1 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
4 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
6 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
9 * lto-cgraph.c (lto_output_node): Use
10 symtab_get_symbol_partitioning_class.
11 (lto_output_varpool_node): likewise.
12 (symtab_get_symbol_partitioning_class): Move here from
14 * cgraph.h (symbol_partitioning_class): Likewise.
15 (symtab_get_symbol_partitioning_class): Declare.
17 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
19 * ggc.h (ggc_internal_cleared_alloc): New macro.
20 * vec.h (vec_safe_copy): Handle memory stats.
21 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
22 * target-globals.c (save_target_globals): Likewise.
24 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
27 * expr.c (emit_move_resolve_push): Export; be bit more selective
28 on when to clear alias set.
29 * expr.h (emit_move_resolve_push): Declare.
30 * function.h (struct function): Add tail_call_marked.
31 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
32 * config/i386/i386-protos.h (ix86_expand_push): Remove.
33 * config/i386/i386.md (TImode move expander): De not call
35 (FP push expanders): Preserve memory attributes.
36 * config/i386/sse.md (push<mode>1): Remove.
37 * config/i386/i386.c (ix86_expand_vector_move): Handle push
39 (ix86_expand_push): Remove.
40 * config/i386/mmx.md (push<mode>1): Remove.
42 2014-02-06 Jakub Jelinek <jakub@redhat.com>
44 PR rtl-optimization/60030
45 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
46 lopart with paradoxical subreg before shifting it up by hprec.
48 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
50 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
51 Remove extra newline at end of file.
52 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
53 (arm_issue_rate): Handle cortexa57.
54 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
55 (cortex-a57.cortex-a53): Likewise.
57 2014-02-06 Jakub Jelinek <jakub@redhat.com>
60 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
61 don't record in REG_FRAME_RELATED_EXPR registers not set in that
63 (arm_expand_prologue): Adjust all callers.
64 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
65 info, registers also at the lowest numbered registers side. Use
66 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
70 * var-tracking.c (adjust_mems): Before adding a SET
71 to amd->side_effects, adjust it's SET_SRC using
72 simplify_replace_fn_rtx.
74 2014-02-06 Alan Modra <amodra@gmail.com>
77 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
78 change SDmode to DDmode when lra_in_progress.
80 2014-02-06 Jakub Jelinek <jakub@redhat.com>
83 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
84 free_data_ref on the dr first, and before goto again also set dr
85 to the next dr. For simd_lane_access, free old datarefs[i] before
86 overwriting it. For get_vectype_for_scalar_type failure, don't
87 free_data_ref if simd_lane_access.
89 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
92 * tree.h (opts_for_fn): New inline function.
94 * config/i386/i386.c (ix86_function_regparm): Use
95 opt_for_fn (decl, optimize) instead of optimize.
97 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
99 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
100 for SYMBOL_REF in large memory model.
102 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
104 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
106 (cortex-a57): Likewise.
107 (cortex-a57.cortex-a53): Likewise.
109 2014-02-06 Yury Gribov <y.gribov@samsung.com>
110 Kugan Vivekanandarajah <kuganv@linaro.org>
112 * config/arm/arm.c (arm_vector_alignment_reachable): Check
114 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
116 2014-02-06 Richard Biener <rguenther@suse.de>
118 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
119 set_loop_copy and initialize_original_copy_tables.
121 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
123 * config/aarch64/aarch64-simd.md
124 (aarch64_ashr_simddi): Change QI to SI.
126 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
127 Jakub Jelinek <jakub@redhat.com>
130 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
133 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
135 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
136 CODE_FOR_altivec_vpku[hw]um to
137 CODE_FOR_altivec_vpku[hw]um_direct.
138 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
139 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
140 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
141 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
143 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
145 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
146 generation for -maltivec=be.
147 (altivec_vsumsws): Simplify redundant test.
149 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
151 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
152 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
153 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
154 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
156 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
158 (altivec_vpks<VI_char>ss): Likewise.
159 (altivec_vpks<VI_char>us): Likewise.
160 (altivec_vpku<VI_char>us): Likewise.
161 (altivec_vpku<VI_char>um): Likewise.
162 (altivec_vpku<VI_char>um_direct): New (copy of
163 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
165 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
166 target is little endian and -maltivec=be is not specified.
167 (*altivec_vupkhs<VU_char>_direct): New (copy of
168 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
170 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
171 target is little endian and -maltivec=be is not specified.
172 (*altivec_vupkls<VU_char>_direct): New (copy of
173 altivec_vupkls<VU_char> that always emits vupkls*, for internal
175 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
176 little endian and -maltivec=be is not specified.
177 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
178 little endian and -maltivec=be is not specified.
180 2014-02-05 Richard Henderson <rth@redhat.com>
183 * combine-stack-adj.c: Revert r206943.
184 * sched-int.h (struct deps_desc): Add last_args_size.
185 * sched-deps.c (init_deps): Initialize it.
186 (sched_analyze_insn): Add OUTPUT dependencies between insns that
187 contain REG_ARGS_SIZE notes.
189 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
191 * lto-cgraph.c (asm_nodes_output): Make global.
192 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
193 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
195 (driver_handle_option): Handle OPT_fwpa.
197 2014-02-05 Jakub Jelinek <jakub@redhat.com>
200 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
201 a comment typo and formatting issue. If odr_hash hasn't been
202 created, return vNULL and set *completep to false.
205 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
206 bb with no successors.
208 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
211 * doc/invoke.texi (-march): Clarify documentation for ARM.
215 2014-02-05 Richard Biener <rguenther@suse.de>
217 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
218 when not vectorizing because of too many alias checks.
219 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
220 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
222 2014-02-05 Nick Clifton <nickc@redhat.com>
224 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
225 accept extended registers in any mode when compiling for the
228 2014-02-05 Yury Gribov <y.gribov@samsung.com>
230 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
231 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
232 sanitization attributes.
233 (can_inline_edge_p): Likewise.
234 (sanitize_attrs_match_for_inline_p): New function.
236 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
238 * ipa-prop.c (detect_type_change): Shor circuit testing of
239 type changes on THIS pointer.
241 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
244 * config/pa/pa.c (legitimize_tls_address): Return original address
245 if not passed a SYMBOL_REF rtx.
246 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
248 (pa_emit_move_sequence): Simplify TLS source operands.
249 (pa_legitimate_constant_p): Reject all TLS constants.
250 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
251 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
253 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
255 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
256 groups when we know they are controlled by LTO.
257 * varasm.c (default_binds_local_p_1): If object is in other partition,
258 it will be resolved locally.
260 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
262 * gcc/config/host-linux.c (linux_gt_pch_use_address): Don't
263 use SSIZE_MAX because it is not always defined.
265 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
268 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
269 threshold for pseudo splitting.
270 (update_ebb_live_info): Process call argument hard registers and
271 hard registers from insn definition too.
272 (max_small_class_regs_num): New constant.
273 (inherit_in_ebb): Update live hard regs through EBBs. Update
274 reloads_num only for small register classes. Don't split for
277 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
280 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
283 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
285 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
287 2014-02-04 Marek Polacek <polacek@redhat.com>
289 * gdbinit.in (pel): Define.
291 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
293 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
296 2014-02-04 Richard Biener <rguenther@suse.de>
299 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
300 in function context local.
301 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
302 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
303 similar to LTO_imported_decl_ref.
305 2014-02-04 Jakub Jelinek <jakub@redhat.com>
307 PR tree-optimization/60002
308 * cgraphclones.c (build_function_decl_skip_args): Clear
311 PR tree-optimization/60023
312 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
313 false to gsi_replace.
314 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
315 has been in some EH region and vec_stmt could throw, add
316 vec_stmt into the same EH region.
317 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
318 has no lhs, ignore it.
319 * internal-fn.c (expand_MASK_LOAD): Likewise.
322 * tree-inline.c (copy_forbidden): Fail for
323 __attribute__((optimize (0))) functions.
326 * omp-low.c (simd_clone_struct_copy): If from->inbranch
327 is set, copy one less argument.
328 (expand_simd_clones): Don't subtract clone_info->inbranch
329 from simd_clone_struct_alloc argument.
331 PR rtl-optimization/57915
332 * recog.c (simplify_while_replacing): If all unary/binary/relational
333 operation arguments are constant, attempt to simplify those.
336 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
337 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
339 2014-02-04 Richard Biener <rguenther@suse.de>
341 PR tree-optimization/60012
342 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
343 TBAA disambiguation to all DDRs.
345 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
348 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
349 (LINK_SPEC): Use it for -shared, -shared-libgcc.
351 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
354 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
356 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
358 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
359 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
361 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
364 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
365 to figure out targets of polymorphic calls with known decl.
366 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
367 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
368 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
369 (get_polymorphic_call_info): ... here.
370 (get_polymorphic_call_info_from_invariant): New function.
372 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
374 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
375 lookup via vtable pointer; check for type consistency
376 and turn inconsitent facts into UNREACHABLE.
377 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
378 * gimple-fold.c (gimple_get_virt_method_for_vtable):
379 Do not ICE on type inconsistent querries; return UNREACHABLE
382 2014-02-03 Richard Henderson <rth@twiddle.net>
385 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
386 already processed this node.
387 (normalize_one_pred_1): Pass along mark_set.
388 (normalize_one_pred): Create and destroy a pointer_set_t.
389 (normalize_one_pred_chain): Likewise.
391 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
393 PR gcov-profile/58602
394 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
396 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
399 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
400 Try to devirtualize by the knowledge of virtual table pointer given by
401 aggregate propagation.
402 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
403 ipa_print_node_jump_functions): Dump also offset that
404 is relevant for polymorphic calls.
405 (determine_known_aggregate_parts): Add arg_type parameter; use it
406 instead of determining the type from pointer type.
407 (ipa_compute_jump_functions_for_edge): Update call of
408 determine_known_aggregate_parts.
409 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
410 (gimple_get_virt_method_for_binfo): ... here; simplify using
411 vtable_pointer_value_to_vtable.
