]> gcc.gnu.org Git - gcc.git/blob - gcc/ChangeLog
re PR lto/59468 (ICE on invalid C++ code with LTO in gimple_get_virt_method_for_binfo...
[gcc.git] / gcc / ChangeLog
1 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2
3 PR lto/59468
4 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
5 and wrapper.
6 * ipa-devirt.c: Include demangle.h
7 (odr_violation_reported): New static variable.
8 (add_type_duplicate): Update odr_violations.
9 (maybe_record_node): Add completep parameter; update it.
10 (record_target_from_binfo): Add COMPLETEP parameter;
11 update it as needed.
12 (possible_polymorphic_call_targets_1): Likewise.
13 (struct polymorphic_call_target_d): Add nonconstruction_targets;
14 rename FINAL to COMPLETE.
15 (record_targets_from_bases): Sanity check we found the binfo;
16 fix COMPLETEP updating.
17 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
18 parameter, fix computing of COMPLETEP.
19 (dump_possible_polymorphic_call_targets): Imrove readability of dump; at
20 LTO time do demangling.
21 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
22 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
23 parameter.
24 (gimple_get_virt_method_for_binfo): Likewise.
25 * gimple-fold.h (gimple_get_virt_method_for_binfo,
26 gimple_get_virt_method_for_vtable): Update prototypes.
27
28 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
29
30 PR target/49008
31 * genautomata.c (add_presence_absence): Fix typo with
32 {final_}presence_list.
33
34 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
35
36 PR target/60137
37 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
38 for VSX/Altivec vectors that land in GPR registers.
39
40 2014-02-11 Richard Henderson <rth@redhat.com>
41 Jakub Jelinek <jakub@redhat.com>
42
43 PR debug/59776
44 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
45 around drhs if type conversion to lacc->type is not useless.
46
47 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
48
49 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
50 tuning struct.
51 (cortex-a57.cortex-a53): Likewise.
52 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
53
54 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
55
56 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
57 arm_restrict_it.
58
59 2014-02-11 Renlin Li <Renlin.Li@arm.com>
60
61 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
62 add_options_for_arm_vfp3.
63
64 2014-02-11 Jeff Law <law@redhat.com>
65
66 PR middle-end/54041
67 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
68 object with an undesirable mode.
69
70 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
71
72 PR libgomp/60107
73 * config/i386/sol2-9.h: New file.
74 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
75 *-*-solaris2.9*): Use it.
76
77 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
78
79 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
80 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
81
82 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
83
84 * config/microblaze/microblaze.c: Extend mcpu version format
85
86 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
87
88 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
89
90 2014-02-10 Richard Henderson <rth@redhat.com>
91
92 PR target/59927
93 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
94 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
95 ms-abi vs -mno-accumulate-outgoing-args.
96 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
97 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
98 respect to ms-abi.
99
100 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
101
102 PR middle-end/60080
103 * cfgexpand.c (expand_asm_operands): Attach source location to
104 ASM_INPUT rtx objects.
105 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
106
107 2014-02-10 Nick Clifton <nickc@redhat.com>
108
109 * config/mn10300/mn10300.c (popcount): New function.
110 (mn10300_expand_prologue): Include saved registers in stack usage
111 count.
112
113 2014-02-10 Jeff Law <law@redhat.com>
114
115 PR middle-end/52306
116 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
117 when changing the SET_DEST of a prior insn to avoid an input
118 reload.
119
120 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
121
122 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
123 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
124 -mcall-openbsd, or -mcall-linux.
125 (CC1_ENDIAN_BIG_SPEC): Remove.
126 (CC1_ENDIAN_LITTLE_SPEC): Remove.
127 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
128 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
129 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
130 and %cc1_endian_default.
131 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
132
133 2014-02-10 Richard Biener <rguenther@suse.de>
134
135 PR tree-optimization/60115
136 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
137 MEM_REF handling. Properly verify that the accesses are not
138 out of the objects bound.
139
140 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
141
142 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
143 coretex to cortex.
144
145 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
146
147 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
148 proper constants and fix formatting.
149 (possible_polymorphic_call_targets): Fix formatting.
150
151 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
152 Ilya Tocar <ilya.tocar@intel.com>
153
154 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
155 (_mm512_loadu_epi32): Renamed into...
