| blob | 83d021a4748374bfb1047b9a3c4698c573bea5f1 |
1 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2 Jakub Jelinek <jakub@redhat.com>
4 PR middle-end/60013
5 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
6 of the dataflow.
8 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
11 CODE_FOR_altivec_vpku[hw]um to
12 CODE_FOR_altivec_vpku[hw]um_direct.
13 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
14 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
15 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
16 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
18 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
21 generation for -maltivec=be.
22 (altivec_vsumsws): Simplify redundant test.
24 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
27 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
28 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
29 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
30 gen_altivec_vpkuwum.
31 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
32 BYTES_BIG_ENDIAN.
33 (altivec_vpks<VI_char>ss): Likewise.
34 (altivec_vpks<VI_char>us): Likewise.
35 (altivec_vpku<VI_char>us): Likewise.
36 (altivec_vpku<VI_char>um): Likewise.
37 (altivec_vpku<VI_char>um_direct): New (copy of
38 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
39 internal use).
40 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
41 target is little endian and -maltivec=be is not specified.
42 (*altivec_vupkhs<VU_char>_direct): New (copy of
43 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
44 use).
45 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
46 target is little endian and -maltivec=be is not specified.
47 (*altivec_vupkls<VU_char>_direct): New (copy of
48 altivec_vupkls<VU_char> that always emits vupkls*, for internal
49 use).
50 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
51 little endian and -maltivec=be is not specified.
52 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
53 little endian and -maltivec=be is not specified.
55 2014-02-05 Richard Henderson <rth@redhat.com>
57 PR debug/52727
58 * combine-stack-adj.c: Revert r206943.
59 * sched-int.h (struct deps_desc): Add last_args_size.
60 * sched-deps.c (init_deps): Initialize it.
61 (sched_analyze_insn): Add OUTPUT dependencies between insns that
62 contain REG_ARGS_SIZE notes.
64 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
66 * lto-cgraph.c (asm_nodes_output): Make global.
67 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
68 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
69 parameter
70 (driver_handle_option): Handle OPT_fwpa.
72 2014-02-05 Jakub Jelinek <jakub@redhat.com>
74 PR ipa/59947
75 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
76 a comment typo and formatting issue. If odr_hash hasn't been
77 created, return vNULL and set *completep to false.
79 PR middle-end/57499
80 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
81 bb with no successors.
83 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
85 PR target/59718
86 * doc/invoke.texi (-march): Clarify documentation for ARM.
87 (-mtune): Likewise.
88 (-mcpu): Likewise.
90 2014-02-05 Richard Biener <rguenther@suse.de>
92 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
93 when not vectorizing because of too many alias checks.
94 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
95 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
97 2014-02-05 Nick Clifton <nickc@redhat.com>
99 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
100 accept extended registers in any mode when compiling for the
101 MN10300.
103 2014-02-05 Yury Gribov <y.gribov@samsung.com>
105 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
106 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
107 sanitization attributes.
108 (can_inline_edge_p): Likewise.
109 (sanitize_attrs_match_for_inline_p): New function.
111 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
113 * ipa-prop.c (detect_type_change): Shor circuit testing of
114 type changes on THIS pointer.
116 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
118 PR target/59777
119 * config/pa/pa.c (legitimize_tls_address): Return original address
120 if not passed a SYMBOL_REF rtx.
121 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
122 addresses.
123 (pa_emit_move_sequence): Simplify TLS source operands.
124 (pa_legitimate_constant_p): Reject all TLS constants.
125 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
126 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
128 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
130 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
131 groups when we know they are controlled by LTO.
132 * varasm.c (default_binds_local_p_1): If object is in other partition,
133 it will be resolved locally.
135 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
137 * gcc/config/host-linux.c (linux_gt_pch_use_address): Don't
138 use SSIZE_MAX because it is not always defined.
140 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
142 PR bootstrap/59913
143 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
144 threshold for pseudo splitting.
145 (update_ebb_live_info): Process call argument hard registers and
146 hard registers from insn definition too.
147 (max_small_class_regs_num): New constant.
148 (inherit_in_ebb): Update live hard regs through EBBs. Update
149 reloads_num only for small register classes. Don't split for
150 outputs of jumps.
152 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
154 PR ipa/60058
155 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
156 is non-null.
158 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
160 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
162 2014-02-04 Marek Polacek <polacek@redhat.com>
164 * gdbinit.in (pel): Define.
166 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
168 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
169 behavior.
171 2014-02-04 Richard Biener <rguenther@suse.de>
173 PR lto/59723
174 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
175 in function context local.
176 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
177 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
178 similar to LTO_imported_decl_ref.
180 2014-02-04 Jakub Jelinek <jakub@redhat.com>
182 PR tree-optimization/60002
183 * cgraphclones.c (build_function_decl_skip_args): Clear
184 DECL_LANG_SPECIFIC.
186 PR tree-optimization/60023
187 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
188 false to gsi_replace.
189 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
190 has been in some EH region and vec_stmt could throw, add
191 vec_stmt into the same EH region.
192 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
193 has no lhs, ignore it.
194 * internal-fn.c (expand_MASK_LOAD): Likewise.
196 PR ipa/60026
197 * tree-inline.c (copy_forbidden): Fail for
198 __attribute__((optimize (0))) functions.
200 PR other/58712
201 * omp-low.c (simd_clone_struct_copy): If from->inbranch
202 is set, copy one less argument.
203 (expand_simd_clones): Don't subtract clone_info->inbranch
204 from simd_clone_struct_alloc argument.
206 PR rtl-optimization/57915
207 * recog.c (simplify_while_replacing): If all unary/binary/relational
208 operation arguments are constant, attempt to simplify those.
210 PR middle-end/59261
211 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
212 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
214 2014-02-04 Richard Biener <rguenther@suse.de>
216 PR tree-optimization/60012
217 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
218 TBAA disambiguation to all DDRs.
220 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
222 PR target/59788
223 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
224 (LINK_SPEC): Use it for -shared, -shared-libgcc.
226 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
228 PR ipa/59882
229 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
231 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
233 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
234 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
236 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
238 PR ipa/59831
239 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
240 to figure out targets of polymorphic calls with known decl.
241 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
242 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
243 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
244 (get_polymorphic_call_info): ... here.
245 (get_polymorphic_call_info_from_invariant): New function.
247 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
249 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
250 lookup via vtable pointer; check for type consistency
251 and turn inconsitent facts into UNREACHABLE.
252 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
253 * gimple-fold.c (gimple_get_virt_method_for_vtable):
254 Do not ICE on type inconsistent querries; return UNREACHABLE
255 instead.
257 2014-02-03 Richard Henderson <rth@twiddle.net>
259 PR tree-opt/59924
260 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
261 already processed this node.
262 (normalize_one_pred_1): Pass along mark_set.
263 (normalize_one_pred): Create and destroy a pointer_set_t.
264 (normalize_one_pred_chain): Likewise.
266 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
268 PR gcov-profile/58602
269 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
271 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
273 PR ipa/59831
274 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
275 Try to devirtualize by the knowledge of virtual table pointer given by
276 aggregate propagation.
277 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
278 ipa_print_node_jump_functions): Dump also offset that
279 is relevant for polymorphic calls.
280 (determine_known_aggregate_parts): Add arg_type parameter; use it
281 instead of determining the type from pointer type.
282 (ipa_compute_jump_functions_for_edge): Update call of
283 determine_known_aggregate_parts.
284 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
285 (gimple_get_virt_method_for_binfo): ... here; simplify using
286 vtable_pointer_value_to_vtable.
287 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
288 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
289 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
290 (vtable_pointer_value_to_vtable): Break out from ...; handle also
291 POINTER_PLUS_EXPR.
292 (vtable_pointer_value_to_binfo): ... here.
293 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
295 2014-02-03 Teresa Johnson <tejohnson@google.com>
297 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
298 redef of outer loop index variable.
300 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
302 PR c++/53017
303 PR c++/59211
304 * doc/extend.texi (Function Attributes): Typo.
306 2014-02-03 Cong Hou <congh@google.com>
308 PR tree-optimization/60000
309 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
310 if the vectorized statement is a store. A store statement can only
311 appear at the end of pattern statements.
313 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
315 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
316 (ix86_option_override_internal): Default long double to 64-bit for
317 32-bit Bionic and to 128-bit for 64-bit Bionic.
319 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
320 TARGET_LONG_DOUBLE_128 is true.
321 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
323 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
324 (mlong-double-64): Negate -mlong-double-128.
325 (mlong-double-128): New option.
327 * config/i386/i386-c.c (ix86_target_macros): Define
328 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
330 * doc/invoke.texi: Document -mlong-double-128.
332 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
334 PR rtl-optimization/60024
335 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
337 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
339 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
341 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
343 PR rtl-optimization/57662
344 * sel-sched.c (code_motion_path_driver): Do not mark already not
345 existing blocks in the visiting bitmap.
347 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
349 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
350 on the insn being emitted.
352 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
353 Will Deacon <will.deacon@arm.com>
355 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
357 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
359 * config/arm/arm-tables.opt: Regenerate.
361 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
363 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
364 for vector types other than V16QImode.
365 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
366 define_expand, and call altivec_expand_vec_perm_le when producing
367 code with little endian element order.
368 (*altivec_vperm_<mode>_internal): New insn having previous
369 behavior of altivec_vperm_<mode>.
370 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
371 altivec_expand_vec_perm_le when producing code with little endian
372 element order.
373 (*altivec_vperm_<mode>_uns_internal): New insn having previous
374 behavior of altivec_vperm_<mode>_uns.
376 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
378 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
379 (altivec_vsumsws): Add handling for -maltivec=be with a little
380 endian target.
381 (altivec_vsumsws_direct): New.
382 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
383 gen_altivec_vsumsws.
385 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
387 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
388 vtable_pointer_value_to_binfo): New functions.
389 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
390 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
392 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
394 * config/nios2/nios2.md (load_got_register): Initialize GOT
395 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
396 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
398 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
400 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
401 preserverd by passthrough, do not propagate the type.
403 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
405 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
406 (mips_atomic_assign_expand_fenv): New function.
407 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
409 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
411 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
412 (__builtin_mips_set_fcsr): Likewise.
413 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
414 MIPS_USI_FTYPE_VOID.
415 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
416 (mips16_expand_set_fcsr): Likewise.
417 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
418 (mips16_set_fcsr_stub): Likewise.