412 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
413 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
414 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
415 (vtable_pointer_value_to_vtable): Break out from ...; handle also
417 (vtable_pointer_value_to_binfo): ... here.
418 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
420 2014-02-03 Teresa Johnson <tejohnson@google.com>
422 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
423 redef of outer loop index variable.
425 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
429 * doc/extend.texi (Function Attributes): Typo.
431 2014-02-03 Cong Hou <congh@google.com>
433 PR tree-optimization/60000
434 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
435 if the vectorized statement is a store. A store statement can only
436 appear at the end of pattern statements.
438 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
440 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
441 (ix86_option_override_internal): Default long double to 64-bit for
442 32-bit Bionic and to 128-bit for 64-bit Bionic.
444 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
445 TARGET_LONG_DOUBLE_128 is true.
446 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
448 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
449 (mlong-double-64): Negate -mlong-double-128.
450 (mlong-double-128): New option.
452 * config/i386/i386-c.c (ix86_target_macros): Define
453 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
455 * doc/invoke.texi: Document -mlong-double-128.
457 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
459 PR rtl-optimization/60024
460 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
462 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
464 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
466 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
468 PR rtl-optimization/57662
469 * sel-sched.c (code_motion_path_driver): Do not mark already not
470 existing blocks in the visiting bitmap.
472 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
474 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
475 on the insn being emitted.
477 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
478 Will Deacon <will.deacon@arm.com>
480 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
482 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
484 * config/arm/arm-tables.opt: Regenerate.
486 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
488 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
489 for vector types other than V16QImode.
490 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
491 define_expand, and call altivec_expand_vec_perm_le when producing
492 code with little endian element order.
493 (*altivec_vperm_<mode>_internal): New insn having previous
494 behavior of altivec_vperm_<mode>.
495 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
496 altivec_expand_vec_perm_le when producing code with little endian
498 (*altivec_vperm_<mode>_uns_internal): New insn having previous
499 behavior of altivec_vperm_<mode>_uns.
501 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
503 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
504 (altivec_vsumsws): Add handling for -maltivec=be with a little
506 (altivec_vsumsws_direct): New.
507 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
510 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
512 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
513 vtable_pointer_value_to_binfo): New functions.
514 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
515 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
517 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
519 * config/nios2/nios2.md (load_got_register): Initialize GOT
520 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
521 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
523 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
525 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
526 preserverd by passthrough, do not propagate the type.
528 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
530 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
531 (mips_atomic_assign_expand_fenv): New function.
532 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
534 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
536 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
537 (__builtin_mips_set_fcsr): Likewise.
538 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
540 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
541 (mips16_expand_set_fcsr): Likewise.
542 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
543 (mips16_set_fcsr_stub): Likewise.
544 (mips16_get_fcsr_one_only_stub): New class.
545 (mips16_set_fcsr_one_only_stub): Likewise.
546 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
547 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
548 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
549 (hard_float): New availability predicate.
550 (mips_builtins): Add get_fcsr and set_fcsr.
551 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
552 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
553 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
554 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
555 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
558 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
560 * config/mips/mips.c (mips_one_only_stub): New class.
561 (mips_need_mips16_rdhwr_p): Replace with...
562 (mips16_rdhwr_stub): ...this new variable.
563 (mips16_stub_call_address): New function.
564 (mips16_rdhwr_one_only_stub): New class.
565 (mips_expand_thread_pointer): Use mips16_stub_call_address.
566 (mips_output_mips16_rdhwr): Delete.
567 (mips_finish_stub): New function.
568 (mips_code_end): Use it to handle rdhwr stubs.
570 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
573 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
574 when calculating size of integer atomic types.
576 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
578 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
580 2014-02-01 Jakub Jelinek <jakub@redhat.com>
582 PR tree-optimization/60003
583 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
584 * profile.c (branch_prob): Use gimple_call_builtin_p
585 to check for BUILT_IN_SETJMP_RECEIVER.
586 * tree-inline.c (copy_bb): Call notice_special_calls.
588 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
591 * lra-constraints.c (process_alt_operands): Update reload_sum only
594 2014-01-31 Richard Henderson <rth@redhat.com>
597 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
598 until after else_eh is processed.
600 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
602 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
603 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
604 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
605 in smmintrin.h, remove them.
606 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
607 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
608 * config/i386/i386.md (ROUND_SAE): Fix value.
609 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
610 (const48_operand): New.
611 * config/i386/subst.md (round), (round_expand): Use
612 const_4_or_8_to_11_operand.
613 (round_saeonly), (round_saeonly_expand): Use const48_operand.
615 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
617 * config/i386/constraints.md (Yk): Swap meaning with k.
618 * config/i386/i386.md (movhi_internal): Change Yk to k.
619 (movqi_internal): Ditto.
620 (*k<logic><mode>): Ditto.
623 (kandn<mode>): Ditto.
624 (*<code>hi_1): Ditto.
625 (*<code>qi_1): Ditto.
626 (kxnor<mode>): Ditto.
630 (*one_cmplhi2_1): Ditto.
631 (*one_cmplqi2_1): Ditto.
632 * config/i386/sse.md (): Change k to Yk.
633 (avx512f_load<mode>_mask): Ditto.
634 (avx512f_blendm<mode>): Ditto.
635 (avx512f_store<mode>_mask): Ditto.
636 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
637 (avx512f_storedqu<mode>_mask): Ditto.
638 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
640 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
641 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
642 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
643 (avx512f_maskcmp<mode>3): Ditto.
644 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
645 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
646 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
647 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
648 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
649 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
650 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
651 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
652 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
653 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
654 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
655 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
656 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
657 (vec_extract_lo_<mode>_maskm): Ditto.
658 (vec_extract_hi_<mode>_maskm): Ditto.
659 (avx512f_vternlog<mode>_mask): Ditto.
660 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
661 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
662 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
663 (avx512f_<code>v8div16qi2_mask): Ditto.
664 (avx512f_<code>v8div16qi2_mask_store): Ditto.
665 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
666 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
667 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
668 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
669 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
670 (*avx512pf_gatherpf<mode>df_mask): Ditto.
671 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
672 (*avx512pf_scatterpf<mode>df_mask): Ditto.
673 (avx512cd_maskb_vec_dupv8di): Ditto.
674 (avx512cd_maskw_vec_dupv16si): Ditto.
675 (avx512f_vpermi2var<mode>3_maskz): Ditto.
676 (avx512f_vpermi2var<mode>3_mask): Ditto.
677 (avx512f_vpermi2var<mode>3_mask): Ditto.
678 (avx512f_vpermt2var<mode>3_maskz): Ditto.
679 (*avx512f_gathersi<mode>): Ditto.
680 (*avx512f_gathersi<mode>_2): Ditto.
681 (*avx512f_gatherdi<mode>): Ditto.
682 (*avx512f_gatherdi<mode>_2): Ditto.
683 (*avx512f_scattersi<mode>): Ditto.
684 (*avx512f_scatterdi<mode>): Ditto.
685 (avx512f_compress<mode>_mask): Ditto.
686 (avx512f_compressstore<mode>_mask): Ditto.
687 (avx512f_expand<mode>_mask): Ditto.
688 * config/i386/subst.md (mask): Change k to Yk.
689 (mask_scalar_merge): Ditto.
692 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
694 * doc/extend.texi (Vector Extensions): Document ?: in C++.
696 2014-01-31 Richard Biener <rguenther@suse.de>
699 * builtins.c (fold_builtin_memory_op): Make sure to not
700 use a floating-point mode or a boolean or enumeral type for
703 2014-01-30 DJ Delorie <dj@redhat.com>
705 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
706 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
707 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
708 whenever main() has an epilogue.
710 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
712 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
713 unused variable "field".
714 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
715 (vsx_mergeh_<mode>): Likewise.
716 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
717 (altivec_vmrghh): Likewise.
718 (altivec_vmrghw): Likewise.
719 (altivec_vmrglb): Likewise.
720 (altivec_vmrglh): Likewise.
721 (altivec_vmrglw): Likewise.
722 (altivec_vspltb): Add missing uses.
723 (altivec_vsplth): Likewise.
724 (altivec_vspltw): Likewise.
725 (altivec_vspltsf): Likewise.
727 2014-01-30 Jakub Jelinek <jakub@redhat.com>
730 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
731 frame related instructions.
733 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
735 PR rtl-optimization/59959
736 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
737 any reload of register whose subreg is invalid.
739 2014-01-30 Jakub Jelinek <jakub@redhat.com>
741 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
742 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
743 Add missing return type - void.
745 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
747 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
748 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
749 remove element index adjustment for endian (now handled in vsx.md
751 (altivec_expand_vec_perm_const): Use
752 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
753 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
754 (vsx_xxspltw_<mode>): Adjust element index for little endian.
755 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
756 define_expand and a new define_insn *altivec_vspltb_internal;
757 adjust for -maltivec=be on a little endian target.
758 (altivec_vspltb_direct): New.
759 (altivec_vsplth): Divide into a define_expand and a new
760 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
761 little endian target.
762 (altivec_vsplth_direct): New.
763 (altivec_vspltw): Divide into a define_expand and a new
764 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
765 little endian target.
766 (altivec_vspltw_direct): New.
767 (altivec_vspltsf): Divide into a define_expand and a new
768 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
769 a little endian target.
771 2014-01-30 Richard Biener <rguenther@suse.de>
773 PR tree-optimization/59993
774 * tree-ssa-forwprop.c (associate_pointerplus): Check we
775 can propagate form the earlier stmt and avoid the transform
776 when the intermediate result is needed.
778 2014-01-30 Alangi Derick <alangiderick@gmail.com>
780 * README.Portability: Fix typo.
782 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
784 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
785 comparison_operator with ordered_comparison_operator.
787 2014-01-30 Nick Clifton <nickc@redhat.com>
789 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
790 Rename to mn10300_store_multiple_regs.
791 * config/mn10300/mn10300.c: Likewise.
