| blob | b69091eec05344d07ee317f4f4fb791ee19e11d0 |
1 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3 * varpool.c: Include pointer-set.h.
4 (varpool_remove_unreferenced_decls): Variables in other partitions
5 will not be output; be however careful to not lose information
6 about partitioning.
8 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
10 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
11 lookup in the vtable constructor.
13 2014-02-07 Jeff Law <law@redhat.com>
15 PR target/40977
16 * config/m68k/m68k.md (ashldi_extsi): Turn into a
17 define_insn_and_split.
19 * ipa-inline.c (inline_small_functions): Fix typos.
21 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
23 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
24 (s390_can_use_return_insn): Declare.
25 * config/s390/s390.h (EPILOGUE_USES): Define.
26 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
27 instructions.
28 (s390_chunkify_start): Handle return JUMP_LABELs.
29 (s390_early_mach): Emit a main_pool instruction on the entry edge.
30 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
31 (s390_can_use_return_insn): New functions.
32 (s390_fix_long_loop_prediction): Handle conditional returns.
33 (TARGET_SET_UP_BY_PROLOGUE): Define.
34 * config/s390/s390.md (ANY_RETURN): New code iterator.
35 (*creturn, *csimple_return, return, simple_return): New patterns.
37 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
39 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
40 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
41 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
42 REG_CFA_RESTORE list when deciding not to restore a register.
44 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
46 * config/s390/s390.c: Include tree-pass.h and context.h.
47 (s390_early_mach): New function, split out from...
48 (s390_emit_prologue): ...here.
49 (pass_data_s390_early_mach): New pass structure.
50 (pass_s390_early_mach): New class.
51 (s390_option_override): Create and register early_mach pass.
52 Move to end of file.
54 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
56 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
57 to match for the exit block.
59 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
61 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
62 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
63 Reject misaligned operands.
65 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
67 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
68 fail.
70 2014-02-07 Richard Biener <rguenther@suse.de>
72 PR middle-end/60092
73 * gimple-low.c (lower_builtin_posix_memalign): New function.
74 (lower_stmt): Call it to lower posix_memalign in a way
75 to make alignment info accessible.
77 2014-02-07 Jakub Jelinek <jakub@redhat.com>
79 PR c++/60082
80 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
81 __builtin_setjmp_receiver.
83 2014-02-07 Richard Biener <rguenther@suse.de>
85 PR middle-end/60092
86 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
87 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
88 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
89 Handle BUILT_IN_POSIX_MEMALIGN.
90 (find_func_clobbers): Likewise.
91 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
92 (call_may_clobber_ref_p_1): Likewise.
94 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
96 PR ipa/59918
97 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
99 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
101 PR ipa/59469
102 * lto-cgraph.c (lto_output_node): Use
103 symtab_get_symbol_partitioning_class.
104 (lto_output_varpool_node): likewise.
105 (symtab_get_symbol_partitioning_class): Move here from
106 lto/lto-partition.c
107 * cgraph.h (symbol_partitioning_class): Likewise.
108 (symtab_get_symbol_partitioning_class): Declare.
110 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
112 * ggc.h (ggc_internal_cleared_alloc): New macro.
113 * vec.h (vec_safe_copy): Handle memory stats.
114 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
115 * target-globals.c (save_target_globals): Likewise.
117 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
119 PR target/60077
120 * expr.c (emit_move_resolve_push): Export; be bit more selective
121 on when to clear alias set.
122 * expr.h (emit_move_resolve_push): Declare.
123 * function.h (struct function): Add tail_call_marked.
124 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
125 * config/i386/i386-protos.h (ix86_expand_push): Remove.
126 * config/i386/i386.md (TImode move expander): De not call
127 ix86_expand_push.
128 (FP push expanders): Preserve memory attributes.
129 * config/i386/sse.md (push<mode>1): Remove.
130 * config/i386/i386.c (ix86_expand_vector_move): Handle push
131 operation.
132 (ix86_expand_push): Remove.
133 * config/i386/mmx.md (push<mode>1): Remove.
135 2014-02-06 Jakub Jelinek <jakub@redhat.com>
137 PR rtl-optimization/60030
138 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
139 lopart with paradoxical subreg before shifting it up by hprec.
141 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
143 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
144 Remove extra newline at end of file.
145 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
146 (arm_issue_rate): Handle cortexa57.
147 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
148 (cortex-a57.cortex-a53): Likewise.
150 2014-02-06 Jakub Jelinek <jakub@redhat.com>
152 PR target/59575
153 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
154 don't record in REG_FRAME_RELATED_EXPR registers not set in that
155 bitmask.
156 (arm_expand_prologue): Adjust all callers.
157 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
158 info, registers also at the lowest numbered registers side. Use
159 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
160 XEXP.
162 PR debug/59992
163 * var-tracking.c (adjust_mems): Before adding a SET
164 to amd->side_effects, adjust it's SET_SRC using
165 simplify_replace_fn_rtx.
167 2014-02-06 Alan Modra <amodra@gmail.com>
169 PR target/60032
170 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
171 change SDmode to DDmode when lra_in_progress.
173 2014-02-06 Jakub Jelinek <jakub@redhat.com>
175 PR middle-end/59150
176 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
177 free_data_ref on the dr first, and before goto again also set dr
178 to the next dr. For simd_lane_access, free old datarefs[i] before
179 overwriting it. For get_vectype_for_scalar_type failure, don't
180 free_data_ref if simd_lane_access.
182 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
184 PR target/60062
185 * tree.h (opts_for_fn): New inline function.
186 (opt_for_fn): Define.
187 * config/i386/i386.c (ix86_function_regparm): Use
188 opt_for_fn (decl, optimize) instead of optimize.
190 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
192 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
193 for SYMBOL_REF in large memory model.
195 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
197 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
198 and crypto support.
199 (cortex-a57): Likewise.
200 (cortex-a57.cortex-a53): Likewise.
202 2014-02-06 Yury Gribov <y.gribov@samsung.com>
203 Kugan Vivekanandarajah <kuganv@linaro.org>
205 * config/arm/arm.c (arm_vector_alignment_reachable): Check
206 unaligned_access.
207 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
209 2014-02-06 Richard Biener <rguenther@suse.de>
211 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
212 set_loop_copy and initialize_original_copy_tables.
214 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
216 * config/aarch64/aarch64-simd.md
217 (aarch64_ashr_simddi): Change QI to SI.
219 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
220 Jakub Jelinek <jakub@redhat.com>
222 PR middle-end/60013
223 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
224 of the dataflow.
226 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
228 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
229 CODE_FOR_altivec_vpku[hw]um to
230 CODE_FOR_altivec_vpku[hw]um_direct.
231 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
232 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
233 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
234 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
236 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
238 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
239 generation for -maltivec=be.
240 (altivec_vsumsws): Simplify redundant test.
242 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
244 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
245 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
246 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
247 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
248 gen_altivec_vpkuwum.
249 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
250 BYTES_BIG_ENDIAN.
251 (altivec_vpks<VI_char>ss): Likewise.
252 (altivec_vpks<VI_char>us): Likewise.
253 (altivec_vpku<VI_char>us): Likewise.
254 (altivec_vpku<VI_char>um): Likewise.
255 (altivec_vpku<VI_char>um_direct): New (copy of
256 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
257 internal use).
258 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
259 target is little endian and -maltivec=be is not specified.
260 (*altivec_vupkhs<VU_char>_direct): New (copy of
261 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
262 use).
263 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
264 target is little endian and -maltivec=be is not specified.
265 (*altivec_vupkls<VU_char>_direct): New (copy of
266 altivec_vupkls<VU_char> that always emits vupkls*, for internal
267 use).
268 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
269 little endian and -maltivec=be is not specified.
270 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
271 little endian and -maltivec=be is not specified.
273 2014-02-05 Richard Henderson <rth@redhat.com>
275 PR debug/52727
276 * combine-stack-adj.c: Revert r206943.
277 * sched-int.h (struct deps_desc): Add last_args_size.
278 * sched-deps.c (init_deps): Initialize it.
279 (sched_analyze_insn): Add OUTPUT dependencies between insns that
280 contain REG_ARGS_SIZE notes.
282 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
284 * lto-cgraph.c (asm_nodes_output): Make global.
285 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
286 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
287 parameter
288 (driver_handle_option): Handle OPT_fwpa.
290 2014-02-05 Jakub Jelinek <jakub@redhat.com>
292 PR ipa/59947
293 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
294 a comment typo and formatting issue. If odr_hash hasn't been
295 created, return vNULL and set *completep to false.
297 PR middle-end/57499
298 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
299 bb with no successors.
301 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
303 PR target/59718
304 * doc/invoke.texi (-march): Clarify documentation for ARM.
305 (-mtune): Likewise.
306 (-mcpu): Likewise.
308 2014-02-05 Richard Biener <rguenther@suse.de>
310 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
311 when not vectorizing because of too many alias checks.
312 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
313 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
315 2014-02-05 Nick Clifton <nickc@redhat.com>
317 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
318 accept extended registers in any mode when compiling for the
319 MN10300.
321 2014-02-05 Yury Gribov <y.gribov@samsung.com>
323 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
324 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
325 sanitization attributes.
326 (can_inline_edge_p): Likewise.
327 (sanitize_attrs_match_for_inline_p): New function.
329 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
331 * ipa-prop.c (detect_type_change): Shor circuit testing of
332 type changes on THIS pointer.
334 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
336 PR target/59777
337 * config/pa/pa.c (legitimize_tls_address): Return original address
338 if not passed a SYMBOL_REF rtx.
339 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
340 addresses.
341 (pa_emit_move_sequence): Simplify TLS source operands.
342 (pa_legitimate_constant_p): Reject all TLS constants.
343 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
344 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
346 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
348 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
349 groups when we know they are controlled by LTO.
350 * varasm.c (default_binds_local_p_1): If object is in other partition,
351 it will be resolved locally.
353 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
355 * gcc/config/host-linux.c (linux_gt_pch_use_address): Don't
356 use SSIZE_MAX because it is not always defined.
358 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
360 PR bootstrap/59913
361 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
362 threshold for pseudo splitting.
363 (update_ebb_live_info): Process call argument hard registers and
364 hard registers from insn definition too.
365 (max_small_class_regs_num): New constant.
366 (inherit_in_ebb): Update live hard regs through EBBs. Update
367 reloads_num only for small register classes. Don't split for
368 outputs of jumps.
370 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
372 PR ipa/60058
373 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
374 is non-null.
376 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
378 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
380 2014-02-04 Marek Polacek <polacek@redhat.com>
382 * gdbinit.in (pel): Define.
384 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
386 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
387 behavior.
389 2014-02-04 Richard Biener <rguenther@suse.de>
391 PR lto/59723
392 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
393 in function context local.
394 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
395 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
396 similar to LTO_imported_decl_ref.
