and normal subreg's of TDmode are
2061 not allowed.
2062 (mov<mode>_64bit_dm): Likewise.
2063 (movtd_64bit_nodm): Likewise.
2065 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2067 PR tree-optimization/60174
2068 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2069 statement of an SSA_NAME that occurs in an abnormal PHI node.
2071 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2073 PR sanitizer/60142
2074 * final.c (SEEN_BB): Remove.
2075 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2076 (final_scan_insn): Don't force_source_line on second
2077 NOTE_INSN_BASIC_BLOCK.
2079 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2081 PR target/60205
2082 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2083 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2084 (type_natural_mode): Warn ABI change when %zmm register is not
2085 available for AVX512F vector value passing.
2087 2014-02-18 Kai Tietz <ktietz@redhat.com>
2089 PR target/60193
2090 * config/i386/i386.c (ix86_expand_prologue): Use value in
2091 rax register as displacement when restoring %r10 or %rax.
2092 Fix wrong offset when restoring both registers.
2094 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2096 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2097 assertion with conditional return.
2099 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2100 Uros Bizjak <ubizjak@gmail.com>
2102 PR driver/60233
2103 * config/i386/driver-i386.c (host_detect_local_cpu): If
2104 YMM state is not saved by the OS, also clear has_f16c. Move
2105 CPUID 0x80000001 handling before YMM state saving checking.
2107 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2109 PR rtl-optimization/58960
2110 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2111 factored out from ...
2112 (sched_init): ... here.
2113 (free_global_sched_pressure_data): New, factored out from ...
2114 (sched_finish): ... here.
2115 * sched-int.h (free_global_sched_pressure_data): Declare.
2116 * sched-rgn.c (nr_regions_initial): New static global.
2117 (haifa_find_rgns): Initialize it.
2118 (schedule_region): Disable sched-pressure for the newly
2119 generated regions.
2121 2014-02-17 Richard Biener <rguenther@suse.de>
2123 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2124 release SSA defs of pattern stmts.
2126 2014-02-17 Richard Biener <rguenther@suse.de>
2128 * tree-inline.c (expand_call_inline): Release the virtual
2129 operand defined by the call we are about to inline.
2131 2014-02-17 Richard Biener <rguenther@suse.de>
2133 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2135 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2136 Ilya Tocar <ilya.tocar@intel.com>
2138 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2139 arguments order in builtin.
2140 (_mm512_permutexvar_epi64): Ditto.
2141 (_mm512_mask_permutexvar_epi64): Ditto
2142 (_mm512_maskz_permutexvar_epi32): Ditto
2143 (_mm512_permutexvar_epi32): Ditto
2144 (_mm512_mask_permutexvar_epi32): Ditto
2146 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2148 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2149 (p8_vmrgow): Likewise.
2151 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2153 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2154 endian targets.
2156 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2158 PR target/60203
2159 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2160 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2161 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2162 using direct move instructions on ISA 2.07. Also adjust
2163 instruction length for 64-bit.
2164 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2165 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2167 2014-02-15 Alan Modra <amodra@gmail.com>
2169 PR target/58675
2170 PR target/57935
2171 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2172 find_replacement on parts of insn rtl that might be reloaded.
2174 2014-02-15 Richard Biener <rguenther@suse.de>
2176 PR tree-optimization/60183
2177 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2178 (tree_ssa_phiprop): Calculate and free post-dominators.
2180 2014-02-14 Jeff Law <law@redhat.com>
2182 PR rtl-optimization/60131
2183 * ree.c (get_extended_src_reg): New function.
2184 (combine_reaching_defs): Use it rather than assuming location of REG.
2185 (find_and_remove_re): Verify first operand of extension is
2186 a REG before adding the insns to the copy list.
2188 2014-02-14 Roland McGrath <mcgrathr@google.com>
2190 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2191 * configure: Regenerated.
2192 * config.in: Regenerated.
2193 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2194 instead of ASM_SHORT.
2196 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2197 Richard Earnshaw <rearnsha@arm.com>
2199 PR rtl-optimization/59535
2200 * lra-constraints.c (process_alt_operands): Encourage alternative
2201 when unassigned pseudo class is superset of the alternative class.
2202 (inherit_reload_reg): Don't inherit when optimizing for code size.
2203 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2204 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2205 modes not less than 4 for Thumb1.
2207 2014-02-14 Kyle McMartin <kyle@redhat.com>
2209 PR pch/60010
2210 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2212 2014-02-14 Richard Biener <rguenther@suse.de>
2214 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2215 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2216 Do not strip INDIRECT_REFs.
2218 2014-02-14 Richard Biener <rguenther@suse.de>
2220 PR lto/60179
2221 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2222 DECL_FUNCTION_SPECIFIC_TARGET.
2223 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2224 * tree-streamer-out.c (pack_ts_target_option): Remove.
2225 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2226 (write_ts_function_decl_tree_pointers): Do not stream
2227 DECL_FUNCTION_SPECIFIC_TARGET.
2228 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2229 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2230 (lto_input_ts_function_decl_tree_pointers): Do not stream
2231 DECL_FUNCTION_SPECIFIC_TARGET.
2233 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2235 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2236 (get_initial_def_for_induction, vectorizable_induction): Ignore
2237 debug stmts when looking for exit_phi.
2238 (vectorizable_live_operation): Fix up condition.
2240 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2242 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2243 nreverse() because it changes the content of original tree list.
2245 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2247 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2248 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2250 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2252 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2253 GNU coding standards.
2255 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2257 PR debug/60152
2258 * dwarf2out.c (gen_subprogram_die): Don't call
2259 add_calling_convention_attribute if subr_die is old_die.
2261 2014-02-13 Sharad Singhai <singhai@google.com>
2263 * doc/optinfo.texi: Fix order of nodes.
2265 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2267 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2268 operands[2], not operands[3].
2270 2014-02-13 Richard Biener <rguenther@suse.de>
2272 PR bootstrap/59878
2273 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2274 mention the possibility of an in-tree build.
2275 (CLooG): Update recommended version to 0.18.1, mention the
2276 possibility of an in-tree build and clarify that the ISL
2277 bundled with CLooG does not work.
2279 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2281 PR target/43546
2282 * expr.c (compress_float_constant): If x is a hard register,
2283 extend into a pseudo and then move to x.
2285 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2287 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2288 caused by bad second argument to warning_at() with -mhotpatch and
2289 nested functions (e.g. with gfortran).
2291 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2293 * opts.c (option_name): Remove "enabled by default" rider.
2295 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2297 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2299 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2300 Uros Bizjak <ubizjak@gmail.com>
2302 PR target/60151
2303 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2304 * configure: Regenerated.
2306 2014-02-12 Richard Biener <rguenther@suse.de>
2308 * vec.c (vec_prefix::calculate_allocation): Move as
2309 inline variant to vec.h.
2310 (vec_prefix::calculate_allocation_1): New out-of-line version.
2311 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2312 (vec_prefix::m_has_auto_buf): Rename to ...
2313 (vec_prefix::m_using_auto_storage): ... this.
2314 (vec_prefix::calculate_allocation): Inline the easy cases
2315 and dispatch to calculate_allocation_1 which doesn't need the
2316 prefix address.
2317 (va_heap::reserve): Use gcc_checking_assert.
2318 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2319 m_using_auto_storage.
2320 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2321 member and adjust.
2322 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2323 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2324 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2326 2014-02-12 Richard Biener <rguenther@suse.de>
2328 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2329 when we found a dependence.
2331 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2333 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2334 common code...
2335 (maybe_fold_stmt): ... into this new function.
2336 * omp-low.c (lower_omp): Update comment.
2338 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2339 last use.
2341 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2342 dereference.
2344 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2346 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2347 identifiers in comments.
2348 (cortexa53_extra_costs): Likewise.
2349 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2350 (cortexa7_extra_costs): Likewise.
2351 (cortexa12_extra_costs): Likewise.
2352 (cortexa15_extra_costs): Likewise.
2353 (v7m_extra_costs): Likewise.
2355 2014-02-12 Richard Biener <rguenther@suse.de>
2357 PR middle-end/60092
2358 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2359 of posix_memalign being successful.
2360 (lower_stmt): Restrict lowering of posix_memalign to when
2361 -ftree-bit-ccp is enabled.
2363 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2365 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2366 arg_loc.
2367 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2369 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2371 PR rtl-optimization/60116
2372 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2373 other_insn once the combination has been validated.
2375 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2377 PR lto/59468
2378 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2379 and wrapper.
2380 * ipa-devirt.c: Include demangle.h
2381 (odr_violation_reported): New static variable.
2382 (add_type_duplicate): Update odr_violations.
2383 (maybe_record_node): Add completep parameter; update it.
2384 (record_target_from_binfo): Add COMPLETEP parameter;
2385 update it as needed.
2386 (possible_polymorphic_call_targets_1): Likewise.
2387 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2388 rename FINAL to COMPLETE.
2389 (record_targets_from_bases): Sanity check we found the binfo;
2390 fix COMPLETEP updating.
2391 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2392 parameter, fix computing of COMPLETEP.
2393 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2394 at LTO time do demangling.
2395 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2396 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2397 parameter.
2398 (gimple_get_virt_method_for_binfo): Likewise.
2399 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2400 gimple_get_virt_method_for_vtable): Update prototypes.
2402 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2404 PR target/49008
2405 * genautomata.c (add_presence_absence): Fix typo with
2406 {final_}presence_list.
2408 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2410 PR target/60137
2411 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2412 for VSX/Altivec vectors that land in GPR registers.
2414 2014-02-11 Richard Henderson <rth@redhat.com>
2415 Jakub Jelinek <jakub@redhat.com>
2417 PR debug/59776
2418 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2419 around drhs if type conversion to lacc->type is not useless.
2421 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2423 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2424 tuning struct.
2425 (cortex-a57.cortex-a53): Likewise.
2426 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2428 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2430 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2431 arm_restrict_it.
2433 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2435 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2436 add_options_for_arm_vfp3.
2438 2014-02-11 Jeff Law <law@redhat.com>
2440 PR middle-end/54041
2441 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2442 object with an undesirable mode.
2444 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2446 PR libgomp/60107
2447 * config/i386/sol2-9.h: New file.
2448 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2449 *-*-solaris2.9*): Use it.
2451 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2453 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2454 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2456 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2458 * config/microblaze/microblaze.c: Extend mcpu version format
2460 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2462 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2464 2014-02-10 Richard Henderson <rth@redhat.com>
2466 PR target/59927
2467 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2468 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2469 ms-abi vs -mno-accumulate-outgoing-args.
2470 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2471 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2472 respect to ms-abi.
2474 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2476 PR middle-end/60080
2477 * cfgexpand.c (expand_asm_operands): Attach source location to
2478 ASM_INPUT rtx objects.
2479 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2481 2014-02-10 Nick Clifton <nickc@redhat.com>
2483 * config/mn10300/mn10300.c (popcount): New function.
2484 (mn10300_expand_prologue): Include saved registers in stack usage
2485 count.
2487 2014-02-10 Jeff Law <law@redhat.com>
2489 PR middle-end/52306
2490 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2491 when changing the SET_DEST of a prior insn to avoid an input reload.
2493 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2495 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2496 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2497 -mcall-openbsd, or -mcall-linux.
2498 (CC1_ENDIAN_BIG_SPEC): Remove.
2499 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2500 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2501 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2502 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2503 and %cc1_endian_default.
2504 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2506 2014-02-10 Richard Biener <rguenther@suse.de>
2508 PR tree-optimization/60115
2509 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2510 MEM_REF handling. Properly verify that the accesses are not
2511 out of the objects bound.
2513 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2515 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2516 coretex to cortex.
2518 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2520 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2521 proper constants and fix formatting.
2522 (possible_polymorphic_call_targets): Fix formatting.
2524 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2525 Ilya Tocar <ilya.tocar@intel.com>
2527 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2528 (_mm512_loadu_epi32): Renamed into...
2529 (_mm512_loadu_si512): This.
2530 (_mm512_storeu_epi32): Renamed into...
2531 (_mm512_storeu_si512): This.
2532 (_mm512_maskz_ceil_ps): Removed.
2533 (_mm512_maskz_ceil_pd): Ditto.
2534 (_mm512_maskz_floor_ps): Ditto.
2535 (_mm512_maskz_floor_pd): Ditto.
2536 (_mm512_floor_round_ps): Ditto.
2537 (_mm512_floor_round_pd): Ditto.
2538 (_mm512_ceil_round_ps): Ditto.
2539 (_mm512_ceil_round_pd): Ditto.
2540 (_mm512_mask_floor_round_ps): Ditto.