419 (mips16_get_fcsr_one_only_stub): New class.
420 (mips16_set_fcsr_one_only_stub): Likewise.
421 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
422 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
423 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
424 (hard_float): New availability predicate.
425 (mips_builtins): Add get_fcsr and set_fcsr.
426 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
427 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
428 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
429 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
430 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
431 patterns.
433 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
435 * config/mips/mips.c (mips_one_only_stub): New class.
436 (mips_need_mips16_rdhwr_p): Replace with...
437 (mips16_rdhwr_stub): ...this new variable.
438 (mips16_stub_call_address): New function.
439 (mips16_rdhwr_one_only_stub): New class.
440 (mips_expand_thread_pointer): Use mips16_stub_call_address.
441 (mips_output_mips16_rdhwr): Delete.
442 (mips_finish_stub): New function.
443 (mips_code_end): Use it to handle rdhwr stubs.
445 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
447 PR target/60017
448 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
449 when calculating size of integer atomic types.
451 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
453 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
455 2014-02-01 Jakub Jelinek <jakub@redhat.com>
457 PR tree-optimization/60003
458 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
459 * profile.c (branch_prob): Use gimple_call_builtin_p
460 to check for BUILT_IN_SETJMP_RECEIVER.
461 * tree-inline.c (copy_bb): Call notice_special_calls.
463 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
465 PR bootstrap/59985
466 * lra-constraints.c (process_alt_operands): Update reload_sum only
467 on the first pass.
469 2014-01-31 Richard Henderson <rth@redhat.com>
471 PR middle-end/60004
472 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
473 until after else_eh is processed.
475 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
477 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
478 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
479 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
480 in smmintrin.h, remove them.
481 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
482 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
483 * config/i386/i386.md (ROUND_SAE): Fix value.
484 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
485 (const48_operand): New.
486 * config/i386/subst.md (round), (round_expand): Use
487 const_4_or_8_to_11_operand.
488 (round_saeonly), (round_saeonly_expand): Use const48_operand.
490 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
492 * config/i386/constraints.md (Yk): Swap meaning with k.
493 * config/i386/i386.md (movhi_internal): Change Yk to k.
494 (movqi_internal): Ditto.
495 (*k<logic><mode>): Ditto.
496 (*andhi_1): Ditto.
497 (*andqi_1): Ditto.
498 (kandn<mode>): Ditto.
499 (*<code>hi_1): Ditto.
500 (*<code>qi_1): Ditto.
501 (kxnor<mode>): Ditto.
502 (kortestzhi): Ditto.
503 (kortestchi): Ditto.
504 (kunpckhi): Ditto.
505 (*one_cmplhi2_1): Ditto.
506 (*one_cmplqi2_1): Ditto.
507 * config/i386/sse.md (): Change k to Yk.
508 (avx512f_load<mode>_mask): Ditto.
509 (avx512f_blendm<mode>): Ditto.
510 (avx512f_store<mode>_mask): Ditto.
511 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
512 (avx512f_storedqu<mode>_mask): Ditto.
513 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
514 Ditto.
515 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
516 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
517 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
518 (avx512f_maskcmp<mode>3): Ditto.
519 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
520 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
521 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
522 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
523 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
524 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
525 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
526 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
527 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
528 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
529 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
530 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
531 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
532 (vec_extract_lo_<mode>_maskm): Ditto.
533 (vec_extract_hi_<mode>_maskm): Ditto.
534 (avx512f_vternlog<mode>_mask): Ditto.
535 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
536 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
537 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
538 (avx512f_<code>v8div16qi2_mask): Ditto.
539 (avx512f_<code>v8div16qi2_mask_store): Ditto.
540 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
541 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
542 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
543 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
544 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
545 (*avx512pf_gatherpf<mode>df_mask): Ditto.
546 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
547 (*avx512pf_scatterpf<mode>df_mask): Ditto.
548 (avx512cd_maskb_vec_dupv8di): Ditto.
549 (avx512cd_maskw_vec_dupv16si): Ditto.
550 (avx512f_vpermi2var<mode>3_maskz): Ditto.
551 (avx512f_vpermi2var<mode>3_mask): Ditto.
552 (avx512f_vpermi2var<mode>3_mask): Ditto.
553 (avx512f_vpermt2var<mode>3_maskz): Ditto.
554 (*avx512f_gathersi<mode>): Ditto.
555 (*avx512f_gathersi<mode>_2): Ditto.
556 (*avx512f_gatherdi<mode>): Ditto.
557 (*avx512f_gatherdi<mode>_2): Ditto.
558 (*avx512f_scattersi<mode>): Ditto.
559 (*avx512f_scatterdi<mode>): Ditto.
560 (avx512f_compress<mode>_mask): Ditto.
561 (avx512f_compressstore<mode>_mask): Ditto.
562 (avx512f_expand<mode>_mask): Ditto.
563 * config/i386/subst.md (mask): Change k to Yk.
564 (mask_scalar_merge): Ditto.
565 (sd): Ditto.
567 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
569 * doc/extend.texi (Vector Extensions): Document ?: in C++.
571 2014-01-31 Richard Biener <rguenther@suse.de>
573 PR middle-end/59990
574 * builtins.c (fold_builtin_memory_op): Make sure to not
575 use a floating-point mode or a boolean or enumeral type for
576 the copy operation.
578 2014-01-30 DJ Delorie <dj@redhat.com>
580 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
581 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
582 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
583 whenever main() has an epilogue.
585 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
587 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
588 unused variable "field".
589 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
590 (vsx_mergeh_<mode>): Likewise.
591 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
592 (altivec_vmrghh): Likewise.
593 (altivec_vmrghw): Likewise.
594 (altivec_vmrglb): Likewise.
595 (altivec_vmrglh): Likewise.
596 (altivec_vmrglw): Likewise.
597 (altivec_vspltb): Add missing uses.
598 (altivec_vsplth): Likewise.
599 (altivec_vspltw): Likewise.
600 (altivec_vspltsf): Likewise.
602 2014-01-30 Jakub Jelinek <jakub@redhat.com>
604 PR target/59923
605 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
606 frame related instructions.
608 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
610 PR rtl-optimization/59959
611 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
612 any reload of register whose subreg is invalid.
614 2014-01-30 Jakub Jelinek <jakub@redhat.com>
616 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
617 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
618 Add missing return type - void.
620 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
622 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
623 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
624 remove element index adjustment for endian (now handled in vsx.md
625 and altivec.md).
626 (altivec_expand_vec_perm_const): Use
627 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
628 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
629 (vsx_xxspltw_<mode>): Adjust element index for little endian.
630 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
631 define_expand and a new define_insn *altivec_vspltb_internal;
632 adjust for -maltivec=be on a little endian target.
633 (altivec_vspltb_direct): New.
634 (altivec_vsplth): Divide into a define_expand and a new
635 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
636 little endian target.
637 (altivec_vsplth_direct): New.
638 (altivec_vspltw): Divide into a define_expand and a new
639 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
640 little endian target.
641 (altivec_vspltw_direct): New.
642 (altivec_vspltsf): Divide into a define_expand and a new
643 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
644 a little endian target.
646 2014-01-30 Richard Biener <rguenther@suse.de>
648 PR tree-optimization/59993
649 * tree-ssa-forwprop.c (associate_pointerplus): Check we
650 can propagate form the earlier stmt and avoid the transform
651 when the intermediate result is needed.
653 2014-01-30 Alangi Derick <alangiderick@gmail.com>
655 * README.Portability: Fix typo.
657 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
659 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
660 comparison_operator with ordered_comparison_operator.
662 2014-01-30 Nick Clifton <nickc@redhat.com>
664 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
665 Rename to mn10300_store_multiple_regs.
666 * config/mn10300/mn10300.c: Likewise.
667 * config/mn10300/mn10300.md (store_movm): Fix typo: call
668 store_multiple_regs.
669 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
670 Call mn10300_store_multiple_regs.
672 2014-01-30 Nick Clifton <nickc@redhat.com>
673 DJ Delorie <dj@redhat.com>
675 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
676 %fp 2 to keep registers after it properly word-aligned.
677 (rl78_alloc_physical_registers_umul): Handle the case where both
678 input operands are the same.
680 2014-01-30 Richard Biener <rguenther@suse.de>
682 PR tree-optimization/59903
683 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
684 check properly.
686 2014-01-30 Jason Merrill <jason@redhat.com>
688 PR c++/59633
689 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
691 PR c++/59645
692 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
694 2014-01-30 Richard Biener <rguenther@suse.de>
696 PR tree-optimization/59951
697 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
699 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
701 PR target/59784
702 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
703 SFmode to DFmode case.
705 2014-01-29 DJ Delorie <dj@redhat.com>
707 * config/msp430/msp430.opt (-minrt): New.
708 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
709 if -minrt given.
710 (ENDFILE_SPEC): Likewise.
712 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
714 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
715 (estimate_function_body_sizes): Use it.
717 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
719 PR c++/58561
720 * dwarf2out.c (is_cxx_auto): New.
721 (is_base_type): Use it.
722 (gen_type_die_with_usage): Likewise.
724 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
726 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
727 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
728 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
729 -maltivec=be with LE targets.
730 (vsx_mergeh_<mode>): Likewise.
731 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
732 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
733 (altivec_vmrghb): Replace with define_expand and new
734 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
735 (altivec_vmrghb_direct): New define_insn.
736 (altivec_vmrghh): Replace with define_expand and new
737 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
738 (altivec_vmrghh_direct): New define_insn.
739 (altivec_vmrghw): Replace with define_expand and new
740 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
741 (altivec_vmrghw_direct): New define_insn.
742 (*altivec_vmrghsf): Adjust for endianness.
743 (altivec_vmrglb): Replace with define_expand and new
744 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
745 (altivec_vmrglb_direct): New define_insn.
746 (altivec_vmrglh): Replace with define_expand and new
747 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
748 (altivec_vmrglh_direct): New define_insn.
749 (altivec_vmrglw): Replace with define_expand and new
750 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
751 (altivec_vmrglw_direct): New define_insn.
752 (*altivec_vmrglsf): Adjust for endianness.
753 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
754 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
755 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
756 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
757 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
758 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
759 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
760 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
762 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
764 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
765 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
766 whitespace.
768 2014-01-29 Richard Biener <rguenther@suse.de>
770 PR tree-optimization/58742
771 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
772 associate_pointerplus_align.