792 * config/mn10300/mn10300.md (store_movm): Fix typo: call
794 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
795 Call mn10300_store_multiple_regs.
797 2014-01-30 Nick Clifton <nickc@redhat.com>
798 DJ Delorie <dj@redhat.com>
800 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
801 %fp 2 to keep registers after it properly word-aligned.
802 (rl78_alloc_physical_registers_umul): Handle the case where both
803 input operands are the same.
805 2014-01-30 Richard Biener <rguenther@suse.de>
807 PR tree-optimization/59903
808 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
811 2014-01-30 Jason Merrill <jason@redhat.com>
814 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
817 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
819 2014-01-30 Richard Biener <rguenther@suse.de>
821 PR tree-optimization/59951
822 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
824 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
827 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
828 SFmode to DFmode case.
830 2014-01-29 DJ Delorie <dj@redhat.com>
832 * config/msp430/msp430.opt (-minrt): New.
833 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
835 (ENDFILE_SPEC): Likewise.
837 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
839 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
840 (estimate_function_body_sizes): Use it.
842 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
845 * dwarf2out.c (is_cxx_auto): New.
846 (is_base_type): Use it.
847 (gen_type_die_with_usage): Likewise.
849 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
851 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
852 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
853 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
854 -maltivec=be with LE targets.
855 (vsx_mergeh_<mode>): Likewise.
856 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
857 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
858 (altivec_vmrghb): Replace with define_expand and new
859 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
860 (altivec_vmrghb_direct): New define_insn.
861 (altivec_vmrghh): Replace with define_expand and new
862 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
863 (altivec_vmrghh_direct): New define_insn.
864 (altivec_vmrghw): Replace with define_expand and new
865 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
866 (altivec_vmrghw_direct): New define_insn.
867 (*altivec_vmrghsf): Adjust for endianness.
868 (altivec_vmrglb): Replace with define_expand and new
869 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
870 (altivec_vmrglb_direct): New define_insn.
871 (altivec_vmrglh): Replace with define_expand and new
872 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
873 (altivec_vmrglh_direct): New define_insn.
874 (altivec_vmrglw): Replace with define_expand and new
875 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
876 (altivec_vmrglw_direct): New define_insn.
877 (*altivec_vmrglsf): Adjust for endianness.
878 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
879 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
880 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
881 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
882 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
883 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
884 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
885 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
887 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
889 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
890 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
893 2014-01-29 Richard Biener <rguenther@suse.de>
895 PR tree-optimization/58742
896 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
897 associate_pointerplus_align.
898 (associate_pointerplus_diff): New function.
899 (associate_pointerplus): Likewise. Call associate_pointerplus_align
900 and associate_pointerplus_diff.
902 2014-01-29 Richard Biener <rguenther@suse.de>
904 * lto-streamer.h (LTO_major_version): Bump to 3.
905 (LTO_minor_version): Reset to 0.
907 2014-01-29 Renlin Li <Renlin.Li@arm.com>
909 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
910 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
911 (arm_file_start): Generate correct asm header for armv7ve.
912 * config/arm/bpabi.h: Add multilib support for armv7ve.
913 * config/arm/driver-arm.c: Change the architectures of cortex-a7
914 and cortex-a15 to armv7ve.
915 * config/arm/t-aprofile: Add multilib support for armv7ve.
916 * doc/invoke.texi: Document -march=armv7ve.
918 2014-01-29 Richard Biener <rguenther@suse.de>
920 PR tree-optimization/58742
921 * tree-ssa-forwprop.c (associate_plusminus): Return true
922 if we changed sth, defer EH cleanup to ...
923 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
924 (simplify_mult): New function.
926 2014-01-29 Jakub Jelinek <jakub@redhat.com>
929 PR tree-optimization/59920
930 * tree.c (build_common_builtin_nodes): Remove
931 __builtin_setjmp_dispatcher initialization.
932 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
933 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
934 instead of gsi_after_labels + manually skipping debug stmts.
935 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
936 ignore bbs with IFN_ABNORMAL_DISPATCHER.
937 * tree-inline.c (copy_edges_for_bb): Remove
938 can_make_abnormal_goto argument, instead add abnormal_goto_dest
939 argument. Ignore computed_goto_p stmts. Don't call
940 make_abnormal_goto_edges. If a call might need abnormal edges
941 for non-local gotos, see if it already has an edge to
942 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
943 with true argument, don't do anything then, otherwise add
944 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
945 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
947 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
948 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
949 (lower_stmt): Don't set data->calls_builtin_setjmp.
950 (lower_builtin_setjmp): Adjust comment.
951 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
952 * tree-cfg.c (found_computed_goto): Remove.
953 (factor_computed_gotos): Remove.
954 (make_goto_expr_edges): Return bool, true for computed gotos.
955 Don't call make_abnormal_goto_edges.
956 (build_gimple_cfg): Don't set found_computed_goto, don't call
957 factor_computed_gotos.
958 (computed_goto_p): No longer static.
959 (make_blocks): Don't set found_computed_goto.
960 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
961 (make_edges): If make_goto_expr_edges returns true, push bb
962 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
963 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
964 vector. Record mapping between bbs and OpenMP regions if there
965 are any, adjust make_gimple_omp_edges caller. Call
966 handle_abnormal_edges.
967 (make_abnormal_goto_edges): Remove.
968 * tree-cfg.h (make_abnormal_goto_edges): Remove.
969 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
970 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
971 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
972 * internal-fn.def (ABNORMAL_DISPATCHER): New.
973 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
974 filling *region also set *region_idx to (*region)->entry->index.
977 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
978 For REGs set ORIGINAL_REGNO.
980 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
982 * doc/md.texi: Mention that a target shouldn't implement
983 vec_widen_(s|u)mul_even/odd pair if it is less efficient
986 2014-01-29 Jakub Jelinek <jakub@redhat.com>
988 PR tree-optimization/59594
989 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
990 a copy of the datarefs vector rather than the vector itself.
992 2014-01-28 Jason Merrill <jason@redhat.com>
995 * dwarf2out.c (auto_die): New static.
996 (gen_type_die_with_usage): Handle C++1y 'auto'.
997 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1000 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1003 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1004 (SPEC_X32): Likewise.
1005 (SPEC_64): Likewise.
1006 * config/i386/i386.c (ix86_option_override_internal): Turn off
1007 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1009 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1010 * config/i386/i386.h (TARGET_16BIT): New macro.
1011 (TARGET_16BIT_P): Likewise.
1012 * config/i386/i386.opt: Add m16.
1013 * doc/invoke.texi: Document -m16.
1015 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1017 PR preprocessor/59935
1018 * input.c (location_get_source_line): Bail out on when line number
1019 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1021 2014-01-28 Richard Biener <rguenther@suse.de>
1023 PR tree-optimization/58742
1024 * tree-ssa-forwprop.c (associate_plusminus): Handle
1025 pointer subtraction of the form (T)(P + A) - (T)P.
1027 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1029 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1032 2014-01-28 Richard Biener <rguenther@suse.de>
1035 2014-01-28 Richard Biener <rguenther@suse.de>
1037 PR rtl-optimization/45364
1038 PR rtl-optimization/59890
1039 * var-tracking.c (local_get_addr_clear_given_value): Handle
1040 already cleared slot.
1041 (val_reset): Handle not allocated local_get_addr_cache.
1042 (vt_find_locations): Use post-order on the inverted CFG.
1044 2014-01-28 Richard Biener <rguenther@suse.de>
1046 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1048 2014-01-28 Richard Biener <rguenther@suse.de>
1050 PR rtl-optimization/45364
1051 PR rtl-optimization/59890
1052 * var-tracking.c (local_get_addr_clear_given_value): Handle
1053 already cleared slot.
1054 (val_reset): Handle not allocated local_get_addr_cache.
1055 (vt_find_locations): Use post-order on the inverted CFG.
1057 2014-01-28 Alan Modra <amodra@gmail.com>
1059 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1060 * configure.ac <recursive call for build != host>: Define
1061 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1062 and LD_FOR_BUILD too.
1063 * configure: Regenerate.
1065 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1067 * config/i386/i386.c (get_builtin_code_for_version): Separate
1068 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1069 Broadwell from Haswell.
1071 2014-01-27 Steve Ellcey <sellcey@mips.com>
1073 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1074 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1075 * config/mips/mips.c (mips_option_override): Change setting
1077 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1078 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1079 Change from Mask to Var.
1081 2014-01-27 Jeff Law <law@redhat.com>
1083 * ipa-inline.c (inline_small_functions): Fix typo.
1085 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1087 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1088 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1089 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1090 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1091 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1092 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1093 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1094 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1095 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1096 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1097 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1098 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1099 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1100 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1101 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1102 (_mm512_storeu_epi64): Ditto.
1103 (_mm512_cmpge_epi32_mask): Ditto.
1104 (_mm512_cmpge_epu32_mask): Ditto.
1105 (_mm512_cmpge_epi64_mask): Ditto.
1106 (_mm512_cmpge_epu64_mask): Ditto.
1107 (_mm512_cmple_epi32_mask): Ditto.
1108 (_mm512_cmple_epu32_mask): Ditto.
1109 (_mm512_cmple_epi64_mask): Ditto.
1110 (_mm512_cmple_epu64_mask): Ditto.
1111 (_mm512_cmplt_epi32_mask): Ditto.
1112 (_mm512_cmplt_epu32_mask): Ditto.
1113 (_mm512_cmplt_epi64_mask): Ditto.
1114 (_mm512_cmplt_epu64_mask): Ditto.
1115 (_mm512_cmpneq_epi32_mask): Ditto.
1116 (_mm512_cmpneq_epu32_mask): Ditto.
1117 (_mm512_cmpneq_epi64_mask): Ditto.
1118 (_mm512_cmpneq_epu64_mask): Ditto.
1119 (_mm512_expand_pd): Ditto.
1120 (_mm512_expand_ps): Ditto.