2541 (_mm512_mask_floor_round_pd): Ditto.
2542 (_mm512_mask_ceil_round_ps): Ditto.
2543 (_mm512_mask_ceil_round_pd): Ditto.
2544 (_mm512_maskz_floor_round_ps): Ditto.
2545 (_mm512_maskz_floor_round_pd): Ditto.
2546 (_mm512_maskz_ceil_round_ps): Ditto.
2547 (_mm512_maskz_ceil_round_pd): Ditto.
2548 (_mm512_expand_pd): Ditto.
2549 (_mm512_expand_ps): Ditto.
2550 * config/i386/i386.c (ix86_builtins): Remove
2551 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2552 (bdesc_args): Ditto.
2553 * config/i386/predicates.md (const1256_operand): New.
2554 (const_1_to_2_operand): Ditto.
2555 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2556 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2557 (*avx512pf_gatherpf<mode>sf): Ditto.
2558 (avx512pf_gatherpf<mode>df): Ditto.
2559 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2560 (*avx512pf_gatherpf<mode>df): Ditto.
2561 (avx512pf_scatterpf<mode>sf): Ditto.
2562 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2563 (*avx512pf_scatterpf<mode>sf): Ditto.
2564 (avx512pf_scatterpf<mode>df): Ditto.
2565 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2566 (*avx512pf_scatterpf<mode>df): Ditto.
2567 (avx512f_expand<mode>): Removed.
2568 (<shift_insn><mode>3<mask_name>): Change predicate type.
2570 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2572 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2573 not at the end of datarefs vector use ordered_remove to avoid
2574 reordering datarefs vector.
2576 PR c/59984
2577 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2578 mark local addressable non-static vars as GOVD_PRIVATE
2579 instead of GOVD_LOCAL.
2580 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2581 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2582 than copying them.
2584 PR middle-end/60092
2585 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2586 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2587 assume_aligned or alloc_align attributes.
2588 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2589 arguments. Handle also assume_aligned and alloc_align attributes.
2590 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2591 calls to functions with assume_aligned or alloc_align attributes.
2592 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2594 2014-02-08 Terry Guo <terry.guo@arm.com>
2596 * doc/invoke.texi: Document ARM -march=armv7e-m.
2598 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2600 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2601 flag on __cilkrts_rethrow builtin.
2603 PR ipa/60026
2604 * ipa-cp.c (determine_versionability): Fail at -O0
2605 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2606 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2608 Revert:
2609 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2611 PR ipa/60026
2612 * tree-inline.c (copy_forbidden): Fail for
2613 __attribute__((optimize (0))) functions.
2615 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2617 * varpool.c: Include pointer-set.h.
2618 (varpool_remove_unreferenced_decls): Variables in other partitions
2619 will not be output; be however careful to not lose information
2620 about partitioning.
2622 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2624 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2625 lookup in the vtable constructor.
2627 2014-02-07 Jeff Law <law@redhat.com>
2629 PR target/40977
2630 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2631 define_insn_and_split.
2633 * ipa-inline.c (inline_small_functions): Fix typos.
2635 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2637 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2638 (s390_can_use_return_insn): Declare.
2639 * config/s390/s390.h (EPILOGUE_USES): Define.
2640 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2641 instructions.
2642 (s390_chunkify_start): Handle return JUMP_LABELs.
2643 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2644 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2645 (s390_can_use_return_insn): New functions.
2646 (s390_fix_long_loop_prediction): Handle conditional returns.
2647 (TARGET_SET_UP_BY_PROLOGUE): Define.
2648 * config/s390/s390.md (ANY_RETURN): New code iterator.
2649 (*creturn, *csimple_return, return, simple_return): New patterns.
2651 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2653 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2654 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2655 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2656 REG_CFA_RESTORE list when deciding not to restore a register.
2658 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2660 * config/s390/s390.c: Include tree-pass.h and context.h.
2661 (s390_early_mach): New function, split out from...
2662 (s390_emit_prologue): ...here.
2663 (pass_data_s390_early_mach): New pass structure.
2664 (pass_s390_early_mach): New class.
2665 (s390_option_override): Create and register early_mach pass.
2666 Move to end of file.
2668 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2670 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2671 to match for the exit block.
2673 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2675 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2676 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2677 Reject misaligned operands.
2679 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2681 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2683 2014-02-07 Richard Biener <rguenther@suse.de>
2685 PR middle-end/60092
2686 * gimple-low.c (lower_builtin_posix_memalign): New function.
2687 (lower_stmt): Call it to lower posix_memalign in a way
2688 to make alignment info accessible.
2690 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2692 PR c++/60082
2693 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2694 __builtin_setjmp_receiver.
2696 2014-02-07 Richard Biener <rguenther@suse.de>
2698 PR middle-end/60092
2699 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2700 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2701 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2702 Handle BUILT_IN_POSIX_MEMALIGN.
2703 (find_func_clobbers): Likewise.
2704 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2705 (call_may_clobber_ref_p_1): Likewise.
2707 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2709 PR ipa/59918
2710 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2711 sanity check.
2713 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2715 PR ipa/59469
2716 * lto-cgraph.c (lto_output_node): Use
2717 symtab_get_symbol_partitioning_class.
2718 (lto_output_varpool_node): likewise.
2719 (symtab_get_symbol_partitioning_class): Move here from
2720 lto/lto-partition.c
2721 * cgraph.h (symbol_partitioning_class): Likewise.
2722 (symtab_get_symbol_partitioning_class): Declare.
2724 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2726 * ggc.h (ggc_internal_cleared_alloc): New macro.
2727 * vec.h (vec_safe_copy): Handle memory stats.
2728 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2729 * target-globals.c (save_target_globals): Likewise.
2731 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2733 PR target/60077
2734 * expr.c (emit_move_resolve_push): Export; be bit more selective
2735 on when to clear alias set.
2736 * expr.h (emit_move_resolve_push): Declare.
2737 * function.h (struct function): Add tail_call_marked.
2738 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2739 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2740 * config/i386/i386.md (TImode move expander): De not call
2741 ix86_expand_push.
2742 (FP push expanders): Preserve memory attributes.
2743 * config/i386/sse.md (push<mode>1): Remove.
2744 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2745 (ix86_expand_push): Remove.
2746 * config/i386/mmx.md (push<mode>1): Remove.
2748 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2750 PR rtl-optimization/60030
2751 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2752 lopart with paradoxical subreg before shifting it up by hprec.
2754 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2756 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2757 Remove extra newline at end of file.
2758 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2759 (arm_issue_rate): Handle cortexa57.
2760 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2761 (cortex-a57.cortex-a53): Likewise.
2763 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2765 PR target/59575
2766 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2767 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2768 bitmask.
2769 (arm_expand_prologue): Adjust all callers.
2770 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2771 info, registers also at the lowest numbered registers side. Use
2772 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2773 XEXP.
2775 PR debug/59992
2776 * var-tracking.c (adjust_mems): Before adding a SET to
2777 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2779 2014-02-06 Alan Modra <amodra@gmail.com>
2781 PR target/60032
2782 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2783 change SDmode to DDmode when lra_in_progress.
2785 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2787 PR middle-end/59150
2788 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2789 free_data_ref on the dr first, and before goto again also set dr
2790 to the next dr. For simd_lane_access, free old datarefs[i] before
2791 overwriting it. For get_vectype_for_scalar_type failure, don't
2792 free_data_ref if simd_lane_access.
2794 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2796 PR target/60062
2797 * tree.h (opts_for_fn): New inline function.
2798 (opt_for_fn): Define.
2799 * config/i386/i386.c (ix86_function_regparm): Use
2800 opt_for_fn (decl, optimize) instead of optimize.
2802 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2804 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2805 for SYMBOL_REF in large memory model.
2807 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2809 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2810 and crypto support.
2811 (cortex-a57): Likewise.
2812 (cortex-a57.cortex-a53): Likewise.
2814 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2815 Kugan Vivekanandarajah <kuganv@linaro.org>
2817 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2818 unaligned_access.
2819 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2821 2014-02-06 Richard Biener <rguenther@suse.de>
2823 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2824 set_loop_copy and initialize_original_copy_tables.
2826 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2828 * config/aarch64/aarch64-simd.md
2829 (aarch64_ashr_simddi): Change QI to SI.
2831 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2832 Jakub Jelinek <jakub@redhat.com>
2834 PR middle-end/60013
2835 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2836 of the dataflow.
2838 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2840 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2841 CODE_FOR_altivec_vpku[hw]um to
2842 CODE_FOR_altivec_vpku[hw]um_direct.
2843 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2844 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2845 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2846 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2848 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2850 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2851 generation for -maltivec=be.
2852 (altivec_vsumsws): Simplify redundant test.
2854 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2856 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2857 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2858 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2859 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2860 gen_altivec_vpkuwum.
2861 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2862 BYTES_BIG_ENDIAN.
2863 (altivec_vpks<VI_char>ss): Likewise.
2864 (altivec_vpks<VI_char>us): Likewise.
2865 (altivec_vpku<VI_char>us): Likewise.
2866 (altivec_vpku<VI_char>um): Likewise.
2867 (altivec_vpku<VI_char>um_direct): New (copy of
2868 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2869 internal use).
2870 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2871 target is little endian and -maltivec=be is not specified.
2872 (*altivec_vupkhs<VU_char>_direct): New (copy of
2873 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2874 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2875 target is little endian and -maltivec=be is not specified.
2876 (*altivec_vupkls<VU_char>_direct): New (copy of
2877 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2878 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2879 little endian and -maltivec=be is not specified.
2880 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2881 little endian and -maltivec=be is not specified.
2883 2014-02-05 Richard Henderson <rth@redhat.com>
2885 PR debug/52727
2886 * combine-stack-adj.c: Revert r206943.
2887 * sched-int.h (struct deps_desc): Add last_args_size.
2888 * sched-deps.c (init_deps): Initialize it.
2889 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2890 contain REG_ARGS_SIZE notes.
2892 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2894 * lto-cgraph.c (asm_nodes_output): Make global.
2895 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2896 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2897 (driver_handle_option): Handle OPT_fwpa.
2899 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2901 PR ipa/59947
2902 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2903 a comment typo and formatting issue. If odr_hash hasn't been
2904 created, return vNULL and set *completep to false.
2906 PR middle-end/57499
2907 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2908 bb with no successors.
2910 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2912 PR target/59718
2913 * doc/invoke.texi (-march): Clarify documentation for ARM.
2914 (-mtune): Likewise.
2915 (-mcpu): Likewise.
2917 2014-02-05 Richard Biener <rguenther@suse.de>
2919 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2920 when not vectorizing because of too many alias checks.
2921 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2922 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2924 2014-02-05 Nick Clifton <nickc@redhat.com>
2926 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2927 accept extended registers in any mode when compiling for the MN10300.
2929 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2931 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2932 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2933 sanitization attributes.
2934 (can_inline_edge_p): Likewise.
2935 (sanitize_attrs_match_for_inline_p): New function.
2937 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2939 * ipa-prop.c (detect_type_change): Shor circuit testing of
2940 type changes on THIS pointer.
2942 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2944 PR target/59777
2945 * config/pa/pa.c (legitimize_tls_address): Return original address
2946 if not passed a SYMBOL_REF rtx.
2947 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2948 addresses.
2949 (pa_emit_move_sequence): Simplify TLS source operands.
2950 (pa_legitimate_constant_p): Reject all TLS constants.
2951 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2952 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2954 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2956 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2957 groups when we know they are controlled by LTO.
2958 * varasm.c (default_binds_local_p_1): If object is in other partition,
2959 it will be resolved locally.
2961 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2963 * config/host-linux.c (linux_gt_pch_use_address): Don't
2964 use SSIZE_MAX because it is not always defined.
2966 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2968 PR bootstrap/59913
2969 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2970 threshold for pseudo splitting.
2971 (update_ebb_live_info): Process call argument hard registers and
2972 hard registers from insn definition too.
2973 (max_small_class_regs_num): New constant.
2974 (inherit_in_ebb): Update live hard regs through EBBs. Update
2975 reloads_num only for small register classes. Don't split for
2976 outputs of jumps.
2978 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2980 PR ipa/60058
2981 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2982 is non-null.
2984 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2986 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2987 visibility is safe.
2989 2014-02-04 Marek Polacek <polacek@redhat.com>
2991 * gdbinit.in (pel): Define.
2993 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2995 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2996 behavior.
2998 2014-02-04 Richard Biener <rguenther@suse.de>
3000 PR lto/59723
3001 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3002 in function context local.
3003 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3004 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3005 similar to LTO_imported_decl_ref.
