1 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
4 for SYMBOL_REF in large memory model.
6 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
10 (cortex-a57): Likewise.
11 (cortex-a57.cortex-a53): Likewise.
13 2014-02-06 Yury Gribov <y.gribov@samsung.com>
14 Kugan Vivekanandarajah <kuganv@linaro.org>
16 * config/arm/arm.c (arm_vector_alignment_reachable): Check
18 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
20 2014-02-06 Richard Biener <rguenther@suse.de>
22 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
23 set_loop_copy and initialize_original_copy_tables.
25 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
27 * config/aarch64/aarch64-simd.md
28 (aarch64_ashr_simddi): Change QI to SI.
30 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
31 Jakub Jelinek <jakub@redhat.com>
34 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
37 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
40 CODE_FOR_altivec_vpku[hw]um to
41 CODE_FOR_altivec_vpku[hw]um_direct.
42 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
43 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
44 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
45 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
47 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
49 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
50 generation for -maltivec=be.
51 (altivec_vsumsws): Simplify redundant test.
53 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
55 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
56 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
57 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
58 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
60 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
62 (altivec_vpks<VI_char>ss): Likewise.
63 (altivec_vpks<VI_char>us): Likewise.
64 (altivec_vpku<VI_char>us): Likewise.
65 (altivec_vpku<VI_char>um): Likewise.
66 (altivec_vpku<VI_char>um_direct): New (copy of
67 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
69 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
70 target is little endian and -maltivec=be is not specified.
71 (*altivec_vupkhs<VU_char>_direct): New (copy of
72 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
74 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
75 target is little endian and -maltivec=be is not specified.
76 (*altivec_vupkls<VU_char>_direct): New (copy of
77 altivec_vupkls<VU_char> that always emits vupkls*, for internal
79 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
80 little endian and -maltivec=be is not specified.
81 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
82 little endian and -maltivec=be is not specified.
84 2014-02-05 Richard Henderson <rth@redhat.com>
87 * combine-stack-adj.c: Revert r206943.
88 * sched-int.h (struct deps_desc): Add last_args_size.
89 * sched-deps.c (init_deps): Initialize it.
90 (sched_analyze_insn): Add OUTPUT dependencies between insns that
91 contain REG_ARGS_SIZE notes.
93 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
95 * lto-cgraph.c (asm_nodes_output): Make global.
96 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
97 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
99 (driver_handle_option): Handle OPT_fwpa.
101 2014-02-05 Jakub Jelinek <jakub@redhat.com>
104 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
105 a comment typo and formatting issue. If odr_hash hasn't been
106 created, return vNULL and set *completep to false.
109 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
110 bb with no successors.
112 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
115 * doc/invoke.texi (-march): Clarify documentation for ARM.
119 2014-02-05 Richard Biener <rguenther@suse.de>
121 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
122 when not vectorizing because of too many alias checks.
123 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
124 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
126 2014-02-05 Nick Clifton <nickc@redhat.com>
128 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
129 accept extended registers in any mode when compiling for the
132 2014-02-05 Yury Gribov <y.gribov@samsung.com>
134 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
135 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
136 sanitization attributes.
137 (can_inline_edge_p): Likewise.
138 (sanitize_attrs_match_for_inline_p): New function.
140 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
142 * ipa-prop.c (detect_type_change): Shor circuit testing of
143 type changes on THIS pointer.
145 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
148 * config/pa/pa.c (legitimize_tls_address): Return original address
149 if not passed a SYMBOL_REF rtx.
150 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
152 (pa_emit_move_sequence): Simplify TLS source operands.
153 (pa_legitimate_constant_p): Reject all TLS constants.
154 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
155 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
157 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
159 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
160 groups when we know they are controlled by LTO.
161 * varasm.c (default_binds_local_p_1): If object is in other partition,
162 it will be resolved locally.
164 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
166 * gcc/config/host-linux.c (linux_gt_pch_use_address): Don't
167 use SSIZE_MAX because it is not always defined.
169 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
172 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
173 threshold for pseudo splitting.
174 (update_ebb_live_info): Process call argument hard registers and
175 hard registers from insn definition too.
176 (max_small_class_regs_num): New constant.
177 (inherit_in_ebb): Update live hard regs through EBBs. Update
178 reloads_num only for small register classes. Don't split for
181 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
184 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
187 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
189 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
191 2014-02-04 Marek Polacek <polacek@redhat.com>
193 * gdbinit.in (pel): Define.
195 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
197 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
200 2014-02-04 Richard Biener <rguenther@suse.de>
203 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
204 in function context local.
205 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
206 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
207 similar to LTO_imported_decl_ref.
209 2014-02-04 Jakub Jelinek <jakub@redhat.com>
211 PR tree-optimization/60002
212 * cgraphclones.c (build_function_decl_skip_args): Clear
215 PR tree-optimization/60023
216 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
217 false to gsi_replace.
218 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
219 has been in some EH region and vec_stmt could throw, add
220 vec_stmt into the same EH region.
221 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
222 has no lhs, ignore it.
223 * internal-fn.c (expand_MASK_LOAD): Likewise.
226 * tree-inline.c (copy_forbidden): Fail for
227 __attribute__((optimize (0))) functions.
230 * omp-low.c (simd_clone_struct_copy): If from->inbranch
231 is set, copy one less argument.
232 (expand_simd_clones): Don't subtract clone_info->inbranch
233 from simd_clone_struct_alloc argument.
235 PR rtl-optimization/57915
236 * recog.c (simplify_while_replacing): If all unary/binary/relational
237 operation arguments are constant, attempt to simplify those.
240 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
241 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
243 2014-02-04 Richard Biener <rguenther@suse.de>
245 PR tree-optimization/60012
246 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
247 TBAA disambiguation to all DDRs.
249 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
252 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
253 (LINK_SPEC): Use it for -shared, -shared-libgcc.
255 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
258 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
260 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
262 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
263 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
265 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
268 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
269 to figure out targets of polymorphic calls with known decl.
270 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
271 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
272 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
273 (get_polymorphic_call_info): ... here.
274 (get_polymorphic_call_info_from_invariant): New function.
276 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
278 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
279 lookup via vtable pointer; check for type consistency
280 and turn inconsitent facts into UNREACHABLE.
281 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
282 * gimple-fold.c (gimple_get_virt_method_for_vtable):
283 Do not ICE on type inconsistent querries; return UNREACHABLE
286 2014-02-03 Richard Henderson <rth@twiddle.net>
289 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
290 already processed this node.
291 (normalize_one_pred_1): Pass along mark_set.
292 (normalize_one_pred): Create and destroy a pointer_set_t.
293 (normalize_one_pred_chain): Likewise.
295 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
297 PR gcov-profile/58602
298 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
300 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
303 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
304 Try to devirtualize by the knowledge of virtual table pointer given by
305 aggregate propagation.
306 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
307 ipa_print_node_jump_functions): Dump also offset that
308 is relevant for polymorphic calls.
309 (determine_known_aggregate_parts): Add arg_type parameter; use it
310 instead of determining the type from pointer type.
311 (ipa_compute_jump_functions_for_edge): Update call of
312 determine_known_aggregate_parts.
313 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
314 (gimple_get_virt_method_for_binfo): ... here; simplify using
315 vtable_pointer_value_to_vtable.
316 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
317 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
318 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
319 (vtable_pointer_value_to_vtable): Break out from ...; handle also
321 (vtable_pointer_value_to_binfo): ... here.
322 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
324 2014-02-03 Teresa Johnson <tejohnson@google.com>
326 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
327 redef of outer loop index variable.
329 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
333 * doc/extend.texi (Function Attributes): Typo.
335 2014-02-03 Cong Hou <congh@google.com>
337 PR tree-optimization/60000
338 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
339 if the vectorized statement is a store. A store statement can only
340 appear at the end of pattern statements.
342 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
344 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
345 (ix86_option_override_internal): Default long double to 64-bit for
346 32-bit Bionic and to 128-bit for 64-bit Bionic.
348 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
349 TARGET_LONG_DOUBLE_128 is true.
350 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
352 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
353 (mlong-double-64): Negate -mlong-double-128.
354 (mlong-double-128): New option.
356 * config/i386/i386-c.c (ix86_target_macros): Define
357 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
359 * doc/invoke.texi: Document -mlong-double-128.
361 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
363 PR rtl-optimization/60024
364 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
366 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
368 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
370 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
372 PR rtl-optimization/57662
373 * sel-sched.c (code_motion_path_driver): Do not mark already not
374 existing blocks in the visiting bitmap.
376 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
378 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
379 on the insn being emitted.
381 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
382 Will Deacon <will.deacon@arm.com>
384 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
386 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
388 * config/arm/arm-tables.opt: Regenerate.
390 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
392 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
393 for vector types other than V16QImode.
394 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
395 define_expand, and call altivec_expand_vec_perm_le when producing
396 code with little endian element order.
397 (*altivec_vperm_<mode>_internal): New insn having previous
398 behavior of altivec_vperm_<mode>.
399 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
400 altivec_expand_vec_perm_le when producing code with little endian
402 (*altivec_vperm_<mode>_uns_internal): New insn having previous
403 behavior of altivec_vperm_<mode>_uns.
405 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
407 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
408 (altivec_vsumsws): Add handling for -maltivec=be with a little
410 (altivec_vsumsws_direct): New.
411 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
414 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
416 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
417 vtable_pointer_value_to_binfo): New functions.
418 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
419 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
421 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
423 * config/nios2/nios2.md (load_got_register): Initialize GOT
424 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
425 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
427 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
429 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
430 preserverd by passthrough, do not propagate the type.