773 (associate_pointerplus_diff): New function.
774 (associate_pointerplus): Likewise. Call associate_pointerplus_align
775 and associate_pointerplus_diff.
777 2014-01-29 Richard Biener <rguenther@suse.de>
779 * lto-streamer.h (LTO_major_version): Bump to 3.
780 (LTO_minor_version): Reset to 0.
782 2014-01-29 Renlin Li <Renlin.Li@arm.com>
784 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
785 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
786 (arm_file_start): Generate correct asm header for armv7ve.
787 * config/arm/bpabi.h: Add multilib support for armv7ve.
788 * config/arm/driver-arm.c: Change the architectures of cortex-a7
789 and cortex-a15 to armv7ve.
790 * config/arm/t-aprofile: Add multilib support for armv7ve.
791 * doc/invoke.texi: Document -march=armv7ve.
793 2014-01-29 Richard Biener <rguenther@suse.de>
795 PR tree-optimization/58742
796 * tree-ssa-forwprop.c (associate_plusminus): Return true
797 if we changed sth, defer EH cleanup to ...
798 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
799 (simplify_mult): New function.
801 2014-01-29 Jakub Jelinek <jakub@redhat.com>
803 PR middle-end/59917
804 PR tree-optimization/59920
805 * tree.c (build_common_builtin_nodes): Remove
806 __builtin_setjmp_dispatcher initialization.
807 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
808 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
809 instead of gsi_after_labels + manually skipping debug stmts.
810 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
811 ignore bbs with IFN_ABNORMAL_DISPATCHER.
812 * tree-inline.c (copy_edges_for_bb): Remove
813 can_make_abnormal_goto argument, instead add abnormal_goto_dest
814 argument. Ignore computed_goto_p stmts. Don't call
815 make_abnormal_goto_edges. If a call might need abnormal edges
816 for non-local gotos, see if it already has an edge to
817 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
818 with true argument, don't do anything then, otherwise add
819 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
820 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
821 caller.
822 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
823 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
824 (lower_stmt): Don't set data->calls_builtin_setjmp.
825 (lower_builtin_setjmp): Adjust comment.
826 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
827 * tree-cfg.c (found_computed_goto): Remove.
828 (factor_computed_gotos): Remove.
829 (make_goto_expr_edges): Return bool, true for computed gotos.
830 Don't call make_abnormal_goto_edges.
831 (build_gimple_cfg): Don't set found_computed_goto, don't call
832 factor_computed_gotos.
833 (computed_goto_p): No longer static.
834 (make_blocks): Don't set found_computed_goto.
835 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
836 (make_edges): If make_goto_expr_edges returns true, push bb
837 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
838 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
839 vector. Record mapping between bbs and OpenMP regions if there
840 are any, adjust make_gimple_omp_edges caller. Call
841 handle_abnormal_edges.
842 (make_abnormal_goto_edges): Remove.
843 * tree-cfg.h (make_abnormal_goto_edges): Remove.
844 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
845 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
846 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
847 * internal-fn.def (ABNORMAL_DISPATCHER): New.
848 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
849 filling *region also set *region_idx to (*region)->entry->index.
851 PR other/58712
852 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
853 For REGs set ORIGINAL_REGNO.
855 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
857 * doc/md.texi: Mention that a target shouldn't implement
858 vec_widen_(s|u)mul_even/odd pair if it is less efficient
859 than hi/lo pair.
861 2014-01-29 Jakub Jelinek <jakub@redhat.com>
863 PR tree-optimization/59594
864 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
865 a copy of the datarefs vector rather than the vector itself.
867 2014-01-28 Jason Merrill <jason@redhat.com>
869 PR c++/53756
870 * dwarf2out.c (auto_die): New static.
871 (gen_type_die_with_usage): Handle C++1y 'auto'.
872 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
873 on definition.
875 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
877 PR target/59672
878 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
879 (SPEC_X32): Likewise.
880 (SPEC_64): Likewise.
881 * config/i386/i386.c (ix86_option_override_internal): Turn off
882 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
883 for TARGET_16BIT.
884 (x86_file_start): Output .code16gcc for TARGET_16BIT.
885 * config/i386/i386.h (TARGET_16BIT): New macro.
886 (TARGET_16BIT_P): Likewise.
887 * config/i386/i386.opt: Add m16.
888 * doc/invoke.texi: Document -m16.
890 2014-01-28 Jakub Jelinek <jakub@redhat.com>
892 PR preprocessor/59935
893 * input.c (location_get_source_line): Bail out on when line number
894 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
896 2014-01-28 Richard Biener <rguenther@suse.de>
898 PR tree-optimization/58742
899 * tree-ssa-forwprop.c (associate_plusminus): Handle
900 pointer subtraction of the form (T)(P + A) - (T)P.
902 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
904 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
905 at const_int_cost.
907 2014-01-28 Richard Biener <rguenther@suse.de>
909 Revert
910 2014-01-28 Richard Biener <rguenther@suse.de>
912 PR rtl-optimization/45364
913 PR rtl-optimization/59890
914 * var-tracking.c (local_get_addr_clear_given_value): Handle
915 already cleared slot.
916 (val_reset): Handle not allocated local_get_addr_cache.
917 (vt_find_locations): Use post-order on the inverted CFG.
919 2014-01-28 Richard Biener <rguenther@suse.de>
921 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
923 2014-01-28 Richard Biener <rguenther@suse.de>
925 PR rtl-optimization/45364
926 PR rtl-optimization/59890
927 * var-tracking.c (local_get_addr_clear_given_value): Handle
928 already cleared slot.
929 (val_reset): Handle not allocated local_get_addr_cache.
930 (vt_find_locations): Use post-order on the inverted CFG.
932 2014-01-28 Alan Modra <amodra@gmail.com>
934 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
935 * configure.ac <recursive call for build != host>: Define
936 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
937 and LD_FOR_BUILD too.
938 * configure: Regenerate.
940 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
942 * config/i386/i386.c (get_builtin_code_for_version): Separate
943 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
944 Broadwell from Haswell.
946 2014-01-27 Steve Ellcey <sellcey@mips.com>
948 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
949 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
950 * config/mips/mips.c (mips_option_override): Change setting
951 of TARGET_DSP.
952 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
953 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
954 Change from Mask to Var.
956 2014-01-27 Jeff Law <law@redhat.com>
958 * ipa-inline.c (inline_small_functions): Fix typo.
960 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
962 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
963 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
964 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
965 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
966 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
967 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
968 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
969 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
970 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
971 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
972 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
973 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
974 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
975 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
976 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
977 (_mm512_storeu_epi64): Ditto.
978 (_mm512_cmpge_epi32_mask): Ditto.
979 (_mm512_cmpge_epu32_mask): Ditto.
980 (_mm512_cmpge_epi64_mask): Ditto.
981 (_mm512_cmpge_epu64_mask): Ditto.
982 (_mm512_cmple_epi32_mask): Ditto.
983 (_mm512_cmple_epu32_mask): Ditto.
984 (_mm512_cmple_epi64_mask): Ditto.
985 (_mm512_cmple_epu64_mask): Ditto.
986 (_mm512_cmplt_epi32_mask): Ditto.
987 (_mm512_cmplt_epu32_mask): Ditto.
988 (_mm512_cmplt_epi64_mask): Ditto.
989 (_mm512_cmplt_epu64_mask): Ditto.
990 (_mm512_cmpneq_epi32_mask): Ditto.
991 (_mm512_cmpneq_epu32_mask): Ditto.
992 (_mm512_cmpneq_epi64_mask): Ditto.
993 (_mm512_cmpneq_epu64_mask): Ditto.
994 (_mm512_expand_pd): Ditto.
995 (_mm512_expand_ps): Ditto.
996 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
997 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
998 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
999 * config/i386/i386.c (ix86_builtins): Add
1000 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1001 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1002 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1003 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1004 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1005 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1006 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1007 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1008 IX86_BUILTIN_PMOVUSQW512_MEM.
1009 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1010 __builtin_ia32_pmovsqd512mem_mask,
1011 __builtin_ia32_pmovqd512mem_mask,
1012 __builtin_ia32_pmovusqw512mem_mask,
1013 __builtin_ia32_pmovsqw512mem_mask,
1014 __builtin_ia32_pmovqw512mem_mask,
1015 __builtin_ia32_pmovusdw512mem_mask,
1016 __builtin_ia32_pmovsdw512mem_mask,
1017 __builtin_ia32_pmovdw512mem_mask,
1018 __builtin_ia32_pmovqb512mem_mask,
1019 __builtin_ia32_pmovusqb512mem_mask,
1020 __builtin_ia32_pmovsqb512mem_mask,
1021 __builtin_ia32_pmovusdb512mem_mask,
1022 __builtin_ia32_pmovsdb512mem_mask,
1023 __builtin_ia32_pmovdb512mem_mask.
1024 (bdesc_args): Add __builtin_ia32_expanddf512,
1025 __builtin_ia32_expandsf512.
1026 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1027 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1028 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1029 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1030 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1031 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1032 (avx512f_<code>v8div16qi2_mask_store): This.
1033 (avx512f_expand<mode>): New.
1035 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1037 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1038 New.
1039 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1040 (_mm512_prefetch_i32scatter_pd): Ditto.
1041 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1042 (_mm512_prefetch_i64scatter_pd): Ditto.
1043 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1044 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1045 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1046 (_mm512_prefetch_i32scatter_ps): Ditto.
1047 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1048 (_mm512_prefetch_i64scatter_ps): Ditto.
1049 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1050 * config/i386/i386-builtin-types.def: Define
1051 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1052 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1053 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1054 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1055 IX86_BUILTIN_SCATTERPFQPD.
1056 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1057 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1058 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1059 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1060 __builtin_ia32_scatterpfqps.
1061 (ix86_expand_builtin): Expand new built-ins.
1062 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1063 fix memory access data type.
1064 (*avx512pf_gatherpf<mode>_mask): Ditto.
1065 (*avx512pf_gatherpf<mode>): Ditto.
1066 (avx512pf_scatterpf<mode>): Ditto.
1067 (*avx512pf_scatterpf<mode>_mask): Ditto.
1068 (*avx512pf_scatterpf<mode>): Ditto.
1069 (GATHER_SCATTER_SF_MEM_MODE): New.
1070 (avx512pf_gatherpf<mode>df): Ditto.
1071 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1072 (*avx512pf_scatterpf<mode>df): Ditto.