398 2014-02-04 Jakub Jelinek <jakub@redhat.com>
400 PR tree-optimization/60002
401 * cgraphclones.c (build_function_decl_skip_args): Clear
402 DECL_LANG_SPECIFIC.
404 PR tree-optimization/60023
405 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
406 false to gsi_replace.
407 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
408 has been in some EH region and vec_stmt could throw, add
409 vec_stmt into the same EH region.
410 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
411 has no lhs, ignore it.
412 * internal-fn.c (expand_MASK_LOAD): Likewise.
414 PR ipa/60026
415 * tree-inline.c (copy_forbidden): Fail for
416 __attribute__((optimize (0))) functions.
418 PR other/58712
419 * omp-low.c (simd_clone_struct_copy): If from->inbranch
420 is set, copy one less argument.
421 (expand_simd_clones): Don't subtract clone_info->inbranch
422 from simd_clone_struct_alloc argument.
424 PR rtl-optimization/57915
425 * recog.c (simplify_while_replacing): If all unary/binary/relational
426 operation arguments are constant, attempt to simplify those.
428 PR middle-end/59261
429 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
430 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
432 2014-02-04 Richard Biener <rguenther@suse.de>
434 PR tree-optimization/60012
435 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
436 TBAA disambiguation to all DDRs.
438 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
440 PR target/59788
441 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
442 (LINK_SPEC): Use it for -shared, -shared-libgcc.
444 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
446 PR ipa/59882
447 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
449 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
451 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
452 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
454 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
456 PR ipa/59831
457 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
458 to figure out targets of polymorphic calls with known decl.
459 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
460 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
461 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
462 (get_polymorphic_call_info): ... here.
463 (get_polymorphic_call_info_from_invariant): New function.
465 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
467 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
468 lookup via vtable pointer; check for type consistency
469 and turn inconsitent facts into UNREACHABLE.
470 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
471 * gimple-fold.c (gimple_get_virt_method_for_vtable):
472 Do not ICE on type inconsistent querries; return UNREACHABLE
473 instead.
475 2014-02-03 Richard Henderson <rth@twiddle.net>
477 PR tree-opt/59924
478 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
479 already processed this node.
480 (normalize_one_pred_1): Pass along mark_set.
481 (normalize_one_pred): Create and destroy a pointer_set_t.
482 (normalize_one_pred_chain): Likewise.
484 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
486 PR gcov-profile/58602
487 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
489 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
491 PR ipa/59831
492 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
493 Try to devirtualize by the knowledge of virtual table pointer given by
494 aggregate propagation.
495 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
496 ipa_print_node_jump_functions): Dump also offset that
497 is relevant for polymorphic calls.
498 (determine_known_aggregate_parts): Add arg_type parameter; use it
499 instead of determining the type from pointer type.
500 (ipa_compute_jump_functions_for_edge): Update call of
501 determine_known_aggregate_parts.
502 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
503 (gimple_get_virt_method_for_binfo): ... here; simplify using
504 vtable_pointer_value_to_vtable.
505 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
506 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
507 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
508 (vtable_pointer_value_to_vtable): Break out from ...; handle also
509 POINTER_PLUS_EXPR.
510 (vtable_pointer_value_to_binfo): ... here.
511 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
513 2014-02-03 Teresa Johnson <tejohnson@google.com>
515 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
516 redef of outer loop index variable.
518 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
520 PR c++/53017
521 PR c++/59211
522 * doc/extend.texi (Function Attributes): Typo.
524 2014-02-03 Cong Hou <congh@google.com>
526 PR tree-optimization/60000
527 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
528 if the vectorized statement is a store. A store statement can only
529 appear at the end of pattern statements.
531 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
533 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
534 (ix86_option_override_internal): Default long double to 64-bit for
535 32-bit Bionic and to 128-bit for 64-bit Bionic.
537 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
538 TARGET_LONG_DOUBLE_128 is true.
539 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
541 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
542 (mlong-double-64): Negate -mlong-double-128.
543 (mlong-double-128): New option.
545 * config/i386/i386-c.c (ix86_target_macros): Define
546 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
548 * doc/invoke.texi: Document -mlong-double-128.
550 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
552 PR rtl-optimization/60024
553 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
555 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
557 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
559 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
561 PR rtl-optimization/57662
562 * sel-sched.c (code_motion_path_driver): Do not mark already not
563 existing blocks in the visiting bitmap.
565 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
567 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
568 on the insn being emitted.
570 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
571 Will Deacon <will.deacon@arm.com>
573 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
575 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
577 * config/arm/arm-tables.opt: Regenerate.
579 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
581 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
582 for vector types other than V16QImode.
583 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
584 define_expand, and call altivec_expand_vec_perm_le when producing
585 code with little endian element order.
586 (*altivec_vperm_<mode>_internal): New insn having previous
587 behavior of altivec_vperm_<mode>.
588 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
589 altivec_expand_vec_perm_le when producing code with little endian
590 element order.
591 (*altivec_vperm_<mode>_uns_internal): New insn having previous
592 behavior of altivec_vperm_<mode>_uns.
594 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
596 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
597 (altivec_vsumsws): Add handling for -maltivec=be with a little
598 endian target.
599 (altivec_vsumsws_direct): New.
600 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
601 gen_altivec_vsumsws.
603 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
605 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
606 vtable_pointer_value_to_binfo): New functions.
607 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
608 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
610 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
612 * config/nios2/nios2.md (load_got_register): Initialize GOT
613 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
614 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
616 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
618 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
619 preserverd by passthrough, do not propagate the type.
621 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
623 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
624 (mips_atomic_assign_expand_fenv): New function.
625 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
627 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
629 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
630 (__builtin_mips_set_fcsr): Likewise.
631 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
632 MIPS_USI_FTYPE_VOID.
633 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
634 (mips16_expand_set_fcsr): Likewise.
635 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
636 (mips16_set_fcsr_stub): Likewise.
637 (mips16_get_fcsr_one_only_stub): New class.
638 (mips16_set_fcsr_one_only_stub): Likewise.
639 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
640 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
641 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
642 (hard_float): New availability predicate.
643 (mips_builtins): Add get_fcsr and set_fcsr.
644 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
645 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
646 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
647 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
648 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
649 patterns.
651 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
653 * config/mips/mips.c (mips_one_only_stub): New class.
654 (mips_need_mips16_rdhwr_p): Replace with...
655 (mips16_rdhwr_stub): ...this new variable.
656 (mips16_stub_call_address): New function.
657 (mips16_rdhwr_one_only_stub): New class.
658 (mips_expand_thread_pointer): Use mips16_stub_call_address.
659 (mips_output_mips16_rdhwr): Delete.
660 (mips_finish_stub): New function.
661 (mips_code_end): Use it to handle rdhwr stubs.
663 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
665 PR target/60017
666 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
667 when calculating size of integer atomic types.
669 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
671 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
673 2014-02-01 Jakub Jelinek <jakub@redhat.com>
675 PR tree-optimization/60003
676 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
677 * profile.c (branch_prob): Use gimple_call_builtin_p
678 to check for BUILT_IN_SETJMP_RECEIVER.
679 * tree-inline.c (copy_bb): Call notice_special_calls.
681 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
683 PR bootstrap/59985
684 * lra-constraints.c (process_alt_operands): Update reload_sum only
685 on the first pass.
687 2014-01-31 Richard Henderson <rth@redhat.com>
689 PR middle-end/60004
690 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
691 until after else_eh is processed.
693 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
695 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
696 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
697 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
698 in smmintrin.h, remove them.
699 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
700 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
701 * config/i386/i386.md (ROUND_SAE): Fix value.
702 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
703 (const48_operand): New.
704 * config/i386/subst.md (round), (round_expand): Use
705 const_4_or_8_to_11_operand.
706 (round_saeonly), (round_saeonly_expand): Use const48_operand.
708 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
710 * config/i386/constraints.md (Yk): Swap meaning with k.
711 * config/i386/i386.md (movhi_internal): Change Yk to k.
712 (movqi_internal): Ditto.
713 (*k<logic><mode>): Ditto.
714 (*andhi_1): Ditto.
715 (*andqi_1): Ditto.
716 (kandn<mode>): Ditto.
717 (*<code>hi_1): Ditto.
718 (*<code>qi_1): Ditto.
719 (kxnor<mode>): Ditto.
720 (kortestzhi): Ditto.
721 (kortestchi): Ditto.
722 (kunpckhi): Ditto.
723 (*one_cmplhi2_1): Ditto.
724 (*one_cmplqi2_1): Ditto.
725 * config/i386/sse.md (): Change k to Yk.
726 (avx512f_load<mode>_mask): Ditto.
727 (avx512f_blendm<mode>): Ditto.
728 (avx512f_store<mode>_mask): Ditto.
729 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
730 (avx512f_storedqu<mode>_mask): Ditto.
731 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
732 Ditto.
733 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
734 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
735 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
736 (avx512f_maskcmp<mode>3): Ditto.
737 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
738 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
739 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
740 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
741 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
742 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
743 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
744 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
745 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
746 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
747 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
748 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
749 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
750 (vec_extract_lo_<mode>_maskm): Ditto.
751 (vec_extract_hi_<mode>_maskm): Ditto.
752 (avx512f_vternlog<mode>_mask): Ditto.
753 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
754 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
755 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
756 (avx512f_<code>v8div16qi2_mask): Ditto.
757 (avx512f_<code>v8div16qi2_mask_store): Ditto.
758 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
759 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
760 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
761 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
762 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
763 (*avx512pf_gatherpf<mode>df_mask): Ditto.
764 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
765 (*avx512pf_scatterpf<mode>df_mask): Ditto.
766 (avx512cd_maskb_vec_dupv8di): Ditto.
767 (avx512cd_maskw_vec_dupv16si): Ditto.
768 (avx512f_vpermi2var<mode>3_maskz): Ditto.
769 (avx512f_vpermi2var<mode>3_mask): Ditto.
770 (avx512f_vpermi2var<mode>3_mask): Ditto.
771 (avx512f_vpermt2var<mode>3_maskz): Ditto.
772 (*avx512f_gathersi<mode>): Ditto.
773 (*avx512f_gathersi<mode>_2): Ditto.
774 (*avx512f_gatherdi<mode>): Ditto.
775 (*avx512f_gatherdi<mode>_2): Ditto.
776 (*avx512f_scattersi<mode>): Ditto.
777 (*avx512f_scatterdi<mode>): Ditto.
778 (avx512f_compress<mode>_mask): Ditto.
779 (avx512f_compressstore<mode>_mask): Ditto.
780 (avx512f_expand<mode>_mask): Ditto.
781 * config/i386/subst.md (mask): Change k to Yk.
782 (mask_scalar_merge): Ditto.
783 (sd): Ditto.
785 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
787 * doc/extend.texi (Vector Extensions): Document ?: in C++.
789 2014-01-31 Richard Biener <rguenther@suse.de>
791 PR middle-end/59990
792 * builtins.c (fold_builtin_memory_op): Make sure to not
793 use a floating-point mode or a boolean or enumeral type for
794 the copy operation.