1121 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1122 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1123 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1124 * config/i386/i386.c (ix86_builtins): Add
1125 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1126 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1127 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1128 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1129 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1130 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1131 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1132 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1133 IX86_BUILTIN_PMOVUSQW512_MEM.
1134 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1135 __builtin_ia32_pmovsqd512mem_mask,
1136 __builtin_ia32_pmovqd512mem_mask,
1137 __builtin_ia32_pmovusqw512mem_mask,
1138 __builtin_ia32_pmovsqw512mem_mask,
1139 __builtin_ia32_pmovqw512mem_mask,
1140 __builtin_ia32_pmovusdw512mem_mask,
1141 __builtin_ia32_pmovsdw512mem_mask,
1142 __builtin_ia32_pmovdw512mem_mask,
1143 __builtin_ia32_pmovqb512mem_mask,
1144 __builtin_ia32_pmovusqb512mem_mask,
1145 __builtin_ia32_pmovsqb512mem_mask,
1146 __builtin_ia32_pmovusdb512mem_mask,
1147 __builtin_ia32_pmovsdb512mem_mask,
1148 __builtin_ia32_pmovdb512mem_mask.
1149 (bdesc_args): Add __builtin_ia32_expanddf512,
1150 __builtin_ia32_expandsf512.
1151 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1152 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1153 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1154 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1155 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1156 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1157 (avx512f_<code>v8div16qi2_mask_store): This.
1158 (avx512f_expand<mode>): New.
1160 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1162 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1164 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1165 (_mm512_prefetch_i32scatter_pd): Ditto.
1166 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1167 (_mm512_prefetch_i64scatter_pd): Ditto.
1168 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1169 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1170 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1171 (_mm512_prefetch_i32scatter_ps): Ditto.
1172 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1173 (_mm512_prefetch_i64scatter_ps): Ditto.
1174 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1175 * config/i386/i386-builtin-types.def: Define
1176 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1177 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1178 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1179 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1180 IX86_BUILTIN_SCATTERPFQPD.
1181 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1182 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1183 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1184 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1185 __builtin_ia32_scatterpfqps.
1186 (ix86_expand_builtin): Expand new built-ins.
1187 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1188 fix memory access data type.
1189 (*avx512pf_gatherpf<mode>_mask): Ditto.
1190 (*avx512pf_gatherpf<mode>): Ditto.
1191 (avx512pf_scatterpf<mode>): Ditto.
1192 (*avx512pf_scatterpf<mode>_mask): Ditto.
1193 (*avx512pf_scatterpf<mode>): Ditto.
1194 (GATHER_SCATTER_SF_MEM_MODE): New.
1195 (avx512pf_gatherpf<mode>df): Ditto.
1196 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1197 (*avx512pf_scatterpf<mode>df): Ditto.
1199 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1202 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1203 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1206 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1208 * common/config/arm/arm-common.c
1209 (arm_rewrite_mcpu): Handle multiple names.
1211 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1213 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1215 * gimple-builder.h (create_gimple_tmp): Delete.
1217 2014-01-27 Christian Bruel <christian.bruel@st.com>
1219 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1222 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1224 * config/pa/pa.md (call): Generate indirect long calls to non-local
1225 functions when outputing 32-bit code.
1226 (call_value): Likewise except for special call to buggy powf function.
1228 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1229 portable runtime and PIC indirect calls.
1230 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1231 and PIC call sequences. Use ldo instead of blr to set return register
1232 in PIC call sequence.
1234 2014-01-25 Walter Lee <walt@tilera.com>
1236 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1237 avoid clobbering a live register.
1239 2014-01-25 Walter Lee <walt@tilera.com>
1241 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1242 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1243 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1244 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1246 2014-01-25 Walter Lee <walt@tilera.com>
1248 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1249 arguments on even registers.
1250 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1252 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1253 (BIGGEST_ALIGNMENT): Ditto.
1254 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1256 2014-01-25 Walter Lee <walt@tilera.com>
1258 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1259 insns before bundling.
1260 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1262 2014-01-25 Walter Lee <walt@tilera.com>
1264 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1265 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1266 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1268 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1270 * config/mips/constraints.md (kl): Delete.
1271 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1272 define expands, using...
1273 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1274 instructions for MIPS16.
1275 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1276 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1278 2014-01-25 Walter Lee <walt@tilera.com>
1280 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1284 2014-01-25 Walter Lee <walt@tilera.com>
1286 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1287 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1289 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1291 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1294 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1296 * print-rtl.c (in_call_function_usage): New var.
1297 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1298 EXPR_LIST mode as mode and not as reg note name.
1301 * cfgloopmanip.c (copy_loop_info): If
1302 loop->warned_aggressive_loop_optimizations, make sure
1303 the flag is set in target loop too.
1305 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1307 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1309 * builtins.def: Likewise.
1310 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1311 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1312 * ira.c (ira_setup_eliminable_regset): Likewise.
1313 * omp-low.c (gate_expand_omp): Likewise.
1314 (execute_lower_omp): Likewise.
1315 (diagnose_sb_0): Likewise.
1316 (gate_diagnose_omp_blocks): Likewise.
1317 (simd_clone_clauses_extract): Likewise.
1320 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1322 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1323 correction for little endian...
1324 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1327 2014-01-24 Jeff Law <law@redhat.com>
1329 PR tree-optimization/59919
1330 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1331 for non-returning calls.
1333 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1335 * common/config/aarch64/aarch64-common.c
1336 (aarch64_rewrite_mcpu): Handle multiple names.
1337 * config/aarch64/aarch64.h
1338 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1340 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1342 * input.c (add_file_to_cache_tab): Handle the case where fopen
1345 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1348 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1349 from push operand if code of push isn't PRE_DEC.
1351 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1354 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1355 -mquad-memory-atomic. Update -mquad-memory documentation to say
1356 it is only used for non-atomic loads/stores.
1358 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1359 -mquad-memory or -mquad-memory-atomic switches.
1361 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1362 -mquad-memory-atomic to ISA 2.07 support.
1364 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1365 to separate support of normal quad word memory operations (ldq, stq)
1366 from the atomic quad word memory operations.
1368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1369 support to separate non-atomic quad word operations from atomic
1370 quad word operations. Disable non-atomic quad word operations in
1371 little endian mode so that we don't have to swap words after the
1372 load and before the store.
1373 (quad_load_store_p): Add comment about atomic quad word support.
1374 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1375 options printed with -mdebug=reg.
1377 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1378 -mquad-memory-atomic as the test for whether we have quad word
1379 atomic instructions.
1380 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1381 or -mp8-vector are used, allow byte/half-word atomic operations.
1383 * config/rs6000/sync.md (load_lockedti): Insure that the address
1384 is a proper indexed or indirect address for the lqarx instruction.
1385 On little endian systems, swap the hi/lo registers after the lqarx
1387 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1388 insure the address is valid for the lqarx instruction.
1389 (store_conditionalti): Insure that the address is a proper indexed
1390 or indirect address for the stqcrx. instruction. On little endian
1391 systems, swap the hi/lo registers before doing the stqcrx.
1393 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1394 insure the address is valid for the stqcrx. instruction.
1396 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1397 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1398 type of quad memory support is available.
1400 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1403 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1404 there is a danger of looping.
1406 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1408 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1409 force flag_ira_loop_pressure if set via command line.
1411 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1413 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1414 (ashr_simd): New builtin handling DI mode.
1415 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1416 (aarch64_sshr_simddi): New match pattern.
1417 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1418 (vshrd_n_s64): Likewise.
1419 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1421 2014-01-23 Nick Clifton <nickc@redhat.com>
1423 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1424 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1425 favour of mcu specific scripts.
1426 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1429 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1430 Alex Velenko <Alex.Velenko@arm.com>
1432 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1433 (vaddv_s16): Likewise.
1434 (vaddv_s32): Likewise.
1435 (vaddv_u8): Likewise.
1436 (vaddv_u16): Likewise.
1437 (vaddv_u32): Likewise.
1438 (vaddvq_s8): Likewise.
1439 (vaddvq_s16): Likewise.
1440 (vaddvq_s32): Likewise.
1441 (vaddvq_s64): Likewise.
1442 (vaddvq_u8): Likewise.
1443 (vaddvq_u16): Likewise.
1444 (vaddvq_u32): Likewise.
1445 (vaddvq_u64): Likewise.
1446 (vaddv_f32): Likewise.
1447 (vaddvq_f32): Likewise.
1448 (vaddvq_f64): Likewise.
1449 (vmaxv_f32): Likewise.
1450 (vmaxv_s8): Likewise.
1451 (vmaxv_s16): Likewise.
1452 (vmaxv_s32): Likewise.
1453 (vmaxv_u8): Likewise.
1454 (vmaxv_u16): Likewise.
1455 (vmaxv_u32): Likewise.
1456 (vmaxvq_f32): Likewise.
1457 (vmaxvq_f64): Likewise.
1458 (vmaxvq_s8): Likewise.
1459 (vmaxvq_s16): Likewise.
1460 (vmaxvq_s32): Likewise.
1461 (vmaxvq_u8): Likewise.
1462 (vmaxvq_u16): Likewise.
1463 (vmaxvq_u32): Likewise.
1464 (vmaxnmv_f32): Likewise.
1465 (vmaxnmvq_f32): Likewise.
1466 (vmaxnmvq_f64): Likewise.
1467 (vminv_f32): Likewise.
1468 (vminv_s8): Likewise.
1469 (vminv_s16): Likewise.
1470 (vminv_s32): Likewise.
1471 (vminv_u8): Likewise.
1472 (vminv_u16): Likewise.
1473 (vminv_u32): Likewise.
1474 (vminvq_f32): Likewise.
1475 (vminvq_f64): Likewise.
1476 (vminvq_s8): Likewise.
1477 (vminvq_s16): Likewise.
1478 (vminvq_s32): Likewise.
1479 (vminvq_u8): Likewise.