1074 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1076 PR bootstrap/59934
1077 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1078 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1079 reached.
1081 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1083 * common/config/arm/arm-common.c
1084 (arm_rewrite_mcpu): Handle multiple names.
1085 * config/arm/arm.h
1086 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1088 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1090 * gimple-builder.h (create_gimple_tmp): Delete.
1092 2014-01-27 Christian Bruel <christian.bruel@st.com>
1094 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1095 words comparisons.
1097 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1099 * config/pa/pa.md (call): Generate indirect long calls to non-local
1100 functions when outputing 32-bit code.
1101 (call_value): Likewise except for special call to buggy powf function.
1103 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1104 portable runtime and PIC indirect calls.
1105 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1106 and PIC call sequences. Use ldo instead of blr to set return register
1107 in PIC call sequence.
1109 2014-01-25 Walter Lee <walt@tilera.com>
1111 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1112 avoid clobbering a live register.
1114 2014-01-25 Walter Lee <walt@tilera.com>
1116 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1117 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1118 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1119 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1121 2014-01-25 Walter Lee <walt@tilera.com>
1123 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1124 arguments on even registers.
1125 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1126 STACK_BOUNDARY.
1127 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1128 (BIGGEST_ALIGNMENT): Ditto.
1129 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1131 2014-01-25 Walter Lee <walt@tilera.com>
1133 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1134 insns before bundling.
1135 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1137 2014-01-25 Walter Lee <walt@tilera.com>
1139 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1140 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1141 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1143 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1145 * config/mips/constraints.md (kl): Delete.
1146 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1147 define expands, using...
1148 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1149 instructions for MIPS16.
1150 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1151 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1153 2014-01-25 Walter Lee <walt@tilera.com>
1155 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1156 (clzdi2): Ditto.
1157 (ffsdi2): Ditto.
1159 2014-01-25 Walter Lee <walt@tilera.com>
1161 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1162 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1164 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1166 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1167 Handle XOR.
1169 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1171 * print-rtl.c (in_call_function_usage): New var.
1172 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1173 EXPR_LIST mode as mode and not as reg note name.
1175 PR middle-end/59561
1176 * cfgloopmanip.c (copy_loop_info): If
1177 loop->warned_aggressive_loop_optimizations, make sure
1178 the flag is set in target loop too.
1180 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1182 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1183 flag_cilkplus.
1184 * builtins.def: Likewise.
1185 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1186 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1187 * ira.c (ira_setup_eliminable_regset): Likewise.
1188 * omp-low.c (gate_expand_omp): Likewise.
1189 (execute_lower_omp): Likewise.
1190 (diagnose_sb_0): Likewise.
1191 (gate_diagnose_omp_blocks): Likewise.
1192 (simd_clone_clauses_extract): Likewise.
1193 (gate): Likewise.
1195 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1197 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1198 correction for little endian...
1199 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1200 here.
1202 2014-01-24 Jeff Law <law@redhat.com>
1204 PR tree-optimization/59919
1205 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1206 for non-returning calls.
1208 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1210 * common/config/aarch64/aarch64-common.c
1211 (aarch64_rewrite_mcpu): Handle multiple names.
1212 * config/aarch64/aarch64.h
1213 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1215 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1217 * input.c (add_file_to_cache_tab): Handle the case where fopen
1218 returns NULL.
1220 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1222 PR target/59929
1223 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1224 from push operand if code of push isn't PRE_DEC.
1226 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1228 PR target/59909
1229 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1230 -mquad-memory-atomic. Update -mquad-memory documentation to say
1231 it is only used for non-atomic loads/stores.
1233 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1234 -mquad-memory or -mquad-memory-atomic switches.
1236 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1237 -mquad-memory-atomic to ISA 2.07 support.
1239 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1240 to separate support of normal quad word memory operations (ldq, stq)
1241 from the atomic quad word memory operations.
1243 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1244 support to separate non-atomic quad word operations from atomic
1245 quad word operations. Disable non-atomic quad word operations in
1246 little endian mode so that we don't have to swap words after the
1247 load and before the store.
1248 (quad_load_store_p): Add comment about atomic quad word support.
1249 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1250 options printed with -mdebug=reg.
1252 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1253 -mquad-memory-atomic as the test for whether we have quad word
1254 atomic instructions.
1255 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1256 or -mp8-vector are used, allow byte/half-word atomic operations.
1258 * config/rs6000/sync.md (load_lockedti): Insure that the address
1259 is a proper indexed or indirect address for the lqarx instruction.
1260 On little endian systems, swap the hi/lo registers after the lqarx
1261 instruction.
1262 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1263 insure the address is valid for the lqarx instruction.
1264 (store_conditionalti): Insure that the address is a proper indexed
1265 or indirect address for the stqcrx. instruction. On little endian
1266 systems, swap the hi/lo registers before doing the stqcrx.
1267 instruction.
1268 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1269 insure the address is valid for the stqcrx. instruction.
1271 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1272 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1273 type of quad memory support is available.
1275 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1277 PR regression/59915
1278 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1279 there is a danger of looping.
1281 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1283 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1284 force flag_ira_loop_pressure if set via command line.
1286 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1288 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1289 (ashr_simd): New builtin handling DI mode.
1290 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1291 (aarch64_sshr_simddi): New match pattern.
1292 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1293 (vshrd_n_s64): Likewise.
1294 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1296 2014-01-23 Nick Clifton <nickc@redhat.com>
1298 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1299 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1300 favour of mcu specific scripts.
1301 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1302 430x multilibs.
1304 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1305 Alex Velenko <Alex.Velenko@arm.com>
1307 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1308 (vaddv_s16): Likewise.
1309 (vaddv_s32): Likewise.
1310 (vaddv_u8): Likewise.
1311 (vaddv_u16): Likewise.
1312 (vaddv_u32): Likewise.
1313 (vaddvq_s8): Likewise.
1314 (vaddvq_s16): Likewise.
1315 (vaddvq_s32): Likewise.
1316 (vaddvq_s64): Likewise.
1317 (vaddvq_u8): Likewise.
1318 (vaddvq_u16): Likewise.
1319 (vaddvq_u32): Likewise.
1320 (vaddvq_u64): Likewise.
1321 (vaddv_f32): Likewise.
1322 (vaddvq_f32): Likewise.
1323 (vaddvq_f64): Likewise.
1324 (vmaxv_f32): Likewise.
1325 (vmaxv_s8): Likewise.
1326 (vmaxv_s16): Likewise.
1327 (vmaxv_s32): Likewise.
1328 (vmaxv_u8): Likewise.
1329 (vmaxv_u16): Likewise.
1330 (vmaxv_u32): Likewise.
1331 (vmaxvq_f32): Likewise.
1332 (vmaxvq_f64): Likewise.
1333 (vmaxvq_s8): Likewise.
1334 (vmaxvq_s16): Likewise.
1335 (vmaxvq_s32): Likewise.
1336 (vmaxvq_u8): Likewise.
1337 (vmaxvq_u16): Likewise.
1338 (vmaxvq_u32): Likewise.
1339 (vmaxnmv_f32): Likewise.
1340 (vmaxnmvq_f32): Likewise.
1341 (vmaxnmvq_f64): Likewise.
1342 (vminv_f32): Likewise.
1343 (vminv_s8): Likewise.
1344 (vminv_s16): Likewise.
1345 (vminv_s32): Likewise.
1346 (vminv_u8): Likewise.
1347 (vminv_u16): Likewise.
1348 (vminv_u32): Likewise.
1349 (vminvq_f32): Likewise.
1350 (vminvq_f64): Likewise.
1351 (vminvq_s8): Likewise.
1352 (vminvq_s16): Likewise.
1353 (vminvq_s32): Likewise.
1354 (vminvq_u8): Likewise.
1355 (vminvq_u16): Likewise.
1356 (vminvq_u32): Likewise.
1357 (vminnmv_f32): Likewise.
1358 (vminnmvq_f32): Likewise.
1359 (vminnmvq_f64): Likewise.
1361 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1363 * config/aarch64/aarch64-simd.md
1364 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1365 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1366 (*aarch64_mul3_elt<mode>): Likewise.
1367 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1368 (*aarch64_mul3_elt_to_64v2df): Likewise.
1369 (*aarch64_mla_elt<mode>): Likewise.
1370 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1371 (*aarch64_mls_elt<mode>): Likewise.
1372 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1373 (*aarch64_fma4_elt<mode>): Likewise.
1374 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1375 (*aarch64_fma4_elt_to_64v2df): Likewise.
1376 (*aarch64_fnma4_elt<mode>): Likewise.
1377 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1378 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1379 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1380 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1381 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1382 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1383 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1384 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1385 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1387 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1389 * config/aarch64/aarch64-simd.md
1390 (aarch64_be_checked_get_lane<mode>): New define_expand.
1391 * config/aarch64/aarch64-simd-builtins.def
1392 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1393 New builtin definition.
1394 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1395 Use new safe be builtin.
1397 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1399 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1400 New define_insn.
1401 (aarch64_be_st1<mode>): Likewise.
1402 (aarch_ld1<VALL:mode>): Define_expand modified.
1403 (aarch_st1<VALL:mode>): Likewise.
1404 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1405 (UNSPEC_ST1): Likewise.
1407 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1409 * config/microblaze/microblaze.md: Add trap insn and attribute
1411 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1413 PR preprocessor/58580
1414 * input.h (location_get_source_line): Take an additional line_size
1415 parameter.
1416 (void diagnostics_file_cache_fini): Declare new function.
1417 * input.c (struct fcache): New type.
1418 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1419 New static constants.
1420 (diagnostic_file_cache_init, total_lines_num)
1421 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1422 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1423 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1424 (get_next_line, read_next_line, goto_next_line, read_line_num):
1425 New static function definitions.
1426 (diagnostic_file_cache_fini): New function.
1427 (location_get_source_line): Take an additional output line_len
1428 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1429 read_line_num.
1430 * diagnostic.c (diagnostic_finish): Call
1431 diagnostic_file_cache_fini.
1432 (adjust_line): Take an additional input parameter for the length
1433 of the line, rather than calculating it with strlen.
1434 (diagnostic_show_locus): Adjust the use of
1435 location_get_source_line and adjust_line with respect to their new
1436 signature. While displaying a line now, do not stop at the first
1437 null byte. Rather, display the zero byte as a space and keep
1438 going until we reach the size of the line.