432 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
434 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
435 (mips_atomic_assign_expand_fenv): New function.
436 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
438 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
440 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
441 (__builtin_mips_set_fcsr): Likewise.
442 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
444 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
445 (mips16_expand_set_fcsr): Likewise.
446 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
447 (mips16_set_fcsr_stub): Likewise.
448 (mips16_get_fcsr_one_only_stub): New class.
449 (mips16_set_fcsr_one_only_stub): Likewise.
450 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
451 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
452 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
453 (hard_float): New availability predicate.
454 (mips_builtins): Add get_fcsr and set_fcsr.
455 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
456 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
457 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
458 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
459 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
462 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
464 * config/mips/mips.c (mips_one_only_stub): New class.
465 (mips_need_mips16_rdhwr_p): Replace with...
466 (mips16_rdhwr_stub): ...this new variable.
467 (mips16_stub_call_address): New function.
468 (mips16_rdhwr_one_only_stub): New class.
469 (mips_expand_thread_pointer): Use mips16_stub_call_address.
470 (mips_output_mips16_rdhwr): Delete.
471 (mips_finish_stub): New function.
472 (mips_code_end): Use it to handle rdhwr stubs.
474 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
477 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
478 when calculating size of integer atomic types.
480 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
482 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
484 2014-02-01 Jakub Jelinek <jakub@redhat.com>
486 PR tree-optimization/60003
487 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
488 * profile.c (branch_prob): Use gimple_call_builtin_p
489 to check for BUILT_IN_SETJMP_RECEIVER.
490 * tree-inline.c (copy_bb): Call notice_special_calls.
492 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
495 * lra-constraints.c (process_alt_operands): Update reload_sum only
498 2014-01-31 Richard Henderson <rth@redhat.com>
501 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
502 until after else_eh is processed.
504 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
506 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
507 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
508 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
509 in smmintrin.h, remove them.
510 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
511 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
512 * config/i386/i386.md (ROUND_SAE): Fix value.
513 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
514 (const48_operand): New.
515 * config/i386/subst.md (round), (round_expand): Use
516 const_4_or_8_to_11_operand.
517 (round_saeonly), (round_saeonly_expand): Use const48_operand.
519 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
521 * config/i386/constraints.md (Yk): Swap meaning with k.
522 * config/i386/i386.md (movhi_internal): Change Yk to k.
523 (movqi_internal): Ditto.
524 (*k<logic><mode>): Ditto.
527 (kandn<mode>): Ditto.
528 (*<code>hi_1): Ditto.
529 (*<code>qi_1): Ditto.
530 (kxnor<mode>): Ditto.
534 (*one_cmplhi2_1): Ditto.
535 (*one_cmplqi2_1): Ditto.
536 * config/i386/sse.md (): Change k to Yk.
537 (avx512f_load<mode>_mask): Ditto.
538 (avx512f_blendm<mode>): Ditto.
539 (avx512f_store<mode>_mask): Ditto.
540 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
541 (avx512f_storedqu<mode>_mask): Ditto.
542 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
544 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
545 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
546 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
547 (avx512f_maskcmp<mode>3): Ditto.
548 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
549 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
550 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
551 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
552 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
553 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
554 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
555 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
556 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
557 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
558 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
559 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
560 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
561 (vec_extract_lo_<mode>_maskm): Ditto.
562 (vec_extract_hi_<mode>_maskm): Ditto.
563 (avx512f_vternlog<mode>_mask): Ditto.
564 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
565 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
566 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
567 (avx512f_<code>v8div16qi2_mask): Ditto.
568 (avx512f_<code>v8div16qi2_mask_store): Ditto.
569 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
570 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
571 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
572 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
573 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
574 (*avx512pf_gatherpf<mode>df_mask): Ditto.
575 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
576 (*avx512pf_scatterpf<mode>df_mask): Ditto.
577 (avx512cd_maskb_vec_dupv8di): Ditto.
578 (avx512cd_maskw_vec_dupv16si): Ditto.
579 (avx512f_vpermi2var<mode>3_maskz): Ditto.
580 (avx512f_vpermi2var<mode>3_mask): Ditto.
581 (avx512f_vpermi2var<mode>3_mask): Ditto.
582 (avx512f_vpermt2var<mode>3_maskz): Ditto.
583 (*avx512f_gathersi<mode>): Ditto.
584 (*avx512f_gathersi<mode>_2): Ditto.
585 (*avx512f_gatherdi<mode>): Ditto.
586 (*avx512f_gatherdi<mode>_2): Ditto.
587 (*avx512f_scattersi<mode>): Ditto.
588 (*avx512f_scatterdi<mode>): Ditto.
589 (avx512f_compress<mode>_mask): Ditto.
590 (avx512f_compressstore<mode>_mask): Ditto.
591 (avx512f_expand<mode>_mask): Ditto.
592 * config/i386/subst.md (mask): Change k to Yk.
593 (mask_scalar_merge): Ditto.
596 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
598 * doc/extend.texi (Vector Extensions): Document ?: in C++.
600 2014-01-31 Richard Biener <rguenther@suse.de>
603 * builtins.c (fold_builtin_memory_op): Make sure to not
604 use a floating-point mode or a boolean or enumeral type for
607 2014-01-30 DJ Delorie <dj@redhat.com>
609 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
610 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
611 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
612 whenever main() has an epilogue.
614 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
616 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
617 unused variable "field".
618 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
619 (vsx_mergeh_<mode>): Likewise.
620 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
621 (altivec_vmrghh): Likewise.
622 (altivec_vmrghw): Likewise.
623 (altivec_vmrglb): Likewise.
624 (altivec_vmrglh): Likewise.
625 (altivec_vmrglw): Likewise.
626 (altivec_vspltb): Add missing uses.
627 (altivec_vsplth): Likewise.
628 (altivec_vspltw): Likewise.
629 (altivec_vspltsf): Likewise.
631 2014-01-30 Jakub Jelinek <jakub@redhat.com>
634 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
635 frame related instructions.
637 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
639 PR rtl-optimization/59959
640 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
641 any reload of register whose subreg is invalid.
643 2014-01-30 Jakub Jelinek <jakub@redhat.com>
645 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
646 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
647 Add missing return type - void.
649 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
651 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
652 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
653 remove element index adjustment for endian (now handled in vsx.md
655 (altivec_expand_vec_perm_const): Use
656 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
657 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
658 (vsx_xxspltw_<mode>): Adjust element index for little endian.
659 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
660 define_expand and a new define_insn *altivec_vspltb_internal;
661 adjust for -maltivec=be on a little endian target.
662 (altivec_vspltb_direct): New.
663 (altivec_vsplth): Divide into a define_expand and a new
664 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
665 little endian target.
666 (altivec_vsplth_direct): New.
667 (altivec_vspltw): Divide into a define_expand and a new
668 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
669 little endian target.
670 (altivec_vspltw_direct): New.
671 (altivec_vspltsf): Divide into a define_expand and a new
672 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
673 a little endian target.
675 2014-01-30 Richard Biener <rguenther@suse.de>
677 PR tree-optimization/59993
678 * tree-ssa-forwprop.c (associate_pointerplus): Check we
679 can propagate form the earlier stmt and avoid the transform
680 when the intermediate result is needed.
682 2014-01-30 Alangi Derick <alangiderick@gmail.com>
684 * README.Portability: Fix typo.
686 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
688 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
689 comparison_operator with ordered_comparison_operator.
691 2014-01-30 Nick Clifton <nickc@redhat.com>
693 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
694 Rename to mn10300_store_multiple_regs.
695 * config/mn10300/mn10300.c: Likewise.
696 * config/mn10300/mn10300.md (store_movm): Fix typo: call
698 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
699 Call mn10300_store_multiple_regs.
701 2014-01-30 Nick Clifton <nickc@redhat.com>
702 DJ Delorie <dj@redhat.com>
704 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
705 %fp 2 to keep registers after it properly word-aligned.
706 (rl78_alloc_physical_registers_umul): Handle the case where both
707 input operands are the same.
709 2014-01-30 Richard Biener <rguenther@suse.de>
711 PR tree-optimization/59903
712 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
715 2014-01-30 Jason Merrill <jason@redhat.com>
718 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
721 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
723 2014-01-30 Richard Biener <rguenther@suse.de>
725 PR tree-optimization/59951
726 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
728 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
731 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
732 SFmode to DFmode case.
734 2014-01-29 DJ Delorie <dj@redhat.com>
736 * config/msp430/msp430.opt (-minrt): New.
737 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
739 (ENDFILE_SPEC): Likewise.
741 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
743 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
744 (estimate_function_body_sizes): Use it.
746 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
749 * dwarf2out.c (is_cxx_auto): New.
750 (is_base_type): Use it.
751 (gen_type_die_with_usage): Likewise.
753 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
755 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
756 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
757 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
758 -maltivec=be with LE targets.
759 (vsx_mergeh_<mode>): Likewise.
760 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
761 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
762 (altivec_vmrghb): Replace with define_expand and new
763 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
764 (altivec_vmrghb_direct): New define_insn.
765 (altivec_vmrghh): Replace with define_expand and new
766 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
767 (altivec_vmrghh_direct): New define_insn.
768 (altivec_vmrghw): Replace with define_expand and new
769 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
770 (altivec_vmrghw_direct): New define_insn.
771 (*altivec_vmrghsf): Adjust for endianness.
772 (altivec_vmrglb): Replace with define_expand and new
773 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
774 (altivec_vmrglb_direct): New define_insn.
775 (altivec_vmrglh): Replace with define_expand and new
776 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
777 (altivec_vmrglh_direct): New define_insn.