156 (_mm512_loadu_si512): This.
157 (_mm512_storeu_epi32): Renamed into...
158 (_mm512_storeu_si512): This.
159 (_mm512_maskz_ceil_ps): Removed.
160 (_mm512_maskz_ceil_pd): Ditto.
161 (_mm512_maskz_floor_ps): Ditto.
162 (_mm512_maskz_floor_pd): Ditto.
163 (_mm512_floor_round_ps): Ditto.
164 (_mm512_floor_round_pd): Ditto.
165 (_mm512_ceil_round_ps): Ditto.
166 (_mm512_ceil_round_pd): Ditto.
167 (_mm512_mask_floor_round_ps): Ditto.
168 (_mm512_mask_floor_round_pd): Ditto.
169 (_mm512_mask_ceil_round_ps): Ditto.
170 (_mm512_mask_ceil_round_pd): Ditto.
171 (_mm512_maskz_floor_round_ps): Ditto.
172 (_mm512_maskz_floor_round_pd): Ditto.
173 (_mm512_maskz_ceil_round_ps): Ditto.
174 (_mm512_maskz_ceil_round_pd): Ditto.
175 (_mm512_expand_pd): Ditto.
176 (_mm512_expand_ps): Ditto.
177 * config/i386/i386.c (ix86_builtins): Remove
178 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
179 (bdesc_args): Ditto.
180 * config/i386/predicates.md (const1256_operand): New.
181 (const_1_to_2_operand): Ditto.
182 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
183 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
184 (*avx512pf_gatherpf<mode>sf): Ditto.
185 (avx512pf_gatherpf<mode>df): Ditto.
186 (*avx512pf_gatherpf<mode>df_mask): Ditto.
187 (*avx512pf_gatherpf<mode>df): Ditto.
188 (avx512pf_scatterpf<mode>sf): Ditto.
189 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
190 (*avx512pf_scatterpf<mode>sf): Ditto.
191 (avx512pf_scatterpf<mode>df): Ditto.
192 (*avx512pf_scatterpf<mode>df_mask): Ditto.
193 (*avx512pf_scatterpf<mode>df): Ditto.
194 (avx512f_expand<mode>): Removed.
195 (<shift_insn><mode>3<mask_name>): Change predicate type.
196
197 2014-02-08 Jakub Jelinek <jakub@redhat.com>
198
199 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
200 not at the end of datarefs vector use ordered_remove to avoid
201 reordering datarefs vector.
202
203 PR c/59984
204 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
205 mark local addressable non-static vars as GOVD_PRIVATE
206 instead of GOVD_LOCAL.
207 * omp-low.c (lower_omp_for): Move gimple_bind_vars
208 and BLOCK_VARS of gimple_bind_block to new_stmt rather
209 than copying them.
210
211 PR middle-end/60092
212 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
213 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
214 assume_aligned or alloc_align attributes.
215 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
216 arguments. Handle also assume_aligned and alloc_align attributes.
217 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
218 Handle calls to functions with assume_aligned or alloc_align
219 attributes.
220 * doc/extend.texi: Document assume_aligned and alloc_align
221 attributes.
222
223 2014-02-08 Terry Guo <terry.guo@arm.com>
224
225 * doc/invoke.texi: Document ARM -march=armv7e-m.
226
227 2014-02-08 Jakub Jelinek <jakub@redhat.com>
228
229 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
230 flag on __cilkrts_rethrow builtin.
231
232 PR ipa/60026
233 * ipa-cp.c (determine_versionability): Fail at -O0
234 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
235 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
236
237 Revert:
238 2014-02-04 Jakub Jelinek <jakub@redhat.com>
239
240 PR ipa/60026
241 * tree-inline.c (copy_forbidden): Fail for
242 __attribute__((optimize (0))) functions.
243
244 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
245
246 * varpool.c: Include pointer-set.h.
247 (varpool_remove_unreferenced_decls): Variables in other partitions
248 will not be output; be however careful to not lose information
249 about partitioning.
250
251 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
252
253 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
254 lookup in the vtable constructor.
255
256 2014-02-07 Jeff Law <law@redhat.com>
257
258 PR target/40977
259 * config/m68k/m68k.md (ashldi_extsi): Turn into a
260 define_insn_and_split.