796 2014-01-30 DJ Delorie <dj@redhat.com>
798 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
799 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
800 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
801 whenever main() has an epilogue.
803 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
805 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
806 unused variable "field".
807 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
808 (vsx_mergeh_<mode>): Likewise.
809 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
810 (altivec_vmrghh): Likewise.
811 (altivec_vmrghw): Likewise.
812 (altivec_vmrglb): Likewise.
813 (altivec_vmrglh): Likewise.
814 (altivec_vmrglw): Likewise.
815 (altivec_vspltb): Add missing uses.
816 (altivec_vsplth): Likewise.
817 (altivec_vspltw): Likewise.
818 (altivec_vspltsf): Likewise.
820 2014-01-30 Jakub Jelinek <jakub@redhat.com>
822 PR target/59923
823 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
824 frame related instructions.
826 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
828 PR rtl-optimization/59959
829 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
830 any reload of register whose subreg is invalid.
832 2014-01-30 Jakub Jelinek <jakub@redhat.com>
834 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
835 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
836 Add missing return type - void.
838 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
840 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
841 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
842 remove element index adjustment for endian (now handled in vsx.md
843 and altivec.md).
844 (altivec_expand_vec_perm_const): Use
845 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
846 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
847 (vsx_xxspltw_<mode>): Adjust element index for little endian.
848 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
849 define_expand and a new define_insn *altivec_vspltb_internal;
850 adjust for -maltivec=be on a little endian target.
851 (altivec_vspltb_direct): New.
852 (altivec_vsplth): Divide into a define_expand and a new
853 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
854 little endian target.
855 (altivec_vsplth_direct): New.
856 (altivec_vspltw): Divide into a define_expand and a new
857 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
858 little endian target.
859 (altivec_vspltw_direct): New.
860 (altivec_vspltsf): Divide into a define_expand and a new
861 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
862 a little endian target.
864 2014-01-30 Richard Biener <rguenther@suse.de>
866 PR tree-optimization/59993
867 * tree-ssa-forwprop.c (associate_pointerplus): Check we
868 can propagate form the earlier stmt and avoid the transform
869 when the intermediate result is needed.
871 2014-01-30 Alangi Derick <alangiderick@gmail.com>
873 * README.Portability: Fix typo.
875 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
877 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
878 comparison_operator with ordered_comparison_operator.
880 2014-01-30 Nick Clifton <nickc@redhat.com>
882 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
883 Rename to mn10300_store_multiple_regs.
884 * config/mn10300/mn10300.c: Likewise.
885 * config/mn10300/mn10300.md (store_movm): Fix typo: call
886 store_multiple_regs.
887 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
888 Call mn10300_store_multiple_regs.
890 2014-01-30 Nick Clifton <nickc@redhat.com>
891 DJ Delorie <dj@redhat.com>
893 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
894 %fp 2 to keep registers after it properly word-aligned.
895 (rl78_alloc_physical_registers_umul): Handle the case where both
896 input operands are the same.
898 2014-01-30 Richard Biener <rguenther@suse.de>
900 PR tree-optimization/59903
901 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
902 check properly.
904 2014-01-30 Jason Merrill <jason@redhat.com>
906 PR c++/59633
907 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
909 PR c++/59645
910 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
912 2014-01-30 Richard Biener <rguenther@suse.de>
914 PR tree-optimization/59951
915 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
917 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
919 PR target/59784
920 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
921 SFmode to DFmode case.
923 2014-01-29 DJ Delorie <dj@redhat.com>
925 * config/msp430/msp430.opt (-minrt): New.
926 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
927 if -minrt given.
928 (ENDFILE_SPEC): Likewise.
930 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
932 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
933 (estimate_function_body_sizes): Use it.
935 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
937 PR c++/58561
938 * dwarf2out.c (is_cxx_auto): New.
939 (is_base_type): Use it.
940 (gen_type_die_with_usage): Likewise.
942 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
944 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
945 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
946 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
947 -maltivec=be with LE targets.
948 (vsx_mergeh_<mode>): Likewise.
949 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
950 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
951 (altivec_vmrghb): Replace with define_expand and new
952 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
953 (altivec_vmrghb_direct): New define_insn.
954 (altivec_vmrghh): Replace with define_expand and new
955 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
956 (altivec_vmrghh_direct): New define_insn.
957 (altivec_vmrghw): Replace with define_expand and new
958 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
959 (altivec_vmrghw_direct): New define_insn.
960 (*altivec_vmrghsf): Adjust for endianness.
961 (altivec_vmrglb): Replace with define_expand and new
962 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
963 (altivec_vmrglb_direct): New define_insn.
964 (altivec_vmrglh): Replace with define_expand and new
965 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
966 (altivec_vmrglh_direct): New define_insn.
967 (altivec_vmrglw): Replace with define_expand and new
968 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
969 (altivec_vmrglw_direct): New define_insn.
970 (*altivec_vmrglsf): Adjust for endianness.
971 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
972 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
973 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
974 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
975 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
976 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
977 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
978 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
980 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
982 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
983 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
984 whitespace.
986 2014-01-29 Richard Biener <rguenther@suse.de>
988 PR tree-optimization/58742
989 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
990 associate_pointerplus_align.
991 (associate_pointerplus_diff): New function.
992 (associate_pointerplus): Likewise. Call associate_pointerplus_align
993 and associate_pointerplus_diff.
995 2014-01-29 Richard Biener <rguenther@suse.de>
997 * lto-streamer.h (LTO_major_version): Bump to 3.
998 (LTO_minor_version): Reset to 0.
1000 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1002 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1003 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1004 (arm_file_start): Generate correct asm header for armv7ve.
1005 * config/arm/bpabi.h: Add multilib support for armv7ve.
1006 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1007 and cortex-a15 to armv7ve.
1008 * config/arm/t-aprofile: Add multilib support for armv7ve.
1009 * doc/invoke.texi: Document -march=armv7ve.
1011 2014-01-29 Richard Biener <rguenther@suse.de>
1013 PR tree-optimization/58742
1014 * tree-ssa-forwprop.c (associate_plusminus): Return true
1015 if we changed sth, defer EH cleanup to ...
1016 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1017 (simplify_mult): New function.
1019 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1021 PR middle-end/59917
1022 PR tree-optimization/59920
1023 * tree.c (build_common_builtin_nodes): Remove
1024 __builtin_setjmp_dispatcher initialization.
1025 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1026 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1027 instead of gsi_after_labels + manually skipping debug stmts.
1028 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1029 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1030 * tree-inline.c (copy_edges_for_bb): Remove
1031 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1032 argument. Ignore computed_goto_p stmts. Don't call
1033 make_abnormal_goto_edges. If a call might need abnormal edges
1034 for non-local gotos, see if it already has an edge to
1035 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1036 with true argument, don't do anything then, otherwise add
1037 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1038 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1039 caller.
1040 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1041 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1042 (lower_stmt): Don't set data->calls_builtin_setjmp.
1043 (lower_builtin_setjmp): Adjust comment.
1044 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1045 * tree-cfg.c (found_computed_goto): Remove.
1046 (factor_computed_gotos): Remove.
1047 (make_goto_expr_edges): Return bool, true for computed gotos.
1048 Don't call make_abnormal_goto_edges.
1049 (build_gimple_cfg): Don't set found_computed_goto, don't call
1050 factor_computed_gotos.
1051 (computed_goto_p): No longer static.
1052 (make_blocks): Don't set found_computed_goto.
1053 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1054 (make_edges): If make_goto_expr_edges returns true, push bb
1055 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1056 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1057 vector. Record mapping between bbs and OpenMP regions if there
1058 are any, adjust make_gimple_omp_edges caller. Call
1059 handle_abnormal_edges.
1060 (make_abnormal_goto_edges): Remove.
1061 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1062 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1063 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1064 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1065 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1066 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1067 filling *region also set *region_idx to (*region)->entry->index.
1069 PR other/58712
1070 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1071 For REGs set ORIGINAL_REGNO.
1073 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1075 * doc/md.texi: Mention that a target shouldn't implement
1076 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1077 than hi/lo pair.
1079 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1081 PR tree-optimization/59594
1082 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1083 a copy of the datarefs vector rather than the vector itself.
1085 2014-01-28 Jason Merrill <jason@redhat.com>
1087 PR c++/53756
1088 * dwarf2out.c (auto_die): New static.
1089 (gen_type_die_with_usage): Handle C++1y 'auto'.
1090 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1091 on definition.
1093 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1095 PR target/59672
1096 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1097 (SPEC_X32): Likewise.
1098 (SPEC_64): Likewise.
1099 * config/i386/i386.c (ix86_option_override_internal): Turn off
1100 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1101 for TARGET_16BIT.
1102 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1103 * config/i386/i386.h (TARGET_16BIT): New macro.
1104 (TARGET_16BIT_P): Likewise.
1105 * config/i386/i386.opt: Add m16.
1106 * doc/invoke.texi: Document -m16.
1108 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1110 PR preprocessor/59935
1111 * input.c (location_get_source_line): Bail out on when line number
1112 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1114 2014-01-28 Richard Biener <rguenther@suse.de>
1116 PR tree-optimization/58742
1117 * tree-ssa-forwprop.c (associate_plusminus): Handle
1118 pointer subtraction of the form (T)(P + A) - (T)P.
1120 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1122 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1123 at const_int_cost.
1125 2014-01-28 Richard Biener <rguenther@suse.de>
1127 Revert
1128 2014-01-28 Richard Biener <rguenther@suse.de>
1130 PR rtl-optimization/45364
1131 PR rtl-optimization/59890
1132 * var-tracking.c (local_get_addr_clear_given_value): Handle
1133 already cleared slot.
1134 (val_reset): Handle not allocated local_get_addr_cache.
1135 (vt_find_locations): Use post-order on the inverted CFG.
1137 2014-01-28 Richard Biener <rguenther@suse.de>
1139 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1141 2014-01-28 Richard Biener <rguenther@suse.de>
1143 PR rtl-optimization/45364
1144 PR rtl-optimization/59890
1145 * var-tracking.c (local_get_addr_clear_given_value): Handle
1146 already cleared slot.
1147 (val_reset): Handle not allocated local_get_addr_cache.
1148 (vt_find_locations): Use post-order on the inverted CFG.
1150 2014-01-28 Alan Modra <amodra@gmail.com>
1152 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1153 * configure.ac <recursive call for build != host>: Define
1154 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1155 and LD_FOR_BUILD too.
1156 * configure: Regenerate.
1158 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1160 * config/i386/i386.c (get_builtin_code_for_version): Separate
1161 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1162 Broadwell from Haswell.
1164 2014-01-27 Steve Ellcey <sellcey@mips.com>
1166 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1167 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1168 * config/mips/mips.c (mips_option_override): Change setting
1169 of TARGET_DSP.
1170 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1171 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1172 Change from Mask to Var.