778 (altivec_vmrglw): Replace with define_expand and new
779 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
780 (altivec_vmrglw_direct): New define_insn.
781 (*altivec_vmrglsf): Adjust for endianness.
782 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
783 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
784 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
785 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
786 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
787 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
788 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
789 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
791 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
793 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
794 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
797 2014-01-29 Richard Biener <rguenther@suse.de>
799 PR tree-optimization/58742
800 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
801 associate_pointerplus_align.
802 (associate_pointerplus_diff): New function.
803 (associate_pointerplus): Likewise. Call associate_pointerplus_align
804 and associate_pointerplus_diff.
806 2014-01-29 Richard Biener <rguenther@suse.de>
808 * lto-streamer.h (LTO_major_version): Bump to 3.
809 (LTO_minor_version): Reset to 0.
811 2014-01-29 Renlin Li <Renlin.Li@arm.com>
813 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
814 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
815 (arm_file_start): Generate correct asm header for armv7ve.
816 * config/arm/bpabi.h: Add multilib support for armv7ve.
817 * config/arm/driver-arm.c: Change the architectures of cortex-a7
818 and cortex-a15 to armv7ve.
819 * config/arm/t-aprofile: Add multilib support for armv7ve.
820 * doc/invoke.texi: Document -march=armv7ve.
822 2014-01-29 Richard Biener <rguenther@suse.de>
824 PR tree-optimization/58742
825 * tree-ssa-forwprop.c (associate_plusminus): Return true
826 if we changed sth, defer EH cleanup to ...
827 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
828 (simplify_mult): New function.
830 2014-01-29 Jakub Jelinek <jakub@redhat.com>
833 PR tree-optimization/59920
834 * tree.c (build_common_builtin_nodes): Remove
835 __builtin_setjmp_dispatcher initialization.
836 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
837 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
838 instead of gsi_after_labels + manually skipping debug stmts.
839 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
840 ignore bbs with IFN_ABNORMAL_DISPATCHER.
841 * tree-inline.c (copy_edges_for_bb): Remove
842 can_make_abnormal_goto argument, instead add abnormal_goto_dest
843 argument. Ignore computed_goto_p stmts. Don't call
844 make_abnormal_goto_edges. If a call might need abnormal edges
845 for non-local gotos, see if it already has an edge to
846 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
847 with true argument, don't do anything then, otherwise add
848 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
849 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
851 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
852 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
853 (lower_stmt): Don't set data->calls_builtin_setjmp.
854 (lower_builtin_setjmp): Adjust comment.
855 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
856 * tree-cfg.c (found_computed_goto): Remove.
857 (factor_computed_gotos): Remove.
858 (make_goto_expr_edges): Return bool, true for computed gotos.
859 Don't call make_abnormal_goto_edges.
860 (build_gimple_cfg): Don't set found_computed_goto, don't call
861 factor_computed_gotos.
862 (computed_goto_p): No longer static.
863 (make_blocks): Don't set found_computed_goto.
864 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
865 (make_edges): If make_goto_expr_edges returns true, push bb
866 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
867 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
868 vector. Record mapping between bbs and OpenMP regions if there
869 are any, adjust make_gimple_omp_edges caller. Call
870 handle_abnormal_edges.
871 (make_abnormal_goto_edges): Remove.
872 * tree-cfg.h (make_abnormal_goto_edges): Remove.
873 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
874 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
875 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
876 * internal-fn.def (ABNORMAL_DISPATCHER): New.
877 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
878 filling *region also set *region_idx to (*region)->entry->index.
881 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
882 For REGs set ORIGINAL_REGNO.
884 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
886 * doc/md.texi: Mention that a target shouldn't implement
887 vec_widen_(s|u)mul_even/odd pair if it is less efficient
890 2014-01-29 Jakub Jelinek <jakub@redhat.com>
892 PR tree-optimization/59594
893 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
894 a copy of the datarefs vector rather than the vector itself.
896 2014-01-28 Jason Merrill <jason@redhat.com>
899 * dwarf2out.c (auto_die): New static.
900 (gen_type_die_with_usage): Handle C++1y 'auto'.
901 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
904 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
907 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
908 (SPEC_X32): Likewise.
910 * config/i386/i386.c (ix86_option_override_internal): Turn off
911 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
913 (x86_file_start): Output .code16gcc for TARGET_16BIT.
914 * config/i386/i386.h (TARGET_16BIT): New macro.
915 (TARGET_16BIT_P): Likewise.
916 * config/i386/i386.opt: Add m16.
917 * doc/invoke.texi: Document -m16.
919 2014-01-28 Jakub Jelinek <jakub@redhat.com>
921 PR preprocessor/59935
922 * input.c (location_get_source_line): Bail out on when line number
923 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
925 2014-01-28 Richard Biener <rguenther@suse.de>
927 PR tree-optimization/58742
928 * tree-ssa-forwprop.c (associate_plusminus): Handle
929 pointer subtraction of the form (T)(P + A) - (T)P.
931 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
933 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
936 2014-01-28 Richard Biener <rguenther@suse.de>
939 2014-01-28 Richard Biener <rguenther@suse.de>
941 PR rtl-optimization/45364
942 PR rtl-optimization/59890
943 * var-tracking.c (local_get_addr_clear_given_value): Handle
944 already cleared slot.
945 (val_reset): Handle not allocated local_get_addr_cache.
946 (vt_find_locations): Use post-order on the inverted CFG.
948 2014-01-28 Richard Biener <rguenther@suse.de>
950 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
952 2014-01-28 Richard Biener <rguenther@suse.de>
954 PR rtl-optimization/45364
955 PR rtl-optimization/59890
956 * var-tracking.c (local_get_addr_clear_given_value): Handle
957 already cleared slot.
958 (val_reset): Handle not allocated local_get_addr_cache.
959 (vt_find_locations): Use post-order on the inverted CFG.
961 2014-01-28 Alan Modra <amodra@gmail.com>
963 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
964 * configure.ac <recursive call for build != host>: Define
965 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
966 and LD_FOR_BUILD too.
967 * configure: Regenerate.
969 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
971 * config/i386/i386.c (get_builtin_code_for_version): Separate
972 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
973 Broadwell from Haswell.
975 2014-01-27 Steve Ellcey <sellcey@mips.com>
977 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
978 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
979 * config/mips/mips.c (mips_option_override): Change setting
981 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
982 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
983 Change from Mask to Var.
985 2014-01-27 Jeff Law <law@redhat.com>
987 * ipa-inline.c (inline_small_functions): Fix typo.
989 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
991 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
992 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
993 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
994 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
995 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
996 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
997 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
998 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
999 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1000 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1001 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1002 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1003 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1004 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1005 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1006 (_mm512_storeu_epi64): Ditto.
1007 (_mm512_cmpge_epi32_mask): Ditto.
1008 (_mm512_cmpge_epu32_mask): Ditto.
1009 (_mm512_cmpge_epi64_mask): Ditto.
1010 (_mm512_cmpge_epu64_mask): Ditto.
1011 (_mm512_cmple_epi32_mask): Ditto.
1012 (_mm512_cmple_epu32_mask): Ditto.
1013 (_mm512_cmple_epi64_mask): Ditto.
1014 (_mm512_cmple_epu64_mask): Ditto.
1015 (_mm512_cmplt_epi32_mask): Ditto.
1016 (_mm512_cmplt_epu32_mask): Ditto.
1017 (_mm512_cmplt_epi64_mask): Ditto.
1018 (_mm512_cmplt_epu64_mask): Ditto.
1019 (_mm512_cmpneq_epi32_mask): Ditto.
1020 (_mm512_cmpneq_epu32_mask): Ditto.
1021 (_mm512_cmpneq_epi64_mask): Ditto.
1022 (_mm512_cmpneq_epu64_mask): Ditto.
1023 (_mm512_expand_pd): Ditto.
1024 (_mm512_expand_ps): Ditto.
1025 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1026 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1027 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1028 * config/i386/i386.c (ix86_builtins): Add
1029 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1030 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1031 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1032 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1033 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1034 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1035 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1036 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1037 IX86_BUILTIN_PMOVUSQW512_MEM.
1038 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1039 __builtin_ia32_pmovsqd512mem_mask,
1040 __builtin_ia32_pmovqd512mem_mask,
1041 __builtin_ia32_pmovusqw512mem_mask,
1042 __builtin_ia32_pmovsqw512mem_mask,
1043 __builtin_ia32_pmovqw512mem_mask,
1044 __builtin_ia32_pmovusdw512mem_mask,
1045 __builtin_ia32_pmovsdw512mem_mask,
1046 __builtin_ia32_pmovdw512mem_mask,
1047 __builtin_ia32_pmovqb512mem_mask,
1048 __builtin_ia32_pmovusqb512mem_mask,
1049 __builtin_ia32_pmovsqb512mem_mask,
1050 __builtin_ia32_pmovusdb512mem_mask,
1051 __builtin_ia32_pmovsdb512mem_mask,
1052 __builtin_ia32_pmovdb512mem_mask.
1053 (bdesc_args): Add __builtin_ia32_expanddf512,
1054 __builtin_ia32_expandsf512.
1055 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1056 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1057 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1058 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1059 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1060 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1061 (avx512f_<code>v8div16qi2_mask_store): This.
1062 (avx512f_expand<mode>): New.
1064 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1066 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1068 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1069 (_mm512_prefetch_i32scatter_pd): Ditto.
1070 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1071 (_mm512_prefetch_i64scatter_pd): Ditto.
1072 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1073 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1074 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1075 (_mm512_prefetch_i32scatter_ps): Ditto.