3007 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3009 PR tree-optimization/60002
3010 * cgraphclones.c (build_function_decl_skip_args): Clear
3011 DECL_LANG_SPECIFIC.
3013 PR tree-optimization/60023
3014 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3015 false to gsi_replace.
3016 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3017 has been in some EH region and vec_stmt could throw, add
3018 vec_stmt into the same EH region.
3019 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3020 has no lhs, ignore it.
3021 * internal-fn.c (expand_MASK_LOAD): Likewise.
3023 PR ipa/60026
3024 * tree-inline.c (copy_forbidden): Fail for
3025 __attribute__((optimize (0))) functions.
3027 PR other/58712
3028 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3029 is set, copy one less argument.
3030 (expand_simd_clones): Don't subtract clone_info->inbranch
3031 from simd_clone_struct_alloc argument.
3033 PR rtl-optimization/57915
3034 * recog.c (simplify_while_replacing): If all unary/binary/relational
3035 operation arguments are constant, attempt to simplify those.
3037 PR middle-end/59261
3038 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3039 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3041 2014-02-04 Richard Biener <rguenther@suse.de>
3043 PR tree-optimization/60012
3044 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3045 TBAA disambiguation to all DDRs.
3047 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3049 PR target/59788
3050 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3051 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3053 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3055 PR ipa/59882
3056 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3058 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3060 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3061 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3063 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3065 PR ipa/59831
3066 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3067 to figure out targets of polymorphic calls with known decl.
3068 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3069 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3070 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3071 (get_polymorphic_call_info): ... here.
3072 (get_polymorphic_call_info_from_invariant): New function.
3074 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3076 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3077 lookup via vtable pointer; check for type consistency
3078 and turn inconsitent facts into UNREACHABLE.
3079 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3080 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3081 type inconsistent querries; return UNREACHABLE instead.
3083 2014-02-03 Richard Henderson <rth@twiddle.net>
3085 PR tree-opt/59924
3086 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3087 already processed this node.
3088 (normalize_one_pred_1): Pass along mark_set.
3089 (normalize_one_pred): Create and destroy a pointer_set_t.
3090 (normalize_one_pred_chain): Likewise.
3092 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3094 PR gcov-profile/58602
3095 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3097 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3099 PR ipa/59831
3100 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3101 -fno-devirtualize; try to devirtualize by the knowledge of
3102 virtual table pointer given by aggregate propagation.
3103 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3104 (ipa_print_node_jump_functions): Dump also offset that
3105 is relevant for polymorphic calls.
3106 (determine_known_aggregate_parts): Add arg_type parameter; use it
3107 instead of determining the type from pointer type.
3108 (ipa_compute_jump_functions_for_edge): Update call of
3109 determine_known_aggregate_parts.
3110 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3111 (gimple_get_virt_method_for_binfo): ... here; simplify using
3112 vtable_pointer_value_to_vtable.
3113 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3114 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3115 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3116 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3117 POINTER_PLUS_EXPR.
3118 (vtable_pointer_value_to_binfo): ... here.
3119 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3121 2014-02-03 Teresa Johnson <tejohnson@google.com>
3123 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3124 redef of outer loop index variable.
3126 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3128 PR c++/53017
3129 PR c++/59211
3130 * doc/extend.texi (Function Attributes): Typo.
3132 2014-02-03 Cong Hou <congh@google.com>
3134 PR tree-optimization/60000
3135 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3136 if the vectorized statement is a store. A store statement can only
3137 appear at the end of pattern statements.
3139 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3141 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3142 (ix86_option_override_internal): Default long double to 64-bit for
3143 32-bit Bionic and to 128-bit for 64-bit Bionic.
3145 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3146 TARGET_LONG_DOUBLE_128 is true.
3147 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3149 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3150 (mlong-double-64): Negate -mlong-double-128.
3151 (mlong-double-128): New option.
3153 * config/i386/i386-c.c (ix86_target_macros): Define
3154 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3156 * doc/invoke.texi: Document -mlong-double-128.
3158 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3160 PR rtl-optimization/60024
3161 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3163 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3165 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3167 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3169 PR rtl-optimization/57662
3170 * sel-sched.c (code_motion_path_driver): Do not mark already not
3171 existing blocks in the visiting bitmap.
3173 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3175 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3176 on the insn being emitted.
3178 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3179 Will Deacon <will.deacon@arm.com>
3181 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3183 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3185 * config/arm/arm-tables.opt: Regenerate.
3187 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3189 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3190 for vector types other than V16QImode.
3191 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3192 define_expand, and call altivec_expand_vec_perm_le when producing
3193 code with little endian element order.
3194 (*altivec_vperm_<mode>_internal): New insn having previous
3195 behavior of altivec_vperm_<mode>.
3196 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3197 altivec_expand_vec_perm_le when producing code with little endian
3198 element order.
3199 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3200 behavior of altivec_vperm_<mode>_uns.
3202 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3204 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3205 (altivec_vsumsws): Add handling for -maltivec=be with a little
3206 endian target.
3207 (altivec_vsumsws_direct): New.
3208 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3209 gen_altivec_vsumsws.
3211 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3213 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3214 vtable_pointer_value_to_binfo): New functions.
3215 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3216 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3218 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3220 * config/nios2/nios2.md (load_got_register): Initialize GOT
3221 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3222 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3224 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3226 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3227 preserverd by passthrough, do not propagate the type.
3229 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3231 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3232 (mips_atomic_assign_expand_fenv): New function.
3233 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3235 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3237 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3238 (__builtin_mips_set_fcsr): Likewise.
3239 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3240 MIPS_USI_FTYPE_VOID.
3241 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3242 (mips16_expand_set_fcsr): Likewise.
3243 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3244 (mips16_set_fcsr_stub): Likewise.
3245 (mips16_get_fcsr_one_only_stub): New class.
3246 (mips16_set_fcsr_one_only_stub): Likewise.
3247 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3248 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3249 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3250 (hard_float): New availability predicate.
3251 (mips_builtins): Add get_fcsr and set_fcsr.
3252 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3253 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3254 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3255 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3256 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3257 patterns.
3259 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3261 * config/mips/mips.c (mips_one_only_stub): New class.
3262 (mips_need_mips16_rdhwr_p): Replace with...
3263 (mips16_rdhwr_stub): ...this new variable.
3264 (mips16_stub_call_address): New function.
3265 (mips16_rdhwr_one_only_stub): New class.
3266 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3267 (mips_output_mips16_rdhwr): Delete.
3268 (mips_finish_stub): New function.
3269 (mips_code_end): Use it to handle rdhwr stubs.
3271 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3273 PR target/60017
3274 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3275 when calculating size of integer atomic types.
3277 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3279 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3281 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3283 PR tree-optimization/60003
3284 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3285 * profile.c (branch_prob): Use gimple_call_builtin_p
3286 to check for BUILT_IN_SETJMP_RECEIVER.
3287 * tree-inline.c (copy_bb): Call notice_special_calls.
3289 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3291 PR bootstrap/59985
3292 * lra-constraints.c (process_alt_operands): Update reload_sum only
3293 on the first pass.
3295 2014-01-31 Richard Henderson <rth@redhat.com>
3297 PR middle-end/60004
3298 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3299 until after else_eh is processed.
3301 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3303 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3304 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3305 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3306 in smmintrin.h, remove them.
3307 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3308 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3309 * config/i386/i386.md (ROUND_SAE): Fix value.
3310 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3311 (const48_operand): New.
3312 * config/i386/subst.md (round), (round_expand): Use
3313 const_4_or_8_to_11_operand.
3314 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3316 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3318 * config/i386/constraints.md (Yk): Swap meaning with k.
3319 * config/i386/i386.md (movhi_internal): Change Yk to k.
3320 (movqi_internal): Ditto.
3321 (*k<logic><mode>): Ditto.
3322 (*andhi_1): Ditto.
3323 (*andqi_1): Ditto.
3324 (kandn<mode>): Ditto.
3325 (*<code>hi_1): Ditto.
3326 (*<code>qi_1): Ditto.
3327 (kxnor<mode>): Ditto.
3328 (kortestzhi): Ditto.
3329 (kortestchi): Ditto.
3330 (kunpckhi): Ditto.
3331 (*one_cmplhi2_1): Ditto.
3332 (*one_cmplqi2_1): Ditto.
3333 * config/i386/sse.md (): Change k to Yk.
3334 (avx512f_load<mode>_mask): Ditto.
3335 (avx512f_blendm<mode>): Ditto.
3336 (avx512f_store<mode>_mask): Ditto.
3337 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3338 (avx512f_storedqu<mode>_mask): Ditto.
3339 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3340 Ditto.
3341 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3342 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3343 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3344 (avx512f_maskcmp<mode>3): Ditto.
3345 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3346 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3347 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3348 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3349 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3350 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3351 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3352 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3353 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3354 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3355 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3356 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3357 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3358 (vec_extract_lo_<mode>_maskm): Ditto.
3359 (vec_extract_hi_<mode>_maskm): Ditto.
3360 (avx512f_vternlog<mode>_mask): Ditto.
3361 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3362 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3363 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3364 (avx512f_<code>v8div16qi2_mask): Ditto.
3365 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3366 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3367 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3368 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3369 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3370 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3371 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3372 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3373 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3374 (avx512cd_maskb_vec_dupv8di): Ditto.
3375 (avx512cd_maskw_vec_dupv16si): Ditto.
3376 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3377 (avx512f_vpermi2var<mode>3_mask): Ditto.
3378 (avx512f_vpermi2var<mode>3_mask): Ditto.
3379 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3380 (*avx512f_gathersi<mode>): Ditto.
3381 (*avx512f_gathersi<mode>_2): Ditto.
3382 (*avx512f_gatherdi<mode>): Ditto.
3383 (*avx512f_gatherdi<mode>_2): Ditto.
3384 (*avx512f_scattersi<mode>): Ditto.
3385 (*avx512f_scatterdi<mode>): Ditto.
3386 (avx512f_compress<mode>_mask): Ditto.
3387 (avx512f_compressstore<mode>_mask): Ditto.
3388 (avx512f_expand<mode>_mask): Ditto.
3389 * config/i386/subst.md (mask): Change k to Yk.
3390 (mask_scalar_merge): Ditto.
3391 (sd): Ditto.
3393 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3395 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3397 2014-01-31 Richard Biener <rguenther@suse.de>
3399 PR middle-end/59990
3400 * builtins.c (fold_builtin_memory_op): Make sure to not
3401 use a floating-point mode or a boolean or enumeral type for
3402 the copy operation.
3404 2014-01-30 DJ Delorie <dj@redhat.com>
3406 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3407 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3408 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3409 whenever main() has an epilogue.
3411 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3413 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3414 unused variable "field".
3415 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3416 (vsx_mergeh_<mode>): Likewise.
3417 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3418 (altivec_vmrghh): Likewise.
3419 (altivec_vmrghw): Likewise.
3420 (altivec_vmrglb): Likewise.
3421 (altivec_vmrglh): Likewise.
3422 (altivec_vmrglw): Likewise.
3423 (altivec_vspltb): Add missing uses.
3424 (altivec_vsplth): Likewise.
3425 (altivec_vspltw): Likewise.
3426 (altivec_vspltsf): Likewise.
3428 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3430 PR target/59923
3431 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3432 frame related instructions.
3434 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3436 PR rtl-optimization/59959
3437 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3438 any reload of register whose subreg is invalid.
3440 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3442 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3443 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3444 Add missing return type - void.
3446 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3448 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3449 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3450 remove element index adjustment for endian (now handled in vsx.md
3451 and altivec.md).
3452 (altivec_expand_vec_perm_const): Use
3453 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3454 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3455 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3456 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3457 define_expand and a new define_insn *altivec_vspltb_internal;
3458 adjust for -maltivec=be on a little endian target.
3459 (altivec_vspltb_direct): New.
3460 (altivec_vsplth): Divide into a define_expand and a new
3461 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3462 little endian target.
3463 (altivec_vsplth_direct): New.
3464 (altivec_vspltw): Divide into a define_expand and a new
3465 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3466 little endian target.
3467 (altivec_vspltw_direct): New.
3468 (altivec_vspltsf): Divide into a define_expand and a new
3469 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3470 a little endian target.
3472 2014-01-30 Richard Biener <rguenther@suse.de>
3474 PR tree-optimization/59993
3475 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3476 can propagate form the earlier stmt and avoid the transform
3477 when the intermediate result is needed.
3479 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3481 * README.Portability: Fix typo.
3483 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3485 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3486 comparison_operator with ordered_comparison_operator.
3488 2014-01-30 Nick Clifton <nickc@redhat.com>
3490 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3491 Rename to mn10300_store_multiple_regs.