1439 * Makefile.in: Add vec.o to OBJS-libcommon
1441 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1442 Ilya Tocar <ilya.tocar@intel.com>
1444 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1445 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1446 (__builtin_ia32_kmov16): Ditto.
1447 * config/i386/i386.md (UNSPEC_KMOV): New.
1448 (kmovw): Ditto.
1450 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1452 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1453 (_mm512_storeu_si512): Ditto.
1455 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1457 PR target/52125
1458 * rtl.h (get_referenced_operands): Declare.
1459 * recog.c (get_referenced_operands): New function.
1460 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1461 operands have been referenced when recording LO_SUM references.
1463 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1465 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1467 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1469 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1470 Enable for generic and recent AMD targets.
1472 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1474 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1475 ARG_SIZE note when adjustment was eliminated.
1477 2014-01-22 Jeff Law <law@redhat.com>
1479 PR tree-optimization/59597
1480 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1481 in file. Accept new argument REGISTERING and use it to modify
1482 dump output appropriately.
1483 (register_jump_thread): Corresponding changes.
1484 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1485 thread paths involving joiner blocks. Add code to dump cancelled
1486 jump threading paths.
1488 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1490 PR rtl-optimization/59477
1491 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1492 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1494 2014-01-22 Tom Tromey <tromey@redhat.com>
1496 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1497 PARAMS.
1498 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1500 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1502 PR rtl-optimization/59896
1503 * lra-constraints.c (process_alt_operands): Check unused note for
1504 matched operands of insn with no output reloads.
1506 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1508 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1509 (mips_move_from_gpr_cost): Likewise.
1511 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1513 PR rtl-optimization/59858
1514 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1515 ira_class_hard_regs_num.
1516 (process_alt_operands): Increase reject for dying matched operand.
1518 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1520 PR target/59003
1521 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1522 smaller than size, perform several stores or loads and stores
1523 at dst + count - size to store or copy all of size bytes, rather
1524 than just last modesize bytes.
1526 2014-01-20 DJ Delorie <dj@redhat.com>
1528 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1529 that CLOBBERs are REGs before propogating their values.
1531 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1533 PR middle-end/59789
1534 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1535 (cgraph_inline_failed_type): New function.
1536 * cgraph.h (DEFCIFCODE): Add type.
1537 (cgraph_inline_failed_type_t): New enum.
1538 (cgraph_inline_failed_type): New prototype.
1539 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1540 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1541 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1542 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1543 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1544 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1545 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1546 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1547 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1548 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1549 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1550 OPTIMIZATION_MISMATCH.
1551 * tree-inline.c (expand_call_inline): Emit errors during
1552 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1554 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1556 PR target/59685
1557 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1558 mode attribute in insn output.
1560 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1562 * output.h (output_constant): Delete.
1563 * varasm.c (output_constant): Make private.
1565 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1567 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1569 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1571 PR middle-end/59860
1572 * tree.h (fold_builtin_strcat): New prototype.
1573 * builtins.c (fold_builtin_strcat): No longer static. Add len
1574 argument, if non-NULL, don't call c_strlen. Optimize
1575 directly into __builtin_memcpy instead of __builtin_strcpy.
1576 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1577 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1579 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1581 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1582 for SImode_address_operand operands, having only a REG argument.
1584 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1586 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1587 loader name using mbig-endian.
1588 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1590 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1592 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1593 (-mtune): Likewise.
1594 (-mcpu): Likewise.
1596 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1598 * config/aarch64/aarch64-protos.h
1599 (aarch64_cannot_change_mode_class_ptr): Declare.
1600 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1601 aarch64_cannot_change_mode_class_ptr): New.
1602 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1603 backend hook aarch64_cannot_change_mode_class.
1605 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1607 * common/config/aarch64/aarch64-common.c
1608 (aarch64_handle_option): Don't handle any option order logic here.
1609 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1610 selected_cpu, warn on architecture version mismatch.
1611 (aarch64_override_options): Fix parsing order for option strings.
1613 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1614 Iain Sandoe <iain@codesourcery.com>
1616 PR bootstrap/59496
1617 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1618 warning. Amend comment to reflect current functionality.
1620 2014-01-20 Richard Biener <rguenther@suse.de>
1622 PR middle-end/59860
1623 * builtins.c (fold_builtin_strcat): Remove case better handled
1624 by tree-ssa-strlen.c.
1626 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1628 * config/aarch64/aarch64.opt
1629 (mcpu, march, mtune): Make case-insensitive.
1631 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1633 PR target/59880
1634 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1635 if operands[1] is a REG or ZERO_EXTEND of a REG.
1637 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
1639 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1641 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1643 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1644 long non-pic millicode calls.
1646 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1648 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1650 2014-01-19 Kito Cheng <kito@0xlab.org>
1652 * builtins.c (expand_movstr): Check movstr expand done or fail.
1654 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1655 H.J. Lu <hongjiu.lu@intel.com>
1657 PR target/59379
1658 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1659 to DImode for zero-extended addresses.
1661 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1663 PR rtl-optimization/57763
1664 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1665 on the new indirect jump_insn and increment LABEL_NUSES (label).
1667 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1669 PR bootstrap/59580
1670 PR bootstrap/59583
1671 * config.gcc (x86_archs): New variable.
1672 (x86_64_archs): Likewise.
1673 (x86_cpus): Likewise.
1674 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1675 --with-arch/--with-cpu= options.
1676 Support --with-arch=/--with-cpu={nehalem,westmere,
1677 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1679 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1681 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1682 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
1684 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1686 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
1688 2014-01-18 Jakub Jelinek <jakub@redhat.com>
1690 PR target/58944
1691 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
1692 clear cpp_get_options (parse_in)->warn_unused_macros for
1693 ix86_target_macros_internal with cpp_define.
1695 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
1697 * jump.c (delete_related_insns): Keep (use (insn))s.
1698 * reorg.c (redundant_insn): Check for barriers too.
1700 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1702 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
1704 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
1706 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
1707 call to $$dyncall when TARGET_LONG_CALLS is true.
1709 2014-01-17 Jeff Law <law@redhat.com>
1711 * ree.c (combine_set_extension): Temporarily disable test for
1712 changing number of hard registers.
1714 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1716 PR middle-end/58125
1717 * ipa-inline-analysis.c (inline_free_summary):
1718 Do not free summary of aliases.
1720 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1722 PR middle-end/59706
1723 * gimplify.c (gimplify_expr): Use create_tmp_var
1724 instead of create_tmp_var_raw. If cond doesn't have
1725 integral type, don't add the IFN_ANNOTATE builtin at all.
1727 2014-01-17 Martin Jambor <mjambor@suse.cz>
1729 PR ipa/59736
1730 * ipa-cp.c (prev_edge_clone): New variable.
1731 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
1732 Also resize prev_edge_clone vector.
1733 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
1734 (ipcp_edge_removal_hook): New function.
1735 (ipcp_driver): Register ipcp_edge_removal_hook.
1737 2014-01-17 Andrew Pinski <apinski@cavium.com>
1738 Steve Ellcey <sellcey@mips.com>
1740 PR target/59462
1741 * config/mips/mips.c (mips_print_operand): Check operand mode instead
1742 of operator mode.
1744 2014-01-17 Jeff Law <law@redhat.com>
1746 PR middle-end/57904
1747 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
1748 so that pass_ccp runs first.
1750 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1752 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
1753 (ix86_adjust_cost): Use !TARGET_XXX.
1754 (do_reorder_for_imul): Likewise.
1755 (swap_top_of_ready_list): Likewise.
1756 (ix86_sched_reorder): Likewise.
1758 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
1760 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1761 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
1762 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
1763 (intel_memset): New. Duplicate slm_memset.
1764 (intel_cost): New. Duplicate slm_cost.
1765 (m_INTEL): New macro.
1766 (processor_target_table): Add "intel".
1767 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
1768 with PROCESSOR_INTEL for "intel".
1769 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
1770 PROCESSOR_SILVERMONT.
1771 (ix86_issue_rate): Likewise.
1772 (ix86_adjust_cost): Likewise.
1773 (ia32_multipass_dfa_lookahead): Likewise.
1774 (swap_top_of_ready_list): Likewise.
1775 (ix86_sched_reorder): Likewise.
1776 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
1777 instead of TARGET_OPT_AGU.
1778 * config/i386/i386.h (TARGET_INTEL): New.
1779 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
1780 (processor_type): Add PROCESSOR_INTEL.
1781 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
1782 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
1784 2014-01-17 Marek Polacek <polacek@redhat.com>
1786 PR c/58346
1787 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
1788 size is zero.
1790 2014-01-17 Richard Biener <rguenther@suse.de>
1792 PR tree-optimization/46590
1793 * opts.c (default_options_table): Add entries for
1794 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
1795 all enabled at -O1 but not for -Og.
1796 * common.opt (fbranch-count-reg): Remove Init(1).
1797 (fmove-loop-invariants): Likewise.
1798 (ftree-pta): Likewise.
1800 2014-01-17 Jakub Jelinek <jakub@redhat.com>
1802 * config/i386/i386.c (ix86_data_alignment): For compatibility with
1803 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
1804 decls to at least the GCC 4.8 used alignments.
1806 PR fortran/59440
1807 * tree-nested.c (convert_nonlocal_reference_stmt,
1808 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
1809 of GIMPLE_BIND stmts, adjust associated decls.
1811 2014-01-17 Richard Biener <rguenther@suse.de>
1813 PR tree-optimization/46590
1814 * vec.h (vec<>::bseach): New member function implementing
1815 binary search according to C89 bsearch.
1816 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
1817 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
1818 bitmap pointer again. Make accesses_in_loop a flat array.
1819 (mem_ref_obstack): New global.
1820 (outermost_indep_loop): Adjust for mem_ref->stored changes.
1821 (mark_ref_stored): Likewise.
1822 (ref_indep_loop_p_2): Likewise.
1823 (set_ref_stored_in_loop): New helper function.
1824 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
1825 (memref_free): Adjust.
1826 (record_mem_ref_loc): Simplify.
1827 (gather_mem_refs_stmt): Adjust.
1828 (sort_locs_in_loop_postorder_cmp): New function.
1829 (analyze_memory_references): Sort accesses_in_loop after
1830 loop postorder number.
1831 (find_ref_loc_in_loop_cmp): New function.