1076 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1077 (_mm512_prefetch_i64scatter_ps): Ditto.
1078 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1079 * config/i386/i386-builtin-types.def: Define
1080 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1081 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1082 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1083 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1084 IX86_BUILTIN_SCATTERPFQPD.
1085 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1086 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1087 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1088 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1089 __builtin_ia32_scatterpfqps.
1090 (ix86_expand_builtin): Expand new built-ins.
1091 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1092 fix memory access data type.
1093 (*avx512pf_gatherpf<mode>_mask): Ditto.
1094 (*avx512pf_gatherpf<mode>): Ditto.
1095 (avx512pf_scatterpf<mode>): Ditto.
1096 (*avx512pf_scatterpf<mode>_mask): Ditto.
1097 (*avx512pf_scatterpf<mode>): Ditto.
1098 (GATHER_SCATTER_SF_MEM_MODE): New.
1099 (avx512pf_gatherpf<mode>df): Ditto.
1100 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1101 (*avx512pf_scatterpf<mode>df): Ditto.
1103 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1106 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1107 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1110 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1112 * common/config/arm/arm-common.c
1113 (arm_rewrite_mcpu): Handle multiple names.
1115 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1117 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1119 * gimple-builder.h (create_gimple_tmp): Delete.
1121 2014-01-27 Christian Bruel <christian.bruel@st.com>
1123 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1126 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1128 * config/pa/pa.md (call): Generate indirect long calls to non-local
1129 functions when outputing 32-bit code.
1130 (call_value): Likewise except for special call to buggy powf function.
1132 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1133 portable runtime and PIC indirect calls.
1134 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1135 and PIC call sequences. Use ldo instead of blr to set return register
1136 in PIC call sequence.
1138 2014-01-25 Walter Lee <walt@tilera.com>
1140 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1141 avoid clobbering a live register.
1143 2014-01-25 Walter Lee <walt@tilera.com>
1145 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1146 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1147 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1148 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1150 2014-01-25 Walter Lee <walt@tilera.com>
1152 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1153 arguments on even registers.
1154 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1156 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1157 (BIGGEST_ALIGNMENT): Ditto.
1158 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1160 2014-01-25 Walter Lee <walt@tilera.com>
1162 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1163 insns before bundling.
1164 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1166 2014-01-25 Walter Lee <walt@tilera.com>
1168 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1169 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1170 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1172 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1174 * config/mips/constraints.md (kl): Delete.
1175 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1176 define expands, using...
1177 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1178 instructions for MIPS16.
1179 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1180 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1182 2014-01-25 Walter Lee <walt@tilera.com>
1184 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1188 2014-01-25 Walter Lee <walt@tilera.com>
1190 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1191 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1193 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1195 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1198 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1200 * print-rtl.c (in_call_function_usage): New var.
1201 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1202 EXPR_LIST mode as mode and not as reg note name.
1205 * cfgloopmanip.c (copy_loop_info): If
1206 loop->warned_aggressive_loop_optimizations, make sure
1207 the flag is set in target loop too.
1209 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1211 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1213 * builtins.def: Likewise.
1214 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1215 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1216 * ira.c (ira_setup_eliminable_regset): Likewise.
1217 * omp-low.c (gate_expand_omp): Likewise.
1218 (execute_lower_omp): Likewise.
1219 (diagnose_sb_0): Likewise.
1220 (gate_diagnose_omp_blocks): Likewise.
1221 (simd_clone_clauses_extract): Likewise.
1224 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1226 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1227 correction for little endian...
1228 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1231 2014-01-24 Jeff Law <law@redhat.com>
1233 PR tree-optimization/59919
1234 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1235 for non-returning calls.
1237 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1239 * common/config/aarch64/aarch64-common.c
1240 (aarch64_rewrite_mcpu): Handle multiple names.
1241 * config/aarch64/aarch64.h
1242 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1244 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1246 * input.c (add_file_to_cache_tab): Handle the case where fopen
1249 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1252 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1253 from push operand if code of push isn't PRE_DEC.
1255 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1258 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1259 -mquad-memory-atomic. Update -mquad-memory documentation to say
1260 it is only used for non-atomic loads/stores.
1262 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1263 -mquad-memory or -mquad-memory-atomic switches.
1265 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1266 -mquad-memory-atomic to ISA 2.07 support.
1268 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1269 to separate support of normal quad word memory operations (ldq, stq)
1270 from the atomic quad word memory operations.
1272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1273 support to separate non-atomic quad word operations from atomic
1274 quad word operations. Disable non-atomic quad word operations in
1275 little endian mode so that we don't have to swap words after the
1276 load and before the store.
1277 (quad_load_store_p): Add comment about atomic quad word support.
1278 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1279 options printed with -mdebug=reg.
1281 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1282 -mquad-memory-atomic as the test for whether we have quad word
1283 atomic instructions.
1284 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1285 or -mp8-vector are used, allow byte/half-word atomic operations.
1287 * config/rs6000/sync.md (load_lockedti): Insure that the address
1288 is a proper indexed or indirect address for the lqarx instruction.
1289 On little endian systems, swap the hi/lo registers after the lqarx
1291 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1292 insure the address is valid for the lqarx instruction.
1293 (store_conditionalti): Insure that the address is a proper indexed
1294 or indirect address for the stqcrx. instruction. On little endian
1295 systems, swap the hi/lo registers before doing the stqcrx.
1297 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1298 insure the address is valid for the stqcrx. instruction.
1300 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1301 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1302 type of quad memory support is available.
1304 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1307 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1308 there is a danger of looping.
1310 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1312 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1313 force flag_ira_loop_pressure if set via command line.
1315 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1317 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1318 (ashr_simd): New builtin handling DI mode.
1319 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1320 (aarch64_sshr_simddi): New match pattern.
1321 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1322 (vshrd_n_s64): Likewise.
1323 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1325 2014-01-23 Nick Clifton <nickc@redhat.com>
1327 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1328 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1329 favour of mcu specific scripts.
1330 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1333 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1334 Alex Velenko <Alex.Velenko@arm.com>
1336 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1337 (vaddv_s16): Likewise.
1338 (vaddv_s32): Likewise.
1339 (vaddv_u8): Likewise.
1340 (vaddv_u16): Likewise.
1341 (vaddv_u32): Likewise.
1342 (vaddvq_s8): Likewise.
1343 (vaddvq_s16): Likewise.
1344 (vaddvq_s32): Likewise.
1345 (vaddvq_s64): Likewise.
1346 (vaddvq_u8): Likewise.
1347 (vaddvq_u16): Likewise.
1348 (vaddvq_u32): Likewise.
1349 (vaddvq_u64): Likewise.
1350 (vaddv_f32): Likewise.
1351 (vaddvq_f32): Likewise.
1352 (vaddvq_f64): Likewise.
1353 (vmaxv_f32): Likewise.
1354 (vmaxv_s8): Likewise.
1355 (vmaxv_s16): Likewise.
1356 (vmaxv_s32): Likewise.
1357 (vmaxv_u8): Likewise.
1358 (vmaxv_u16): Likewise.
1359 (vmaxv_u32): Likewise.
1360 (vmaxvq_f32): Likewise.
1361 (vmaxvq_f64): Likewise.
1362 (vmaxvq_s8): Likewise.
1363 (vmaxvq_s16): Likewise.
1364 (vmaxvq_s32): Likewise.
1365 (vmaxvq_u8): Likewise.
1366 (vmaxvq_u16): Likewise.
1367 (vmaxvq_u32): Likewise.
1368 (vmaxnmv_f32): Likewise.
1369 (vmaxnmvq_f32): Likewise.
1370 (vmaxnmvq_f64): Likewise.
1371 (vminv_f32): Likewise.
1372 (vminv_s8): Likewise.
1373 (vminv_s16): Likewise.
1374 (vminv_s32): Likewise.
1375 (vminv_u8): Likewise.
1376 (vminv_u16): Likewise.
1377 (vminv_u32): Likewise.
1378 (vminvq_f32): Likewise.
1379 (vminvq_f64): Likewise.
1380 (vminvq_s8): Likewise.
1381 (vminvq_s16): Likewise.
1382 (vminvq_s32): Likewise.
1383 (vminvq_u8): Likewise.
1384 (vminvq_u16): Likewise.
1385 (vminvq_u32): Likewise.
1386 (vminnmv_f32): Likewise.
1387 (vminnmvq_f32): Likewise.
1388 (vminnmvq_f64): Likewise.
1390 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1392 * config/aarch64/aarch64-simd.md
1393 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1394 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1395 (*aarch64_mul3_elt<mode>): Likewise.
1396 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1397 (*aarch64_mul3_elt_to_64v2df): Likewise.
1398 (*aarch64_mla_elt<mode>): Likewise.
1399 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1400 (*aarch64_mls_elt<mode>): Likewise.
1401 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1402 (*aarch64_fma4_elt<mode>): Likewise.
1403 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1404 (*aarch64_fma4_elt_to_64v2df): Likewise.
1405 (*aarch64_fnma4_elt<mode>): Likewise.
1406 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1407 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1408 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1409 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1410 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1411 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1412 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1413 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1414 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1416 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1418 * config/aarch64/aarch64-simd.md
1419 (aarch64_be_checked_get_lane<mode>): New define_expand.
1420 * config/aarch64/aarch64-simd-builtins.def
1421 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1422 New builtin definition.
1423 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1424 Use new safe be builtin.
1426 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1428 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1430 (aarch64_be_st1<mode>): Likewise.
1431 (aarch_ld1<VALL:mode>): Define_expand modified.
1432 (aarch_st1<VALL:mode>): Likewise.