3492 * config/mn10300/mn10300.c: Likewise.
3493 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3494 store_multiple_regs.
3495 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3496 Call mn10300_store_multiple_regs.
3498 2014-01-30 Nick Clifton <nickc@redhat.com>
3499 DJ Delorie <dj@redhat.com>
3501 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3502 %fp 2 to keep registers after it properly word-aligned.
3503 (rl78_alloc_physical_registers_umul): Handle the case where both
3504 input operands are the same.
3506 2014-01-30 Richard Biener <rguenther@suse.de>
3508 PR tree-optimization/59903
3509 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3510 check properly.
3512 2014-01-30 Jason Merrill <jason@redhat.com>
3514 PR c++/59633
3515 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3517 PR c++/59645
3518 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3520 2014-01-30 Richard Biener <rguenther@suse.de>
3522 PR tree-optimization/59951
3523 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3525 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3527 PR target/59784
3528 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3529 SFmode to DFmode case.
3531 2014-01-29 DJ Delorie <dj@redhat.com>
3533 * config/msp430/msp430.opt (-minrt): New.
3534 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3535 if -minrt given.
3536 (ENDFILE_SPEC): Likewise.
3538 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3540 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3541 (estimate_function_body_sizes): Use it.
3543 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3545 PR c++/58561
3546 * dwarf2out.c (is_cxx_auto): New.
3547 (is_base_type): Use it.
3548 (gen_type_die_with_usage): Likewise.
3550 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3552 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3553 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3554 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3555 -maltivec=be with LE targets.
3556 (vsx_mergeh_<mode>): Likewise.
3557 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3558 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3559 (altivec_vmrghb): Replace with define_expand and new
3560 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3561 (altivec_vmrghb_direct): New define_insn.
3562 (altivec_vmrghh): Replace with define_expand and new
3563 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3564 (altivec_vmrghh_direct): New define_insn.
3565 (altivec_vmrghw): Replace with define_expand and new
3566 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3567 (altivec_vmrghw_direct): New define_insn.
3568 (*altivec_vmrghsf): Adjust for endianness.
3569 (altivec_vmrglb): Replace with define_expand and new
3570 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3571 (altivec_vmrglb_direct): New define_insn.
3572 (altivec_vmrglh): Replace with define_expand and new
3573 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3574 (altivec_vmrglh_direct): New define_insn.
3575 (altivec_vmrglw): Replace with define_expand and new
3576 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3577 (altivec_vmrglw_direct): New define_insn.
3578 (*altivec_vmrglsf): Adjust for endianness.
3579 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3580 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3581 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3582 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3583 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3584 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3585 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3586 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3588 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3590 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3591 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3592 whitespace.
3594 2014-01-29 Richard Biener <rguenther@suse.de>
3596 PR tree-optimization/58742
3597 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3598 associate_pointerplus_align.
3599 (associate_pointerplus_diff): New function.
3600 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3601 and associate_pointerplus_diff.
3603 2014-01-29 Richard Biener <rguenther@suse.de>
3605 * lto-streamer.h (LTO_major_version): Bump to 3.
3606 (LTO_minor_version): Reset to 0.
3608 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3610 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3611 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3612 (arm_file_start): Generate correct asm header for armv7ve.
3613 * config/arm/bpabi.h: Add multilib support for armv7ve.
3614 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3615 and cortex-a15 to armv7ve.
3616 * config/arm/t-aprofile: Add multilib support for armv7ve.
3617 * doc/invoke.texi: Document -march=armv7ve.
3619 2014-01-29 Richard Biener <rguenther@suse.de>
3621 PR tree-optimization/58742
3622 * tree-ssa-forwprop.c (associate_plusminus): Return true
3623 if we changed sth, defer EH cleanup to ...
3624 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3625 (simplify_mult): New function.
3627 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3629 PR middle-end/59917
3630 PR tree-optimization/59920
3631 * tree.c (build_common_builtin_nodes): Remove
3632 __builtin_setjmp_dispatcher initialization.
3633 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3634 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3635 instead of gsi_after_labels + manually skipping debug stmts.
3636 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3637 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3638 * tree-inline.c (copy_edges_for_bb): Remove
3639 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3640 argument. Ignore computed_goto_p stmts. Don't call
3641 make_abnormal_goto_edges. If a call might need abnormal edges
3642 for non-local gotos, see if it already has an edge to
3643 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3644 with true argument, don't do anything then, otherwise add
3645 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3646 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3647 caller.
3648 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3649 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3650 (lower_stmt): Don't set data->calls_builtin_setjmp.
3651 (lower_builtin_setjmp): Adjust comment.
3652 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3653 * tree-cfg.c (found_computed_goto): Remove.
3654 (factor_computed_gotos): Remove.
3655 (make_goto_expr_edges): Return bool, true for computed gotos.
3656 Don't call make_abnormal_goto_edges.
3657 (build_gimple_cfg): Don't set found_computed_goto, don't call
3658 factor_computed_gotos.
3659 (computed_goto_p): No longer static.
3660 (make_blocks): Don't set found_computed_goto.
3661 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3662 (make_edges): If make_goto_expr_edges returns true, push bb
3663 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3664 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3665 vector. Record mapping between bbs and OpenMP regions if there
3666 are any, adjust make_gimple_omp_edges caller. Call
3667 handle_abnormal_edges.
3668 (make_abnormal_goto_edges): Remove.
3669 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3670 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3671 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3672 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3673 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3674 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3675 filling *region also set *region_idx to (*region)->entry->index.
3677 PR other/58712
3678 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3679 For REGs set ORIGINAL_REGNO.
3681 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3683 * doc/md.texi: Mention that a target shouldn't implement
3684 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3685 than hi/lo pair.
3687 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3689 PR tree-optimization/59594
3690 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3691 a copy of the datarefs vector rather than the vector itself.
3693 2014-01-28 Jason Merrill <jason@redhat.com>
3695 PR c++/53756
3696 * dwarf2out.c (auto_die): New static.
3697 (gen_type_die_with_usage): Handle C++1y 'auto'.
3698 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3699 on definition.
3701 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3703 PR target/59672
3704 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3705 (SPEC_X32): Likewise.
3706 (SPEC_64): Likewise.
3707 * config/i386/i386.c (ix86_option_override_internal): Turn off
3708 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3709 for TARGET_16BIT.
3710 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3711 * config/i386/i386.h (TARGET_16BIT): New macro.
3712 (TARGET_16BIT_P): Likewise.
3713 * config/i386/i386.opt: Add m16.
3714 * doc/invoke.texi: Document -m16.
3716 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3718 PR preprocessor/59935
3719 * input.c (location_get_source_line): Bail out on when line number
3720 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3722 2014-01-28 Richard Biener <rguenther@suse.de>
3724 PR tree-optimization/58742
3725 * tree-ssa-forwprop.c (associate_plusminus): Handle
3726 pointer subtraction of the form (T)(P + A) - (T)P.
3728 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3730 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3731 at const_int_cost.
3733 2014-01-28 Richard Biener <rguenther@suse.de>
3735 Revert
3736 2014-01-28 Richard Biener <rguenther@suse.de>
3738 PR rtl-optimization/45364
3739 PR rtl-optimization/59890
3740 * var-tracking.c (local_get_addr_clear_given_value): Handle
3741 already cleared slot.
3742 (val_reset): Handle not allocated local_get_addr_cache.
3743 (vt_find_locations): Use post-order on the inverted CFG.
3745 2014-01-28 Richard Biener <rguenther@suse.de>
3747 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3749 2014-01-28 Richard Biener <rguenther@suse.de>
3751 PR rtl-optimization/45364
3752 PR rtl-optimization/59890
3753 * var-tracking.c (local_get_addr_clear_given_value): Handle
3754 already cleared slot.
3755 (val_reset): Handle not allocated local_get_addr_cache.
3756 (vt_find_locations): Use post-order on the inverted CFG.
3758 2014-01-28 Alan Modra <amodra@gmail.com>
3760 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3761 * configure.ac <recursive call for build != host>: Define
3762 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3763 and LD_FOR_BUILD too.
3764 * configure: Regenerate.
3766 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3768 * config/i386/i386.c (get_builtin_code_for_version): Separate
3769 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3770 Broadwell from Haswell.
3772 2014-01-27 Steve Ellcey <sellcey@mips.com>
3774 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3775 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3776 * config/mips/mips.c (mips_option_override): Change setting
3777 of TARGET_DSP.
3778 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3779 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3780 Change from Mask to Var.
3782 2014-01-27 Jeff Law <law@redhat.com>
3784 * ipa-inline.c (inline_small_functions): Fix typo.
3786 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3788 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3789 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3790 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3791 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3792 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3793 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3794 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3795 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3796 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3797 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3798 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3799 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3800 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3801 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3802 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3803 (_mm512_storeu_epi64): Ditto.
3804 (_mm512_cmpge_epi32_mask): Ditto.
3805 (_mm512_cmpge_epu32_mask): Ditto.
3806 (_mm512_cmpge_epi64_mask): Ditto.
3807 (_mm512_cmpge_epu64_mask): Ditto.
3808 (_mm512_cmple_epi32_mask): Ditto.
3809 (_mm512_cmple_epu32_mask): Ditto.
3810 (_mm512_cmple_epi64_mask): Ditto.
3811 (_mm512_cmple_epu64_mask): Ditto.
3812 (_mm512_cmplt_epi32_mask): Ditto.
3813 (_mm512_cmplt_epu32_mask): Ditto.
3814 (_mm512_cmplt_epi64_mask): Ditto.
3815 (_mm512_cmplt_epu64_mask): Ditto.
3816 (_mm512_cmpneq_epi32_mask): Ditto.
3817 (_mm512_cmpneq_epu32_mask): Ditto.
3818 (_mm512_cmpneq_epi64_mask): Ditto.
3819 (_mm512_cmpneq_epu64_mask): Ditto.
3820 (_mm512_expand_pd): Ditto.
3821 (_mm512_expand_ps): Ditto.
3822 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3823 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3824 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3825 * config/i386/i386.c (ix86_builtins): Add
3826 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3827 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3828 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3829 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3830 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3831 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3832 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3833 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3834 IX86_BUILTIN_PMOVUSQW512_MEM.
3835 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3836 __builtin_ia32_pmovsqd512mem_mask,
3837 __builtin_ia32_pmovqd512mem_mask,
3838 __builtin_ia32_pmovusqw512mem_mask,
3839 __builtin_ia32_pmovsqw512mem_mask,
3840 __builtin_ia32_pmovqw512mem_mask,
3841 __builtin_ia32_pmovusdw512mem_mask,
3842 __builtin_ia32_pmovsdw512mem_mask,
3843 __builtin_ia32_pmovdw512mem_mask,
3844 __builtin_ia32_pmovqb512mem_mask,
3845 __builtin_ia32_pmovusqb512mem_mask,
3846 __builtin_ia32_pmovsqb512mem_mask,
3847 __builtin_ia32_pmovusdb512mem_mask,
3848 __builtin_ia32_pmovsdb512mem_mask,
3849 __builtin_ia32_pmovdb512mem_mask.
3850 (bdesc_args): Add __builtin_ia32_expanddf512,
3851 __builtin_ia32_expandsf512.
3852 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3853 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3854 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3855 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3856 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3857 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3858 (avx512f_<code>v8div16qi2_mask_store): This.
3859 (avx512f_expand<mode>): New.
3861 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3863 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3864 New.
3865 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3866 (_mm512_prefetch_i32scatter_pd): Ditto.
3867 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3868 (_mm512_prefetch_i64scatter_pd): Ditto.
3869 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3870 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3871 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3872 (_mm512_prefetch_i32scatter_ps): Ditto.
3873 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3874 (_mm512_prefetch_i64scatter_ps): Ditto.
3875 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3876 * config/i386/i386-builtin-types.def: Define
3877 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3878 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3879 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3880 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3881 IX86_BUILTIN_SCATTERPFQPD.
3882 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3883 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3884 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3885 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3886 __builtin_ia32_scatterpfqps.
3887 (ix86_expand_builtin): Expand new built-ins.
3888 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3889 fix memory access data type.
3890 (*avx512pf_gatherpf<mode>_mask): Ditto.
3891 (*avx512pf_gatherpf<mode>): Ditto.
3892 (avx512pf_scatterpf<mode>): Ditto.
3893 (*avx512pf_scatterpf<mode>_mask): Ditto.
3894 (*avx512pf_scatterpf<mode>): Ditto.
3895 (GATHER_SCATTER_SF_MEM_MODE): New.