1832 (for_all_locs_in_loop): Find relevant cluster of locs in
1833 accesses_in_loop and iterate without recursion.
1834 (execute_sm): Avoid uninit warning.
1835 (struct ref_always_accessed): Simplify.
1836 (ref_always_accessed::operator ()): Likewise.
1837 (ref_always_accessed_p): Likewise.
1838 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
1839 loop postorder numbers here.
1840 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
1841 numbers.
1843 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
1845 PR c++/57945
1846 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
1847 on decls for which assemble_alias has been called.
1849 2014-01-17 Nick Clifton <nickc@redhat.com>
1851 * config/msp430/msp430.opt: (mcpu): New option.
1852 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
1853 (msp430_option_override): Parse target_cpu. If the MCU name
1854 matches a generic string, clear target_mcu.
1855 (msp430_attr): Allow numeric interrupt values up to 63.
1856 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
1857 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
1858 option.
1859 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
1860 Add mcpu matches.
1861 * config/msp430/msp430.md (popm): Use %J rather than %I.
1862 (addsi3): Use msp430_nonimmediate_operand for operand 2.
1863 (addhi_cy_i): Use immediate_operand for operand 2.
1864 * doc/invoke.texi: Document -mcpu option.
1866 2014-01-17 Richard Biener <rguenther@suse.de>
1868 PR rtl-optimization/38518
1869 * df.h (df_analyze_loop): Declare.
1870 * df-core.c: Include cfgloop.h.
1871 (df_analyze_1): Split out main part of df_analyze.
1872 (df_analyze): Adjust.
1873 (loop_inverted_post_order_compute): New function.
1874 (loop_post_order_compute): Likewise.
1875 (df_analyze_loop): New function avoiding whole-function
1876 postorder computes.
1877 * loop-invariant.c (find_defs): Use df_analyze_loop.
1878 (find_invariants): Adjust.
1879 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
1881 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
1883 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
1884 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
1886 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
1888 * ipa-ref.c (ipa_remove_stmt_references): Fix references
1889 traversal when removing references.
1891 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
1893 PR ipa/59775
1894 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
1896 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
1898 PR middle-end/56791
1899 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
1900 pushing a reload for an autoinc when we had previously reloaded an
1901 inner part of the address.
1903 2014-01-16 Jakub Jelinek <jakub@redhat.com>
1905 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
1906 field.
1907 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
1908 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
1909 when not giving up or versioning for alias only because of
1910 loop->safelen.
1911 (vect_analyze_data_ref_dependences): Set to true.
1912 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
1913 is a GIMPLE_PHI.
1914 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
1915 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
1916 to the condition.
1918 PR middle-end/58344
1919 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
1921 PR target/59839
1922 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
1923 operand 0 predicate for gathers, use a new pseudo as subtarget.
1925 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1927 PR middle-end/59609
1928 * lra-constraints.c (process_alt_operands): Add printing debug info.
1929 Check absence of input/output reloads for matched operands too.
1931 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
1933 PR rtl-optimization/59835
1934 * ira.c (ira_init_register_move_cost): Increase cost for
1935 impossible modes.
1937 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
1939 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
1941 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
1943 PR target/59780
1944 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
1945 non-register objects. Use gen_(high/low)part more consistently.
1946 Fix assertions.
1948 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1950 PR target/59844
1951 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
1952 endian support, remove tests for WORDS_BIG_ENDIAN.
1953 (p8_mfvsrd_3_<mode>): Likewise.
1954 (reload_gpr_from_vsx<mode>): Likewise.
1955 (reload_gpr_from_vsxsf): Likewise.
1956 (p8_mfvsrd_4_disf): Likewise.
1958 2014-01-16 Richard Biener <rguenther@suse.de>
1960 PR rtl-optimization/46590
1961 * lcm.c (compute_antinout_edge): Use postorder iteration.
1962 (compute_laterin): Use inverted postorder iteration.
1964 2014-01-16 Nick Clifton <nickc@redhat.com>
1966 PR middle-end/28865
1967 * varasm.c (output_constant): Return the number of bytes actually
1968 emitted.
1969 (output_constructor_array_range): Update the field size with the
1970 number of bytes emitted by output_constant.
1971 (output_constructor_regular_field): Likewise. Also do not
1972 complain if the total number of bytes emitted is now greater
1973 than the expected fieldpos.
1974 * output.h (output_constant): Update prototype and descriptive comment.
1976 2014-01-16 Marek Polacek <polacek@redhat.com>
1978 PR middle-end/59827
1979 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
1980 it is error_mark_node.
1982 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
1984 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
1985 VALID_AVX256_REG_OR_OI_MODE.
1987 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
1989 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
1990 current procedure should be profiled.
1992 2014-01-15 Andrew Pinski <apinski@cavium.com>
1994 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
1995 of moving from/to the STACK_REG register class.
1997 2014-01-15 Richard Henderson <rth@redhat.com>
1999 PR debug/54694
2000 * reginfo.c (global_regs_decl): Globalize.
2001 * rtl.h (global_regs_decl): Declare.
2002 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2003 reserved via global_regs.
2005 2014-01-15 Teresa Johnson <tejohnson@google.com>
2007 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2009 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2011 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2012 and vmulosh rather than call gen_vec_widen_smult_*.
2013 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2014 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2015 (vec_widen_smult_even_v16qi): Likewise.
2016 (vec_widen_umult_even_v8hi): Likewise.
2017 (vec_widen_smult_even_v8hi): Likewise.
2018 (vec_widen_umult_odd_v16qi): Likewise.
2019 (vec_widen_smult_odd_v16qi): Likewise.
2020 (vec_widen_umult_odd_v8hi): Likewise.
2021 (vec_widen_smult_odd_v8hi): Likewise.
2022 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2023 vmuloub rather than call gen_vec_widen_umult_*.
2024 (vec_widen_umult_lo_v16qi): Likewise.
2025 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2026 vmulosb rather than call gen_vec_widen_smult_*.
2027 (vec_widen_smult_lo_v16qi): Likewise.
2028 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2029 rather than call gen_vec_widen_umult_*.
2030 (vec_widen_umult_lo_v8hi): Likewise.
2031 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2032 rather than call gen_vec_widen_smult_*.
2033 (vec_widen_smult_lo_v8hi): Likewise.
2035 2014-01-15 Jeff Law <law@redhat.com>
2037 PR tree-optimization/59747
2038 * ree.c (find_and_remove_re): Properly handle case where a second
2039 eliminated extension requires widening a copy created for elimination
2040 of a prior extension.
2041 (combine_set_extension): Ensure that the number of hard regs needed
2042 for a destination register does not change when we widen it.
2044 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2046 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2047 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2048 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2049 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2050 (avr-*-rtems*): Likewise.
2051 (bfin*-rtems*): Likewise.
2052 (moxie-*-rtems*): Likewise.
2053 (h8300-*-rtems*): Likewise.
2054 (i[34567]86-*-rtems*): Likewise.
2055 (lm32-*-rtems*): Likewise.
2056 (m32r-*-rtems*): Likewise.
2057 (m68k-*-rtems*): Likewise.
2058 (microblaze*-*-rtems*): Likewise.
2059 (mips*-*-rtems*): Likewise.
2060 (powerpc-*-rtems*): Likewise.
2061 (sh-*-rtems*): Likewise.
2062 (sparc-*-rtems*): Likewise.
2063 (sparc64-*-rtems*): Likewise.
2064 (v850-*-rtems*): Likewise.
2065 (m32c-*-rtems*): Likewise.
2067 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2069 PR rtl-optimization/59511
2070 * ira.c (ira_init_register_move_cost): Use memory costs for some
2071 cases of register move cost calculations.
2072 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2073 instead of BB frequency.
2074 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2075 * lra-assigns.c (find_hard_regno_for): Ditto.
2077 2014-01-15 Richard Biener <rguenther@suse.de>
2079 PR tree-optimization/59822
2080 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2081 (vectorizable_load): Use it to hoist defs of uses of invariant
2082 loads out of the loop.
2084 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2085 Kugan Vivekanandarajah <kuganv@linaro.org>
2087 PR target/59695
2088 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2089 truncation.
2091 2014-01-15 Richard Biener <rguenther@suse.de>
2093 PR rtl-optimization/59802
2094 * lcm.c (compute_available): Use inverted postorder to seed
2095 the initial worklist.
2097 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2099 PR target/59803
2100 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2101 ADDR_REGS for invalid symrefs in non-PIC code.
2103 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2105 PR other/58712
2106 * builtins.c (determine_block_size): Initialize *probable_max_size
2107 even if len_rtx is CONST_INT.
2109 2014-01-14 Andrew Pinski <apinski@cavium.com>
2111 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2112 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2113 (cortexa53_tunings): Likewise.
2114 (aarch64_sched_issue_rate): New function.
2115 (TARGET_SCHED_ISSUE_RATE): Define.
2117 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2119 * ira-costs.c (find_costs_and_classes): Add missed
2120 ira_init_register_move_cost_if_necessary.
2122 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2124 PR target/59787
2125 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2127 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2129 PR target/59794
2130 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2131 to indicate if type is used for function return value. Warn ABI
2132 change if the vector mode isn't available for function return value.
2133 (ix86_function_arg_advance): Pass false to type_natural_mode.
2134 (ix86_function_arg): Likewise.
2135 (ix86_gimplify_va_arg): Likewise.
2136 (function_arg_32): Don't warn ABI change.
2137 (ix86_function_value): Pass true to type_natural_mode.
2138 (ix86_return_in_memory): Likewise.
2139 (ix86_struct_value_rtx): Removed.
2140 (TARGET_STRUCT_VALUE_RTX): Likewise.
2142 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2144 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2145 converting a conditional jump into a conditional return.
2147 2014-01-14 Richard Biener <rguenther@suse.de>
2149 PR tree-optimization/58921
2150 PR tree-optimization/59006
2151 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2152 hoisting invariant stmts.
2153 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2154 invariant loads on the preheader edge if possible.
2156 2014-01-14 Joey Ye <joey.ye@arm.com>
2158 * doc/plugin.texi (Building GCC plugins): Update to C++.
2160 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2162 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2163 (_mm_rcp28_round_ss): Ditto.
2164 (_mm_rsqrt28_round_sd): Ditto.
2165 (_mm_rsqrt28_round_ss): Ditto.
2166 (_mm_rcp28_sd): Ditto.
2167 (_mm_rcp28_ss): Ditto.