1433 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1434 (UNSPEC_ST1): Likewise.
1436 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1438 * config/microblaze/microblaze.md: Add trap insn and attribute
1440 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1442 PR preprocessor/58580
1443 * input.h (location_get_source_line): Take an additional line_size
1445 (void diagnostics_file_cache_fini): Declare new function.
1446 * input.c (struct fcache): New type.
1447 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1448 New static constants.
1449 (diagnostic_file_cache_init, total_lines_num)
1450 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1451 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1452 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1453 (get_next_line, read_next_line, goto_next_line, read_line_num):
1454 New static function definitions.
1455 (diagnostic_file_cache_fini): New function.
1456 (location_get_source_line): Take an additional output line_len
1457 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1459 * diagnostic.c (diagnostic_finish): Call
1460 diagnostic_file_cache_fini.
1461 (adjust_line): Take an additional input parameter for the length
1462 of the line, rather than calculating it with strlen.
1463 (diagnostic_show_locus): Adjust the use of
1464 location_get_source_line and adjust_line with respect to their new
1465 signature. While displaying a line now, do not stop at the first
1466 null byte. Rather, display the zero byte as a space and keep
1467 going until we reach the size of the line.
1468 * Makefile.in: Add vec.o to OBJS-libcommon
1470 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1471 Ilya Tocar <ilya.tocar@intel.com>
1473 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1474 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1475 (__builtin_ia32_kmov16): Ditto.
1476 * config/i386/i386.md (UNSPEC_KMOV): New.
1479 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1481 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1482 (_mm512_storeu_si512): Ditto.
1484 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1487 * rtl.h (get_referenced_operands): Declare.
1488 * recog.c (get_referenced_operands): New function.
1489 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1490 operands have been referenced when recording LO_SUM references.
1492 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1494 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1496 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1498 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1499 Enable for generic and recent AMD targets.
1501 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1503 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1504 ARG_SIZE note when adjustment was eliminated.
1506 2014-01-22 Jeff Law <law@redhat.com>
1508 PR tree-optimization/59597
1509 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1510 in file. Accept new argument REGISTERING and use it to modify
1511 dump output appropriately.
1512 (register_jump_thread): Corresponding changes.
1513 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1514 thread paths involving joiner blocks. Add code to dump cancelled
1515 jump threading paths.
1517 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1519 PR rtl-optimization/59477
1520 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1521 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1523 2014-01-22 Tom Tromey <tromey@redhat.com>
1525 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1527 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1529 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1531 PR rtl-optimization/59896
1532 * lra-constraints.c (process_alt_operands): Check unused note for
1533 matched operands of insn with no output reloads.
1535 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1537 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1538 (mips_move_from_gpr_cost): Likewise.
1540 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1542 PR rtl-optimization/59858
1543 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1544 ira_class_hard_regs_num.
1545 (process_alt_operands): Increase reject for dying matched operand.
1547 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1550 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1551 smaller than size, perform several stores or loads and stores
1552 at dst + count - size to store or copy all of size bytes, rather
1553 than just last modesize bytes.
1555 2014-01-20 DJ Delorie <dj@redhat.com>
1557 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1558 that CLOBBERs are REGs before propogating their values.
1560 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1563 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1564 (cgraph_inline_failed_type): New function.
1565 * cgraph.h (DEFCIFCODE): Add type.
1566 (cgraph_inline_failed_type_t): New enum.
1567 (cgraph_inline_failed_type): New prototype.
1568 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1569 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1570 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1571 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1572 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1573 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1574 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1575 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1576 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1577 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1578 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1579 OPTIMIZATION_MISMATCH.
1580 * tree-inline.c (expand_call_inline): Emit errors during
1581 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1583 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1586 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1587 mode attribute in insn output.
1589 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1591 * output.h (output_constant): Delete.
1592 * varasm.c (output_constant): Make private.
1594 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1596 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1598 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1601 * tree.h (fold_builtin_strcat): New prototype.
1602 * builtins.c (fold_builtin_strcat): No longer static. Add len
1603 argument, if non-NULL, don't call c_strlen. Optimize
1604 directly into __builtin_memcpy instead of __builtin_strcpy.
1605 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1606 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1608 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1610 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1611 for SImode_address_operand operands, having only a REG argument.
1613 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1615 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1616 loader name using mbig-endian.
1617 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1619 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1621 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1625 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1627 * config/aarch64/aarch64-protos.h
1628 (aarch64_cannot_change_mode_class_ptr): Declare.
1629 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1630 aarch64_cannot_change_mode_class_ptr): New.
1631 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1632 backend hook aarch64_cannot_change_mode_class.
1634 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1636 * common/config/aarch64/aarch64-common.c
1637 (aarch64_handle_option): Don't handle any option order logic here.
1638 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1639 selected_cpu, warn on architecture version mismatch.
1640 (aarch64_override_options): Fix parsing order for option strings.
1642 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1643 Iain Sandoe <iain@codesourcery.com>
1646 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1647 warning. Amend comment to reflect current functionality.
1649 2014-01-20 Richard Biener <rguenther@suse.de>
1652 * builtins.c (fold_builtin_strcat): Remove case better handled
1653 by tree-ssa-strlen.c.
1655 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1657 * config/aarch64/aarch64.opt
1658 (mcpu, march, mtune): Make case-insensitive.
1660 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1663 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1664 if operands[1] is a REG or ZERO_EXTEND of a REG.
1666 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
1668 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1670 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1672 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1673 long non-pic millicode calls.
1675 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1677 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1679 2014-01-19 Kito Cheng <kito@0xlab.org>
1681 * builtins.c (expand_movstr): Check movstr expand done or fail.
1683 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1684 H.J. Lu <hongjiu.lu@intel.com>
1687 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1688 to DImode for zero-extended addresses.
1690 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1692 PR rtl-optimization/57763
1693 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1694 on the new indirect jump_insn and increment LABEL_NUSES (label).
1696 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1700 * config.gcc (x86_archs): New variable.
1701 (x86_64_archs): Likewise.
1702 (x86_cpus): Likewise.
1703 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1704 --with-arch/--with-cpu= options.
1705 Support --with-arch=/--with-cpu={nehalem,westmere,
1706 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1708 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1710 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1711 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1713 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1715 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1717 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1720 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1721 clear cpp_get_options (parse_in)->warn_unused_macros for
1722 ix86_target_macros_internal with cpp_define.
1724 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1726 * jump.c (delete_related_insns): Keep (use (insn))s.
1727 * reorg.c (redundant_insn): Check for barriers too.
1729 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1731 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1733 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1735 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1736 call to $$dyncall when TARGET_LONG_CALLS is true.
1738 2014-01-17 Jeff Law <law@redhat.com>
1740 * ree.c (combine_set_extension): Temporarily disable test for
1741 changing number of hard registers.
1743 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1746 * ipa-inline-analysis.c (inline_free_summary):
1747 Do not free summary of aliases.
1749 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1752 * gimplify.c (gimplify_expr): Use create_tmp_var
1753 instead of create_tmp_var_raw. If cond doesn't have
1754 integral type, don't add the IFN_ANNOTATE builtin at all.
1756 2014-01-17 Martin Jambor <mjambor@suse.cz>
1759 * ipa-cp.c (prev_edge_clone): New variable.
1760 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1761 Also resize prev_edge_clone vector.
1762 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1763 (ipcp_edge_removal_hook): New function.
1764 (ipcp_driver): Register ipcp_edge_removal_hook.
1766 2014-01-17 Andrew Pinski <apinski@cavium.com>
1767 Steve Ellcey <sellcey@mips.com>
1770 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1773 2014-01-17 Jeff Law <law@redhat.com>
1776 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1777 so that pass_ccp runs first.
1779 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1781 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1782 (ix86_adjust_cost): Use !TARGET_XXX.
1783 (do_reorder_for_imul): Likewise.
1784 (swap_top_of_ready_list): Likewise.
1785 (ix86_sched_reorder): Likewise.
1787 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1789 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1790 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1791 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1792 (intel_memset): New. Duplicate slm_memset.
1793 (intel_cost): New. Duplicate slm_cost.
1794 (m_INTEL): New macro.
1795 (processor_target_table): Add "intel".
1796 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1797 with PROCESSOR_INTEL for "intel".
1798 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1799 PROCESSOR_SILVERMONT.
1800 (ix86_issue_rate): Likewise.
1801 (ix86_adjust_cost): Likewise.
1802 (ia32_multipass_dfa_lookahead): Likewise.
1803 (swap_top_of_ready_list): Likewise.
1804 (ix86_sched_reorder): Likewise.
1805 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1806 instead of TARGET_OPT_AGU.
1807 * config/i386/i386.h (TARGET_INTEL): New.
1808 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1809 (processor_type): Add PROCESSOR_INTEL.
1810 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1811 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1813 2014-01-17 Marek Polacek <polacek@redhat.com>
1816 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1819 2014-01-17 Richard Biener <rguenther@suse.de>
1821 PR tree-optimization/46590
1822 * opts.c (default_options_table): Add entries for
1823 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1824 all enabled at -O1 but not for -Og.
1825 * common.opt (fbranch-count-reg): Remove Init(1).
1826 (fmove-loop-invariants): Likewise.
1827 (ftree-pta): Likewise.
1829 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1831 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1832 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1833 decls to at least the GCC 4.8 used alignments.
1836 * tree-nested.c (convert_nonlocal_reference_stmt,
1837 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1838 of GIMPLE_BIND stmts, adjust associated decls.