261
262 * ipa-inline.c (inline_small_functions): Fix typos.
263
264 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
265
266 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
267 (s390_can_use_return_insn): Declare.
268 * config/s390/s390.h (EPILOGUE_USES): Define.
269 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
270 instructions.
271 (s390_chunkify_start): Handle return JUMP_LABELs.
272 (s390_early_mach): Emit a main_pool instruction on the entry edge.
273 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
274 (s390_can_use_return_insn): New functions.
275 (s390_fix_long_loop_prediction): Handle conditional returns.
276 (TARGET_SET_UP_BY_PROLOGUE): Define.
277 * config/s390/s390.md (ANY_RETURN): New code iterator.
278 (*creturn, *csimple_return, return, simple_return): New patterns.
279
280 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
281
282 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
283 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
284 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
285 REG_CFA_RESTORE list when deciding not to restore a register.
286
287 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
288
289 * config/s390/s390.c: Include tree-pass.h and context.h.
290 (s390_early_mach): New function, split out from...
291 (s390_emit_prologue): ...here.
292 (pass_data_s390_early_mach): New pass structure.
293 (pass_s390_early_mach): New class.
294 (s390_option_override): Create and register early_mach pass.
295 Move to end of file.
296
297 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
298
299 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
300 to match for the exit block.
301
302 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
303
304 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
305 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
306 Reject misaligned operands.
307
308 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
309
310 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
311 fail.
312
313 2014-02-07 Richard Biener <rguenther@suse.de>
314
315 PR middle-end/60092
316 * gimple-low.c (lower_builtin_posix_memalign): New function.
317 (lower_stmt): Call it to lower posix_memalign in a way
318 to make alignment info accessible.
319
320 2014-02-07 Jakub Jelinek <jakub@redhat.com>
321
322 PR c++/60082
323 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
324 __builtin_setjmp_receiver.
325
326 2014-02-07 Richard Biener <rguenther@suse.de>
327
328 PR middle-end/60092
329 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
330 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
331 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
332 Handle BUILT_IN_POSIX_MEMALIGN.
333 (find_func_clobbers): Likewise.
334 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
335 (call_may_clobber_ref_p_1): Likewise.
336
337 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
338
339 PR ipa/59918
340 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
341
342 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
343
344 PR ipa/59469
345 * lto-cgraph.c (lto_output_node): Use
346 symtab_get_symbol_partitioning_class.
347 (lto_output_varpool_node): likewise.
348 (symtab_get_symbol_partitioning_class): Move here from
349 lto/lto-partition.c
350 * cgraph.h (symbol_partitioning_class): Likewise.
351 (symtab_get_symbol_partitioning_class): Declare.
352
353 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
354
355 * ggc.h (ggc_internal_cleared_alloc): New macro.
356 * vec.h (vec_safe_copy): Handle memory stats.
357 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
358 * target-globals.c (save_target_globals): Likewise.
359
360 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
361
362 PR target/60077
363 * expr.c (emit_move_resolve_push): Export; be bit more selective
364 on when to clear alias set.
365 * expr.h (emit_move_resolve_push): Declare.
366 * function.h (struct function): Add tail_call_marked.
367 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
368 * config/i386/i386-protos.h (ix86_expand_push): Remove.
369 * config/i386/i386.md (TImode move expander): De not call
370 ix86_expand_push.
371 (FP push expanders): Preserve memory attributes.
372 * config/i386/sse.md (push<mode>1): Remove.
373 * config/i386/i386.c (ix86_expand_vector_move): Handle push
374 operation.
375 (ix86_expand_push): Remove.
376 * config/i386/mmx.md (push<mode>1): Remove.
377
378 2014-02-06 Jakub Jelinek <jakub@redhat.com>
379
380 PR rtl-optimization/60030
381 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
382 lopart with paradoxical subreg before shifting it up by hprec.
383
384 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
385
386 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
387 Remove extra newline at end of file.
388 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
389 (arm_issue_rate): Handle cortexa57.
390 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
391 (cortex-a57.cortex-a53): Likewise.
392
393 2014-02-06 Jakub Jelinek <jakub@redhat.com>
394
395 PR target/59575
396 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
397 don't record in REG_FRAME_RELATED_EXPR registers not set in that
398 bitmask.
399 (arm_expand_prologue): Adjust all callers.