2168 (_mm_rsqrt28_sd): Ditto.
2169 (_mm_rsqrt28_ss): Ditto.
2170 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2171 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2172 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2173 (IX86_BUILTIN_RCP28SD): Ditto.
2174 (IX86_BUILTIN_RCP28SS): Ditto.
2175 (IX86_BUILTIN_RSQRT28SD): Ditto.
2176 (IX86_BUILTIN_RSQRT28SS): Ditto.
2177 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2178 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2179 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2180 (ix86_expand_special_args_builtin): Expand new FTYPE.
2181 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2182 (srcp14<mode>): Make insn unary.
2183 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2184 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2185 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2186 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2187 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2188 Fix rounding: make it SAE only.
2189 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2190 Ditto.
2191 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2192 Ditto.
2193 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2194 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2195 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2196 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2197 (round_saeonly_mask_scalar_operand4): Ditto.
2198 (round_saeonly_mask_scalar_op3): Ditto.
2199 (round_saeonly_mask_scalar_op4): Ditto.
2201 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2203 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2204 Implement -maltivec=be for vec_insert and vec_extract.
2206 2014-01-10 DJ Delorie <dj@redhat.com>
2208 * config/msp430/msp430.md (call_internal): Don't allow memory
2209 references with SP as the base register.
2210 (call_value_internal): Likewise.
2211 * config/msp430/constraints.md (Yc): New. For memory references
2212 that don't use SP as a base register.
2214 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2215 "an integer without a # prefix"
2216 * config/msp430/msp430.md (epilogue_helper): Use it.
2218 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2220 PR target/59617
2221 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2222 AVX512F gather builtins.
2223 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2224 on gather decls with INTEGER_TYPE masktype.
2225 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2226 directly into the builtin rather than hoisting it before loop.
2228 PR tree-optimization/59387
2229 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2230 (scev_const_prop): If folded_casts and type has undefined overflow,
2231 use force_gimple_operand instead of force_gimple_operand_gsi and
2232 for each added stmt if it is assign with
2233 arith_code_with_undefined_signed_overflow, call
2234 rewrite_to_defined_overflow.
2235 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2236 gimple-fold.h instead.
2237 (arith_code_with_undefined_signed_overflow,
2238 rewrite_to_defined_overflow): Moved to ...
2239 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2240 rewrite_to_defined_overflow): ... here. No longer static.
2241 Include gimplify-me.h.
2242 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2243 rewrite_to_defined_overflow): New prototypes.
2245 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2247 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2249 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2251 * builtins.c (get_object_alignment_2): Minor tweak.
2252 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2254 2014-01-13 Christian Bruel <christian.bruel@st.com>
2256 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2257 optimized non constant lengths.
2259 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2261 PR libgomp/59194
2262 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2263 load as __atomic_load_N if possible.
2265 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2267 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2268 target parameter.
2269 (rs6000_expand_builtin): Adjust call.
2271 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2273 PR target/58115
2274 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2275 * config/rs6000/rs6000.c: Include target-globals.h.
2276 (rs6000_set_current_function): Instead of doing target_reinit
2277 unconditionally, use save_target_globals_default_opts and
2278 restore_target_globals.
2280 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2281 FPSCR.
2282 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2283 (rs6000_expand_builtin): Handle mffs and mtfsf.
2284 (rs6000_init_builtins): Define mffs and mtfsf.
2285 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2286 (rs6000_mffs): New pattern.
2287 (rs6000_mtfsf): New pattern.
2289 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2291 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2292 Start narrowing with START. Apply candidate-use pair
2293 and check overall cost in narrowing.
2294 (iv_ca_prune): Pass new argument.
2296 2014-01-10 Jeff Law <law@redhat.com>
2298 PR middle-end/59743
2299 * ree.c (combine_reaching_defs): Ensure the defining statement
2300 occurs before the extension when optimizing extensions with
2301 different source and destination hard registers.
2303 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2305 PR ipa/58585
2306 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2307 vtables into the type inheritance graph.
2309 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2311 PR rtl-optimization/59754
2312 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2313 modes in the REGNO != REGNO case.
2315 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2317 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2319 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2321 PR tree-optimization/59745
2322 * tree-predcom.c (tree_predictive_commoning_loop): Call
2323 free_affine_expand_cache if giving up because components is NULL.
2325 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2326 GC in payload of target_globals struct instead of allocating them on
2327 the heap and the larger structs separately using GC.
2328 * target-globals.h (struct target_globals): Make regs, hard_regs,
2329 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2330 of GTY((skip)) and change type to void *.
2331 (reset_target_globals): Cast loads from those fields to corresponding
2332 types.
2334 2014-01-10 Steve Ellcey <sellcey@mips.com>
2336 PR plugins/59335
2337 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2338 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2339 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2341 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2343 PR target/59744
2344 * aarch64-modes.def (CC_Zmode): New flags mode.
2345 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2346 represents an equality.
2347 (aarch64_get_condition_code): Handle CC_Zmode.
2348 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2350 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2352 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2353 extraction in good case.
2355 2014-01-10 Richard Biener <rguenther@suse.de>
2357 PR tree-optimization/59374
2358 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2359 checking after SLP discovery. Mark stmts not participating
2360 in any SLP instance properly.
2362 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2364 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2365 when handling a SET rtx.
2367 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2369 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2370 (cortex-a57): Likewise.
2371 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2373 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2375 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2376 non-iwmmxt builtins.
2378 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2380 PR ipa/58252
2381 PR ipa/59226
2382 * ipa-devirt.c record_target_from_binfo): Take as argument
2383 stack of binfos and lookup matching one for virtual inheritance.
2384 (possible_polymorphic_call_targets_1): Update.
2386 2014-01-10 Huacai Chen <chenhc@lemote.com>
2388 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2389 kernel strings for Loongson-2E/2F/3A.
2391 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2393 PR middle-end/59670
2394 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2395 is_gimple_call before calling gimple_call_internal_p.
2397 2014-01-09 Steve Ellcey <sellcey@mips.com>
2399 * Makefile.in (TREE_FLOW_H): Remove.
2400 (TREE_SSA_H): Add file names from tree-flow.h.
2401 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2402 * tree.h: Remove tree-flow.h reference.
2403 * hash-table.h: Remove tree-flow.h reference.
2404 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2405 reference with tree-ssa-loop.h.
2407 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2409 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2410 default element-order behavior for -maltivec.
2411 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2412 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2413 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2414 when targeting big endian, at least for now.
2415 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2417 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2419 PR middle-end/47735
2420 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2421 var satisfies use_register_for_decl, just take into account type
2422 alignment, rather than decl alignment.
2424 PR tree-optimization/59622
2425 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2426 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2427 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2428 Don't devirtualize for inplace at all. For targets.length () == 1,
2429 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2431 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2433 * config/i386/i386.md (cpu): Remove the unused btver1.
2435 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2437 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2439 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2441 PR target/58115
2442 * tree-core.h (struct target_globals): New forward declaration.
2443 (struct tree_target_option): Add globals field.
2444 * tree.h (TREE_TARGET_GLOBALS): Define.
2445 (prepare_target_option_nodes_for_pch): New prototype.
2446 * target-globals.h (struct target_globals): Define even if
2447 !SWITCHABLE_TARGET.
2448 * tree.c (prepare_target_option_node_for_pch,
2449 prepare_target_option_nodes_for_pch): New functions.
2450 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2451 * config/i386/i386.c: Include target-globals.h.
2452 (ix86_set_current_function): Instead of doing target_reinit
2453 unconditionally, use save_target_globals_default_opts and
2454 restore_target_globals.
2456 2014-01-09 Richard Biener <rguenther@suse.de>
2458 PR tree-optimization/59715
2459 * tree-cfg.h (split_critical_edges): Declare.
2460 * tree-cfg.c (split_critical_edges): Export.
2461 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2463 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2465 * cfgexpand.c (expand_stack_vars): Optionally disable
2466 asan stack protection.
2467 (expand_used_vars): Likewise.
2468 (partition_stack_vars): Likewise.
2469 * asan.c (asan_emit_stack_protection): Optionally disable
2470 after return stack usage.
2471 (instrument_derefs): Optionally disable memory access instrumentation.
2472 (instrument_builtin_call): Likewise.
2473 (instrument_strlen_call): Likewise.
2474 (asan_protect_global): Optionally disable global variables protection.
2475 * doc/invoke.texi: Added doc for new options.
2476 * params.def: Added new options.
2477 * params.h: Likewise.
2479 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2481 PR rtl-optimization/59724
2482 * ifcvt.c (cond_exec_process_if_block): Don't call
2483 flow_find_head_matching_sequence with 0 longest_match.
2484 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2485 non-active insns if !stop_after.
2486 (try_head_merge_bb): Revert 2014-01-07 changes.
2488 2014-01-08 Jeff Law <law@redhat.com>
2490 * ree.c (get_sub_rtx): New function, extracted from...
2491 (merge_def_and_ext): Here.
2492 (combine_reaching_defs): Use get_sub_rtx.
2494 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2496 * cgraph.h (varpool_variable_node): Do not choke on null node.
2498 2014-01-08 Catherine Moore <clm@codesourcery.com>
2500 * config/mips/mips.md (simple_return): Attempt to use JRC
2501 for microMIPS.
2502 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2504 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2506 PR rtl-optimization/59137
2507 * reorg.c (steal_delay_list_from_target): Call update_block for
2508 elided insns.
2509 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2511 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2513 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2514 two duplicate entries.
2516 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2518 Revert:
2519 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2521 * config/mips/mips.c (mips_truncated_op_cost): New function.
2522 (mips_rtx_costs): Adjust test for BADDU.
2523 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2525 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2527 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2528 (*baddu_si): ...this new pattern.
2530 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2532 PR ipa/59722
2533 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2535 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2537 PR middle-end/57748
2538 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2539 inner_reference_p.
2540 (expand_expr, expand_normal): Adjust.
2541 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2542 inner_reference_p. Use inner_reference_p to expand inner references.
2543 (store_expr): Adjust.
2544 * cfgexpand.c (expand_call_stmt): Adjust.
2546 2014-01-08 Rong Xu <xur@google.com>
2548 * gcov-io.c (gcov_var): Move from gcov-io.h.
2549 (gcov_position): Ditto.
2550 (gcov_is_error): Ditto.
2551 (gcov_rewrite): Ditto.
2552 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2553 only part to libgcc/libgcov.h.