1840 2014-01-17 Richard Biener <rguenther@suse.de>
1842 PR tree-optimization/46590
1843 * vec.h (vec<>::bseach): New member function implementing
1844 binary search according to C89 bsearch.
1845 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1846 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1847 bitmap pointer again. Make accesses_in_loop a flat array.
1848 (mem_ref_obstack): New global.
1849 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1850 (mark_ref_stored): Likewise.
1851 (ref_indep_loop_p_2): Likewise.
1852 (set_ref_stored_in_loop): New helper function.
1853 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1854 (memref_free): Adjust.
1855 (record_mem_ref_loc): Simplify.
1856 (gather_mem_refs_stmt): Adjust.
1857 (sort_locs_in_loop_postorder_cmp): New function.
1858 (analyze_memory_references): Sort accesses_in_loop after
1859 loop postorder number.
1860 (find_ref_loc_in_loop_cmp): New function.
1861 (for_all_locs_in_loop): Find relevant cluster of locs in
1862 accesses_in_loop and iterate without recursion.
1863 (execute_sm): Avoid uninit warning.
1864 (struct ref_always_accessed): Simplify.
1865 (ref_always_accessed::operator ()): Likewise.
1866 (ref_always_accessed_p): Likewise.
1867 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1868 loop postorder numbers here.
1869 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1872 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1875 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1876 on decls for which assemble_alias has been called.
1878 2014-01-17 Nick Clifton <nickc@redhat.com>
1880 * config/msp430/msp430.opt: (mcpu): New option.
1881 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1882 (msp430_option_override): Parse target_cpu. If the MCU name
1883 matches a generic string, clear target_mcu.
1884 (msp430_attr): Allow numeric interrupt values up to 63.
1885 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1886 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1888 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1890 * config/msp430/msp430.md (popm): Use %J rather than %I.
1891 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1892 (addhi_cy_i): Use immediate_operand for operand 2.
1893 * doc/invoke.texi: Document -mcpu option.
1895 2014-01-17 Richard Biener <rguenther@suse.de>
1897 PR rtl-optimization/38518
1898 * df.h (df_analyze_loop): Declare.
1899 * df-core.c: Include cfgloop.h.
1900 (df_analyze_1): Split out main part of df_analyze.
1901 (df_analyze): Adjust.
1902 (loop_inverted_post_order_compute): New function.
1903 (loop_post_order_compute): Likewise.
1904 (df_analyze_loop): New function avoiding whole-function
1906 * loop-invariant.c (find_defs): Use df_analyze_loop.
1907 (find_invariants): Adjust.
1908 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1910 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1912 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1913 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1915 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1917 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1918 traversal when removing references.
1920 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
1923 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1925 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1928 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1929 pushing a reload for an autoinc when we had previously reloaded an
1930 inner part of the address.
1932 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1934 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1936 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1937 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1938 when not giving up or versioning for alias only because of
1940 (vect_analyze_data_ref_dependences): Set to true.
1941 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1943 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1944 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1948 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1951 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1952 operand 0 predicate for gathers, use a new pseudo as subtarget.
1954 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1957 * lra-constraints.c (process_alt_operands): Add printing debug info.
1958 Check absence of input/output reloads for matched operands too.
1960 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1962 PR rtl-optimization/59835
1963 * ira.c (ira_init_register_move_cost): Increase cost for
1966 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1968 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1970 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1973 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1974 non-register objects. Use gen_(high/low)part more consistently.
1977 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1980 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1981 endian support, remove tests for WORDS_BIG_ENDIAN.
1982 (p8_mfvsrd_3_<mode>): Likewise.
1983 (reload_gpr_from_vsx<mode>): Likewise.
1984 (reload_gpr_from_vsxsf): Likewise.
1985 (p8_mfvsrd_4_disf): Likewise.
1987 2014-01-16 Richard Biener <rguenther@suse.de>
1989 PR rtl-optimization/46590
1990 * lcm.c (compute_antinout_edge): Use postorder iteration.
1991 (compute_laterin): Use inverted postorder iteration.
1993 2014-01-16 Nick Clifton <nickc@redhat.com>
1996 * varasm.c (output_constant): Return the number of bytes actually
1998 (output_constructor_array_range): Update the field size with the
1999 number of bytes emitted by output_constant.
2000 (output_constructor_regular_field): Likewise. Also do not
2001 complain if the total number of bytes emitted is now greater
2002 than the expected fieldpos.
2003 * output.h (output_constant): Update prototype and descriptive comment.
2005 2014-01-16 Marek Polacek <polacek@redhat.com>
2008 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2009 it is error_mark_node.
2011 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2013 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2014 VALID_AVX256_REG_OR_OI_MODE.
2016 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2018 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2019 current procedure should be profiled.
2021 2014-01-15 Andrew Pinski <apinski@cavium.com>
2023 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2024 of moving from/to the STACK_REG register class.
2026 2014-01-15 Richard Henderson <rth@redhat.com>
2029 * reginfo.c (global_regs_decl): Globalize.
2030 * rtl.h (global_regs_decl): Declare.
2031 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2032 reserved via global_regs.
2034 2014-01-15 Teresa Johnson <tejohnson@google.com>
2036 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2038 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2040 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2041 and vmulosh rather than call gen_vec_widen_smult_*.
2042 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2043 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2044 (vec_widen_smult_even_v16qi): Likewise.
2045 (vec_widen_umult_even_v8hi): Likewise.
2046 (vec_widen_smult_even_v8hi): Likewise.
2047 (vec_widen_umult_odd_v16qi): Likewise.
2048 (vec_widen_smult_odd_v16qi): Likewise.
2049 (vec_widen_umult_odd_v8hi): Likewise.
2050 (vec_widen_smult_odd_v8hi): Likewise.
2051 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2052 vmuloub rather than call gen_vec_widen_umult_*.
2053 (vec_widen_umult_lo_v16qi): Likewise.
2054 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2055 vmulosb rather than call gen_vec_widen_smult_*.
2056 (vec_widen_smult_lo_v16qi): Likewise.
2057 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2058 rather than call gen_vec_widen_umult_*.
2059 (vec_widen_umult_lo_v8hi): Likewise.
2060 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2061 rather than call gen_vec_widen_smult_*.
2062 (vec_widen_smult_lo_v8hi): Likewise.
2064 2014-01-15 Jeff Law <law@redhat.com>
2066 PR tree-optimization/59747
2067 * ree.c (find_and_remove_re): Properly handle case where a second
2068 eliminated extension requires widening a copy created for elimination
2069 of a prior extension.
2070 (combine_set_extension): Ensure that the number of hard regs needed
2071 for a destination register does not change when we widen it.
2073 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2075 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2076 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2077 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2078 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2079 (avr-*-rtems*): Likewise.
2080 (bfin*-rtems*): Likewise.
2081 (moxie-*-rtems*): Likewise.
2082 (h8300-*-rtems*): Likewise.
2083 (i[34567]86-*-rtems*): Likewise.
2084 (lm32-*-rtems*): Likewise.
2085 (m32r-*-rtems*): Likewise.
2086 (m68k-*-rtems*): Likewise.
2087 (microblaze*-*-rtems*): Likewise.
2088 (mips*-*-rtems*): Likewise.
2089 (powerpc-*-rtems*): Likewise.
2090 (sh-*-rtems*): Likewise.
2091 (sparc-*-rtems*): Likewise.
2092 (sparc64-*-rtems*): Likewise.
2093 (v850-*-rtems*): Likewise.
2094 (m32c-*-rtems*): Likewise.
2096 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2098 PR rtl-optimization/59511
2099 * ira.c (ira_init_register_move_cost): Use memory costs for some
2100 cases of register move cost calculations.
2101 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2102 instead of BB frequency.
2103 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2104 * lra-assigns.c (find_hard_regno_for): Ditto.
2106 2014-01-15 Richard Biener <rguenther@suse.de>
2108 PR tree-optimization/59822
2109 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2110 (vectorizable_load): Use it to hoist defs of uses of invariant
2111 loads out of the loop.
2113 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2114 Kugan Vivekanandarajah <kuganv@linaro.org>
2117 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2120 2014-01-15 Richard Biener <rguenther@suse.de>
2122 PR rtl-optimization/59802
2123 * lcm.c (compute_available): Use inverted postorder to seed
2124 the initial worklist.
2126 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2129 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2130 ADDR_REGS for invalid symrefs in non-PIC code.
2132 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2135 * builtins.c (determine_block_size): Initialize *probable_max_size
2136 even if len_rtx is CONST_INT.
2138 2014-01-14 Andrew Pinski <apinski@cavium.com>
2140 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2141 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2142 (cortexa53_tunings): Likewise.
2143 (aarch64_sched_issue_rate): New function.
2144 (TARGET_SCHED_ISSUE_RATE): Define.
2146 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2148 * ira-costs.c (find_costs_and_classes): Add missed
2149 ira_init_register_move_cost_if_necessary.
2151 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2154 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2156 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2159 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2160 to indicate if type is used for function return value. Warn ABI
2161 change if the vector mode isn't available for function return value.
2162 (ix86_function_arg_advance): Pass false to type_natural_mode.
2163 (ix86_function_arg): Likewise.
2164 (ix86_gimplify_va_arg): Likewise.
2165 (function_arg_32): Don't warn ABI change.
2166 (ix86_function_value): Pass true to type_natural_mode.
2167 (ix86_return_in_memory): Likewise.
2168 (ix86_struct_value_rtx): Removed.
2169 (TARGET_STRUCT_VALUE_RTX): Likewise.
2171 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2173 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2174 converting a conditional jump into a conditional return.