3896 (avx512pf_gatherpf<mode>df): Ditto.
3897 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3898 (*avx512pf_scatterpf<mode>df): Ditto.
3900 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3902 PR bootstrap/59934
3903 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3904 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3905 reached.
3907 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3909 * common/config/arm/arm-common.c
3910 (arm_rewrite_mcpu): Handle multiple names.
3911 * config/arm/arm.h
3912 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3914 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3916 * gimple-builder.h (create_gimple_tmp): Delete.
3918 2014-01-27 Christian Bruel <christian.bruel@st.com>
3920 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3921 words comparisons.
3923 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3925 * config/pa/pa.md (call): Generate indirect long calls to non-local
3926 functions when outputing 32-bit code.
3927 (call_value): Likewise except for special call to buggy powf function.
3929 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3930 portable runtime and PIC indirect calls.
3931 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3932 and PIC call sequences. Use ldo instead of blr to set return register
3933 in PIC call sequence.
3935 2014-01-25 Walter Lee <walt@tilera.com>
3937 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3938 avoid clobbering a live register.
3940 2014-01-25 Walter Lee <walt@tilera.com>
3942 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3943 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3944 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3945 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3947 2014-01-25 Walter Lee <walt@tilera.com>
3949 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3950 arguments on even registers.
3951 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3952 STACK_BOUNDARY.
3953 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3954 (BIGGEST_ALIGNMENT): Ditto.
3955 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3957 2014-01-25 Walter Lee <walt@tilera.com>
3959 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3960 insns before bundling.
3961 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3963 2014-01-25 Walter Lee <walt@tilera.com>
3965 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3966 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3967 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3969 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3971 * config/mips/constraints.md (kl): Delete.
3972 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3973 define expands, using...
3974 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3975 instructions for MIPS16.
3976 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3977 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3979 2014-01-25 Walter Lee <walt@tilera.com>
3981 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3982 (clzdi2): Ditto.
3983 (ffsdi2): Ditto.
3985 2014-01-25 Walter Lee <walt@tilera.com>
3987 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3988 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3990 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3992 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3993 Handle XOR.
3995 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3997 * print-rtl.c (in_call_function_usage): New var.
3998 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3999 EXPR_LIST mode as mode and not as reg note name.
4001 PR middle-end/59561
4002 * cfgloopmanip.c (copy_loop_info): If
4003 loop->warned_aggressive_loop_optimizations, make sure
4004 the flag is set in target loop too.
4006 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4008 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4009 flag_cilkplus.
4010 * builtins.def: Likewise.
4011 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4012 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4013 * ira.c (ira_setup_eliminable_regset): Likewise.
4014 * omp-low.c (gate_expand_omp): Likewise.
4015 (execute_lower_omp): Likewise.
4016 (diagnose_sb_0): Likewise.
4017 (gate_diagnose_omp_blocks): Likewise.
4018 (simd_clone_clauses_extract): Likewise.
4019 (gate): Likewise.
4021 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4023 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4024 correction for little endian...
4025 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4026 here.
4028 2014-01-24 Jeff Law <law@redhat.com>
4030 PR tree-optimization/59919
4031 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4032 for non-returning calls.
4034 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4036 * common/config/aarch64/aarch64-common.c
4037 (aarch64_rewrite_mcpu): Handle multiple names.
4038 * config/aarch64/aarch64.h
4039 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4041 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4043 * input.c (add_file_to_cache_tab): Handle the case where fopen
4044 returns NULL.
4046 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4048 PR target/59929
4049 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4050 from push operand if code of push isn't PRE_DEC.
4052 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4054 PR target/59909
4055 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4056 -mquad-memory-atomic. Update -mquad-memory documentation to say
4057 it is only used for non-atomic loads/stores.
4059 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4060 -mquad-memory or -mquad-memory-atomic switches.
4062 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4063 -mquad-memory-atomic to ISA 2.07 support.
4065 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4066 to separate support of normal quad word memory operations (ldq, stq)
4067 from the atomic quad word memory operations.
4069 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4070 support to separate non-atomic quad word operations from atomic
4071 quad word operations. Disable non-atomic quad word operations in
4072 little endian mode so that we don't have to swap words after the
4073 load and before the store.
4074 (quad_load_store_p): Add comment about atomic quad word support.
4075 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4076 options printed with -mdebug=reg.
4078 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4079 -mquad-memory-atomic as the test for whether we have quad word
4080 atomic instructions.
4081 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4082 or -mp8-vector are used, allow byte/half-word atomic operations.
4084 * config/rs6000/sync.md (load_lockedti): Insure that the address
4085 is a proper indexed or indirect address for the lqarx instruction.
4086 On little endian systems, swap the hi/lo registers after the lqarx
4087 instruction.
4088 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4089 insure the address is valid for the lqarx instruction.
4090 (store_conditionalti): Insure that the address is a proper indexed
4091 or indirect address for the stqcrx. instruction. On little endian
4092 systems, swap the hi/lo registers before doing the stqcrx.
4093 instruction.
4094 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4095 insure the address is valid for the stqcrx. instruction.
4097 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4098 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4099 type of quad memory support is available.
4101 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4103 PR regression/59915
4104 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4105 there is a danger of looping.
4107 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4109 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4110 force flag_ira_loop_pressure if set via command line.
4112 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4114 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4115 (ashr_simd): New builtin handling DI mode.
4116 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4117 (aarch64_sshr_simddi): New match pattern.
4118 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4119 (vshrd_n_s64): Likewise.
4120 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4122 2014-01-23 Nick Clifton <nickc@redhat.com>
4124 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4125 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4126 favour of mcu specific scripts.
4127 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4128 430x multilibs.
4130 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4131 Alex Velenko <Alex.Velenko@arm.com>
4133 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4134 (vaddv_s16): Likewise.
4135 (vaddv_s32): Likewise.
4136 (vaddv_u8): Likewise.
4137 (vaddv_u16): Likewise.
4138 (vaddv_u32): Likewise.
4139 (vaddvq_s8): Likewise.
4140 (vaddvq_s16): Likewise.
4141 (vaddvq_s32): Likewise.
4142 (vaddvq_s64): Likewise.
4143 (vaddvq_u8): Likewise.
4144 (vaddvq_u16): Likewise.
4145 (vaddvq_u32): Likewise.
4146 (vaddvq_u64): Likewise.
4147 (vaddv_f32): Likewise.
4148 (vaddvq_f32): Likewise.
4149 (vaddvq_f64): Likewise.
4150 (vmaxv_f32): Likewise.
4151 (vmaxv_s8): Likewise.
4152 (vmaxv_s16): Likewise.
4153 (vmaxv_s32): Likewise.
4154 (vmaxv_u8): Likewise.
4155 (vmaxv_u16): Likewise.
4156 (vmaxv_u32): Likewise.
4157 (vmaxvq_f32): Likewise.
4158 (vmaxvq_f64): Likewise.
4159 (vmaxvq_s8): Likewise.
4160 (vmaxvq_s16): Likewise.
4161 (vmaxvq_s32): Likewise.
4162 (vmaxvq_u8): Likewise.
4163 (vmaxvq_u16): Likewise.
4164 (vmaxvq_u32): Likewise.
4165 (vmaxnmv_f32): Likewise.
4166 (vmaxnmvq_f32): Likewise.
4167 (vmaxnmvq_f64): Likewise.
4168 (vminv_f32): Likewise.
4169 (vminv_s8): Likewise.
4170 (vminv_s16): Likewise.
4171 (vminv_s32): Likewise.
4172 (vminv_u8): Likewise.
4173 (vminv_u16): Likewise.
4174 (vminv_u32): Likewise.
4175 (vminvq_f32): Likewise.
4176 (vminvq_f64): Likewise.
4177 (vminvq_s8): Likewise.
4178 (vminvq_s16): Likewise.
4179 (vminvq_s32): Likewise.
4180 (vminvq_u8): Likewise.
4181 (vminvq_u16): Likewise.
4182 (vminvq_u32): Likewise.
4183 (vminnmv_f32): Likewise.
4184 (vminnmvq_f32): Likewise.
4185 (vminnmvq_f64): Likewise.
4187 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4189 * config/aarch64/aarch64-simd.md
4190 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4191 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4192 (*aarch64_mul3_elt<mode>): Likewise.
4193 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4194 (*aarch64_mul3_elt_to_64v2df): Likewise.
4195 (*aarch64_mla_elt<mode>): Likewise.
4196 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4197 (*aarch64_mls_elt<mode>): Likewise.
4198 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4199 (*aarch64_fma4_elt<mode>): Likewise.
4200 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4201 (*aarch64_fma4_elt_to_64v2df): Likewise.
4202 (*aarch64_fnma4_elt<mode>): Likewise.
4203 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4204 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4205 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4206 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4207 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4208 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4209 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4210 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4211 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4213 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4215 * config/aarch64/aarch64-simd.md
4216 (aarch64_be_checked_get_lane<mode>): New define_expand.
4217 * config/aarch64/aarch64-simd-builtins.def
4218 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4219 New builtin definition.
4220 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4221 Use new safe be builtin.
4223 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4225 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4226 New define_insn.
4227 (aarch64_be_st1<mode>): Likewise.
4228 (aarch_ld1<VALL:mode>): Define_expand modified.
4229 (aarch_st1<VALL:mode>): Likewise.
4230 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4231 (UNSPEC_ST1): Likewise.
4233 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4235 * config/microblaze/microblaze.md: Add trap insn and attribute
4237 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4239 PR preprocessor/58580
4240 * input.h (location_get_source_line): Take an additional line_size
4241 parameter.
4242 (void diagnostics_file_cache_fini): Declare new function.
4243 * input.c (struct fcache): New type.
4244 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4245 New static constants.
4246 (diagnostic_file_cache_init, total_lines_num)
4247 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4248 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4249 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4250 (get_next_line, read_next_line, goto_next_line, read_line_num):
4251 New static function definitions.
4252 (diagnostic_file_cache_fini): New function.
4253 (location_get_source_line): Take an additional output line_len
4254 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4255 read_line_num.
4256 * diagnostic.c (diagnostic_finish): Call
4257 diagnostic_file_cache_fini.
4258 (adjust_line): Take an additional input parameter for the length
4259 of the line, rather than calculating it with strlen.
4260 (diagnostic_show_locus): Adjust the use of
4261 location_get_source_line and adjust_line with respect to their new
4262 signature. While displaying a line now, do not stop at the first
4263 null byte. Rather, display the zero byte as a space and keep
4264 going until we reach the size of the line.
4265 * Makefile.in: Add vec.o to OBJS-libcommon
4267 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4268 Ilya Tocar <ilya.tocar@intel.com>
4270 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4271 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4272 (__builtin_ia32_kmov16): Ditto.
4273 * config/i386/i386.md (UNSPEC_KMOV): New.
4274 (kmovw): Ditto.
4276 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4278 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4279 (_mm512_storeu_si512): Ditto.
4281 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4283 PR target/52125
4284 * rtl.h (get_referenced_operands): Declare.
4285 * recog.c (get_referenced_operands): New function.
4286 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4287 operands have been referenced when recording LO_SUM references.
4289 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4291 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4293 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4295 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4296 Enable for generic and recent AMD targets.
4298 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4300 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4301 ARG_SIZE note when adjustment was eliminated.
4303 2014-01-22 Jeff Law <law@redhat.com>
4305 PR tree-optimization/59597
4306 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4307 in file. Accept new argument REGISTERING and use it to modify
4308 dump output appropriately.
4309 (register_jump_thread): Corresponding changes.
4310 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4311 thread paths involving joiner blocks. Add code to dump cancelled
4312 jump threading paths.
4314 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4316 PR rtl-optimization/59477
4317 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4318 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4320 2014-01-22 Tom Tromey <tromey@redhat.com>
4322 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4323 PARAMS.
4324 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4326 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4328 PR rtl-optimization/59896
4329 * lra-constraints.c (process_alt_operands): Check unused note for
4330 matched operands of insn with no output reloads.
4332 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4334 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4335 (mips_move_from_gpr_cost): Likewise.
4337 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4339 PR rtl-optimization/59858
4340 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4341 ira_class_hard_regs_num.
4342 (process_alt_operands): Increase reject for dying matched operand.
4344 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4346 PR target/59003
4347 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4348 smaller than size, perform several stores or loads and stores
4349 at dst + count - size to store or copy all of size bytes, rather
4350 than just last modesize bytes.
4352 2014-01-20 DJ Delorie <dj@redhat.com>
4354 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4355 that CLOBBERs are REGs before propogating their values.
4357 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4359 PR middle-end/59789
4360 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4361 (cgraph_inline_failed_type): New function.