1480 (vminvq_u16): Likewise.
1481 (vminvq_u32): Likewise.
1482 (vminnmv_f32): Likewise.
1483 (vminnmvq_f32): Likewise.
1484 (vminnmvq_f64): Likewise.
1486 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1488 * config/aarch64/aarch64-simd.md
1489 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1490 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1491 (*aarch64_mul3_elt<mode>): Likewise.
1492 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1493 (*aarch64_mul3_elt_to_64v2df): Likewise.
1494 (*aarch64_mla_elt<mode>): Likewise.
1495 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1496 (*aarch64_mls_elt<mode>): Likewise.
1497 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1498 (*aarch64_fma4_elt<mode>): Likewise.
1499 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1500 (*aarch64_fma4_elt_to_64v2df): Likewise.
1501 (*aarch64_fnma4_elt<mode>): Likewise.
1502 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1503 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1504 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1505 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1506 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1507 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1508 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1509 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1510 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1512 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1514 * config/aarch64/aarch64-simd.md
1515 (aarch64_be_checked_get_lane<mode>): New define_expand.
1516 * config/aarch64/aarch64-simd-builtins.def
1517 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1518 New builtin definition.
1519 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1520 Use new safe be builtin.
1522 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1524 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1526 (aarch64_be_st1<mode>): Likewise.
1527 (aarch_ld1<VALL:mode>): Define_expand modified.
1528 (aarch_st1<VALL:mode>): Likewise.
1529 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1530 (UNSPEC_ST1): Likewise.
1532 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1534 * config/microblaze/microblaze.md: Add trap insn and attribute
1536 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1538 PR preprocessor/58580
1539 * input.h (location_get_source_line): Take an additional line_size
1541 (void diagnostics_file_cache_fini): Declare new function.
1542 * input.c (struct fcache): New type.
1543 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1544 New static constants.
1545 (diagnostic_file_cache_init, total_lines_num)
1546 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1547 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1548 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1549 (get_next_line, read_next_line, goto_next_line, read_line_num):
1550 New static function definitions.
1551 (diagnostic_file_cache_fini): New function.
1552 (location_get_source_line): Take an additional output line_len
1553 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1555 * diagnostic.c (diagnostic_finish): Call
1556 diagnostic_file_cache_fini.
1557 (adjust_line): Take an additional input parameter for the length
1558 of the line, rather than calculating it with strlen.
1559 (diagnostic_show_locus): Adjust the use of
1560 location_get_source_line and adjust_line with respect to their new
1561 signature. While displaying a line now, do not stop at the first
1562 null byte. Rather, display the zero byte as a space and keep
1563 going until we reach the size of the line.
1564 * Makefile.in: Add vec.o to OBJS-libcommon
1566 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1567 Ilya Tocar <ilya.tocar@intel.com>
1569 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1570 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1571 (__builtin_ia32_kmov16): Ditto.
1572 * config/i386/i386.md (UNSPEC_KMOV): New.
1575 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1577 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1578 (_mm512_storeu_si512): Ditto.
1580 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1583 * rtl.h (get_referenced_operands): Declare.
1584 * recog.c (get_referenced_operands): New function.
1585 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1586 operands have been referenced when recording LO_SUM references.
1588 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1590 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1592 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1594 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1595 Enable for generic and recent AMD targets.
1597 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1599 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1600 ARG_SIZE note when adjustment was eliminated.
1602 2014-01-22 Jeff Law <law@redhat.com>
1604 PR tree-optimization/59597
1605 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1606 in file. Accept new argument REGISTERING and use it to modify
1607 dump output appropriately.
1608 (register_jump_thread): Corresponding changes.
1609 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1610 thread paths involving joiner blocks. Add code to dump cancelled
1611 jump threading paths.
1613 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1615 PR rtl-optimization/59477
1616 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1617 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1619 2014-01-22 Tom Tromey <tromey@redhat.com>
1621 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1623 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1625 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1627 PR rtl-optimization/59896
1628 * lra-constraints.c (process_alt_operands): Check unused note for
1629 matched operands of insn with no output reloads.
1631 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1633 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1634 (mips_move_from_gpr_cost): Likewise.
1636 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1638 PR rtl-optimization/59858
1639 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1640 ira_class_hard_regs_num.
1641 (process_alt_operands): Increase reject for dying matched operand.
1643 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1646 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1647 smaller than size, perform several stores or loads and stores
1648 at dst + count - size to store or copy all of size bytes, rather
1649 than just last modesize bytes.
1651 2014-01-20 DJ Delorie <dj@redhat.com>
1653 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1654 that CLOBBERs are REGs before propogating their values.
1656 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1659 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1660 (cgraph_inline_failed_type): New function.
1661 * cgraph.h (DEFCIFCODE): Add type.
1662 (cgraph_inline_failed_type_t): New enum.
1663 (cgraph_inline_failed_type): New prototype.
1664 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1665 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1666 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1667 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1668 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1669 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1670 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1671 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1672 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1673 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1674 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1675 OPTIMIZATION_MISMATCH.
1676 * tree-inline.c (expand_call_inline): Emit errors during
1677 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1679 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1682 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1683 mode attribute in insn output.
1685 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1687 * output.h (output_constant): Delete.
1688 * varasm.c (output_constant): Make private.
1690 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1692 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1694 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1697 * tree.h (fold_builtin_strcat): New prototype.
1698 * builtins.c (fold_builtin_strcat): No longer static. Add len
1699 argument, if non-NULL, don't call c_strlen. Optimize
1700 directly into __builtin_memcpy instead of __builtin_strcpy.
1701 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1702 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1704 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1706 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1707 for SImode_address_operand operands, having only a REG argument.
1709 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1711 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1712 loader name using mbig-endian.
1713 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1715 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1717 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1721 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1723 * config/aarch64/aarch64-protos.h
1724 (aarch64_cannot_change_mode_class_ptr): Declare.
1725 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1726 aarch64_cannot_change_mode_class_ptr): New.
1727 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1728 backend hook aarch64_cannot_change_mode_class.
1730 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1732 * common/config/aarch64/aarch64-common.c
1733 (aarch64_handle_option): Don't handle any option order logic here.
1734 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1735 selected_cpu, warn on architecture version mismatch.
1736 (aarch64_override_options): Fix parsing order for option strings.
1738 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1739 Iain Sandoe <iain@codesourcery.com>
1742 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1743 warning. Amend comment to reflect current functionality.
1745 2014-01-20 Richard Biener <rguenther@suse.de>
1748 * builtins.c (fold_builtin_strcat): Remove case better handled
1749 by tree-ssa-strlen.c.
1751 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1753 * config/aarch64/aarch64.opt
1754 (mcpu, march, mtune): Make case-insensitive.
1756 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1759 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1760 if operands[1] is a REG or ZERO_EXTEND of a REG.
1762 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
1764 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1766 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1768 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1769 long non-pic millicode calls.
1771 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1773 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1775 2014-01-19 Kito Cheng <kito@0xlab.org>
1777 * builtins.c (expand_movstr): Check movstr expand done or fail.
1779 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1780 H.J. Lu <hongjiu.lu@intel.com>
1783 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1784 to DImode for zero-extended addresses.
1786 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1788 PR rtl-optimization/57763
1789 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1790 on the new indirect jump_insn and increment LABEL_NUSES (label).
1792 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1796 * config.gcc (x86_archs): New variable.
1797 (x86_64_archs): Likewise.
1798 (x86_cpus): Likewise.
1799 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1800 --with-arch/--with-cpu= options.
1801 Support --with-arch=/--with-cpu={nehalem,westmere,
1802 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1804 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1806 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1807 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1809 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1811 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1813 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1816 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1817 clear cpp_get_options (parse_in)->warn_unused_macros for
1818 ix86_target_macros_internal with cpp_define.
1820 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1822 * jump.c (delete_related_insns): Keep (use (insn))s.
1823 * reorg.c (redundant_insn): Check for barriers too.
1825 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1827 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1829 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1831 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1832 call to $$dyncall when TARGET_LONG_CALLS is true.
1834 2014-01-17 Jeff Law <law@redhat.com>
1836 * ree.c (combine_set_extension): Temporarily disable test for
1837 changing number of hard registers.
1839 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1842 * ipa-inline-analysis.c (inline_free_summary):
1843 Do not free summary of aliases.
1845 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1848 * gimplify.c (gimplify_expr): Use create_tmp_var
1849 instead of create_tmp_var_raw. If cond doesn't have
1850 integral type, don't add the IFN_ANNOTATE builtin at all.
1852 2014-01-17 Martin Jambor <mjambor@suse.cz>
1855 * ipa-cp.c (prev_edge_clone): New variable.
1856 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1857 Also resize prev_edge_clone vector.
1858 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1859 (ipcp_edge_removal_hook): New function.
1860 (ipcp_driver): Register ipcp_edge_removal_hook.
1862 2014-01-17 Andrew Pinski <apinski@cavium.com>
1863 Steve Ellcey <sellcey@mips.com>
1866 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1869 2014-01-17 Jeff Law <law@redhat.com>
1872 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1873 so that pass_ccp runs first.
1875 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1877 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1878 (ix86_adjust_cost): Use !TARGET_XXX.
1879 (do_reorder_for_imul): Likewise.
1880 (swap_top_of_ready_list): Likewise.
1881 (ix86_sched_reorder): Likewise.
1883 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1885 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1886 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1887 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1888 (intel_memset): New. Duplicate slm_memset.
1889 (intel_cost): New. Duplicate slm_cost.
1890 (m_INTEL): New macro.
1891 (processor_target_table): Add "intel".
1892 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1893 with PROCESSOR_INTEL for "intel".
1894 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1895 PROCESSOR_SILVERMONT.
1896 (ix86_issue_rate): Likewise.
1897 (ix86_adjust_cost): Likewise.
1898 (ia32_multipass_dfa_lookahead): Likewise.
1899 (swap_top_of_ready_list): Likewise.