2176 2014-01-14 Richard Biener <rguenther@suse.de>
2178 PR tree-optimization/58921
2179 PR tree-optimization/59006
2180 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2181 hoisting invariant stmts.
2182 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2183 invariant loads on the preheader edge if possible.
2185 2014-01-14 Joey Ye <joey.ye@arm.com>
2187 * doc/plugin.texi (Building GCC plugins): Update to C++.
2189 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2191 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2192 (_mm_rcp28_round_ss): Ditto.
2193 (_mm_rsqrt28_round_sd): Ditto.
2194 (_mm_rsqrt28_round_ss): Ditto.
2195 (_mm_rcp28_sd): Ditto.
2196 (_mm_rcp28_ss): Ditto.
2197 (_mm_rsqrt28_sd): Ditto.
2198 (_mm_rsqrt28_ss): Ditto.
2199 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2200 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2201 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2202 (IX86_BUILTIN_RCP28SD): Ditto.
2203 (IX86_BUILTIN_RCP28SS): Ditto.
2204 (IX86_BUILTIN_RSQRT28SD): Ditto.
2205 (IX86_BUILTIN_RSQRT28SS): Ditto.
2206 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2207 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2208 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2209 (ix86_expand_special_args_builtin): Expand new FTYPE.
2210 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2211 (srcp14<mode>): Make insn unary.
2212 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2213 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2214 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2215 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2216 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2217 Fix rounding: make it SAE only.
2218 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2220 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2222 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2223 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2224 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2225 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2226 (round_saeonly_mask_scalar_operand4): Ditto.
2227 (round_saeonly_mask_scalar_op3): Ditto.
2228 (round_saeonly_mask_scalar_op4): Ditto.
2230 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2232 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2233 Implement -maltivec=be for vec_insert and vec_extract.
2235 2014-01-10 DJ Delorie <dj@redhat.com>
2237 * config/msp430/msp430.md (call_internal): Don't allow memory
2238 references with SP as the base register.
2239 (call_value_internal): Likewise.
2240 * config/msp430/constraints.md (Yc): New. For memory references
2241 that don't use SP as a base register.
2243 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2244 "an integer without a # prefix"
2245 * config/msp430/msp430.md (epilogue_helper): Use it.
2247 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2250 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2251 AVX512F gather builtins.
2252 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2253 on gather decls with INTEGER_TYPE masktype.
2254 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2255 directly into the builtin rather than hoisting it before loop.
2257 PR tree-optimization/59387
2258 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2259 (scev_const_prop): If folded_casts and type has undefined overflow,
2260 use force_gimple_operand instead of force_gimple_operand_gsi and
2261 for each added stmt if it is assign with
2262 arith_code_with_undefined_signed_overflow, call
2263 rewrite_to_defined_overflow.
2264 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2265 gimple-fold.h instead.
2266 (arith_code_with_undefined_signed_overflow,
2267 rewrite_to_defined_overflow): Moved to ...
2268 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2269 rewrite_to_defined_overflow): ... here. No longer static.
2270 Include gimplify-me.h.
2271 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2272 rewrite_to_defined_overflow): New prototypes.
2274 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2276 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2278 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2280 * builtins.c (get_object_alignment_2): Minor tweak.
2281 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2283 2014-01-13 Christian Bruel <christian.bruel@st.com>
2285 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2286 optimized non constant lengths.
2288 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2291 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2292 load as __atomic_load_N if possible.
2294 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2296 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2298 (rs6000_expand_builtin): Adjust call.
2300 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2303 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2304 * config/rs6000/rs6000.c: Include target-globals.h.
2305 (rs6000_set_current_function): Instead of doing target_reinit
2306 unconditionally, use save_target_globals_default_opts and
2307 restore_target_globals.
2309 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2311 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2312 (rs6000_expand_builtin): Handle mffs and mtfsf.
2313 (rs6000_init_builtins): Define mffs and mtfsf.
2314 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2315 (rs6000_mffs): New pattern.
2316 (rs6000_mtfsf): New pattern.
2318 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2320 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2321 Start narrowing with START. Apply candidate-use pair
2322 and check overall cost in narrowing.
2323 (iv_ca_prune): Pass new argument.
2325 2014-01-10 Jeff Law <law@redhat.com>
2328 * ree.c (combine_reaching_defs): Ensure the defining statement
2329 occurs before the extension when optimizing extensions with
2330 different source and destination hard registers.
2332 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2335 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2336 vtables into the type inheritance graph.
2338 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2340 PR rtl-optimization/59754
2341 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2342 modes in the REGNO != REGNO case.
2344 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2346 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2348 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2350 PR tree-optimization/59745
2351 * tree-predcom.c (tree_predictive_commoning_loop): Call
2352 free_affine_expand_cache if giving up because components is NULL.
2354 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2355 GC in payload of target_globals struct instead of allocating them on
2356 the heap and the larger structs separately using GC.
2357 * target-globals.h (struct target_globals): Make regs, hard_regs,
2358 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2359 of GTY((skip)) and change type to void *.
2360 (reset_target_globals): Cast loads from those fields to corresponding
2363 2014-01-10 Steve Ellcey <sellcey@mips.com>
2366 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2367 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2368 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2370 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2373 * aarch64-modes.def (CC_Zmode): New flags mode.
2374 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2375 represents an equality.
2376 (aarch64_get_condition_code): Handle CC_Zmode.
2377 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2379 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2381 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2382 extraction in good case.
2384 2014-01-10 Richard Biener <rguenther@suse.de>
2386 PR tree-optimization/59374
2387 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2388 checking after SLP discovery. Mark stmts not participating
2389 in any SLP instance properly.
2391 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2393 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2394 when handling a SET rtx.
2396 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2398 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2399 (cortex-a57): Likewise.
2400 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2402 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2404 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2405 non-iwmmxt builtins.
2407 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2411 * ipa-devirt.c record_target_from_binfo): Take as argument
2412 stack of binfos and lookup matching one for virtual inheritance.
2413 (possible_polymorphic_call_targets_1): Update.
2415 2014-01-10 Huacai Chen <chenhc@lemote.com>
2417 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2418 kernel strings for Loongson-2E/2F/3A.
2420 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2423 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2424 is_gimple_call before calling gimple_call_internal_p.
2426 2014-01-09 Steve Ellcey <sellcey@mips.com>
2428 * Makefile.in (TREE_FLOW_H): Remove.
2429 (TREE_SSA_H): Add file names from tree-flow.h.
2430 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2431 * tree.h: Remove tree-flow.h reference.
2432 * hash-table.h: Remove tree-flow.h reference.
2433 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2434 reference with tree-ssa-loop.h.
2436 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2438 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2439 default element-order behavior for -maltivec.
2440 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2441 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2442 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2443 when targeting big endian, at least for now.
2444 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2446 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2449 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2450 var satisfies use_register_for_decl, just take into account type
2451 alignment, rather than decl alignment.
2453 PR tree-optimization/59622
2454 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2455 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2456 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2457 Don't devirtualize for inplace at all. For targets.length () == 1,
2458 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2460 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2462 * config/i386/i386.md (cpu): Remove the unused btver1.
2464 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2466 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2468 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2471 * tree-core.h (struct target_globals): New forward declaration.
2472 (struct tree_target_option): Add globals field.
2473 * tree.h (TREE_TARGET_GLOBALS): Define.
2474 (prepare_target_option_nodes_for_pch): New prototype.
2475 * target-globals.h (struct target_globals): Define even if
2477 * tree.c (prepare_target_option_node_for_pch,
2478 prepare_target_option_nodes_for_pch): New functions.
2479 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2480 * config/i386/i386.c: Include target-globals.h.
2481 (ix86_set_current_function): Instead of doing target_reinit
2482 unconditionally, use save_target_globals_default_opts and
2483 restore_target_globals.
2485 2014-01-09 Richard Biener <rguenther@suse.de>
2487 PR tree-optimization/59715
2488 * tree-cfg.h (split_critical_edges): Declare.
2489 * tree-cfg.c (split_critical_edges): Export.
2490 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2492 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2494 * cfgexpand.c (expand_stack_vars): Optionally disable
2495 asan stack protection.
2496 (expand_used_vars): Likewise.
2497 (partition_stack_vars): Likewise.
2498 * asan.c (asan_emit_stack_protection): Optionally disable
2499 after return stack usage.
2500 (instrument_derefs): Optionally disable memory access instrumentation.
2501 (instrument_builtin_call): Likewise.
2502 (instrument_strlen_call): Likewise.
2503 (asan_protect_global): Optionally disable global variables protection.
2504 * doc/invoke.texi: Added doc for new options.
2505 * params.def: Added new options.
2506 * params.h: Likewise.
2508 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2510 PR rtl-optimization/59724
2511 * ifcvt.c (cond_exec_process_if_block): Don't call
2512 flow_find_head_matching_sequence with 0 longest_match.
2513 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2514 non-active insns if !stop_after.
2515 (try_head_merge_bb): Revert 2014-01-07 changes.
2517 2014-01-08 Jeff Law <law@redhat.com>
2519 * ree.c (get_sub_rtx): New function, extracted from...
2520 (merge_def_and_ext): Here.
2521 (combine_reaching_defs): Use get_sub_rtx.
2523 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2525 * cgraph.h (varpool_variable_node): Do not choke on null node.
2527 2014-01-08 Catherine Moore <clm@codesourcery.com>
2529 * config/mips/mips.md (simple_return): Attempt to use JRC
2531 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2533 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2535 PR rtl-optimization/59137
2536 * reorg.c (steal_delay_list_from_target): Call update_block for
2538 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2540 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2542 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2543 two duplicate entries.
2545 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2548 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2550 * config/mips/mips.c (mips_truncated_op_cost): New function.