4362 * cgraph.h (DEFCIFCODE): Add type.
4363 (cgraph_inline_failed_type_t): New enum.
4364 (cgraph_inline_failed_type): New prototype.
4365 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4366 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4367 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4368 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4369 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4370 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4371 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4372 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4373 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4374 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4375 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4376 OPTIMIZATION_MISMATCH.
4377 * tree-inline.c (expand_call_inline): Emit errors during
4378 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4380 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4382 PR target/59685
4383 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4384 mode attribute in insn output.
4386 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4388 * output.h (output_constant): Delete.
4389 * varasm.c (output_constant): Make private.
4391 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4393 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4395 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4397 PR middle-end/59860
4398 * tree.h (fold_builtin_strcat): New prototype.
4399 * builtins.c (fold_builtin_strcat): No longer static. Add len
4400 argument, if non-NULL, don't call c_strlen. Optimize
4401 directly into __builtin_memcpy instead of __builtin_strcpy.
4402 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4403 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4405 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4407 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4408 for SImode_address_operand operands, having only a REG argument.
4410 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4412 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4413 loader name using mbig-endian.
4414 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4416 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4418 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4419 (-mtune): Likewise.
4420 (-mcpu): Likewise.
4422 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4424 * config/aarch64/aarch64-protos.h
4425 (aarch64_cannot_change_mode_class_ptr): Declare.
4426 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4427 aarch64_cannot_change_mode_class_ptr): New.
4428 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4429 backend hook aarch64_cannot_change_mode_class.
4431 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4433 * common/config/aarch64/aarch64-common.c
4434 (aarch64_handle_option): Don't handle any option order logic here.
4435 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4436 selected_cpu, warn on architecture version mismatch.
4437 (aarch64_override_options): Fix parsing order for option strings.
4439 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4440 Iain Sandoe <iain@codesourcery.com>
4442 PR bootstrap/59496
4443 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4444 warning. Amend comment to reflect current functionality.
4446 2014-01-20 Richard Biener <rguenther@suse.de>
4448 PR middle-end/59860
4449 * builtins.c (fold_builtin_strcat): Remove case better handled
4450 by tree-ssa-strlen.c.
4452 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4454 * config/aarch64/aarch64.opt
4455 (mcpu, march, mtune): Make case-insensitive.
4457 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4459 PR target/59880
4460 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4461 if operands[1] is a REG or ZERO_EXTEND of a REG.
4463 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4465 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4467 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4469 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4470 long non-pic millicode calls.
4472 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4474 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4476 2014-01-19 Kito Cheng <kito@0xlab.org>
4478 * builtins.c (expand_movstr): Check movstr expand done or fail.
4480 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4481 H.J. Lu <hongjiu.lu@intel.com>
4483 PR target/59379
4484 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4485 to DImode for zero-extended addresses.
4487 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4489 PR rtl-optimization/57763
4490 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4491 on the new indirect jump_insn and increment LABEL_NUSES (label).
4493 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4495 PR bootstrap/59580
4496 PR bootstrap/59583
4497 * config.gcc (x86_archs): New variable.
4498 (x86_64_archs): Likewise.
4499 (x86_cpus): Likewise.
4500 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4501 --with-arch/--with-cpu= options.
4502 Support --with-arch=/--with-cpu={nehalem,westmere,
4503 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4505 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4507 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4508 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4510 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4512 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4514 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4516 PR target/58944
4517 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4518 clear cpp_get_options (parse_in)->warn_unused_macros for
4519 ix86_target_macros_internal with cpp_define.
4521 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4523 * jump.c (delete_related_insns): Keep (use (insn))s.
4524 * reorg.c (redundant_insn): Check for barriers too.
4526 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4528 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4530 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4532 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4533 call to $$dyncall when TARGET_LONG_CALLS is true.
4535 2014-01-17 Jeff Law <law@redhat.com>
4537 * ree.c (combine_set_extension): Temporarily disable test for
4538 changing number of hard registers.
4540 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4542 PR middle-end/58125
4543 * ipa-inline-analysis.c (inline_free_summary):
4544 Do not free summary of aliases.
4546 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4548 PR middle-end/59706
4549 * gimplify.c (gimplify_expr): Use create_tmp_var
4550 instead of create_tmp_var_raw. If cond doesn't have
4551 integral type, don't add the IFN_ANNOTATE builtin at all.
4553 2014-01-17 Martin Jambor <mjambor@suse.cz>
4555 PR ipa/59736
4556 * ipa-cp.c (prev_edge_clone): New variable.
4557 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4558 Also resize prev_edge_clone vector.
4559 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4560 (ipcp_edge_removal_hook): New function.
4561 (ipcp_driver): Register ipcp_edge_removal_hook.
4563 2014-01-17 Andrew Pinski <apinski@cavium.com>
4564 Steve Ellcey <sellcey@mips.com>
4566 PR target/59462
4567 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4568 of operator mode.
4570 2014-01-17 Jeff Law <law@redhat.com>
4572 PR middle-end/57904
4573 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4574 so that pass_ccp runs first.
4576 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4578 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4579 (ix86_adjust_cost): Use !TARGET_XXX.
4580 (do_reorder_for_imul): Likewise.
4581 (swap_top_of_ready_list): Likewise.
4582 (ix86_sched_reorder): Likewise.
4584 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4586 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4587 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4588 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4589 (intel_memset): New. Duplicate slm_memset.
4590 (intel_cost): New. Duplicate slm_cost.
4591 (m_INTEL): New macro.
4592 (processor_target_table): Add "intel".
4593 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4594 with PROCESSOR_INTEL for "intel".
4595 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4596 PROCESSOR_SILVERMONT.
4597 (ix86_issue_rate): Likewise.
4598 (ix86_adjust_cost): Likewise.
4599 (ia32_multipass_dfa_lookahead): Likewise.
4600 (swap_top_of_ready_list): Likewise.
4601 (ix86_sched_reorder): Likewise.
4602 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4603 instead of TARGET_OPT_AGU.
4604 * config/i386/i386.h (TARGET_INTEL): New.
4605 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4606 (processor_type): Add PROCESSOR_INTEL.
4607 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4608 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4610 2014-01-17 Marek Polacek <polacek@redhat.com>
4612 PR c/58346
4613 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4614 size is zero.
4616 2014-01-17 Richard Biener <rguenther@suse.de>
4618 PR tree-optimization/46590
4619 * opts.c (default_options_table): Add entries for
4620 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4621 all enabled at -O1 but not for -Og.
4622 * common.opt (fbranch-count-reg): Remove Init(1).
4623 (fmove-loop-invariants): Likewise.
4624 (ftree-pta): Likewise.
4626 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4628 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4629 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4630 decls to at least the GCC 4.8 used alignments.
4632 PR fortran/59440
4633 * tree-nested.c (convert_nonlocal_reference_stmt,
4634 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4635 of GIMPLE_BIND stmts, adjust associated decls.
4637 2014-01-17 Richard Biener <rguenther@suse.de>
4639 PR tree-optimization/46590
4640 * vec.h (vec<>::bseach): New member function implementing
4641 binary search according to C89 bsearch.
4642 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4643 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4644 bitmap pointer again. Make accesses_in_loop a flat array.
4645 (mem_ref_obstack): New global.
4646 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4647 (mark_ref_stored): Likewise.
4648 (ref_indep_loop_p_2): Likewise.
4649 (set_ref_stored_in_loop): New helper function.
4650 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4651 (memref_free): Adjust.
4652 (record_mem_ref_loc): Simplify.
4653 (gather_mem_refs_stmt): Adjust.
4654 (sort_locs_in_loop_postorder_cmp): New function.
4655 (analyze_memory_references): Sort accesses_in_loop after
4656 loop postorder number.
4657 (find_ref_loc_in_loop_cmp): New function.
4658 (for_all_locs_in_loop): Find relevant cluster of locs in
4659 accesses_in_loop and iterate without recursion.
4660 (execute_sm): Avoid uninit warning.
4661 (struct ref_always_accessed): Simplify.
4662 (ref_always_accessed::operator ()): Likewise.
4663 (ref_always_accessed_p): Likewise.
4664 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4665 loop postorder numbers here.
4666 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4667 numbers.
4669 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4671 PR c++/57945
4672 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4673 on decls for which assemble_alias has been called.
4675 2014-01-17 Nick Clifton <nickc@redhat.com>
4677 * config/msp430/msp430.opt: (mcpu): New option.
4678 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4679 (msp430_option_override): Parse target_cpu. If the MCU name
4680 matches a generic string, clear target_mcu.
4681 (msp430_attr): Allow numeric interrupt values up to 63.
4682 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4683 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4684 option.
4685 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4686 Add mcpu matches.
4687 * config/msp430/msp430.md (popm): Use %J rather than %I.
4688 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4689 (addhi_cy_i): Use immediate_operand for operand 2.
4690 * doc/invoke.texi: Document -mcpu option.
4692 2014-01-17 Richard Biener <rguenther@suse.de>
4694 PR rtl-optimization/38518
4695 * df.h (df_analyze_loop): Declare.
4696 * df-core.c: Include cfgloop.h.
4697 (df_analyze_1): Split out main part of df_analyze.
4698 (df_analyze): Adjust.
4699 (loop_inverted_post_order_compute): New function.
4700 (loop_post_order_compute): Likewise.
4701 (df_analyze_loop): New function avoiding whole-function
4702 postorder computes.
4703 * loop-invariant.c (find_defs): Use df_analyze_loop.
4704 (find_invariants): Adjust.
4705 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4707 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4709 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4710 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4712 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4714 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4715 traversal when removing references.
4717 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4719 PR ipa/59775
4720 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4722 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4724 PR middle-end/56791
4725 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4726 pushing a reload for an autoinc when we had previously reloaded an
4727 inner part of the address.
4729 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4731 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4732 field.
4733 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4734 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4735 when not giving up or versioning for alias only because of
4736 loop->safelen.
4737 (vect_analyze_data_ref_dependences): Set to true.
4738 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4739 is a GIMPLE_PHI.
4740 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4741 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4742 to the condition.
4744 PR middle-end/58344
4745 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4747 PR target/59839
4748 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4749 operand 0 predicate for gathers, use a new pseudo as subtarget.
4751 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4753 PR middle-end/59609
4754 * lra-constraints.c (process_alt_operands): Add printing debug info.
4755 Check absence of input/output reloads for matched operands too.
4757 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4759 PR rtl-optimization/59835
4760 * ira.c (ira_init_register_move_cost): Increase cost for
4761 impossible modes.
4763 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4765 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4767 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4769 PR target/59780
4770 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4771 non-register objects. Use gen_(high/low)part more consistently.
4772 Fix assertions.
4774 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4776 PR target/59844
4777 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4778 endian support, remove tests for WORDS_BIG_ENDIAN.
4779 (p8_mfvsrd_3_<mode>): Likewise.
4780 (reload_gpr_from_vsx<mode>): Likewise.
4781 (reload_gpr_from_vsxsf): Likewise.
4782 (p8_mfvsrd_4_disf): Likewise.
4784 2014-01-16 Richard Biener <rguenther@suse.de>
4786 PR rtl-optimization/46590
4787 * lcm.c (compute_antinout_edge): Use postorder iteration.
4788 (compute_laterin): Use inverted postorder iteration.
4790 2014-01-16 Nick Clifton <nickc@redhat.com>
4792 PR middle-end/28865
4793 * varasm.c (output_constant): Return the number of bytes actually
4794 emitted.
4795 (output_constructor_array_range): Update the field size with the
4796 number of bytes emitted by output_constant.
4797 (output_constructor_regular_field): Likewise. Also do not
4798 complain if the total number of bytes emitted is now greater
4799 than the expected fieldpos.
4800 * output.h (output_constant): Update prototype and descriptive comment.
4802 2014-01-16 Marek Polacek <polacek@redhat.com>
4804 PR middle-end/59827
4805 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4806 it is error_mark_node.
4808 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4810 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4811 VALID_AVX256_REG_OR_OI_MODE.
4813 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4815 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4816 current procedure should be profiled.
4818 2014-01-15 Andrew Pinski <apinski@cavium.com>
4820 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4821 of moving from/to the STACK_REG register class.
4823 2014-01-15 Richard Henderson <rth@redhat.com>
4825 PR debug/54694
4826 * reginfo.c (global_regs_decl): Globalize.
4827 * rtl.h (global_regs_decl): Declare.
4828 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4829 reserved via global_regs.
4831 2014-01-15 Teresa Johnson <tejohnson@google.com>
4833 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4835 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4837 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4838 and vmulosh rather than call gen_vec_widen_smult_*.