400 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
401 info, registers also at the lowest numbered registers side. Use
402 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
403 XEXP.
404
405 PR debug/59992
406 * var-tracking.c (adjust_mems): Before adding a SET
407 to amd->side_effects, adjust it's SET_SRC using
408 simplify_replace_fn_rtx.
409
410 2014-02-06 Alan Modra <amodra@gmail.com>
411
412 PR target/60032
413 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
414 change SDmode to DDmode when lra_in_progress.
415
416 2014-02-06 Jakub Jelinek <jakub@redhat.com>
417
418 PR middle-end/59150
419 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
420 free_data_ref on the dr first, and before goto again also set dr
421 to the next dr. For simd_lane_access, free old datarefs[i] before
422 overwriting it. For get_vectype_for_scalar_type failure, don't
423 free_data_ref if simd_lane_access.
424
425 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
426
427 PR target/60062
428 * tree.h (opts_for_fn): New inline function.
429 (opt_for_fn): Define.
430 * config/i386/i386.c (ix86_function_regparm): Use
431 opt_for_fn (decl, optimize) instead of optimize.
432
433 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
434
435 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
436 for SYMBOL_REF in large memory model.
437
438 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
439
440 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
441 and crypto support.
442 (cortex-a57): Likewise.
443 (cortex-a57.cortex-a53): Likewise.
444
445 2014-02-06 Yury Gribov <y.gribov@samsung.com>
446 Kugan Vivekanandarajah <kuganv@linaro.org>
447
448 * config/arm/arm.c (arm_vector_alignment_reachable): Check
449 unaligned_access.
450 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
451
452 2014-02-06 Richard Biener <rguenther@suse.de>
453
454 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
455 set_loop_copy and initialize_original_copy_tables.
456
457 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
458
459 * config/aarch64/aarch64-simd.md
460 (aarch64_ashr_simddi): Change QI to SI.
461
462 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
463 Jakub Jelinek <jakub@redhat.com>
464
465 PR middle-end/60013
466 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
467 of the dataflow.
468
469 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
470
471 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
472 CODE_FOR_altivec_vpku[hw]um to
473 CODE_FOR_altivec_vpku[hw]um_direct.
474 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
475 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
476 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
477 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
478
479 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
480
481 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
482 generation for -maltivec=be.
483 (altivec_vsumsws): Simplify redundant test.
484
485 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
486
487 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
488 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
489 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
490 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
491 gen_altivec_vpkuwum.
492 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
493 BYTES_BIG_ENDIAN.
494 (altivec_vpks<VI_char>ss): Likewise.
495 (altivec_vpks<VI_char>us): Likewise.
496 (altivec_vpku<VI_char>us): Likewise.
497 (altivec_vpku<VI_char>um): Likewise.
498 (altivec_vpku<VI_char>um_direct): New (copy of
499 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
500 internal use).
501 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
502 target is little endian and -maltivec=be is not specified.
503 (*altivec_vupkhs<VU_char>_direct): New (copy of
504 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
505 use).
506 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
507 target is little endian and -maltivec=be is not specified.
508 (*altivec_vupkls<VU_char>_direct): New (copy of
509 altivec_vupkls<VU_char> that always emits vupkls*, for internal
510 use).
511 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
512 little endian and -maltivec=be is not specified.
513 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
514 little endian and -maltivec=be is not specified.
515
516 2014-02-05 Richard Henderson <rth@redhat.com>
517
518 PR debug/52727
519 * combine-stack-adj.c: Revert r206943.
520 * sched-int.h (struct deps_desc): Add last_args_size.
521 * sched-deps.c (init_deps): Initialize it.
522 (sched_analyze_insn): Add OUTPUT dependencies between insns that
523 contain REG_ARGS_SIZE notes.
524
525 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
526
527 * lto-cgraph.c (asm_nodes_output): Make global.
528 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
529 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
530 parameter
531 (driver_handle_option): Handle OPT_fwpa.
532
533 2014-02-05 Jakub Jelinek <jakub@redhat.com>
534
535 PR ipa/59947
536 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
537 a comment typo and formatting issue. If odr_hash hasn't been
538 created, return vNULL and set *completep to false.
539
540 PR middle-end/57499
541 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
542 bb with no successors.