1900 (ix86_sched_reorder): Likewise.
1901 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1902 instead of TARGET_OPT_AGU.
1903 * config/i386/i386.h (TARGET_INTEL): New.
1904 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1905 (processor_type): Add PROCESSOR_INTEL.
1906 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1907 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1909 2014-01-17 Marek Polacek <polacek@redhat.com>
1912 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1915 2014-01-17 Richard Biener <rguenther@suse.de>
1917 PR tree-optimization/46590
1918 * opts.c (default_options_table): Add entries for
1919 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1920 all enabled at -O1 but not for -Og.
1921 * common.opt (fbranch-count-reg): Remove Init(1).
1922 (fmove-loop-invariants): Likewise.
1923 (ftree-pta): Likewise.
1925 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1927 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1928 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1929 decls to at least the GCC 4.8 used alignments.
1932 * tree-nested.c (convert_nonlocal_reference_stmt,
1933 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1934 of GIMPLE_BIND stmts, adjust associated decls.
1936 2014-01-17 Richard Biener <rguenther@suse.de>
1938 PR tree-optimization/46590
1939 * vec.h (vec<>::bseach): New member function implementing
1940 binary search according to C89 bsearch.
1941 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1942 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1943 bitmap pointer again. Make accesses_in_loop a flat array.
1944 (mem_ref_obstack): New global.
1945 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1946 (mark_ref_stored): Likewise.
1947 (ref_indep_loop_p_2): Likewise.
1948 (set_ref_stored_in_loop): New helper function.
1949 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1950 (memref_free): Adjust.
1951 (record_mem_ref_loc): Simplify.
1952 (gather_mem_refs_stmt): Adjust.
1953 (sort_locs_in_loop_postorder_cmp): New function.
1954 (analyze_memory_references): Sort accesses_in_loop after
1955 loop postorder number.
1956 (find_ref_loc_in_loop_cmp): New function.
1957 (for_all_locs_in_loop): Find relevant cluster of locs in
1958 accesses_in_loop and iterate without recursion.
1959 (execute_sm): Avoid uninit warning.
1960 (struct ref_always_accessed): Simplify.
1961 (ref_always_accessed::operator ()): Likewise.
1962 (ref_always_accessed_p): Likewise.
1963 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1964 loop postorder numbers here.
1965 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1968 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1971 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1972 on decls for which assemble_alias has been called.
1974 2014-01-17 Nick Clifton <nickc@redhat.com>
1976 * config/msp430/msp430.opt: (mcpu): New option.
1977 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1978 (msp430_option_override): Parse target_cpu. If the MCU name
1979 matches a generic string, clear target_mcu.
1980 (msp430_attr): Allow numeric interrupt values up to 63.
1981 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1982 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1984 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1986 * config/msp430/msp430.md (popm): Use %J rather than %I.
1987 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1988 (addhi_cy_i): Use immediate_operand for operand 2.
1989 * doc/invoke.texi: Document -mcpu option.
1991 2014-01-17 Richard Biener <rguenther@suse.de>
1993 PR rtl-optimization/38518
1994 * df.h (df_analyze_loop): Declare.
1995 * df-core.c: Include cfgloop.h.
1996 (df_analyze_1): Split out main part of df_analyze.
1997 (df_analyze): Adjust.
1998 (loop_inverted_post_order_compute): New function.
1999 (loop_post_order_compute): Likewise.
2000 (df_analyze_loop): New function avoiding whole-function
2002 * loop-invariant.c (find_defs): Use df_analyze_loop.
2003 (find_invariants): Adjust.
2004 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2006 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2008 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2009 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2011 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2013 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2014 traversal when removing references.
2016 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2019 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2021 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2024 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2025 pushing a reload for an autoinc when we had previously reloaded an
2026 inner part of the address.
2028 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2030 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2032 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2033 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2034 when not giving up or versioning for alias only because of
2036 (vect_analyze_data_ref_dependences): Set to true.
2037 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2039 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2040 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2044 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2047 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2048 operand 0 predicate for gathers, use a new pseudo as subtarget.
2050 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2053 * lra-constraints.c (process_alt_operands): Add printing debug info.
2054 Check absence of input/output reloads for matched operands too.
2056 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2058 PR rtl-optimization/59835
2059 * ira.c (ira_init_register_move_cost): Increase cost for
2062 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2064 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2066 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2069 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2070 non-register objects. Use gen_(high/low)part more consistently.
2073 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2076 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2077 endian support, remove tests for WORDS_BIG_ENDIAN.
2078 (p8_mfvsrd_3_<mode>): Likewise.
2079 (reload_gpr_from_vsx<mode>): Likewise.
2080 (reload_gpr_from_vsxsf): Likewise.
2081 (p8_mfvsrd_4_disf): Likewise.
2083 2014-01-16 Richard Biener <rguenther@suse.de>
2085 PR rtl-optimization/46590
2086 * lcm.c (compute_antinout_edge): Use postorder iteration.
2087 (compute_laterin): Use inverted postorder iteration.
2089 2014-01-16 Nick Clifton <nickc@redhat.com>
2092 * varasm.c (output_constant): Return the number of bytes actually
2094 (output_constructor_array_range): Update the field size with the
2095 number of bytes emitted by output_constant.
2096 (output_constructor_regular_field): Likewise. Also do not
2097 complain if the total number of bytes emitted is now greater
2098 than the expected fieldpos.
2099 * output.h (output_constant): Update prototype and descriptive comment.
2101 2014-01-16 Marek Polacek <polacek@redhat.com>
2104 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2105 it is error_mark_node.
2107 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2109 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2110 VALID_AVX256_REG_OR_OI_MODE.
2112 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2114 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2115 current procedure should be profiled.
2117 2014-01-15 Andrew Pinski <apinski@cavium.com>
2119 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2120 of moving from/to the STACK_REG register class.
2122 2014-01-15 Richard Henderson <rth@redhat.com>
2125 * reginfo.c (global_regs_decl): Globalize.
2126 * rtl.h (global_regs_decl): Declare.
2127 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2128 reserved via global_regs.
2130 2014-01-15 Teresa Johnson <tejohnson@google.com>
2132 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2134 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2136 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2137 and vmulosh rather than call gen_vec_widen_smult_*.
2138 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2139 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2140 (vec_widen_smult_even_v16qi): Likewise.
2141 (vec_widen_umult_even_v8hi): Likewise.
2142 (vec_widen_smult_even_v8hi): Likewise.
2143 (vec_widen_umult_odd_v16qi): Likewise.
2144 (vec_widen_smult_odd_v16qi): Likewise.
2145 (vec_widen_umult_odd_v8hi): Likewise.
2146 (vec_widen_smult_odd_v8hi): Likewise.
2147 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2148 vmuloub rather than call gen_vec_widen_umult_*.
2149 (vec_widen_umult_lo_v16qi): Likewise.
2150 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2151 vmulosb rather than call gen_vec_widen_smult_*.
2152 (vec_widen_smult_lo_v16qi): Likewise.
2153 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2154 rather than call gen_vec_widen_umult_*.
2155 (vec_widen_umult_lo_v8hi): Likewise.
2156 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2157 rather than call gen_vec_widen_smult_*.
2158 (vec_widen_smult_lo_v8hi): Likewise.
2160 2014-01-15 Jeff Law <law@redhat.com>
2162 PR tree-optimization/59747
2163 * ree.c (find_and_remove_re): Properly handle case where a second
2164 eliminated extension requires widening a copy created for elimination
2165 of a prior extension.
2166 (combine_set_extension): Ensure that the number of hard regs needed
2167 for a destination register does not change when we widen it.
2169 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2171 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2172 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2173 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2174 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2175 (avr-*-rtems*): Likewise.
2176 (bfin*-rtems*): Likewise.
2177 (moxie-*-rtems*): Likewise.
2178 (h8300-*-rtems*): Likewise.
2179 (i[34567]86-*-rtems*): Likewise.
2180 (lm32-*-rtems*): Likewise.
2181 (m32r-*-rtems*): Likewise.
2182 (m68k-*-rtems*): Likewise.
2183 (microblaze*-*-rtems*): Likewise.
2184 (mips*-*-rtems*): Likewise.
2185 (powerpc-*-rtems*): Likewise.
2186 (sh-*-rtems*): Likewise.
2187 (sparc-*-rtems*): Likewise.
2188 (sparc64-*-rtems*): Likewise.
2189 (v850-*-rtems*): Likewise.
2190 (m32c-*-rtems*): Likewise.
2192 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2194 PR rtl-optimization/59511
2195 * ira.c (ira_init_register_move_cost): Use memory costs for some
2196 cases of register move cost calculations.
2197 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2198 instead of BB frequency.
2199 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2200 * lra-assigns.c (find_hard_regno_for): Ditto.
2202 2014-01-15 Richard Biener <rguenther@suse.de>
2204 PR tree-optimization/59822
2205 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2206 (vectorizable_load): Use it to hoist defs of uses of invariant
2207 loads out of the loop.
2209 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2210 Kugan Vivekanandarajah <kuganv@linaro.org>
2213 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2216 2014-01-15 Richard Biener <rguenther@suse.de>
2218 PR rtl-optimization/59802
2219 * lcm.c (compute_available): Use inverted postorder to seed
2220 the initial worklist.
2222 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2225 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2226 ADDR_REGS for invalid symrefs in non-PIC code.
2228 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2231 * builtins.c (determine_block_size): Initialize *probable_max_size
2232 even if len_rtx is CONST_INT.
2234 2014-01-14 Andrew Pinski <apinski@cavium.com>
2236 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2237 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2238 (cortexa53_tunings): Likewise.
2239 (aarch64_sched_issue_rate): New function.
2240 (TARGET_SCHED_ISSUE_RATE): Define.
2242 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2244 * ira-costs.c (find_costs_and_classes): Add missed
2245 ira_init_register_move_cost_if_necessary.
2247 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2250 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2252 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2255 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2256 to indicate if type is used for function return value. Warn ABI
2257 change if the vector mode isn't available for function return value.