4839 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4840 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4841 (vec_widen_smult_even_v16qi): Likewise.
4842 (vec_widen_umult_even_v8hi): Likewise.
4843 (vec_widen_smult_even_v8hi): Likewise.
4844 (vec_widen_umult_odd_v16qi): Likewise.
4845 (vec_widen_smult_odd_v16qi): Likewise.
4846 (vec_widen_umult_odd_v8hi): Likewise.
4847 (vec_widen_smult_odd_v8hi): Likewise.
4848 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4849 vmuloub rather than call gen_vec_widen_umult_*.
4850 (vec_widen_umult_lo_v16qi): Likewise.
4851 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4852 vmulosb rather than call gen_vec_widen_smult_*.
4853 (vec_widen_smult_lo_v16qi): Likewise.
4854 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4855 rather than call gen_vec_widen_umult_*.
4856 (vec_widen_umult_lo_v8hi): Likewise.
4857 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4858 rather than call gen_vec_widen_smult_*.
4859 (vec_widen_smult_lo_v8hi): Likewise.
4861 2014-01-15 Jeff Law <law@redhat.com>
4863 PR tree-optimization/59747
4864 * ree.c (find_and_remove_re): Properly handle case where a second
4865 eliminated extension requires widening a copy created for elimination
4866 of a prior extension.
4867 (combine_set_extension): Ensure that the number of hard regs needed
4868 for a destination register does not change when we widen it.
4870 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4872 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4873 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4874 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4875 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4876 (avr-*-rtems*): Likewise.
4877 (bfin*-rtems*): Likewise.
4878 (moxie-*-rtems*): Likewise.
4879 (h8300-*-rtems*): Likewise.
4880 (i[34567]86-*-rtems*): Likewise.
4881 (lm32-*-rtems*): Likewise.
4882 (m32r-*-rtems*): Likewise.
4883 (m68k-*-rtems*): Likewise.
4884 (microblaze*-*-rtems*): Likewise.
4885 (mips*-*-rtems*): Likewise.
4886 (powerpc-*-rtems*): Likewise.
4887 (sh-*-rtems*): Likewise.
4888 (sparc-*-rtems*): Likewise.
4889 (sparc64-*-rtems*): Likewise.
4890 (v850-*-rtems*): Likewise.
4891 (m32c-*-rtems*): Likewise.
4893 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4895 PR rtl-optimization/59511
4896 * ira.c (ira_init_register_move_cost): Use memory costs for some
4897 cases of register move cost calculations.
4898 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4899 instead of BB frequency.
4900 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4901 * lra-assigns.c (find_hard_regno_for): Ditto.
4903 2014-01-15 Richard Biener <rguenther@suse.de>
4905 PR tree-optimization/59822
4906 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4907 (vectorizable_load): Use it to hoist defs of uses of invariant
4908 loads out of the loop.
4910 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4911 Kugan Vivekanandarajah <kuganv@linaro.org>
4913 PR target/59695
4914 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4915 truncation.
4917 2014-01-15 Richard Biener <rguenther@suse.de>
4919 PR rtl-optimization/59802
4920 * lcm.c (compute_available): Use inverted postorder to seed
4921 the initial worklist.
4923 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4925 PR target/59803
4926 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4927 ADDR_REGS for invalid symrefs in non-PIC code.
4929 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4931 PR other/58712
4932 * builtins.c (determine_block_size): Initialize *probable_max_size
4933 even if len_rtx is CONST_INT.
4935 2014-01-14 Andrew Pinski <apinski@cavium.com>
4937 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4938 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4939 (cortexa53_tunings): Likewise.
4940 (aarch64_sched_issue_rate): New function.
4941 (TARGET_SCHED_ISSUE_RATE): Define.
4943 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4945 * ira-costs.c (find_costs_and_classes): Add missed
4946 ira_init_register_move_cost_if_necessary.
4948 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4950 PR target/59787
4951 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4953 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4955 PR target/59794
4956 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4957 to indicate if type is used for function return value. Warn ABI
4958 change if the vector mode isn't available for function return value.
4959 (ix86_function_arg_advance): Pass false to type_natural_mode.
4960 (ix86_function_arg): Likewise.
4961 (ix86_gimplify_va_arg): Likewise.
4962 (function_arg_32): Don't warn ABI change.
4963 (ix86_function_value): Pass true to type_natural_mode.
4964 (ix86_return_in_memory): Likewise.
4965 (ix86_struct_value_rtx): Removed.
4966 (TARGET_STRUCT_VALUE_RTX): Likewise.
4968 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4970 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4971 converting a conditional jump into a conditional return.
4973 2014-01-14 Richard Biener <rguenther@suse.de>
4975 PR tree-optimization/58921
4976 PR tree-optimization/59006
4977 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4978 hoisting invariant stmts.
4979 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4980 invariant loads on the preheader edge if possible.
4982 2014-01-14 Joey Ye <joey.ye@arm.com>
4984 * doc/plugin.texi (Building GCC plugins): Update to C++.
4986 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4988 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4989 (_mm_rcp28_round_ss): Ditto.
4990 (_mm_rsqrt28_round_sd): Ditto.
4991 (_mm_rsqrt28_round_ss): Ditto.
4992 (_mm_rcp28_sd): Ditto.
4993 (_mm_rcp28_ss): Ditto.
4994 (_mm_rsqrt28_sd): Ditto.
4995 (_mm_rsqrt28_ss): Ditto.
4996 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4997 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4998 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4999 (IX86_BUILTIN_RCP28SD): Ditto.
5000 (IX86_BUILTIN_RCP28SS): Ditto.
5001 (IX86_BUILTIN_RSQRT28SD): Ditto.
5002 (IX86_BUILTIN_RSQRT28SS): Ditto.
5003 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5004 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5005 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5006 (ix86_expand_special_args_builtin): Expand new FTYPE.
5007 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5008 (srcp14<mode>): Make insn unary.
5009 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5010 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5011 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5012 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5013 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5014 Fix rounding: make it SAE only.
5015 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5016 Ditto.
5017 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5018 Ditto.
5019 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5020 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5021 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5022 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5023 (round_saeonly_mask_scalar_operand4): Ditto.
5024 (round_saeonly_mask_scalar_op3): Ditto.
5025 (round_saeonly_mask_scalar_op4): Ditto.
5027 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5029 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5030 Implement -maltivec=be for vec_insert and vec_extract.
5032 2014-01-10 DJ Delorie <dj@redhat.com>
5034 * config/msp430/msp430.md (call_internal): Don't allow memory
5035 references with SP as the base register.
5036 (call_value_internal): Likewise.
5037 * config/msp430/constraints.md (Yc): New. For memory references
5038 that don't use SP as a base register.
5040 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5041 "an integer without a # prefix"
5042 * config/msp430/msp430.md (epilogue_helper): Use it.
5044 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5046 PR target/59617
5047 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5048 AVX512F gather builtins.
5049 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5050 on gather decls with INTEGER_TYPE masktype.
5051 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5052 directly into the builtin rather than hoisting it before loop.
5054 PR tree-optimization/59387
5055 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5056 (scev_const_prop): If folded_casts and type has undefined overflow,
5057 use force_gimple_operand instead of force_gimple_operand_gsi and
5058 for each added stmt if it is assign with
5059 arith_code_with_undefined_signed_overflow, call
5060 rewrite_to_defined_overflow.
5061 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5062 gimple-fold.h instead.
5063 (arith_code_with_undefined_signed_overflow,
5064 rewrite_to_defined_overflow): Moved to ...
5065 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5066 rewrite_to_defined_overflow): ... here. No longer static.
5067 Include gimplify-me.h.
5068 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5069 rewrite_to_defined_overflow): New prototypes.
5071 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5073 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5075 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5077 * builtins.c (get_object_alignment_2): Minor tweak.
5078 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5080 2014-01-13 Christian Bruel <christian.bruel@st.com>
5082 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5083 optimized non constant lengths.
5085 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5087 PR libgomp/59194
5088 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5089 load as __atomic_load_N if possible.
5091 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5093 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5094 target parameter.
5095 (rs6000_expand_builtin): Adjust call.
5097 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5099 PR target/58115
5100 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5101 * config/rs6000/rs6000.c: Include target-globals.h.
5102 (rs6000_set_current_function): Instead of doing target_reinit
5103 unconditionally, use save_target_globals_default_opts and
5104 restore_target_globals.
5106 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5107 FPSCR.
5108 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5109 (rs6000_expand_builtin): Handle mffs and mtfsf.
5110 (rs6000_init_builtins): Define mffs and mtfsf.
5111 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5112 (rs6000_mffs): New pattern.
5113 (rs6000_mtfsf): New pattern.
5115 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5117 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5118 Start narrowing with START. Apply candidate-use pair
5119 and check overall cost in narrowing.
5120 (iv_ca_prune): Pass new argument.
5122 2014-01-10 Jeff Law <law@redhat.com>
5124 PR middle-end/59743
5125 * ree.c (combine_reaching_defs): Ensure the defining statement
5126 occurs before the extension when optimizing extensions with
5127 different source and destination hard registers.
5129 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5131 PR ipa/58585
5132 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5133 vtables into the type inheritance graph.
5135 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5137 PR rtl-optimization/59754
5138 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5139 modes in the REGNO != REGNO case.
5141 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5143 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5145 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5147 PR tree-optimization/59745
5148 * tree-predcom.c (tree_predictive_commoning_loop): Call
5149 free_affine_expand_cache if giving up because components is NULL.
5151 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5152 GC in payload of target_globals struct instead of allocating them on
5153 the heap and the larger structs separately using GC.
5154 * target-globals.h (struct target_globals): Make regs, hard_regs,
5155 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5156 of GTY((skip)) and change type to void *.
5157 (reset_target_globals): Cast loads from those fields to corresponding
5158 types.
5160 2014-01-10 Steve Ellcey <sellcey@mips.com>
5162 PR plugins/59335
5163 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5164 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5165 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5167 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5169 PR target/59744
5170 * aarch64-modes.def (CC_Zmode): New flags mode.
5171 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5172 represents an equality.
5173 (aarch64_get_condition_code): Handle CC_Zmode.
5174 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5176 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5178 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5179 extraction in good case.
5181 2014-01-10 Richard Biener <rguenther@suse.de>
5183 PR tree-optimization/59374
5184 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5185 checking after SLP discovery. Mark stmts not participating
5186 in any SLP instance properly.
5188 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5190 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5191 when handling a SET rtx.
5193 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5195 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5196 (cortex-a57): Likewise.
5197 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5199 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5201 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5202 non-iwmmxt builtins.
5204 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5206 PR ipa/58252
5207 PR ipa/59226
5208 * ipa-devirt.c record_target_from_binfo): Take as argument
5209 stack of binfos and lookup matching one for virtual inheritance.
5210 (possible_polymorphic_call_targets_1): Update.
5212 2014-01-10 Huacai Chen <chenhc@lemote.com>
5214 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5215 kernel strings for Loongson-2E/2F/3A.
5217 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5219 PR middle-end/59670
5220 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5221 is_gimple_call before calling gimple_call_internal_p.
5223 2014-01-09 Steve Ellcey <sellcey@mips.com>
5225 * Makefile.in (TREE_FLOW_H): Remove.
5226 (TREE_SSA_H): Add file names from tree-flow.h.
5227 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5228 * tree.h: Remove tree-flow.h reference.
5229 * hash-table.h: Remove tree-flow.h reference.
5230 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5231 reference with tree-ssa-loop.h.
5233 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5235 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5236 default element-order behavior for -maltivec.
5237 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5238 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5239 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5240 when targeting big endian, at least for now.
5241 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5243 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5245 PR middle-end/47735
5246 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5247 var satisfies use_register_for_decl, just take into account type
5248 alignment, rather than decl alignment.
5250 PR tree-optimization/59622
5251 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5252 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5253 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5254 Don't devirtualize for inplace at all. For targets.length () == 1,
5255 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5257 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5259 * config/i386/i386.md (cpu): Remove the unused btver1.
5261 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5263 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5265 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5267 PR target/58115
5268 * tree-core.h (struct target_globals): New forward declaration.
5269 (struct tree_target_option): Add globals field.
5270 * tree.h (TREE_TARGET_GLOBALS): Define.
5271 (prepare_target_option_nodes_for_pch): New prototype.
5272 * target-globals.h (struct target_globals): Define even if
5273 !SWITCHABLE_TARGET.
5274 * tree.c (prepare_target_option_node_for_pch,
5275 prepare_target_option_nodes_for_pch): New functions.
5276 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5277 * config/i386/i386.c: Include target-globals.h.