543
544 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
545
546 PR target/59718
547 * doc/invoke.texi (-march): Clarify documentation for ARM.
548 (-mtune): Likewise.
549 (-mcpu): Likewise.
550
551 2014-02-05 Richard Biener <rguenther@suse.de>
552
553 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
554 when not vectorizing because of too many alias checks.
555 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
556 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
557
558 2014-02-05 Nick Clifton <nickc@redhat.com>
559
560 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
561 accept extended registers in any mode when compiling for the
562 MN10300.
563
564 2014-02-05 Yury Gribov <y.gribov@samsung.com>
565
566 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
567 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
568 sanitization attributes.
569 (can_inline_edge_p): Likewise.
570 (sanitize_attrs_match_for_inline_p): New function.
571
572 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
573
574 * ipa-prop.c (detect_type_change): Shor circuit testing of
575 type changes on THIS pointer.
576
577 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
578
579 PR target/59777
580 * config/pa/pa.c (legitimize_tls_address): Return original address
581 if not passed a SYMBOL_REF rtx.
582 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
583 addresses.
584 (pa_emit_move_sequence): Simplify TLS source operands.
585 (pa_legitimate_constant_p): Reject all TLS constants.
586 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
587 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
588
589 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
590
591 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
592 groups when we know they are controlled by LTO.
593 * varasm.c (default_binds_local_p_1): If object is in other partition,
594 it will be resolved locally.
595
596 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
597
598 * config/host-linux.c (linux_gt_pch_use_address): Don't
599 use SSIZE_MAX because it is not always defined.
600
601 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
602
603 PR bootstrap/59913
604 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
605 threshold for pseudo splitting.
606 (update_ebb_live_info): Process call argument hard registers and
607 hard registers from insn definition too.
608 (max_small_class_regs_num): New constant.
609 (inherit_in_ebb): Update live hard regs through EBBs. Update
610 reloads_num only for small register classes. Don't split for
611 outputs of jumps.
612
613 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
614
615 PR ipa/60058
616 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
617 is non-null.
618
619 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
620
621 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
622
623 2014-02-04 Marek Polacek <polacek@redhat.com>
624
625 * gdbinit.in (pel): Define.
626
627 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
628
629 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
630 behavior.
631
632 2014-02-04 Richard Biener <rguenther@suse.de>
633
634 PR lto/59723
635 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
636 in function context local.
637 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
638 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
639 similar to LTO_imported_decl_ref.
640
641 2014-02-04 Jakub Jelinek <jakub@redhat.com>
642
643 PR tree-optimization/60002
644 * cgraphclones.c (build_function_decl_skip_args): Clear
645 DECL_LANG_SPECIFIC.
646
647 PR tree-optimization/60023
648 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
649 false to gsi_replace.
650 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
651 has been in some EH region and vec_stmt could throw, add
652 vec_stmt into the same EH region.
653 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
654 has no lhs, ignore it.
655 * internal-fn.c (expand_MASK_LOAD): Likewise.
656
657 PR ipa/60026
658 * tree-inline.c (copy_forbidden): Fail for
659 __attribute__((optimize (0))) functions.
660
661 PR other/58712
662 * omp-low.c (simd_clone_struct_copy): If from->inbranch
663 is set, copy one less argument.
664 (expand_simd_clones): Don't subtract clone_info->inbranch
665 from simd_clone_struct_alloc argument.
666
667 PR rtl-optimization/57915
668 * recog.c (simplify_while_replacing): If all unary/binary/relational
669 operation arguments are constant, attempt to simplify those.
670
671 PR middle-end/59261
672 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
673 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
674
675 2014-02-04 Richard Biener <rguenther@suse.de>
676
677 PR tree-optimization/60012
678 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
679 TBAA disambiguation to all DDRs.
680
681 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
682
683 PR target/59788
684 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
685 (LINK_SPEC): Use it for -shared, -shared-libgcc.
686
687 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
688
689 PR ipa/59882
690 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
691
692 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
693
694 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
695 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
696
697 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
698
699 PR ipa/59831
700 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
701 to figure out targets of polymorphic calls with known decl.
702 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
703 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
704 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
705 (get_polymorphic_call_info): ... here.
706 (get_polymorphic_call_info_from_invariant): New function.