2258 (ix86_function_arg_advance): Pass false to type_natural_mode.
2259 (ix86_function_arg): Likewise.
2260 (ix86_gimplify_va_arg): Likewise.
2261 (function_arg_32): Don't warn ABI change.
2262 (ix86_function_value): Pass true to type_natural_mode.
2263 (ix86_return_in_memory): Likewise.
2264 (ix86_struct_value_rtx): Removed.
2265 (TARGET_STRUCT_VALUE_RTX): Likewise.
2267 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2269 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2270 converting a conditional jump into a conditional return.
2272 2014-01-14 Richard Biener <rguenther@suse.de>
2274 PR tree-optimization/58921
2275 PR tree-optimization/59006
2276 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2277 hoisting invariant stmts.
2278 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2279 invariant loads on the preheader edge if possible.
2281 2014-01-14 Joey Ye <joey.ye@arm.com>
2283 * doc/plugin.texi (Building GCC plugins): Update to C++.
2285 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2287 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2288 (_mm_rcp28_round_ss): Ditto.
2289 (_mm_rsqrt28_round_sd): Ditto.
2290 (_mm_rsqrt28_round_ss): Ditto.
2291 (_mm_rcp28_sd): Ditto.
2292 (_mm_rcp28_ss): Ditto.
2293 (_mm_rsqrt28_sd): Ditto.
2294 (_mm_rsqrt28_ss): Ditto.
2295 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2296 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2297 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2298 (IX86_BUILTIN_RCP28SD): Ditto.
2299 (IX86_BUILTIN_RCP28SS): Ditto.
2300 (IX86_BUILTIN_RSQRT28SD): Ditto.
2301 (IX86_BUILTIN_RSQRT28SS): Ditto.
2302 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2303 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2304 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2305 (ix86_expand_special_args_builtin): Expand new FTYPE.
2306 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2307 (srcp14<mode>): Make insn unary.
2308 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2309 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2310 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2311 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2312 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2313 Fix rounding: make it SAE only.
2314 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2316 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2318 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2319 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2320 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2321 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2322 (round_saeonly_mask_scalar_operand4): Ditto.
2323 (round_saeonly_mask_scalar_op3): Ditto.
2324 (round_saeonly_mask_scalar_op4): Ditto.
2326 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2328 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2329 Implement -maltivec=be for vec_insert and vec_extract.
2331 2014-01-10 DJ Delorie <dj@redhat.com>
2333 * config/msp430/msp430.md (call_internal): Don't allow memory
2334 references with SP as the base register.
2335 (call_value_internal): Likewise.
2336 * config/msp430/constraints.md (Yc): New. For memory references
2337 that don't use SP as a base register.
2339 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2340 "an integer without a # prefix"
2341 * config/msp430/msp430.md (epilogue_helper): Use it.
2343 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2346 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2347 AVX512F gather builtins.
2348 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2349 on gather decls with INTEGER_TYPE masktype.
2350 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2351 directly into the builtin rather than hoisting it before loop.
2353 PR tree-optimization/59387
2354 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2355 (scev_const_prop): If folded_casts and type has undefined overflow,
2356 use force_gimple_operand instead of force_gimple_operand_gsi and
2357 for each added stmt if it is assign with
2358 arith_code_with_undefined_signed_overflow, call
2359 rewrite_to_defined_overflow.
2360 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2361 gimple-fold.h instead.
2362 (arith_code_with_undefined_signed_overflow,
2363 rewrite_to_defined_overflow): Moved to ...
2364 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2365 rewrite_to_defined_overflow): ... here. No longer static.
2366 Include gimplify-me.h.
2367 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2368 rewrite_to_defined_overflow): New prototypes.
2370 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2372 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2374 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2376 * builtins.c (get_object_alignment_2): Minor tweak.
2377 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2379 2014-01-13 Christian Bruel <christian.bruel@st.com>
2381 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2382 optimized non constant lengths.
2384 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2387 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2388 load as __atomic_load_N if possible.
2390 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2392 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2394 (rs6000_expand_builtin): Adjust call.
2396 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2399 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2400 * config/rs6000/rs6000.c: Include target-globals.h.
2401 (rs6000_set_current_function): Instead of doing target_reinit
2402 unconditionally, use save_target_globals_default_opts and
2403 restore_target_globals.
2405 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2407 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2408 (rs6000_expand_builtin): Handle mffs and mtfsf.
2409 (rs6000_init_builtins): Define mffs and mtfsf.
2410 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2411 (rs6000_mffs): New pattern.
2412 (rs6000_mtfsf): New pattern.
2414 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2416 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2417 Start narrowing with START. Apply candidate-use pair
2418 and check overall cost in narrowing.
2419 (iv_ca_prune): Pass new argument.
2421 2014-01-10 Jeff Law <law@redhat.com>
2424 * ree.c (combine_reaching_defs): Ensure the defining statement
2425 occurs before the extension when optimizing extensions with
2426 different source and destination hard registers.
2428 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2431 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2432 vtables into the type inheritance graph.
2434 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2436 PR rtl-optimization/59754
2437 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2438 modes in the REGNO != REGNO case.
2440 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2442 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2444 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2446 PR tree-optimization/59745
2447 * tree-predcom.c (tree_predictive_commoning_loop): Call
2448 free_affine_expand_cache if giving up because components is NULL.
2450 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2451 GC in payload of target_globals struct instead of allocating them on
2452 the heap and the larger structs separately using GC.
2453 * target-globals.h (struct target_globals): Make regs, hard_regs,
2454 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2455 of GTY((skip)) and change type to void *.
2456 (reset_target_globals): Cast loads from those fields to corresponding
2459 2014-01-10 Steve Ellcey <sellcey@mips.com>
2462 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2463 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2464 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2466 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2469 * aarch64-modes.def (CC_Zmode): New flags mode.
2470 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2471 represents an equality.
2472 (aarch64_get_condition_code): Handle CC_Zmode.
2473 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2475 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2477 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2478 extraction in good case.
2480 2014-01-10 Richard Biener <rguenther@suse.de>
2482 PR tree-optimization/59374
2483 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2484 checking after SLP discovery. Mark stmts not participating
2485 in any SLP instance properly.
2487 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2489 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2490 when handling a SET rtx.
2492 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2494 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2495 (cortex-a57): Likewise.
2496 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2498 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2500 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2501 non-iwmmxt builtins.
2503 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2507 * ipa-devirt.c record_target_from_binfo): Take as argument
2508 stack of binfos and lookup matching one for virtual inheritance.
2509 (possible_polymorphic_call_targets_1): Update.
2511 2014-01-10 Huacai Chen <chenhc@lemote.com>
2513 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2514 kernel strings for Loongson-2E/2F/3A.
2516 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2519 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2520 is_gimple_call before calling gimple_call_internal_p.
2522 2014-01-09 Steve Ellcey <sellcey@mips.com>
2524 * Makefile.in (TREE_FLOW_H): Remove.
2525 (TREE_SSA_H): Add file names from tree-flow.h.
2526 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2527 * tree.h: Remove tree-flow.h reference.
2528 * hash-table.h: Remove tree-flow.h reference.
2529 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2530 reference with tree-ssa-loop.h.
2532 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2534 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2535 default element-order behavior for -maltivec.
2536 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2537 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2538 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2539 when targeting big endian, at least for now.
2540 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2542 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2545 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2546 var satisfies use_register_for_decl, just take into account type
2547 alignment, rather than decl alignment.
2549 PR tree-optimization/59622
2550 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2551 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2552 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2553 Don't devirtualize for inplace at all. For targets.length () == 1,
2554 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2556 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2558 * config/i386/i386.md (cpu): Remove the unused btver1.
2560 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2562 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2564 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2567 * tree-core.h (struct target_globals): New forward declaration.
2568 (struct tree_target_option): Add globals field.
2569 * tree.h (TREE_TARGET_GLOBALS): Define.
2570 (prepare_target_option_nodes_for_pch): New prototype.
2571 * target-globals.h (struct target_globals): Define even if
2573 * tree.c (prepare_target_option_node_for_pch,
2574 prepare_target_option_nodes_for_pch): New functions.
2575 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2576 * config/i386/i386.c: Include target-globals.h.
2577 (ix86_set_current_function): Instead of doing target_reinit
2578 unconditionally, use save_target_globals_default_opts and
2579 restore_target_globals.
2581 2014-01-09 Richard Biener <rguenther@suse.de>
2583 PR tree-optimization/59715
2584 * tree-cfg.h (split_critical_edges): Declare.
2585 * tree-cfg.c (split_critical_edges): Export.
2586 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2588 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2590 * cfgexpand.c (expand_stack_vars): Optionally disable
2591 asan stack protection.
2592 (expand_used_vars): Likewise.
2593 (partition_stack_vars): Likewise.
2594 * asan.c (asan_emit_stack_protection): Optionally disable
2595 after return stack usage.
2596 (instrument_derefs): Optionally disable memory access instrumentation.
2597 (instrument_builtin_call): Likewise.
2598 (instrument_strlen_call): Likewise.
2599 (asan_protect_global): Optionally disable global variables protection.
2600 * doc/invoke.texi: Added doc for new options.
2601 * params.def: Added new options.
2602 * params.h: Likewise.
2604 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2606 PR rtl-optimization/59724
2607 * ifcvt.c (cond_exec_process_if_block): Don't call
2608 flow_find_head_matching_sequence with 0 longest_match.
2609 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2610 non-active insns if !stop_after.
2611 (try_head_merge_bb): Revert 2014-01-07 changes.
2613 2014-01-08 Jeff Law <law@redhat.com>
2615 * ree.c (get_sub_rtx): New function, extracted from...
2616 (merge_def_and_ext): Here.
2617 (combine_reaching_defs): Use get_sub_rtx.
2619 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2621 * cgraph.h (varpool_variable_node): Do not choke on null node.