1174 2014-01-27 Jeff Law <law@redhat.com>
1176 * ipa-inline.c (inline_small_functions): Fix typo.
1178 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1180 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1181 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1182 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1183 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1184 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1185 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1186 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1187 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1188 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1189 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1190 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1191 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1192 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1193 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1194 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1195 (_mm512_storeu_epi64): Ditto.
1196 (_mm512_cmpge_epi32_mask): Ditto.
1197 (_mm512_cmpge_epu32_mask): Ditto.
1198 (_mm512_cmpge_epi64_mask): Ditto.
1199 (_mm512_cmpge_epu64_mask): Ditto.
1200 (_mm512_cmple_epi32_mask): Ditto.
1201 (_mm512_cmple_epu32_mask): Ditto.
1202 (_mm512_cmple_epi64_mask): Ditto.
1203 (_mm512_cmple_epu64_mask): Ditto.
1204 (_mm512_cmplt_epi32_mask): Ditto.
1205 (_mm512_cmplt_epu32_mask): Ditto.
1206 (_mm512_cmplt_epi64_mask): Ditto.
1207 (_mm512_cmplt_epu64_mask): Ditto.
1208 (_mm512_cmpneq_epi32_mask): Ditto.
1209 (_mm512_cmpneq_epu32_mask): Ditto.
1210 (_mm512_cmpneq_epi64_mask): Ditto.
1211 (_mm512_cmpneq_epu64_mask): Ditto.
1212 (_mm512_expand_pd): Ditto.
1213 (_mm512_expand_ps): Ditto.
1214 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1215 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1216 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1217 * config/i386/i386.c (ix86_builtins): Add
1218 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1219 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1220 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1221 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1222 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1223 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1224 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1225 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1226 IX86_BUILTIN_PMOVUSQW512_MEM.
1227 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1228 __builtin_ia32_pmovsqd512mem_mask,
1229 __builtin_ia32_pmovqd512mem_mask,
1230 __builtin_ia32_pmovusqw512mem_mask,
1231 __builtin_ia32_pmovsqw512mem_mask,
1232 __builtin_ia32_pmovqw512mem_mask,
1233 __builtin_ia32_pmovusdw512mem_mask,
1234 __builtin_ia32_pmovsdw512mem_mask,
1235 __builtin_ia32_pmovdw512mem_mask,
1236 __builtin_ia32_pmovqb512mem_mask,
1237 __builtin_ia32_pmovusqb512mem_mask,
1238 __builtin_ia32_pmovsqb512mem_mask,
1239 __builtin_ia32_pmovusdb512mem_mask,
1240 __builtin_ia32_pmovsdb512mem_mask,
1241 __builtin_ia32_pmovdb512mem_mask.
1242 (bdesc_args): Add __builtin_ia32_expanddf512,
1243 __builtin_ia32_expandsf512.
1244 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1245 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1246 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1247 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1248 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1249 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1250 (avx512f_<code>v8div16qi2_mask_store): This.
1251 (avx512f_expand<mode>): New.
1253 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1255 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1256 New.
1257 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1258 (_mm512_prefetch_i32scatter_pd): Ditto.
1259 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1260 (_mm512_prefetch_i64scatter_pd): Ditto.
1261 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1262 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1263 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1264 (_mm512_prefetch_i32scatter_ps): Ditto.
1265 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1266 (_mm512_prefetch_i64scatter_ps): Ditto.
1267 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1268 * config/i386/i386-builtin-types.def: Define
1269 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1270 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1271 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1272 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1273 IX86_BUILTIN_SCATTERPFQPD.
1274 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1275 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1276 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1277 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1278 __builtin_ia32_scatterpfqps.
1279 (ix86_expand_builtin): Expand new built-ins.
1280 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1281 fix memory access data type.
1282 (*avx512pf_gatherpf<mode>_mask): Ditto.
1283 (*avx512pf_gatherpf<mode>): Ditto.
1284 (avx512pf_scatterpf<mode>): Ditto.
1285 (*avx512pf_scatterpf<mode>_mask): Ditto.
1286 (*avx512pf_scatterpf<mode>): Ditto.
1287 (GATHER_SCATTER_SF_MEM_MODE): New.
1288 (avx512pf_gatherpf<mode>df): Ditto.
1289 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1290 (*avx512pf_scatterpf<mode>df): Ditto.
1292 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1294 PR bootstrap/59934
1295 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1296 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1297 reached.
1299 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1301 * common/config/arm/arm-common.c
1302 (arm_rewrite_mcpu): Handle multiple names.
1303 * config/arm/arm.h
1304 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1306 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1308 * gimple-builder.h (create_gimple_tmp): Delete.
1310 2014-01-27 Christian Bruel <christian.bruel@st.com>
1312 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1313 words comparisons.
1315 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1317 * config/pa/pa.md (call): Generate indirect long calls to non-local
1318 functions when outputing 32-bit code.
1319 (call_value): Likewise except for special call to buggy powf function.
1321 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1322 portable runtime and PIC indirect calls.
1323 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1324 and PIC call sequences. Use ldo instead of blr to set return register
1325 in PIC call sequence.
1327 2014-01-25 Walter Lee <walt@tilera.com>
1329 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1330 avoid clobbering a live register.
1332 2014-01-25 Walter Lee <walt@tilera.com>
1334 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1335 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1336 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1337 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1339 2014-01-25 Walter Lee <walt@tilera.com>
1341 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1342 arguments on even registers.
1343 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1344 STACK_BOUNDARY.
1345 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1346 (BIGGEST_ALIGNMENT): Ditto.
1347 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1349 2014-01-25 Walter Lee <walt@tilera.com>
1351 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1352 insns before bundling.
1353 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1355 2014-01-25 Walter Lee <walt@tilera.com>
1357 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1358 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1359 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1361 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1363 * config/mips/constraints.md (kl): Delete.
1364 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1365 define expands, using...
1366 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1367 instructions for MIPS16.
1368 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1369 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1371 2014-01-25 Walter Lee <walt@tilera.com>
1373 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1374 (clzdi2): Ditto.
1375 (ffsdi2): Ditto.
1377 2014-01-25 Walter Lee <walt@tilera.com>
1379 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1380 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1382 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1384 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1385 Handle XOR.
1387 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1389 * print-rtl.c (in_call_function_usage): New var.
1390 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1391 EXPR_LIST mode as mode and not as reg note name.
1393 PR middle-end/59561
1394 * cfgloopmanip.c (copy_loop_info): If
1395 loop->warned_aggressive_loop_optimizations, make sure
1396 the flag is set in target loop too.
1398 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1400 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1401 flag_cilkplus.
1402 * builtins.def: Likewise.
1403 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1404 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1405 * ira.c (ira_setup_eliminable_regset): Likewise.
1406 * omp-low.c (gate_expand_omp): Likewise.
1407 (execute_lower_omp): Likewise.
1408 (diagnose_sb_0): Likewise.
1409 (gate_diagnose_omp_blocks): Likewise.
1410 (simd_clone_clauses_extract): Likewise.
1411 (gate): Likewise.
1413 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1415 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1416 correction for little endian...
1417 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1418 here.
1420 2014-01-24 Jeff Law <law@redhat.com>
1422 PR tree-optimization/59919
1423 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1424 for non-returning calls.
1426 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1428 * common/config/aarch64/aarch64-common.c
1429 (aarch64_rewrite_mcpu): Handle multiple names.
1430 * config/aarch64/aarch64.h
1431 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1433 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1435 * input.c (add_file_to_cache_tab): Handle the case where fopen
1436 returns NULL.
1438 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1440 PR target/59929
1441 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1442 from push operand if code of push isn't PRE_DEC.
1444 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1446 PR target/59909
1447 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1448 -mquad-memory-atomic. Update -mquad-memory documentation to say
1449 it is only used for non-atomic loads/stores.
1451 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1452 -mquad-memory or -mquad-memory-atomic switches.
1454 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1455 -mquad-memory-atomic to ISA 2.07 support.
1457 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1458 to separate support of normal quad word memory operations (ldq, stq)
1459 from the atomic quad word memory operations.
1461 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1462 support to separate non-atomic quad word operations from atomic
1463 quad word operations. Disable non-atomic quad word operations in
1464 little endian mode so that we don't have to swap words after the
1465 load and before the store.
1466 (quad_load_store_p): Add comment about atomic quad word support.
1467 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1468 options printed with -mdebug=reg.
1470 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1471 -mquad-memory-atomic as the test for whether we have quad word
1472 atomic instructions.
1473 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1474 or -mp8-vector are used, allow byte/half-word atomic operations.
1476 * config/rs6000/sync.md (load_lockedti): Insure that the address
1477 is a proper indexed or indirect address for the lqarx instruction.
1478 On little endian systems, swap the hi/lo registers after the lqarx
1479 instruction.
1480 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1481 insure the address is valid for the lqarx instruction.
1482 (store_conditionalti): Insure that the address is a proper indexed
1483 or indirect address for the stqcrx. instruction. On little endian
1484 systems, swap the hi/lo registers before doing the stqcrx.
1485 instruction.
1486 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1487 insure the address is valid for the stqcrx. instruction.
1489 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1490 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1491 type of quad memory support is available.
1493 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1495 PR regression/59915
1496 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1497 there is a danger of looping.
1499 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1501 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1502 force flag_ira_loop_pressure if set via command line.
1504 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1506 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1507 (ashr_simd): New builtin handling DI mode.
1508 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1509 (aarch64_sshr_simddi): New match pattern.
1510 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1511 (vshrd_n_s64): Likewise.
1512 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1514 2014-01-23 Nick Clifton <nickc@redhat.com>
1516 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1517 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1518 favour of mcu specific scripts.
1519 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1520 430x multilibs.
1522 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1523 Alex Velenko <Alex.Velenko@arm.com>
1525 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1526 (vaddv_s16): Likewise.
1527 (vaddv_s32): Likewise.
1528 (vaddv_u8): Likewise.
1529 (vaddv_u16): Likewise.
1530 (vaddv_u32): Likewise.
1531 (vaddvq_s8): Likewise.
1532 (vaddvq_s16): Likewise.
1533 (vaddvq_s32): Likewise.
1534 (vaddvq_s64): Likewise.
1535 (vaddvq_u8): Likewise.
1536 (vaddvq_u16): Likewise.
1537 (vaddvq_u32): Likewise.
1538 (vaddvq_u64): Likewise.
1539 (vaddv_f32): Likewise.
1540 (vaddvq_f32): Likewise.
1541 (vaddvq_f64): Likewise.
1542 (vmaxv_f32): Likewise.
1543 (vmaxv_s8): Likewise.
1544 (vmaxv_s16): Likewise.
1545 (vmaxv_s32): Likewise.
1546 (vmaxv_u8): Likewise.
1547 (vmaxv_u16): Likewise.
1548 (vmaxv_u32): Likewise.
1549 (vmaxvq_f32): Likewise.
1550 (vmaxvq_f64): Likewise.
1551 (vmaxvq_s8): Likewise.