707
708 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
709
710 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
711 lookup via vtable pointer; check for type consistency
712 and turn inconsitent facts into UNREACHABLE.
713 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
714 * gimple-fold.c (gimple_get_virt_method_for_vtable):
715 Do not ICE on type inconsistent querries; return UNREACHABLE
716 instead.
717
718 2014-02-03 Richard Henderson <rth@twiddle.net>
719
720 PR tree-opt/59924
721 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
722 already processed this node.
723 (normalize_one_pred_1): Pass along mark_set.
724 (normalize_one_pred): Create and destroy a pointer_set_t.
725 (normalize_one_pred_chain): Likewise.
726
727 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
728
729 PR gcov-profile/58602
730 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
731
732 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
733
734 PR ipa/59831
735 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
736 Try to devirtualize by the knowledge of virtual table pointer given by
737 aggregate propagation.
738 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
739 ipa_print_node_jump_functions): Dump also offset that
740 is relevant for polymorphic calls.
741 (determine_known_aggregate_parts): Add arg_type parameter; use it
742 instead of determining the type from pointer type.
743 (ipa_compute_jump_functions_for_edge): Update call of
744 determine_known_aggregate_parts.
745 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
746 (gimple_get_virt_method_for_binfo): ... here; simplify using
747 vtable_pointer_value_to_vtable.
748 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
749 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
750 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
751 (vtable_pointer_value_to_vtable): Break out from ...; handle also
752 POINTER_PLUS_EXPR.
753 (vtable_pointer_value_to_binfo): ... here.
754 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
755
756 2014-02-03 Teresa Johnson <tejohnson@google.com>
757
758 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
759 redef of outer loop index variable.
760
761 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
762
763 PR c++/53017
764 PR c++/59211
765 * doc/extend.texi (Function Attributes): Typo.
766
767 2014-02-03 Cong Hou <congh@google.com>
768
769 PR tree-optimization/60000
770 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
771 if the vectorized statement is a store. A store statement can only
772 appear at the end of pattern statements.
773
774 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
775
776 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
777 (ix86_option_override_internal): Default long double to 64-bit for
778 32-bit Bionic and to 128-bit for 64-bit Bionic.
779
780 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
781 TARGET_LONG_DOUBLE_128 is true.
782 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
783
784 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
785 (mlong-double-64): Negate -mlong-double-128.
786 (mlong-double-128): New option.
787
788 * config/i386/i386-c.c (ix86_target_macros): Define
789 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
790
791 * doc/invoke.texi: Document -mlong-double-128.
792
793 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
794
795 PR rtl-optimization/60024
796 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
797
798 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
799
800 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
801
802 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
803
804 PR rtl-optimization/57662
805 * sel-sched.c (code_motion_path_driver): Do not mark already not
806 existing blocks in the visiting bitmap.
807
808 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
809
810 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
811 on the insn being emitted.
812
813 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
814 Will Deacon <will.deacon@arm.com>
815
816 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
817
818 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
819
820 * config/arm/arm-tables.opt: Regenerate.
821
822 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
823
824 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
825 for vector types other than V16QImode.
826 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
827 define_expand, and call altivec_expand_vec_perm_le when producing
828 code with little endian element order.
829 (*altivec_vperm_<mode>_internal): New insn having previous
830 behavior of altivec_vperm_<mode>.
831 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
832 altivec_expand_vec_perm_le when producing code with little endian
833 element order.
834 (*altivec_vperm_<mode>_uns_internal): New insn having previous
835 behavior of altivec_vperm_<mode>_uns.
836
837 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
838
839 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
840 (altivec_vsumsws): Add handling for -maltivec=be with a little
841 endian target.
842 (altivec_vsumsws_direct): New.
843 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
844 gen_altivec_vsumsws.
845
846 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
847
848 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
849 vtable_pointer_value_to_binfo): New functions.
850 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
851 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
852
853 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
854
855 * config/nios2/nios2.md (load_got_register): Initialize GOT
856 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
857 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
858
859 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
860
861 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
862 preserverd by passthrough, do not propagate the type.
863
864 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
865
866 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
867 (mips_atomic_assign_expand_fenv): New function.
868 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
869
870 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
871
872 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
873 (__builtin_mips_set_fcsr): Likewise.
874 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
875 MIPS_USI_FTYPE_VOID.
876 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
877 (mips16_expand_set_fcsr): Likewise.