2623 2014-01-08 Catherine Moore <clm@codesourcery.com>
2625 * config/mips/mips.md (simple_return): Attempt to use JRC
2627 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2629 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2631 PR rtl-optimization/59137
2632 * reorg.c (steal_delay_list_from_target): Call update_block for
2634 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2636 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2638 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2639 two duplicate entries.
2641 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2644 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2646 * config/mips/mips.c (mips_truncated_op_cost): New function.
2647 (mips_rtx_costs): Adjust test for BADDU.
2648 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2650 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2652 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2653 (*baddu_si): ...this new pattern.
2655 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2658 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2660 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2663 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2665 (expand_expr, expand_normal): Adjust.
2666 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2667 inner_reference_p. Use inner_reference_p to expand inner references.
2668 (store_expr): Adjust.
2669 * cfgexpand.c (expand_call_stmt): Adjust.
2671 2014-01-08 Rong Xu <xur@google.com>
2673 * gcov-io.c (gcov_var): Move from gcov-io.h.
2674 (gcov_position): Ditto.
2675 (gcov_is_error): Ditto.
2676 (gcov_rewrite): Ditto.
2677 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2678 only part to libgcc/libgcov.h.
2680 2014-01-08 Marek Polacek <polacek@redhat.com>
2683 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2685 2014-01-08 Marek Polacek <polacek@redhat.com>
2688 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2690 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2692 PR rtl-optimization/59649
2693 * stor-layout.c (get_mode_bounds): For BImode return
2694 0 and STORE_FLAG_VALUE.
2696 2014-01-08 Richard Biener <rguenther@suse.de>
2699 * gimple.h (is_gimple_builtin_call): Remove.
2700 (gimple_builtin_call_types_compatible_p): New.
2701 (gimple_call_builtin_p): New overload.
2702 * gimple.c (is_gimple_builtin_call): Remove.
2703 (validate_call): Rename to ...
2704 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2706 (validate_type): New static function.
2707 (gimple_call_builtin_p): New overload and adjust.
2708 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2709 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2710 (gimple_fold_stmt_to_constant_1): Likewise.
2711 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2713 2014-01-08 Richard Biener <rguenther@suse.de>
2716 * gimplify.c (gimplify_expr): Gimplify register-register type
2717 VIEW_CONVERT_EXPRs to separate stmts.
2719 2014-01-07 Jeff Law <law@redhat.com>
2722 * ree.c (combine_set_extension): Handle case where source
2723 and destination registers in an extension insn are different.
2724 (combine_reaching_defs): Allow source and destination registers
2725 in extension to be different under limited circumstances.
2726 (add_removable_extension): Remove restriction that the
2727 source and destination registers in the extension are the same.
2728 (find_and_remove_re): Emit a copy from the extension's
2729 destination to its source after the defining insn if
2730 the source and destination registers are different.
2733 * ifcvt.c (merge_if_block): If we are merging a block with more than
2734 one successor with a block with no successors, remove any BARRIER
2735 after the second block.
2737 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2739 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2741 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2744 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2745 for 14-bit register offsets when INT14_OK_STRICT is false.
2747 2014-01-07 Roland Stigge <stigge@antcom.de>
2748 Michael Meissner <meissner@linux.vnet.ibm.com>
2751 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2752 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2754 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2756 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2759 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2761 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2762 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2763 rtx is const0_rtx or not.
2765 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2768 * target-globals.c (save_target_globals): Remove this_fn_optab
2770 * toplev.c: Include optabs.h.
2771 (target_reinit): Temporarily restore the global options if another
2772 set of options are in force.
2774 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2776 PR rtl-optimization/58668
2777 * cfgcleanup.c (flow_find_cross_jump): Don't count
2778 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2779 to determine what is counted.
2780 (flow_find_head_matching_sequence): Use active_insn_p to determine
2782 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2784 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2785 determine what is counted.
2787 PR tree-optimization/59643
2788 * tree-predcom.c (split_data_refs_to_components): If one dr is
2789 read and one write, determine_offset fails and the write isn't
2790 in the bad component, just put the read into the bad component.
2792 2014-01-07 Mike Stump <mikestump@comcast.net>
2793 Jakub Jelinek <jakub@redhat.com>
2796 * tree-core.h (struct tree_optimization_option): Change optabs
2797 type from unsigned char * to void *.
2798 * optabs.c (init_tree_optimization_optabs): Adjust
2799 TREE_OPTIMIZATION_OPTABS initialization.
2801 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2804 * config/i386/i386.h (struct machine_function): Add
2805 no_drap_save_restore field.
2806 * config/i386/i386.c (ix86_save_reg): Use
2807 !cfun->machine->no_drap_save_restore instead of
2808 crtl->stack_realign_needed.
2809 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2810 this function clears frame_pointer_needed. Set
2811 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2812 and DRAP reg is needed.
2814 2014-01-06 Marek Polacek <polacek@redhat.com>
2817 * doc/implement-c.texi: Mention that other integer types are
2818 permitted as bit-field types in strictly conforming mode.
2820 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2822 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2825 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2827 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2829 2014-01-06 Martin Jambor <mjambor@suse.cz>
2832 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2834 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2836 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2840 * var-tracking.c (add_stores): Preserve the value of the source even if
2841 we don't record the store.
2843 2014-01-06 Terry Guo <terry.guo@arm.com>
2845 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2847 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2850 * config/darwin.c (darwin_function_section): Adjust return values to
2851 correspond to optimisation changes made in r206070.
2853 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2855 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2856 from prefetch_block tune setting.
2857 (nocona_cost): Correct size of prefetch block to 64.
2859 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2861 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2862 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2863 used to save the static chain register in the computation of the offset
2864 from which the FP registers need to be restored.
2866 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2868 PR tree-optimization/59519
2869 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2870 ICE if get_current_def (current_new_name) is already non-NULL, as long
2871 as it is a phi result of some other phi in *new_exit_bb that has
2874 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2875 or vmovdqu* for misaligned_operand.
2876 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2877 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2878 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2879 aligned_mem for AVX512F masked aligned load and store builtins and for
2882 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2884 PR tree-optimization/59651
2885 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2886 Address range for negative step should be added by TYPE_SIZE_UNIT.
2888 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2890 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2891 overlapping registers also for registers other than the stack pointer.
2893 2014-01-03 Marek Polacek <polacek@redhat.com>
2896 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2899 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2902 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2905 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2906 (push splitter): Use <P:MODE_SIZE> instead of
2907 GET_MODE_SIZE (<P:MODE>mode).
2908 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2909 (mov -1, reg peephole2): Likewise.
2910 * config/i386/sse.md (*mov<mode>_internal,
2911 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2912 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2913 *<code><mode>3, *andnot<mode>3<mask_name>,
2914 <mask_codefor><code><mode>3<mask_name>): Likewise.
2915 * config/i386/subst.md (mask_mode512bit_condition,
2916 sd_mask_mode512bit_condition): Likewise.
2918 2014-01-02 Xinliang David Li <davidxl@google.com>
2920 PR tree-optimization/59303
2921 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2922 (dump_predicates): Better output format.
2923 (pred_equal_p): New function.
2924 (is_neq_relop_p): Ditto.
2925 (is_neq_zero_form_p): Ditto.
2926 (pred_expr_equal_p): Ditto.
2927 (pred_neg_p): Ditto.
2928 (simplify_pred): Ditto.
2929 (simplify_preds_2): Ditto.
2930 (simplify_preds_3): Ditto.
2931 (simplify_preds_4): Ditto.
2932 (simplify_preds): Ditto.
2934 (push_to_worklist): Ditto.
2935 (get_pred_info_from_cmp): Ditto.
2936 (is_degenerated_phi): Ditto.
2937 (normalize_one_pred_1): Ditto.
2938 (normalize_one_pred): Ditto.
2939 (normalize_one_pred_chain): Ditto.
2940 (normalize_preds): Ditto.
2941 (normalize_cond_1): Remove function.
2942 (normalize_cond): Ditto.
2943 (is_gcond_subset_of): Ditto.
2944 (is_subset_of_any): Ditto.
2945 (is_or_set_subset_of): Ditto.
2946 (is_and_set_subset_of): Ditto.
2947 (is_norm_cond_subset_of): Ditto.
2948 (pred_chain_length_cmp): Ditto.
2949 (convert_control_dep_chain_into_preds): Type change.
2950 (find_predicates): Ditto.
2951 (find_def_preds): Ditto.
2952 (destroy_predicates_vecs): Ditto.
2953 (find_matching_predicates_in_rest_chains): Ditto.
2954 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2955 (is_pred_expr_subset): Ditto.
2956 (is_pred_chain_subset_of): Ditto.
2957 (is_included_in): Ditto.
2958 (is_superset_of): Ditto.
2960 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2962 Update copyright years.
2964 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2966 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2967 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2968 config/arc/arc.md, config/arc/arc.opt,
2969 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2970 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2971 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2972 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2973 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2974 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2975 vtable-verify.c, vtable-verify.h: Use the standard form for the
2978 2014-01-02 Tobias Burnus <burnus@net-b.de>
2980 * gcc.c (process_command): Update copyright notice dates.
2981 * gcov-dump.c: Ditto.
2983 * doc/cpp.texi: Bump @copying's copyright year.
2984 * doc/cppinternals.texi: Ditto.
2985 * doc/gcc.texi: Ditto.
2986 * doc/gccint.texi: Ditto.
2987 * doc/gcov.texi: Ditto.
2988 * doc/install.texi: Ditto.
2989 * doc/invoke.texi: Ditto.
2991 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2993 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2995 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2997 * config/i386/sse.md (*mov<mode>_internal): Guard
2998 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3000 PR rtl-optimization/59647
3001 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3002 new_rtx into UNSIGNED_FLOAT rtxes.
3004 Copyright (C) 2014 Free Software Foundation, Inc.
3006 Copying and distribution of this file, with or without modification,
3007 are permitted in any medium without royalty provided the copyright
3008 notice and this notice are preserved.