2551 (mips_rtx_costs): Adjust test for BADDU.
2552 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2554 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2556 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2557 (*baddu_si): ...this new pattern.
2559 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2562 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2564 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2567 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2569 (expand_expr, expand_normal): Adjust.
2570 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2571 inner_reference_p. Use inner_reference_p to expand inner references.
2572 (store_expr): Adjust.
2573 * cfgexpand.c (expand_call_stmt): Adjust.
2575 2014-01-08 Rong Xu <xur@google.com>
2577 * gcov-io.c (gcov_var): Move from gcov-io.h.
2578 (gcov_position): Ditto.
2579 (gcov_is_error): Ditto.
2580 (gcov_rewrite): Ditto.
2581 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2582 only part to libgcc/libgcov.h.
2584 2014-01-08 Marek Polacek <polacek@redhat.com>
2587 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2589 2014-01-08 Marek Polacek <polacek@redhat.com>
2592 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2594 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2596 PR rtl-optimization/59649
2597 * stor-layout.c (get_mode_bounds): For BImode return
2598 0 and STORE_FLAG_VALUE.
2600 2014-01-08 Richard Biener <rguenther@suse.de>
2603 * gimple.h (is_gimple_builtin_call): Remove.
2604 (gimple_builtin_call_types_compatible_p): New.
2605 (gimple_call_builtin_p): New overload.
2606 * gimple.c (is_gimple_builtin_call): Remove.
2607 (validate_call): Rename to ...
2608 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2610 (validate_type): New static function.
2611 (gimple_call_builtin_p): New overload and adjust.
2612 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2613 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2614 (gimple_fold_stmt_to_constant_1): Likewise.
2615 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2617 2014-01-08 Richard Biener <rguenther@suse.de>
2620 * gimplify.c (gimplify_expr): Gimplify register-register type
2621 VIEW_CONVERT_EXPRs to separate stmts.
2623 2014-01-07 Jeff Law <law@redhat.com>
2626 * ree.c (combine_set_extension): Handle case where source
2627 and destination registers in an extension insn are different.
2628 (combine_reaching_defs): Allow source and destination registers
2629 in extension to be different under limited circumstances.
2630 (add_removable_extension): Remove restriction that the
2631 source and destination registers in the extension are the same.
2632 (find_and_remove_re): Emit a copy from the extension's
2633 destination to its source after the defining insn if
2634 the source and destination registers are different.
2637 * ifcvt.c (merge_if_block): If we are merging a block with more than
2638 one successor with a block with no successors, remove any BARRIER
2639 after the second block.
2641 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2643 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2645 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2648 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2649 for 14-bit register offsets when INT14_OK_STRICT is false.
2651 2014-01-07 Roland Stigge <stigge@antcom.de>
2652 Michael Meissner <meissner@linux.vnet.ibm.com>
2655 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2656 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2658 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2660 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2663 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2665 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2666 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2667 rtx is const0_rtx or not.
2669 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2672 * target-globals.c (save_target_globals): Remove this_fn_optab
2674 * toplev.c: Include optabs.h.
2675 (target_reinit): Temporarily restore the global options if another
2676 set of options are in force.
2678 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2680 PR rtl-optimization/58668
2681 * cfgcleanup.c (flow_find_cross_jump): Don't count
2682 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2683 to determine what is counted.
2684 (flow_find_head_matching_sequence): Use active_insn_p to determine
2686 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2688 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2689 determine what is counted.
2691 PR tree-optimization/59643
2692 * tree-predcom.c (split_data_refs_to_components): If one dr is
2693 read and one write, determine_offset fails and the write isn't
2694 in the bad component, just put the read into the bad component.
2696 2014-01-07 Mike Stump <mikestump@comcast.net>
2697 Jakub Jelinek <jakub@redhat.com>
2700 * tree-core.h (struct tree_optimization_option): Change optabs
2701 type from unsigned char * to void *.
2702 * optabs.c (init_tree_optimization_optabs): Adjust
2703 TREE_OPTIMIZATION_OPTABS initialization.
2705 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2708 * config/i386/i386.h (struct machine_function): Add
2709 no_drap_save_restore field.
2710 * config/i386/i386.c (ix86_save_reg): Use
2711 !cfun->machine->no_drap_save_restore instead of
2712 crtl->stack_realign_needed.
2713 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2714 this function clears frame_pointer_needed. Set
2715 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2716 and DRAP reg is needed.
2718 2014-01-06 Marek Polacek <polacek@redhat.com>
2721 * doc/implement-c.texi: Mention that other integer types are
2722 permitted as bit-field types in strictly conforming mode.
2724 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2726 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2729 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2731 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2733 2014-01-06 Martin Jambor <mjambor@suse.cz>
2736 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2738 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2740 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2744 * var-tracking.c (add_stores): Preserve the value of the source even if
2745 we don't record the store.
2747 2014-01-06 Terry Guo <terry.guo@arm.com>
2749 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2751 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2754 * config/darwin.c (darwin_function_section): Adjust return values to
2755 correspond to optimisation changes made in r206070.
2757 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2759 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2760 from prefetch_block tune setting.
2761 (nocona_cost): Correct size of prefetch block to 64.
2763 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2765 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2766 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2767 used to save the static chain register in the computation of the offset
2768 from which the FP registers need to be restored.
2770 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2772 PR tree-optimization/59519
2773 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2774 ICE if get_current_def (current_new_name) is already non-NULL, as long
2775 as it is a phi result of some other phi in *new_exit_bb that has
2778 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2779 or vmovdqu* for misaligned_operand.
2780 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2781 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2782 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2783 aligned_mem for AVX512F masked aligned load and store builtins and for
2786 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2788 PR tree-optimization/59651
2789 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2790 Address range for negative step should be added by TYPE_SIZE_UNIT.
2792 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2794 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2795 overlapping registers also for registers other than the stack pointer.
2797 2014-01-03 Marek Polacek <polacek@redhat.com>
2800 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2803 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2806 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2809 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2810 (push splitter): Use <P:MODE_SIZE> instead of
2811 GET_MODE_SIZE (<P:MODE>mode).
2812 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2813 (mov -1, reg peephole2): Likewise.
2814 * config/i386/sse.md (*mov<mode>_internal,
2815 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2816 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2817 *<code><mode>3, *andnot<mode>3<mask_name>,
2818 <mask_codefor><code><mode>3<mask_name>): Likewise.
2819 * config/i386/subst.md (mask_mode512bit_condition,
2820 sd_mask_mode512bit_condition): Likewise.
2822 2014-01-02 Xinliang David Li <davidxl@google.com>
2824 PR tree-optimization/59303
2825 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2826 (dump_predicates): Better output format.
2827 (pred_equal_p): New function.
2828 (is_neq_relop_p): Ditto.
2829 (is_neq_zero_form_p): Ditto.
2830 (pred_expr_equal_p): Ditto.
2831 (pred_neg_p): Ditto.
2832 (simplify_pred): Ditto.
2833 (simplify_preds_2): Ditto.
2834 (simplify_preds_3): Ditto.
2835 (simplify_preds_4): Ditto.
2836 (simplify_preds): Ditto.
2838 (push_to_worklist): Ditto.
2839 (get_pred_info_from_cmp): Ditto.
2840 (is_degenerated_phi): Ditto.
2841 (normalize_one_pred_1): Ditto.
2842 (normalize_one_pred): Ditto.
2843 (normalize_one_pred_chain): Ditto.
2844 (normalize_preds): Ditto.
2845 (normalize_cond_1): Remove function.
2846 (normalize_cond): Ditto.
2847 (is_gcond_subset_of): Ditto.
2848 (is_subset_of_any): Ditto.
2849 (is_or_set_subset_of): Ditto.
2850 (is_and_set_subset_of): Ditto.
2851 (is_norm_cond_subset_of): Ditto.
2852 (pred_chain_length_cmp): Ditto.
2853 (convert_control_dep_chain_into_preds): Type change.
2854 (find_predicates): Ditto.
2855 (find_def_preds): Ditto.
2856 (destroy_predicates_vecs): Ditto.
2857 (find_matching_predicates_in_rest_chains): Ditto.
2858 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2859 (is_pred_expr_subset): Ditto.
2860 (is_pred_chain_subset_of): Ditto.
2861 (is_included_in): Ditto.
2862 (is_superset_of): Ditto.
2864 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2866 Update copyright years.
2868 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2870 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2871 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2872 config/arc/arc.md, config/arc/arc.opt,
2873 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2874 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2875 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2876 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2877 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2878 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2879 vtable-verify.c, vtable-verify.h: Use the standard form for the
2882 2014-01-02 Tobias Burnus <burnus@net-b.de>
2884 * gcc.c (process_command): Update copyright notice dates.
2885 * gcov-dump.c: Ditto.
2887 * doc/cpp.texi: Bump @copying's copyright year.
2888 * doc/cppinternals.texi: Ditto.
2889 * doc/gcc.texi: Ditto.
2890 * doc/gccint.texi: Ditto.
2891 * doc/gcov.texi: Ditto.
2892 * doc/install.texi: Ditto.
2893 * doc/invoke.texi: Ditto.
2895 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2897 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2899 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2901 * config/i386/sse.md (*mov<mode>_internal): Guard
2902 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2904 PR rtl-optimization/59647
2905 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2906 new_rtx into UNSIGNED_FLOAT rtxes.
2908 Copyright (C) 2014 Free Software Foundation, Inc.
2910 Copying and distribution of this file, with or without modification,
2911 are permitted in any medium without royalty provided the copyright
2912 notice and this notice are preserved.