878 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
879 (mips16_set_fcsr_stub): Likewise.
880 (mips16_get_fcsr_one_only_stub): New class.
881 (mips16_set_fcsr_one_only_stub): Likewise.
882 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
883 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
884 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
885 (hard_float): New availability predicate.
886 (mips_builtins): Add get_fcsr and set_fcsr.
887 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
888 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
889 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
890 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
891 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
892 patterns.
893
894 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
895
896 * config/mips/mips.c (mips_one_only_stub): New class.
897 (mips_need_mips16_rdhwr_p): Replace with...
898 (mips16_rdhwr_stub): ...this new variable.
899 (mips16_stub_call_address): New function.
900 (mips16_rdhwr_one_only_stub): New class.
901 (mips_expand_thread_pointer): Use mips16_stub_call_address.
902 (mips_output_mips16_rdhwr): Delete.
903 (mips_finish_stub): New function.
904 (mips_code_end): Use it to handle rdhwr stubs.
905
906 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
907
908 PR target/60017
909 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
910 when calculating size of integer atomic types.
911
912 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
913
914 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
915
916 2014-02-01 Jakub Jelinek <jakub@redhat.com>
917
918 PR tree-optimization/60003
919 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
920 * profile.c (branch_prob): Use gimple_call_builtin_p
921 to check for BUILT_IN_SETJMP_RECEIVER.
922 * tree-inline.c (copy_bb): Call notice_special_calls.
923
924 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
925
926 PR bootstrap/59985
927 * lra-constraints.c (process_alt_operands): Update reload_sum only
928 on the first pass.
929
930 2014-01-31 Richard Henderson <rth@redhat.com>
931
932 PR middle-end/60004
933 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
934 until after else_eh is processed.
935
936 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
937
938 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
939 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
940 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
941 in smmintrin.h, remove them.
942 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
943 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
944 * config/i386/i386.md (ROUND_SAE): Fix value.
945 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
946 (const48_operand): New.
947 * config/i386/subst.md (round), (round_expand): Use
948 const_4_or_8_to_11_operand.
949 (round_saeonly), (round_saeonly_expand): Use const48_operand.
950
951 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
952
953 * config/i386/constraints.md (Yk): Swap meaning with k.
954 * config/i386/i386.md (movhi_internal): Change Yk to k.
955 (movqi_internal): Ditto.
956 (*k<logic><mode>): Ditto.
957 (*andhi_1): Ditto.
958 (*andqi_1): Ditto.
959 (kandn<mode>): Ditto.
960 (*<code>hi_1): Ditto.
961 (*<code>qi_1): Ditto.
962 (kxnor<mode>): Ditto.
963 (kortestzhi): Ditto.
964 (kortestchi): Ditto.
965 (kunpckhi): Ditto.
966 (*one_cmplhi2_1): Ditto.
967 (*one_cmplqi2_1): Ditto.
968 * config/i386/sse.md (): Change k to Yk.
969 (avx512f_load<mode>_mask): Ditto.
970 (avx512f_blendm<mode>): Ditto.
971 (avx512f_store<mode>_mask): Ditto.
972 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
973 (avx512f_storedqu<mode>_mask): Ditto.
974 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
975 Ditto.
976 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
977 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
978 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
979 (avx512f_maskcmp<mode>3): Ditto.
980 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
981 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
982 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
983 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
984 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
985 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
986 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
987 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
988 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
989 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
990 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
991 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
992 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
993 (vec_extract_lo_<mode>_maskm): Ditto.
994 (vec_extract_hi_<mode>_maskm): Ditto.
995 (avx512f_vternlog<mode>_mask): Ditto.
996 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
997 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
998 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
999 (avx512f_<code>v8div16qi2_mask): Ditto.
1000 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1001 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1002 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1003 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1004 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1005 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1006 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1007 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1008 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1009 (avx512cd_maskb_vec_dupv8di): Ditto.
1010 (avx512cd_maskw_vec_dupv16si): Ditto.
1011 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1012 (avx512f_vpermi2var<mode>3_mask): Ditto.
1013 (avx512f_vpermi2var<mode>3_mask): Ditto.
1014 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1015 (*avx512f_gathersi<mode>): Ditto.