1552 (vmaxvq_s16): Likewise.
1553 (vmaxvq_s32): Likewise.
1554 (vmaxvq_u8): Likewise.
1555 (vmaxvq_u16): Likewise.
1556 (vmaxvq_u32): Likewise.
1557 (vmaxnmv_f32): Likewise.
1558 (vmaxnmvq_f32): Likewise.
1559 (vmaxnmvq_f64): Likewise.
1560 (vminv_f32): Likewise.
1561 (vminv_s8): Likewise.
1562 (vminv_s16): Likewise.
1563 (vminv_s32): Likewise.
1564 (vminv_u8): Likewise.
1565 (vminv_u16): Likewise.
1566 (vminv_u32): Likewise.
1567 (vminvq_f32): Likewise.
1568 (vminvq_f64): Likewise.
1569 (vminvq_s8): Likewise.
1570 (vminvq_s16): Likewise.
1571 (vminvq_s32): Likewise.
1572 (vminvq_u8): Likewise.
1573 (vminvq_u16): Likewise.
1574 (vminvq_u32): Likewise.
1575 (vminnmv_f32): Likewise.
1576 (vminnmvq_f32): Likewise.
1577 (vminnmvq_f64): Likewise.
1579 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1581 * config/aarch64/aarch64-simd.md
1582 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1583 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1584 (*aarch64_mul3_elt<mode>): Likewise.
1585 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1586 (*aarch64_mul3_elt_to_64v2df): Likewise.
1587 (*aarch64_mla_elt<mode>): Likewise.
1588 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1589 (*aarch64_mls_elt<mode>): Likewise.
1590 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1591 (*aarch64_fma4_elt<mode>): Likewise.
1592 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1593 (*aarch64_fma4_elt_to_64v2df): Likewise.
1594 (*aarch64_fnma4_elt<mode>): Likewise.
1595 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1596 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1597 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1598 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1599 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1600 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1601 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1602 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1603 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1605 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1607 * config/aarch64/aarch64-simd.md
1608 (aarch64_be_checked_get_lane<mode>): New define_expand.
1609 * config/aarch64/aarch64-simd-builtins.def
1610 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1611 New builtin definition.
1612 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1613 Use new safe be builtin.
1615 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1617 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1618 New define_insn.
1619 (aarch64_be_st1<mode>): Likewise.
1620 (aarch_ld1<VALL:mode>): Define_expand modified.
1621 (aarch_st1<VALL:mode>): Likewise.
1622 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1623 (UNSPEC_ST1): Likewise.
1625 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1627 * config/microblaze/microblaze.md: Add trap insn and attribute
1629 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1631 PR preprocessor/58580
1632 * input.h (location_get_source_line): Take an additional line_size
1633 parameter.
1634 (void diagnostics_file_cache_fini): Declare new function.
1635 * input.c (struct fcache): New type.
1636 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1637 New static constants.
1638 (diagnostic_file_cache_init, total_lines_num)
1639 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1640 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1641 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1642 (get_next_line, read_next_line, goto_next_line, read_line_num):
1643 New static function definitions.
1644 (diagnostic_file_cache_fini): New function.
1645 (location_get_source_line): Take an additional output line_len
1646 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1647 read_line_num.
1648 * diagnostic.c (diagnostic_finish): Call
1649 diagnostic_file_cache_fini.
1650 (adjust_line): Take an additional input parameter for the length
1651 of the line, rather than calculating it with strlen.
1652 (diagnostic_show_locus): Adjust the use of
1653 location_get_source_line and adjust_line with respect to their new
1654 signature. While displaying a line now, do not stop at the first
1655 null byte. Rather, display the zero byte as a space and keep
1656 going until we reach the size of the line.
1657 * Makefile.in: Add vec.o to OBJS-libcommon
1659 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1660 Ilya Tocar <ilya.tocar@intel.com>
1662 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1663 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1664 (__builtin_ia32_kmov16): Ditto.
1665 * config/i386/i386.md (UNSPEC_KMOV): New.
1666 (kmovw): Ditto.
1668 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1670 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1671 (_mm512_storeu_si512): Ditto.
1673 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1675 PR target/52125
1676 * rtl.h (get_referenced_operands): Declare.
1677 * recog.c (get_referenced_operands): New function.
1678 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1679 operands have been referenced when recording LO_SUM references.
1681 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1683 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1685 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1687 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1688 Enable for generic and recent AMD targets.
1690 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1692 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1693 ARG_SIZE note when adjustment was eliminated.
1695 2014-01-22 Jeff Law <law@redhat.com>
1697 PR tree-optimization/59597
1698 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1699 in file. Accept new argument REGISTERING and use it to modify
1700 dump output appropriately.
1701 (register_jump_thread): Corresponding changes.
1702 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1703 thread paths involving joiner blocks. Add code to dump cancelled
1704 jump threading paths.
1706 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1708 PR rtl-optimization/59477
1709 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1710 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1712 2014-01-22 Tom Tromey <tromey@redhat.com>
1714 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1715 PARAMS.
1716 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1718 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1720 PR rtl-optimization/59896
1721 * lra-constraints.c (process_alt_operands): Check unused note for
1722 matched operands of insn with no output reloads.
1724 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1726 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1727 (mips_move_from_gpr_cost): Likewise.
1729 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1731 PR rtl-optimization/59858
1732 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1733 ira_class_hard_regs_num.
1734 (process_alt_operands): Increase reject for dying matched operand.
1736 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1738 PR target/59003
1739 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1740 smaller than size, perform several stores or loads and stores
1741 at dst + count - size to store or copy all of size bytes, rather
1742 than just last modesize bytes.
1744 2014-01-20 DJ Delorie <dj@redhat.com>
1746 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1747 that CLOBBERs are REGs before propogating their values.
1749 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1751 PR middle-end/59789
1752 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1753 (cgraph_inline_failed_type): New function.
1754 * cgraph.h (DEFCIFCODE): Add type.
1755 (cgraph_inline_failed_type_t): New enum.
1756 (cgraph_inline_failed_type): New prototype.
1757 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1758 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1759 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1760 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1761 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1762 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1763 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1764 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1765 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1766 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1767 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1768 OPTIMIZATION_MISMATCH.
1769 * tree-inline.c (expand_call_inline): Emit errors during
1770 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1772 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1774 PR target/59685
1775 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1776 mode attribute in insn output.
1778 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1780 * output.h (output_constant): Delete.
1781 * varasm.c (output_constant): Make private.
1783 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1785 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1787 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1789 PR middle-end/59860
1790 * tree.h (fold_builtin_strcat): New prototype.
1791 * builtins.c (fold_builtin_strcat): No longer static. Add len
1792 argument, if non-NULL, don't call c_strlen. Optimize
1793 directly into __builtin_memcpy instead of __builtin_strcpy.
1794 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1795 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1797 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1799 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1800 for SImode_address_operand operands, having only a REG argument.
1802 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1804 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1805 loader name using mbig-endian.
1806 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1808 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1810 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1811 (-mtune): Likewise.
1812 (-mcpu): Likewise.
1814 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1816 * config/aarch64/aarch64-protos.h
1817 (aarch64_cannot_change_mode_class_ptr): Declare.
1818 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1819 aarch64_cannot_change_mode_class_ptr): New.
1820 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1821 backend hook aarch64_cannot_change_mode_class.
1823 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1825 * common/config/aarch64/aarch64-common.c
1826 (aarch64_handle_option): Don't handle any option order logic here.
1827 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1828 selected_cpu, warn on architecture version mismatch.
1829 (aarch64_override_options): Fix parsing order for option strings.
1831 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1832 Iain Sandoe <iain@codesourcery.com>
1834 PR bootstrap/59496
1835 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1836 warning. Amend comment to reflect current functionality.
1838 2014-01-20 Richard Biener <rguenther@suse.de>
1840 PR middle-end/59860
1841 * builtins.c (fold_builtin_strcat): Remove case better handled
1842 by tree-ssa-strlen.c.
1844 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1846 * config/aarch64/aarch64.opt
1847 (mcpu, march, mtune): Make case-insensitive.
1849 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1851 PR target/59880
1852 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1853 if operands[1] is a REG or ZERO_EXTEND of a REG.
1855 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
1857 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1859 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1861 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1862 long non-pic millicode calls.
1864 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1866 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1868 2014-01-19 Kito Cheng <kito@0xlab.org>
1870 * builtins.c (expand_movstr): Check movstr expand done or fail.
1872 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1873 H.J. Lu <hongjiu.lu@intel.com>
1875 PR target/59379
1876 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1877 to DImode for zero-extended addresses.
1879 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1881 PR rtl-optimization/57763
1882 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1883 on the new indirect jump_insn and increment LABEL_NUSES (label).
1885 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1887 PR bootstrap/59580
1888 PR bootstrap/59583
1889 * config.gcc (x86_archs): New variable.
1890 (x86_64_archs): Likewise.
1891 (x86_cpus): Likewise.
1892 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1893 --with-arch/--with-cpu= options.
1894 Support --with-arch=/--with-cpu={nehalem,westmere,
1895 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1897 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1899 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1900 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1902 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1904 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1906 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1908 PR target/58944
1909 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1910 clear cpp_get_options (parse_in)->warn_unused_macros for
1911 ix86_target_macros_internal with cpp_define.
1913 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1915 * jump.c (delete_related_insns): Keep (use (insn))s.
1916 * reorg.c (redundant_insn): Check for barriers too.
1918 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1920 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1922 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1924 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1925 call to $$dyncall when TARGET_LONG_CALLS is true.
1927 2014-01-17 Jeff Law <law@redhat.com>
1929 * ree.c (combine_set_extension): Temporarily disable test for
1930 changing number of hard registers.
1932 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1934 PR middle-end/58125
1935 * ipa-inline-analysis.c (inline_free_summary):
1936 Do not free summary of aliases.
1938 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1940 PR middle-end/59706
1941 * gimplify.c (gimplify_expr): Use create_tmp_var
1942 instead of create_tmp_var_raw. If cond doesn't have
1943 integral type, don't add the IFN_ANNOTATE builtin at all.
1945 2014-01-17 Martin Jambor <mjambor@suse.cz>
1947 PR ipa/59736
1948 * ipa-cp.c (prev_edge_clone): New variable.
1949 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1950 Also resize prev_edge_clone vector.
1951 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1952 (ipcp_edge_removal_hook): New function.
1953 (ipcp_driver): Register ipcp_edge_removal_hook.
1955 2014-01-17 Andrew Pinski <apinski@cavium.com>
1956 Steve Ellcey <sellcey@mips.com>
1958 PR target/59462
1959 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1960 of operator mode.
1962 2014-01-17 Jeff Law <law@redhat.com>
1964 PR middle-end/57904
1965 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1966 so that pass_ccp runs first.
1968 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1970 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1971 (ix86_adjust_cost): Use !TARGET_XXX.
1972 (do_reorder_for_imul): Likewise.