2555 2014-01-08 Marek Polacek <polacek@redhat.com>
2557 PR middle-end/59669
2558 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2560 2014-01-08 Marek Polacek <polacek@redhat.com>
2562 PR sanitizer/59667
2563 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2565 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2567 PR rtl-optimization/59649
2568 * stor-layout.c (get_mode_bounds): For BImode return
2569 0 and STORE_FLAG_VALUE.
2571 2014-01-08 Richard Biener <rguenther@suse.de>
2573 PR middle-end/59630
2574 * gimple.h (is_gimple_builtin_call): Remove.
2575 (gimple_builtin_call_types_compatible_p): New.
2576 (gimple_call_builtin_p): New overload.
2577 * gimple.c (is_gimple_builtin_call): Remove.
2578 (validate_call): Rename to ...
2579 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2580 check return types.
2581 (validate_type): New static function.
2582 (gimple_call_builtin_p): New overload and adjust.
2583 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2584 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2585 (gimple_fold_stmt_to_constant_1): Likewise.
2586 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2588 2014-01-08 Richard Biener <rguenther@suse.de>
2590 PR middle-end/59471
2591 * gimplify.c (gimplify_expr): Gimplify register-register type
2592 VIEW_CONVERT_EXPRs to separate stmts.
2594 2014-01-07 Jeff Law <law@redhat.com>
2596 PR middle-end/53623
2597 * ree.c (combine_set_extension): Handle case where source
2598 and destination registers in an extension insn are different.
2599 (combine_reaching_defs): Allow source and destination registers
2600 in extension to be different under limited circumstances.
2601 (add_removable_extension): Remove restriction that the
2602 source and destination registers in the extension are the same.
2603 (find_and_remove_re): Emit a copy from the extension's
2604 destination to its source after the defining insn if
2605 the source and destination registers are different.
2607 PR middle-end/59285
2608 * ifcvt.c (merge_if_block): If we are merging a block with more than
2609 one successor with a block with no successors, remove any BARRIER
2610 after the second block.
2612 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2614 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2616 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2618 PR target/59652
2619 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2620 for 14-bit register offsets when INT14_OK_STRICT is false.
2622 2014-01-07 Roland Stigge <stigge@antcom.de>
2623 Michael Meissner <meissner@linux.vnet.ibm.com>
2625 PR 57386/target
2626 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2627 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2629 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2631 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2632 -mcpu.
2634 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2636 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2637 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2638 rtx is const0_rtx or not.
2640 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2642 PR target/58115
2643 * target-globals.c (save_target_globals): Remove this_fn_optab
2644 handling.
2645 * toplev.c: Include optabs.h.
2646 (target_reinit): Temporarily restore the global options if another
2647 set of options are in force.
2649 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2651 PR rtl-optimization/58668
2652 * cfgcleanup.c (flow_find_cross_jump): Don't count
2653 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2654 to determine what is counted.
2655 (flow_find_head_matching_sequence): Use active_insn_p to determine
2656 what is counted.
2657 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2658 counting change.
2659 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2660 determine what is counted.
2662 PR tree-optimization/59643
2663 * tree-predcom.c (split_data_refs_to_components): If one dr is
2664 read and one write, determine_offset fails and the write isn't
2665 in the bad component, just put the read into the bad component.
2667 2014-01-07 Mike Stump <mikestump@comcast.net>
2668 Jakub Jelinek <jakub@redhat.com>
2670 PR pch/59436
2671 * tree-core.h (struct tree_optimization_option): Change optabs
2672 type from unsigned char * to void *.
2673 * optabs.c (init_tree_optimization_optabs): Adjust
2674 TREE_OPTIMIZATION_OPTABS initialization.
2676 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2678 PR target/59644
2679 * config/i386/i386.h (struct machine_function): Add
2680 no_drap_save_restore field.
2681 * config/i386/i386.c (ix86_save_reg): Use
2682 !cfun->machine->no_drap_save_restore instead of
2683 crtl->stack_realign_needed.
2684 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
2685 this function clears frame_pointer_needed. Set
2686 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
2687 and DRAP reg is needed.
2689 2014-01-06 Marek Polacek <polacek@redhat.com>
2691 PR c/57773
2692 * doc/implement-c.texi: Mention that other integer types are
2693 permitted as bit-field types in strictly conforming mode.
2695 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
2697 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
2698 is newly allocated.
2700 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
2702 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
2704 2014-01-06 Martin Jambor <mjambor@suse.cz>
2706 PR ipa/59008
2707 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
2708 to int.
2709 * ipa-prop.c (ipa_print_node_params): Fix indentation.
2711 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
2713 PR debug/59350
2714 PR debug/59510
2715 * var-tracking.c (add_stores): Preserve the value of the source even if
2716 we don't record the store.
2718 2014-01-06 Terry Guo <terry.guo@arm.com>
2720 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
2722 2014-01-05 Iain Sandoe <iain@codesourcery.com>
2724 PR bootstrap/59541
2725 * config/darwin.c (darwin_function_section): Adjust return values to
2726 correspond to optimisation changes made in r206070.
2728 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
2730 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
2731 from prefetch_block tune setting.
2732 (nocona_cost): Correct size of prefetch block to 64.
2734 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
2736 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
2737 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
2738 used to save the static chain register in the computation of the offset
2739 from which the FP registers need to be restored.
2741 2014-01-04 Jakub Jelinek <jakub@redhat.com>
2743 PR tree-optimization/59519
2744 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
2745 ICE if get_current_def (current_new_name) is already non-NULL, as long
2746 as it is a phi result of some other phi in *new_exit_bb that has
2747 the same argument.
2749 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
2750 or vmovdqu* for misaligned_operand.
2751 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
2752 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
2753 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
2754 aligned_mem for AVX512F masked aligned load and store builtins and for
2755 non-temporal moves.
2757 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
2759 PR tree-optimization/59651
2760 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
2761 Address range for negative step should be added by TYPE_SIZE_UNIT.
2763 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
2765 * config/m68k/m68k.c (handle_move_double): Handle pushes with
2766 overlapping registers also for registers other than the stack pointer.
2768 2014-01-03 Marek Polacek <polacek@redhat.com>
2770 PR other/59661
2771 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
2772 __builtin_FILE.
2774 2014-01-03 Jakub Jelinek <jakub@redhat.com>
2776 PR target/59625
2777 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
2778 asm goto as jump.
2780 * config/i386/i386.md (MODE_SIZE): New mode attribute.
2781 (push splitter): Use <P:MODE_SIZE> instead of
2782 GET_MODE_SIZE (<P:MODE>mode).
2783 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
2784 (mov -1, reg peephole2): Likewise.
2785 * config/i386/sse.md (*mov<mode>_internal,
2786 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
2787 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
2788 *<code><mode>3, *andnot<mode>3<mask_name>,
2789 <mask_codefor><code><mode>3<mask_name>): Likewise.
2790 * config/i386/subst.md (mask_mode512bit_condition,
2791 sd_mask_mode512bit_condition): Likewise.
2793 2014-01-02 Xinliang David Li <davidxl@google.com>
2795 PR tree-optimization/59303
2796 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
2797 (dump_predicates): Better output format.
2798 (pred_equal_p): New function.
2799 (is_neq_relop_p): Ditto.
2800 (is_neq_zero_form_p): Ditto.
2801 (pred_expr_equal_p): Ditto.
2802 (pred_neg_p): Ditto.
2803 (simplify_pred): Ditto.
2804 (simplify_preds_2): Ditto.
2805 (simplify_preds_3): Ditto.
2806 (simplify_preds_4): Ditto.
2807 (simplify_preds): Ditto.
2808 (push_pred): Ditto.
2809 (push_to_worklist): Ditto.
2810 (get_pred_info_from_cmp): Ditto.
2811 (is_degenerated_phi): Ditto.
2812 (normalize_one_pred_1): Ditto.
2813 (normalize_one_pred): Ditto.
2814 (normalize_one_pred_chain): Ditto.
2815 (normalize_preds): Ditto.
2816 (normalize_cond_1): Remove function.
2817 (normalize_cond): Ditto.
2818 (is_gcond_subset_of): Ditto.
2819 (is_subset_of_any): Ditto.
2820 (is_or_set_subset_of): Ditto.
2821 (is_and_set_subset_of): Ditto.
2822 (is_norm_cond_subset_of): Ditto.
2823 (pred_chain_length_cmp): Ditto.
2824 (convert_control_dep_chain_into_preds): Type change.
2825 (find_predicates): Ditto.
2826 (find_def_preds): Ditto.
2827 (destroy_predicates_vecs): Ditto.
2828 (find_matching_predicates_in_rest_chains): Ditto.
2829 (use_pred_not_overlap_with_undef_path_pred): Ditto.
2830 (is_pred_expr_subset): Ditto.
2831 (is_pred_chain_subset_of): Ditto.
2832 (is_included_in): Ditto.
2833 (is_superset_of): Ditto.
2835 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2837 Update copyright years.
2839 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
2841 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
2842 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
2843 config/arc/arc.md, config/arc/arc.opt,
2844 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
2845 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
2846 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
2847 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
2848 config/linux-protos.h, config/linux.c, config/winnt-c.c,
2849 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
2850 vtable-verify.c, vtable-verify.h: Use the standard form for the
2851 copyright notice.
2853 2014-01-02 Tobias Burnus <burnus@net-b.de>
2855 * gcc.c (process_command): Update copyright notice dates.
2856 * gcov-dump.c: Ditto.
2857 * gcov.c: Ditto.
2858 * doc/cpp.texi: Bump @copying's copyright year.
2859 * doc/cppinternals.texi: Ditto.
2860 * doc/gcc.texi: Ditto.
2861 * doc/gccint.texi: Ditto.
2862 * doc/gcov.texi: Ditto.
2863 * doc/install.texi: Ditto.
2864 * doc/invoke.texi: Ditto.
2866 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2868 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
2870 2014-01-01 Jakub Jelinek <jakub@redhat.com>
2872 * config/i386/sse.md (*mov<mode>_internal): Guard
2873 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
2875 PR rtl-optimization/59647
2876 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
2877 new_rtx into UNSIGNED_FLOAT rtxes.
2878 \f
2879 Copyright (C) 2014 Free Software Foundation, Inc.
2881 Copying and distribution of this file, with or without modification,
2882 are permitted in any medium without royalty provided the copyright
2883 notice and this notice are preserved.