5278 (ix86_set_current_function): Instead of doing target_reinit
5279 unconditionally, use save_target_globals_default_opts and
5280 restore_target_globals.
5282 2014-01-09 Richard Biener <rguenther@suse.de>
5284 PR tree-optimization/59715
5285 * tree-cfg.h (split_critical_edges): Declare.
5286 * tree-cfg.c (split_critical_edges): Export.
5287 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5289 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5291 * cfgexpand.c (expand_stack_vars): Optionally disable
5292 asan stack protection.
5293 (expand_used_vars): Likewise.
5294 (partition_stack_vars): Likewise.
5295 * asan.c (asan_emit_stack_protection): Optionally disable
5296 after return stack usage.
5297 (instrument_derefs): Optionally disable memory access instrumentation.
5298 (instrument_builtin_call): Likewise.
5299 (instrument_strlen_call): Likewise.
5300 (asan_protect_global): Optionally disable global variables protection.
5301 * doc/invoke.texi: Added doc for new options.
5302 * params.def: Added new options.
5303 * params.h: Likewise.
5305 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5307 PR rtl-optimization/59724
5308 * ifcvt.c (cond_exec_process_if_block): Don't call
5309 flow_find_head_matching_sequence with 0 longest_match.
5310 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5311 non-active insns if !stop_after.
5312 (try_head_merge_bb): Revert 2014-01-07 changes.
5314 2014-01-08 Jeff Law <law@redhat.com>
5316 * ree.c (get_sub_rtx): New function, extracted from...
5317 (merge_def_and_ext): Here.
5318 (combine_reaching_defs): Use get_sub_rtx.
5320 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5322 * cgraph.h (varpool_variable_node): Do not choke on null node.
5324 2014-01-08 Catherine Moore <clm@codesourcery.com>
5326 * config/mips/mips.md (simple_return): Attempt to use JRC
5327 for microMIPS.
5328 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5330 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5332 PR rtl-optimization/59137
5333 * reorg.c (steal_delay_list_from_target): Call update_block for
5334 elided insns.
5335 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5337 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5339 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5340 two duplicate entries.
5342 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5344 Revert:
5345 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5347 * config/mips/mips.c (mips_truncated_op_cost): New function.
5348 (mips_rtx_costs): Adjust test for BADDU.
5349 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5351 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5353 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5354 (*baddu_si): ...this new pattern.
5356 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5358 PR ipa/59722
5359 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5361 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5363 PR middle-end/57748
5364 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5365 inner_reference_p.
5366 (expand_expr, expand_normal): Adjust.
5367 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5368 inner_reference_p. Use inner_reference_p to expand inner references.
5369 (store_expr): Adjust.
5370 * cfgexpand.c (expand_call_stmt): Adjust.
5372 2014-01-08 Rong Xu <xur@google.com>
5374 * gcov-io.c (gcov_var): Move from gcov-io.h.
5375 (gcov_position): Ditto.
5376 (gcov_is_error): Ditto.
5377 (gcov_rewrite): Ditto.
5378 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5379 only part to libgcc/libgcov.h.
5381 2014-01-08 Marek Polacek <polacek@redhat.com>
5383 PR middle-end/59669
5384 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5386 2014-01-08 Marek Polacek <polacek@redhat.com>
5388 PR sanitizer/59667
5389 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5391 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5393 PR rtl-optimization/59649
5394 * stor-layout.c (get_mode_bounds): For BImode return
5395 0 and STORE_FLAG_VALUE.
5397 2014-01-08 Richard Biener <rguenther@suse.de>
5399 PR middle-end/59630
5400 * gimple.h (is_gimple_builtin_call): Remove.
5401 (gimple_builtin_call_types_compatible_p): New.
5402 (gimple_call_builtin_p): New overload.
5403 * gimple.c (is_gimple_builtin_call): Remove.
5404 (validate_call): Rename to ...
5405 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5406 check return types.
5407 (validate_type): New static function.
5408 (gimple_call_builtin_p): New overload and adjust.
5409 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5410 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5411 (gimple_fold_stmt_to_constant_1): Likewise.
5412 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5414 2014-01-08 Richard Biener <rguenther@suse.de>
5416 PR middle-end/59471
5417 * gimplify.c (gimplify_expr): Gimplify register-register type
5418 VIEW_CONVERT_EXPRs to separate stmts.
5420 2014-01-07 Jeff Law <law@redhat.com>
5422 PR middle-end/53623
5423 * ree.c (combine_set_extension): Handle case where source
5424 and destination registers in an extension insn are different.
5425 (combine_reaching_defs): Allow source and destination registers
5426 in extension to be different under limited circumstances.
5427 (add_removable_extension): Remove restriction that the
5428 source and destination registers in the extension are the same.
5429 (find_and_remove_re): Emit a copy from the extension's
5430 destination to its source after the defining insn if
5431 the source and destination registers are different.
5433 PR middle-end/59285
5434 * ifcvt.c (merge_if_block): If we are merging a block with more than
5435 one successor with a block with no successors, remove any BARRIER
5436 after the second block.
5438 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5440 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5442 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5444 PR target/59652
5445 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5446 for 14-bit register offsets when INT14_OK_STRICT is false.
5448 2014-01-07 Roland Stigge <stigge@antcom.de>
5449 Michael Meissner <meissner@linux.vnet.ibm.com>
5451 PR 57386/target
5452 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5453 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5455 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5457 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5458 -mcpu.
5460 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5462 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5463 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5464 rtx is const0_rtx or not.
5466 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5468 PR target/58115
5469 * target-globals.c (save_target_globals): Remove this_fn_optab
5470 handling.
5471 * toplev.c: Include optabs.h.
5472 (target_reinit): Temporarily restore the global options if another
5473 set of options are in force.
5475 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5477 PR rtl-optimization/58668
5478 * cfgcleanup.c (flow_find_cross_jump): Don't count
5479 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5480 to determine what is counted.
5481 (flow_find_head_matching_sequence): Use active_insn_p to determine
5482 what is counted.
5483 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5484 counting change.
5485 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5486 determine what is counted.
5488 PR tree-optimization/59643
5489 * tree-predcom.c (split_data_refs_to_components): If one dr is
5490 read and one write, determine_offset fails and the write isn't
5491 in the bad component, just put the read into the bad component.
5493 2014-01-07 Mike Stump <mikestump@comcast.net>
5494 Jakub Jelinek <jakub@redhat.com>
5496 PR pch/59436
5497 * tree-core.h (struct tree_optimization_option): Change optabs
5498 type from unsigned char * to void *.
5499 * optabs.c (init_tree_optimization_optabs): Adjust
5500 TREE_OPTIMIZATION_OPTABS initialization.
5502 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5504 PR target/59644
5505 * config/i386/i386.h (struct machine_function): Add
5506 no_drap_save_restore field.
5507 * config/i386/i386.c (ix86_save_reg): Use
5508 !cfun->machine->no_drap_save_restore instead of
5509 crtl->stack_realign_needed.
5510 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5511 this function clears frame_pointer_needed. Set
5512 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5513 and DRAP reg is needed.
5515 2014-01-06 Marek Polacek <polacek@redhat.com>
5517 PR c/57773
5518 * doc/implement-c.texi: Mention that other integer types are
5519 permitted as bit-field types in strictly conforming mode.
5521 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5523 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5524 is newly allocated.
5526 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5528 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5530 2014-01-06 Martin Jambor <mjambor@suse.cz>
5532 PR ipa/59008
5533 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5534 to int.
5535 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5537 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5539 PR debug/59350
5540 PR debug/59510
5541 * var-tracking.c (add_stores): Preserve the value of the source even if
5542 we don't record the store.
5544 2014-01-06 Terry Guo <terry.guo@arm.com>
5546 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5548 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5550 PR bootstrap/59541
5551 * config/darwin.c (darwin_function_section): Adjust return values to
5552 correspond to optimisation changes made in r206070.
5554 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5556 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5557 from prefetch_block tune setting.
5558 (nocona_cost): Correct size of prefetch block to 64.
5560 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5562 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5563 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5564 used to save the static chain register in the computation of the offset
5565 from which the FP registers need to be restored.
5567 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5569 PR tree-optimization/59519
5570 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5571 ICE if get_current_def (current_new_name) is already non-NULL, as long
5572 as it is a phi result of some other phi in *new_exit_bb that has
5573 the same argument.
5575 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5576 or vmovdqu* for misaligned_operand.
5577 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5578 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5579 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5580 aligned_mem for AVX512F masked aligned load and store builtins and for
5581 non-temporal moves.
5583 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5585 PR tree-optimization/59651
5586 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5587 Address range for negative step should be added by TYPE_SIZE_UNIT.
5589 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5591 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5592 overlapping registers also for registers other than the stack pointer.
5594 2014-01-03 Marek Polacek <polacek@redhat.com>
5596 PR other/59661
5597 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5598 __builtin_FILE.
5600 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5602 PR target/59625
5603 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5604 asm goto as jump.
5606 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5607 (push splitter): Use <P:MODE_SIZE> instead of
5608 GET_MODE_SIZE (<P:MODE>mode).
5609 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5610 (mov -1, reg peephole2): Likewise.
5611 * config/i386/sse.md (*mov<mode>_internal,
5612 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5613 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5614 *<code><mode>3, *andnot<mode>3<mask_name>,
5615 <mask_codefor><code><mode>3<mask_name>): Likewise.
5616 * config/i386/subst.md (mask_mode512bit_condition,
5617 sd_mask_mode512bit_condition): Likewise.
5619 2014-01-02 Xinliang David Li <davidxl@google.com>
5621 PR tree-optimization/59303
5622 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5623 (dump_predicates): Better output format.
5624 (pred_equal_p): New function.
5625 (is_neq_relop_p): Ditto.
5626 (is_neq_zero_form_p): Ditto.
5627 (pred_expr_equal_p): Ditto.
5628 (pred_neg_p): Ditto.
5629 (simplify_pred): Ditto.
5630 (simplify_preds_2): Ditto.
5631 (simplify_preds_3): Ditto.
5632 (simplify_preds_4): Ditto.
5633 (simplify_preds): Ditto.
5634 (push_pred): Ditto.
5635 (push_to_worklist): Ditto.
5636 (get_pred_info_from_cmp): Ditto.
5637 (is_degenerated_phi): Ditto.
5638 (normalize_one_pred_1): Ditto.
5639 (normalize_one_pred): Ditto.
5640 (normalize_one_pred_chain): Ditto.
5641 (normalize_preds): Ditto.
5642 (normalize_cond_1): Remove function.
5643 (normalize_cond): Ditto.
5644 (is_gcond_subset_of): Ditto.
5645 (is_subset_of_any): Ditto.
5646 (is_or_set_subset_of): Ditto.
5647 (is_and_set_subset_of): Ditto.
5648 (is_norm_cond_subset_of): Ditto.
5649 (pred_chain_length_cmp): Ditto.
5650 (convert_control_dep_chain_into_preds): Type change.
5651 (find_predicates): Ditto.
5652 (find_def_preds): Ditto.
5653 (destroy_predicates_vecs): Ditto.
5654 (find_matching_predicates_in_rest_chains): Ditto.
5655 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5656 (is_pred_expr_subset): Ditto.
5657 (is_pred_chain_subset_of): Ditto.
5658 (is_included_in): Ditto.
5659 (is_superset_of): Ditto.
5661 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5663 Update copyright years.
5665 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5667 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5668 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5669 config/arc/arc.md, config/arc/arc.opt,
5670 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5671 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5672 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5673 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5674 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5675 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5676 vtable-verify.c, vtable-verify.h: Use the standard form for the
5677 copyright notice.
5679 2014-01-02 Tobias Burnus <burnus@net-b.de>
5681 * gcc.c (process_command): Update copyright notice dates.
5682 * gcov-dump.c: Ditto.
5683 * gcov.c: Ditto.
5684 * doc/cpp.texi: Bump @copying's copyright year.
5685 * doc/cppinternals.texi: Ditto.
5686 * doc/gcc.texi: Ditto.
5687 * doc/gccint.texi: Ditto.
5688 * doc/gcov.texi: Ditto.
5689 * doc/install.texi: Ditto.
5690 * doc/invoke.texi: Ditto.
5692 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5694 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5696 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5698 * config/i386/sse.md (*mov<mode>_internal): Guard
5699 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5701 PR rtl-optimization/59647
5702 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5703 new_rtx into UNSIGNED_FLOAT rtxes.
5704 \f
5705 Copyright (C) 2014 Free Software Foundation, Inc.
5707 Copying and distribution of this file, with or without modification,
5708 are permitted in any medium without royalty provided the copyright
5709 notice and this notice are preserved.