1973 (swap_top_of_ready_list): Likewise.
1974 (ix86_sched_reorder): Likewise.
1976 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1978 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1979 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1980 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1981 (intel_memset): New. Duplicate slm_memset.
1982 (intel_cost): New. Duplicate slm_cost.
1983 (m_INTEL): New macro.
1984 (processor_target_table): Add "intel".
1985 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1986 with PROCESSOR_INTEL for "intel".
1987 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1988 PROCESSOR_SILVERMONT.
1989 (ix86_issue_rate): Likewise.
1990 (ix86_adjust_cost): Likewise.
1991 (ia32_multipass_dfa_lookahead): Likewise.
1992 (swap_top_of_ready_list): Likewise.
1993 (ix86_sched_reorder): Likewise.
1994 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1995 instead of TARGET_OPT_AGU.
1996 * config/i386/i386.h (TARGET_INTEL): New.
1997 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1998 (processor_type): Add PROCESSOR_INTEL.
1999 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2000 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2002 2014-01-17 Marek Polacek <polacek@redhat.com>
2004 PR c/58346
2005 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2006 size is zero.
2008 2014-01-17 Richard Biener <rguenther@suse.de>
2010 PR tree-optimization/46590
2011 * opts.c (default_options_table): Add entries for
2012 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2013 all enabled at -O1 but not for -Og.
2014 * common.opt (fbranch-count-reg): Remove Init(1).
2015 (fmove-loop-invariants): Likewise.
2016 (ftree-pta): Likewise.
2018 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2020 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2021 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2022 decls to at least the GCC 4.8 used alignments.
2024 PR fortran/59440
2025 * tree-nested.c (convert_nonlocal_reference_stmt,
2026 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2027 of GIMPLE_BIND stmts, adjust associated decls.
2029 2014-01-17 Richard Biener <rguenther@suse.de>
2031 PR tree-optimization/46590
2032 * vec.h (vec<>::bseach): New member function implementing
2033 binary search according to C89 bsearch.
2034 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2035 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2036 bitmap pointer again. Make accesses_in_loop a flat array.
2037 (mem_ref_obstack): New global.
2038 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2039 (mark_ref_stored): Likewise.
2040 (ref_indep_loop_p_2): Likewise.
2041 (set_ref_stored_in_loop): New helper function.
2042 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2043 (memref_free): Adjust.
2044 (record_mem_ref_loc): Simplify.
2045 (gather_mem_refs_stmt): Adjust.
2046 (sort_locs_in_loop_postorder_cmp): New function.
2047 (analyze_memory_references): Sort accesses_in_loop after
2048 loop postorder number.
2049 (find_ref_loc_in_loop_cmp): New function.
2050 (for_all_locs_in_loop): Find relevant cluster of locs in
2051 accesses_in_loop and iterate without recursion.
2052 (execute_sm): Avoid uninit warning.
2053 (struct ref_always_accessed): Simplify.
2054 (ref_always_accessed::operator ()): Likewise.
2055 (ref_always_accessed_p): Likewise.
2056 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2057 loop postorder numbers here.
2058 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2059 numbers.
2061 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2063 PR c++/57945
2064 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2065 on decls for which assemble_alias has been called.
2067 2014-01-17 Nick Clifton <nickc@redhat.com>
2069 * config/msp430/msp430.opt: (mcpu): New option.
2070 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2071 (msp430_option_override): Parse target_cpu. If the MCU name
2072 matches a generic string, clear target_mcu.
2073 (msp430_attr): Allow numeric interrupt values up to 63.
2074 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2075 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2076 option.
2077 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2078 Add mcpu matches.
2079 * config/msp430/msp430.md (popm): Use %J rather than %I.
2080 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2081 (addhi_cy_i): Use immediate_operand for operand 2.
2082 * doc/invoke.texi: Document -mcpu option.
2084 2014-01-17 Richard Biener <rguenther@suse.de>
2086 PR rtl-optimization/38518
2087 * df.h (df_analyze_loop): Declare.
2088 * df-core.c: Include cfgloop.h.
2089 (df_analyze_1): Split out main part of df_analyze.
2090 (df_analyze): Adjust.
2091 (loop_inverted_post_order_compute): New function.
2092 (loop_post_order_compute): Likewise.
2093 (df_analyze_loop): New function avoiding whole-function
2094 postorder computes.
2095 * loop-invariant.c (find_defs): Use df_analyze_loop.
2096 (find_invariants): Adjust.
2097 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2099 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2101 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2102 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2104 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2106 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2107 traversal when removing references.
2109 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2111 PR ipa/59775
2112 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2114 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2116 PR middle-end/56791
2117 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2118 pushing a reload for an autoinc when we had previously reloaded an
2119 inner part of the address.
2121 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2123 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2124 field.
2125 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2126 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2127 when not giving up or versioning for alias only because of
2128 loop->safelen.
2129 (vect_analyze_data_ref_dependences): Set to true.
2130 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2131 is a GIMPLE_PHI.
2132 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2133 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2134 to the condition.
2136 PR middle-end/58344
2137 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2139 PR target/59839
2140 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2141 operand 0 predicate for gathers, use a new pseudo as subtarget.
2143 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2145 PR middle-end/59609
2146 * lra-constraints.c (process_alt_operands): Add printing debug info.
2147 Check absence of input/output reloads for matched operands too.
2149 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2151 PR rtl-optimization/59835
2152 * ira.c (ira_init_register_move_cost): Increase cost for
2153 impossible modes.
2155 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2157 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2159 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2161 PR target/59780
2162 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2163 non-register objects. Use gen_(high/low)part more consistently.
2164 Fix assertions.
2166 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2168 PR target/59844
2169 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2170 endian support, remove tests for WORDS_BIG_ENDIAN.
2171 (p8_mfvsrd_3_<mode>): Likewise.
2172 (reload_gpr_from_vsx<mode>): Likewise.
2173 (reload_gpr_from_vsxsf): Likewise.
2174 (p8_mfvsrd_4_disf): Likewise.
2176 2014-01-16 Richard Biener <rguenther@suse.de>
2178 PR rtl-optimization/46590
2179 * lcm.c (compute_antinout_edge): Use postorder iteration.
2180 (compute_laterin): Use inverted postorder iteration.
2182 2014-01-16 Nick Clifton <nickc@redhat.com>
2184 PR middle-end/28865
2185 * varasm.c (output_constant): Return the number of bytes actually
2186 emitted.
2187 (output_constructor_array_range): Update the field size with the
2188 number of bytes emitted by output_constant.
2189 (output_constructor_regular_field): Likewise. Also do not
2190 complain if the total number of bytes emitted is now greater
2191 than the expected fieldpos.
2192 * output.h (output_constant): Update prototype and descriptive comment.
2194 2014-01-16 Marek Polacek <polacek@redhat.com>
2196 PR middle-end/59827
2197 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2198 it is error_mark_node.
2200 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2202 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2203 VALID_AVX256_REG_OR_OI_MODE.
2205 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2207 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2208 current procedure should be profiled.
2210 2014-01-15 Andrew Pinski <apinski@cavium.com>
2212 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2213 of moving from/to the STACK_REG register class.
2215 2014-01-15 Richard Henderson <rth@redhat.com>
2217 PR debug/54694
2218 * reginfo.c (global_regs_decl): Globalize.
2219 * rtl.h (global_regs_decl): Declare.
2220 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2221 reserved via global_regs.
2223 2014-01-15 Teresa Johnson <tejohnson@google.com>
2225 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2227 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2229 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2230 and vmulosh rather than call gen_vec_widen_smult_*.
2231 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2232 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2233 (vec_widen_smult_even_v16qi): Likewise.
2234 (vec_widen_umult_even_v8hi): Likewise.
2235 (vec_widen_smult_even_v8hi): Likewise.
2236 (vec_widen_umult_odd_v16qi): Likewise.
2237 (vec_widen_smult_odd_v16qi): Likewise.
2238 (vec_widen_umult_odd_v8hi): Likewise.
2239 (vec_widen_smult_odd_v8hi): Likewise.
2240 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2241 vmuloub rather than call gen_vec_widen_umult_*.
2242 (vec_widen_umult_lo_v16qi): Likewise.
2243 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2244 vmulosb rather than call gen_vec_widen_smult_*.
2245 (vec_widen_smult_lo_v16qi): Likewise.
2246 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2247 rather than call gen_vec_widen_umult_*.
2248 (vec_widen_umult_lo_v8hi): Likewise.
2249 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2250 rather than call gen_vec_widen_smult_*.
2251 (vec_widen_smult_lo_v8hi): Likewise.
2253 2014-01-15 Jeff Law <law@redhat.com>
2255 PR tree-optimization/59747
2256 * ree.c (find_and_remove_re): Properly handle case where a second
2257 eliminated extension requires widening a copy created for elimination
2258 of a prior extension.
2259 (combine_set_extension): Ensure that the number of hard regs needed
2260 for a destination register does not change when we widen it.
2262 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2264 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2265 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2266 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2267 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2268 (avr-*-rtems*): Likewise.
2269 (bfin*-rtems*): Likewise.
2270 (moxie-*-rtems*): Likewise.
2271 (h8300-*-rtems*): Likewise.
2272 (i[34567]86-*-rtems*): Likewise.
2273 (lm32-*-rtems*): Likewise.
2274 (m32r-*-rtems*): Likewise.
2275 (m68k-*-rtems*): Likewise.
2276 (microblaze*-*-rtems*): Likewise.
2277 (mips*-*-rtems*): Likewise.
2278 (powerpc-*-rtems*): Likewise.
2279 (sh-*-rtems*): Likewise.
2280 (sparc-*-rtems*): Likewise.
2281 (sparc64-*-rtems*): Likewise.
2282 (v850-*-rtems*): Likewise.
2283 (m32c-*-rtems*): Likewise.
2285 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2287 PR rtl-optimization/59511
2288 * ira.c (ira_init_register_move_cost): Use memory costs for some
2289 cases of register move cost calculations.
2290 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2291 instead of BB frequency.
2292 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2293 * lra-assigns.c (find_hard_regno_for): Ditto.
2295 2014-01-15 Richard Biener <rguenther@suse.de>
2297 PR tree-optimization/59822
2298 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2299 (vectorizable_load): Use it to hoist defs of uses of invariant
2300 loads out of the loop.
2302 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2303 Kugan Vivekanandarajah <kuganv@linaro.org>
2305 PR target/59695
2306 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2307 truncation.
2309 2014-01-15 Richard Biener <rguenther@suse.de>
2311 PR rtl-optimization/59802
2312 * lcm.c (compute_available): Use inverted postorder to seed
2313 the initial worklist.
2315 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2317 PR target/59803
2318 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2319 ADDR_REGS for invalid symrefs in non-PIC code.
2321 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2323 PR other/58712
2324 * builtins.c (determine_block_size): Initialize *probable_max_size
2325 even if len_rtx is CONST_INT.
2327 2014-01-14 Andrew Pinski <apinski@cavium.com>
2329 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2330 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2331 (cortexa53_tunings): Likewise.
2332 (aarch64_sched_issue_rate): New function.
2333 (TARGET_SCHED_ISSUE_RATE): Define.
2335 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2337 * ira-costs.c (find_costs_and_classes): Add missed
2338 ira_init_register_move_cost_if_necessary.
2340 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2342 PR target/59787
2343 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2345 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2347 PR target/59794
2348 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2349 to indicate if type is used for function return value. Warn ABI
2350 change if the vector mode isn't available for function return value.
2351 (ix86_function_arg_advance): Pass false to type_natural_mode.
2352 (ix86_function_arg): Likewise.
2353 (ix86_gimplify_va_arg): Likewise.
2354 (function_arg_32): Don't warn ABI change.
2355 (ix86_function_value): Pass true to type_natural_mode.
2356 (ix86_return_in_memory): Likewise.
2357 (ix86_struct_value_rtx): Removed.
2358 (TARGET_STRUCT_VALUE_RTX): Likewise.
2360 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2362 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2363 converting a conditional jump into a conditional return.
2365 2014-01-14 Richard Biener <rguenther@suse.de>
2367 PR tree-optimization/58921
2368 PR tree-optimization/59006
2369 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2370 hoisting invariant stmts.
2371 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2372 invariant loads on the preheader edge if possible.
2374 2014-01-14 Joey Ye <joey.ye@arm.com>
2376 * doc/plugin.texi (Building GCC plugins): Update to C++.
2378 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2380 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2381 (_mm_rcp28_round_ss): Ditto.
2382 (_mm_rsqrt28_round_sd): Ditto.
2383 (_mm_rsqrt28_round_ss): Ditto.
2384 (_mm_rcp28_sd): Ditto.
2385 (_mm_rcp28_ss): Ditto.
2386 (_mm_rsqrt28_sd): Ditto.
2387 (_mm_rsqrt28_ss): Ditto.
2388 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2389 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2390 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2391 (IX86_BUILTIN_RCP28SD): Ditto.
2392 (IX86_BUILTIN_RCP28SS): Ditto.
2393 (IX86_BUILTIN_RSQRT28SD): Ditto.
2394 (IX86_BUILTIN_RSQRT28SS): Ditto.
2395 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2396 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2397 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2398 (ix86_expand_special_args_builtin): Expand new FTYPE.
2399 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2400 (srcp14<mode>): Make insn unary.
2401 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2402 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2403 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2404 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2405 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2406 Fix rounding: make it SAE only.
2407 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2408 Ditto.
2409 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2410 Ditto.
2411 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2412 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2413 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2414 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2415 (round_saeonly_mask_scalar_operand4): Ditto.
2416 (round_saeonly_mask_scalar_op3): Ditto.
2417 (round_saeonly_mask_scalar_op4): Ditto.
2419 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2421 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2422 Implement -maltivec=be for vec_insert and vec_extract.
2424 2014-01-10 DJ Delorie <dj@redhat.com>
2426 * config/msp430/msp430.md (call_internal): Don't allow memory
2427 references with SP as the base register.
2428 (call_value_internal): Likewise.
2429 * config/msp430/constraints.md (Yc): New. For memory references
2430 that don't use SP as a base register.
2432 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2433 "an integer without a # prefix"
2434 * config/msp430/msp430.md (epilogue_helper): Use it.
2436 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2438 PR target/59617
2439 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2440 AVX512F gather builtins.
2441 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2442 on gather decls with INTEGER_TYPE masktype.
2443 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2444 directly into the builtin rather than hoisting it before loop.
2446 PR tree-optimization/59387
2447 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2448 (scev_const_prop): If folded_casts and type has undefined overflow,
2449 use force_gimple_operand instead of force_gimple_operand_gsi and
2450 for each added stmt if it is assign with
2451 arith_code_with_undefined_signed_overflow, call
2452 rewrite_to_defined_overflow.
2453 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2454 gimple-fold.h instead.
2455 (arith_code_with_undefined_signed_overflow,
2456 rewrite_to_defined_overflow): Moved to ...
2457 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2458 rewrite_to_defined_overflow): ... here. No longer static.
2459 Include gimplify-me.h.
2460 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2461 rewrite_to_defined_overflow): New prototypes.
2463 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2465 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2467 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2469 * builtins.c (get_object_alignment_2): Minor tweak.
2470 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2472 2014-01-13 Christian Bruel <christian.bruel@st.com>
2474 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2475 optimized non constant lengths.
2477 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2479 PR libgomp/59194
2480 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2481 load as __atomic_load_N if possible.
2483 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2485 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2486 target parameter.
2487 (rs6000_expand_builtin): Adjust call.
2489 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2491 PR target/58115
2492 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2493 * config/rs6000/rs6000.c: Include target-globals.h.
2494 (rs6000_set_current_function): Instead of doing target_reinit
2495 unconditionally, use save_target_globals_default_opts and
2496 restore_target_globals.
2498 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2499 FPSCR.
2500 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2501 (rs6000_expand_builtin): Handle mffs and mtfsf.
2502 (rs6000_init_builtins): Define mffs and mtfsf.
2503 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2504 (rs6000_mffs): New pattern.
2505 (rs6000_mtfsf): New pattern.
2507 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2509 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2510 Start narrowing with START. Apply candidate-use pair
2511 and check overall cost in narrowing.
2512 (iv_ca_prune): Pass new argument.
2514 2014-01-10 Jeff Law <law@redhat.com>
2516 PR middle-end/59743
2517 * ree.c (combine_reaching_defs): Ensure the defining statement
2518 occurs before the extension when optimizing extensions with
2519 different source and destination hard registers.
2521 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2523 PR ipa/58585
2524 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2525 vtables into the type inheritance graph.
2527 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2529 PR rtl-optimization/59754
2530 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2531 modes in the REGNO != REGNO case.
2533 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2535 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2537 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2539 PR tree-optimization/59745
2540 * tree-predcom.c (tree_predictive_commoning_loop): Call
2541 free_affine_expand_cache if giving up because components is NULL.
2543 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2544 GC in payload of target_globals struct instead of allocating them on
2545 the heap and the larger structs separately using GC.
2546 * target-globals.h (struct target_globals): Make regs, hard_regs,
2547 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2548 of GTY((skip)) and change type to void *.
2549 (reset_target_globals): Cast loads from those fields to corresponding
2550 types.
2552 2014-01-10 Steve Ellcey <sellcey@mips.com>
2554 PR plugins/59335
2555 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2556 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2557 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2559 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2561 PR target/59744
2562 * aarch64-modes.def (CC_Zmode): New flags mode.
2563 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2564 represents an equality.
2565 (aarch64_get_condition_code): Handle CC_Zmode.
2566 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2568 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2570 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2571 extraction in good case.
2573 2014-01-10 Richard Biener <rguenther@suse.de>
2575 PR tree-optimization/59374
2576 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2577 checking after SLP discovery. Mark stmts not participating
2578 in any SLP instance properly.
2580 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2582 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2583 when handling a SET rtx.
2585 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2587 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2588 (cortex-a57): Likewise.
2589 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2591 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2593 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2594 non-iwmmxt builtins.
2596 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2598 PR ipa/58252
2599 PR ipa/59226
2600 * ipa-devirt.c record_target_from_binfo): Take as argument
2601 stack of binfos and lookup matching one for virtual inheritance.
2602 (possible_polymorphic_call_targets_1): Update.
2604 2014-01-10 Huacai Chen <chenhc@lemote.com>
2606 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2607 kernel strings for Loongson-2E/2F/3A.
2609 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2611 PR middle-end/59670
2612 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2613 is_gimple_call before calling gimple_call_internal_p.
2615 2014-01-09 Steve Ellcey <sellcey@mips.com>
2617 * Makefile.in (TREE_FLOW_H): Remove.
2618 (TREE_SSA_H): Add file names from tree-flow.h.
2619 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2620 * tree.h: Remove tree-flow.h reference.
2621 * hash-table.h: Remove tree-flow.h reference.
2622 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2623 reference with tree-ssa-loop.h.
2625 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2627 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2628 default element-order behavior for -maltivec.
2629 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2630 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2631 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2632 when targeting big endian, at least for now.
2633 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2635 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2637 PR middle-end/47735
2638 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2639 var satisfies use_register_for_decl, just take into account type
2640 alignment, rather than decl alignment.
2642 PR tree-optimization/59622
2643 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2644 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2645 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2646 Don't devirtualize for inplace at all. For targets.length () == 1,
2647 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2649 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2651 * config/i386/i386.md (cpu): Remove the unused btver1.
2653 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2655 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2657 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2659 PR target/58115
2660 * tree-core.h (struct target_globals): New forward declaration.
2661 (struct tree_target_option): Add globals field.
2662 * tree.h (TREE_TARGET_GLOBALS): Define.
2663 (prepare_target_option_nodes_for_pch): New prototype.
2664 * target-globals.h (struct target_globals): Define even if
2665 !SWITCHABLE_TARGET.
2666 * tree.c (prepare_target_option_node_for_pch,
2667 prepare_target_option_nodes_for_pch): New functions.
2668 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2669 * config/i386/i386.c: Include target-globals.h.
2670 (ix86_set_current_function): Instead of doing target_reinit
2671 unconditionally, use save_target_globals_default_opts and
2672 restore_target_globals.
2674 2014-01-09 Richard Biener <rguenther@suse.de>
2676 PR tree-optimization/59715
2677 * tree-cfg.h (split_critical_edges): Declare.
2678 * tree-cfg.c (split_critical_edges): Export.
2679 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2681 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2683 * cfgexpand.c (expand_stack_vars): Optionally disable
2684 asan stack protection.
2685 (expand_used_vars): Likewise.
2686 (partition_stack_vars): Likewise.
2687 * asan.c (asan_emit_stack_protection): Optionally disable
2688 after return stack usage.
2689 (instrument_derefs): Optionally disable memory access instrumentation.
2690 (instrument_builtin_call): Likewise.
2691 (instrument_strlen_call): Likewise.
2692 (asan_protect_global): Optionally disable global variables protection.
2693 * doc/invoke.texi: Added doc for new options.
2694 * params.def: Added new options.
2695 * params.h: Likewise.
2697 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2699 PR rtl-optimization/59724
2700 * ifcvt.c (cond_exec_process_if_block): Don't call
2701 flow_find_head_matching_sequence with 0 longest_match.
2702 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2703 non-active insns if !stop_after.
2704 (try_head_merge_bb): Revert 2014-01-07 changes.
2706 2014-01-08 Jeff Law <law@redhat.com>
2708 * ree.c (get_sub_rtx): New function, extracted from...
2709 (merge_def_and_ext): Here.
2710 (combine_reaching_defs): Use get_sub_rtx.
2712 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2714 * cgraph.h (varpool_variable_node): Do not choke on null node.
2716 2014-01-08 Catherine Moore <clm@codesourcery.com>
2718 * config/mips/mips.md (simple_return): Attempt to use JRC
2719 for microMIPS.
2720 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2722 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2724 PR rtl-optimization/59137
2725 * reorg.c (steal_delay_list_from_target): Call update_block for
2726 elided insns.
2727 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2729 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2731 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2732 two duplicate entries.
2734 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2736 Revert:
2737 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2739 * config/mips/mips.c (mips_truncated_op_cost): New function.
2740 (mips_rtx_costs): Adjust test for BADDU.
2741 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2743 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2745 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2746 (*baddu_si): ...this new pattern.
2748 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2750 PR ipa/59722
2751 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2753 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2755 PR middle-end/57748
2756 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2757 inner_reference_p.
2758 (expand_expr, expand_normal): Adjust.
2759 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2760 inner_reference_p. Use inner_reference_p to expand inner references.
2761 (store_expr): Adjust.
2762 * cfgexpand.c (expand_call_stmt): Adjust.
2764 2014-01-08 Rong Xu <xur@google.com>
2766 * gcov-io.c (gcov_var): Move from gcov-io.h.
2767 (gcov_position): Ditto.
2768 (gcov_is_error): Ditto.
2769 (gcov_rewrite): Ditto.
2770 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2771 only part to libgcc/libgcov.h.
2773 2014-01-08 Marek Polacek <polacek@redhat.com>
2775 PR middle-end/59669
2776 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2778 2014-01-08 Marek Polacek <polacek@redhat.com>
2780 PR sanitizer/59667
2781 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2783 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2785 PR rtl-optimization/59649
2786 * stor-layout.c (get_mode_bounds): For BImode return
2787 0 and STORE_FLAG_VALUE.
2789 2014-01-08 Richard Biener <rguenther@suse.de>
2791 PR middle-end/59630
2792 * gimple.h (is_gimple_builtin_call): Remove.
2793 (gimple_builtin_call_types_compatible_p): New.
2794 (gimple_call_builtin_p): New overload.
2795 * gimple.c (is_gimple_builtin_call): Remove.
2796 (validate_call): Rename to ...
2797 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2798 check return types.
2799 (validate_type): New static function.
2800 (gimple_call_builtin_p): New overload and adjust.
2801 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2802 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2803 (gimple_fold_stmt_to_constant_1): Likewise.
2804 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2806 2014-01-08 Richard Biener <rguenther@suse.de>
2808 PR middle-end/59471
2809 * gimplify.c (gimplify_expr): Gimplify register-register type
2810 VIEW_CONVERT_EXPRs to separate stmts.
2812 2014-01-07 Jeff Law <law@redhat.com>
2814 PR middle-end/53623
2815 * ree.c (combine_set_extension): Handle case where source
2816 and destination registers in an extension insn are different.
2817 (combine_reaching_defs): Allow source and destination registers
2818 in extension to be different under limited circumstances.
2819 (add_removable_extension): Remove restriction that the
2820 source and destination registers in the extension are the same.
2821 (find_and_remove_re): Emit a copy from the extension's
2822 destination to its source after the defining insn if
2823 the source and destination registers are different.
2825 PR middle-end/59285
2826 * ifcvt.c (merge_if_block): If we are merging a block with more than
2827 one successor with a block with no successors, remove any BARRIER
2828 after the second block.
2830 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2832 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2834 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2836 PR target/59652
2837 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2838 for 14-bit register offsets when INT14_OK_STRICT is false.
2840 2014-01-07 Roland Stigge <stigge@antcom.de>
2841 Michael Meissner <meissner@linux.vnet.ibm.com>
2843 PR 57386/target
2844 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2845 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2847 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2849 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2850 -mcpu.
2852 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2854 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2855 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2856 rtx is const0_rtx or not.
2858 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2860 PR target/58115
2861 * target-globals.c (save_target_globals): Remove this_fn_optab
2862 handling.
2863 * toplev.c: Include optabs.h.
2864 (target_reinit): Temporarily restore the global options if another
2865 set of options are in force.
2867 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2869 PR rtl-optimization/58668
2870 * cfgcleanup.c (flow_find_cross_jump): Don't count
2871 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2872 to determine what is counted.
2873 (flow_find_head_matching_sequence): Use active_insn_p to determine
2874 what is counted.
2875 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2876 counting change.
2877 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2878 determine what is counted.
2880 PR tree-optimization/59643
2881 * tree-predcom.c (split_data_refs_to_components): If one dr is
2882 read and one write, determine_offset fails and the write isn't
2883 in the bad component, just put the read into the bad component.
2885 2014-01-07 Mike Stump <mikestump@comcast.net>
2886 Jakub Jelinek <jakub@redhat.com>
2888 PR pch/59436
2889 * tree-core.h (struct tree_optimization_option): Change optabs
2890 type from unsigned char * to void *.
2891 * optabs.c (init_tree_optimization_optabs): Adjust
2892 TREE_OPTIMIZATION_OPTABS initialization.
2894 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2896 PR target/59644
2897 * config/i386/i386.h (struct machine_function): Add
2898 no_drap_save_restore field.
2899 * config/i386/i386.c (ix86_save_reg): Use
2900 !cfun->machine->no_drap_save_restore instead of
2901 crtl->stack_realign_needed.
2902 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2903 this function clears frame_pointer_needed. Set
2904 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2905 and DRAP reg is needed.
2907 2014-01-06 Marek Polacek <polacek@redhat.com>
2909 PR c/57773
2910 * doc/implement-c.texi: Mention that other integer types are
2911 permitted as bit-field types in strictly conforming mode.
2913 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2915 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2916 is newly allocated.
2918 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2920 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2922 2014-01-06 Martin Jambor <mjambor@suse.cz>
2924 PR ipa/59008
2925 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2926 to int.
2927 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2929 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2931 PR debug/59350
2932 PR debug/59510
2933 * var-tracking.c (add_stores): Preserve the value of the source even if
2934 we don't record the store.
2936 2014-01-06 Terry Guo <terry.guo@arm.com>
2938 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2940 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2942 PR bootstrap/59541
2943 * config/darwin.c (darwin_function_section): Adjust return values to
2944 correspond to optimisation changes made in r206070.
2946 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2948 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2949 from prefetch_block tune setting.
2950 (nocona_cost): Correct size of prefetch block to 64.
2952 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2954 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2955 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2956 used to save the static chain register in the computation of the offset
2957 from which the FP registers need to be restored.
2959 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2961 PR tree-optimization/59519
2962 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2963 ICE if get_current_def (current_new_name) is already non-NULL, as long
2964 as it is a phi result of some other phi in *new_exit_bb that has
2965 the same argument.
2967 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2968 or vmovdqu* for misaligned_operand.
2969 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2970 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2971 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2972 aligned_mem for AVX512F masked aligned load and store builtins and for
2973 non-temporal moves.
2975 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2977 PR tree-optimization/59651
2978 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2979 Address range for negative step should be added by TYPE_SIZE_UNIT.
2981 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2983 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2984 overlapping registers also for registers other than the stack pointer.
2986 2014-01-03 Marek Polacek <polacek@redhat.com>
2988 PR other/59661
2989 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2990 __builtin_FILE.
2992 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2994 PR target/59625
2995 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2996 asm goto as jump.
2998 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2999 (push splitter): Use <P:MODE_SIZE> instead of
3000 GET_MODE_SIZE (<P:MODE>mode).
3001 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3002 (mov -1, reg peephole2): Likewise.
3003 * config/i386/sse.md (*mov<mode>_internal,
3004 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3005 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3006 *<code><mode>3, *andnot<mode>3<mask_name>,
3007 <mask_codefor><code><mode>3<mask_name>): Likewise.
3008 * config/i386/subst.md (mask_mode512bit_condition,
3009 sd_mask_mode512bit_condition): Likewise.
3011 2014-01-02 Xinliang David Li <davidxl@google.com>
3013 PR tree-optimization/59303
3014 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3015 (dump_predicates): Better output format.
3016 (pred_equal_p): New function.
3017 (is_neq_relop_p): Ditto.
3018 (is_neq_zero_form_p): Ditto.
3019 (pred_expr_equal_p): Ditto.
3020 (pred_neg_p): Ditto.
3021 (simplify_pred): Ditto.
3022 (simplify_preds_2): Ditto.
3023 (simplify_preds_3): Ditto.
3024 (simplify_preds_4): Ditto.
3025 (simplify_preds): Ditto.
3026 (push_pred): Ditto.
3027 (push_to_worklist): Ditto.
3028 (get_pred_info_from_cmp): Ditto.
3029 (is_degenerated_phi): Ditto.
3030 (normalize_one_pred_1): Ditto.
3031 (normalize_one_pred): Ditto.
3032 (normalize_one_pred_chain): Ditto.
3033 (normalize_preds): Ditto.
3034 (normalize_cond_1): Remove function.
3035 (normalize_cond): Ditto.
3036 (is_gcond_subset_of): Ditto.
3037 (is_subset_of_any): Ditto.
3038 (is_or_set_subset_of): Ditto.
3039 (is_and_set_subset_of): Ditto.
3040 (is_norm_cond_subset_of): Ditto.
3041 (pred_chain_length_cmp): Ditto.
3042 (convert_control_dep_chain_into_preds): Type change.
3043 (find_predicates): Ditto.
3044 (find_def_preds): Ditto.
3045 (destroy_predicates_vecs): Ditto.
3046 (find_matching_predicates_in_rest_chains): Ditto.
3047 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3048 (is_pred_expr_subset): Ditto.
3049 (is_pred_chain_subset_of): Ditto.
3050 (is_included_in): Ditto.
3051 (is_superset_of): Ditto.
3053 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3055 Update copyright years.
3057 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3059 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3060 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3061 config/arc/arc.md, config/arc/arc.opt,
3062 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3063 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3064 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3065 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3066 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3067 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3068 vtable-verify.c, vtable-verify.h: Use the standard form for the
3069 copyright notice.
3071 2014-01-02 Tobias Burnus <burnus@net-b.de>
3073 * gcc.c (process_command): Update copyright notice dates.
3074 * gcov-dump.c: Ditto.
3075 * gcov.c: Ditto.
3076 * doc/cpp.texi: Bump @copying's copyright year.
3077 * doc/cppinternals.texi: Ditto.
3078 * doc/gcc.texi: Ditto.
3079 * doc/gccint.texi: Ditto.
3080 * doc/gcov.texi: Ditto.
3081 * doc/install.texi: Ditto.
3082 * doc/invoke.texi: Ditto.
3084 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3086 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3088 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3090 * config/i386/sse.md (*mov<mode>_internal): Guard
3091 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3093 PR rtl-optimization/59647
3094 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3095 new_rtx into UNSIGNED_FLOAT rtxes.
3096 \f
3097 Copyright (C) 2014 Free Software Foundation, Inc.
3099 Copying and distribution of this file, with or without modification,
3100 are permitted in any medium without royalty provided the copyright
3101 notice and this notice are preserved.