1016 (*avx512f_gathersi<mode>_2): Ditto.
1017 (*avx512f_gatherdi<mode>): Ditto.
1018 (*avx512f_gatherdi<mode>_2): Ditto.
1019 (*avx512f_scattersi<mode>): Ditto.
1020 (*avx512f_scatterdi<mode>): Ditto.
1021 (avx512f_compress<mode>_mask): Ditto.
1022 (avx512f_compressstore<mode>_mask): Ditto.
1023 (avx512f_expand<mode>_mask): Ditto.
1024 * config/i386/subst.md (mask): Change k to Yk.
1025 (mask_scalar_merge): Ditto.
1026 (sd): Ditto.
1027
1028 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1029
1030 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1031
1032 2014-01-31 Richard Biener <rguenther@suse.de>
1033
1034 PR middle-end/59990
1035 * builtins.c (fold_builtin_memory_op): Make sure to not
1036 use a floating-point mode or a boolean or enumeral type for
1037 the copy operation.
1038
1039 2014-01-30 DJ Delorie <dj@redhat.com>
1040
1041 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1042 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1043 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1044 whenever main() has an epilogue.
1045
1046 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1047
1048 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1049 unused variable "field".
1050 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1051 (vsx_mergeh_<mode>): Likewise.
1052 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1053 (altivec_vmrghh): Likewise.
1054 (altivec_vmrghw): Likewise.
1055 (altivec_vmrglb): Likewise.
1056 (altivec_vmrglh): Likewise.
1057 (altivec_vmrglw): Likewise.
1058 (altivec_vspltb): Add missing uses.
1059 (altivec_vsplth): Likewise.
1060 (altivec_vspltw): Likewise.
1061 (altivec_vspltsf): Likewise.
1062
1063 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1064
1065 PR target/59923
1066 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1067 frame related instructions.
1068
1069 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1070
1071 PR rtl-optimization/59959
1072 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1073 any reload of register whose subreg is invalid.
1074
1075 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1076
1077 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1078 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1079 Add missing return type - void.
1080
1081 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1082
1083 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1084 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1085 remove element index adjustment for endian (now handled in vsx.md
1086 and altivec.md).
1087 (altivec_expand_vec_perm_const): Use
1088 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1089 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1090 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1091 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1092 define_expand and a new define_insn *altivec_vspltb_internal;
1093 adjust for -maltivec=be on a little endian target.
1094 (altivec_vspltb_direct): New.
1095 (altivec_vsplth): Divide into a define_expand and a new
1096 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1097 little endian target.
1098 (altivec_vsplth_direct): New.
1099 (altivec_vspltw): Divide into a define_expand and a new
1100 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1101 little endian target.
1102 (altivec_vspltw_direct): New.
1103 (altivec_vspltsf): Divide into a define_expand and a new
1104 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1105 a little endian target.
1106
1107 2014-01-30 Richard Biener <rguenther@suse.de>
1108
1109 PR tree-optimization/59993
1110 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1111 can propagate form the earlier stmt and avoid the transform
1112 when the intermediate result is needed.
1113
1114 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1115
1116 * README.Portability: Fix typo.
1117
1118 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1119
1120 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1121 comparison_operator with ordered_comparison_operator.
1122
1123 2014-01-30 Nick Clifton <nickc@redhat.com>
1124
1125 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1126 Rename to mn10300_store_multiple_regs.
1127 * config/mn10300/mn10300.c: Likewise.
1128 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1129 store_multiple_regs.
1130 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1131 Call mn10300_store_multiple_regs.
1132
1133 2014-01-30 Nick Clifton <nickc@redhat.com>
1134 DJ Delorie <dj@redhat.com>
1135
1136 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1137 %fp 2 to keep registers after it properly word-aligned.
1138 (rl78_alloc_physical_registers_umul): Handle the case where both
1139 input operands are the same.
1140
1141 2014-01-30 Richard Biener <rguenther@suse.de>
1142
1143 PR tree-optimization/59903
1144 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1145 check properly.
1146
1147 2014-01-30 Jason Merrill <jason@redhat.com>
1148
1149 PR c++/59633
1150 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1151
1152 PR c++/59645
1153 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1154
1155 2014-01-30 Richard Biener <rguenther@suse.de>
1156
1157 PR tree-optimization/59951
1158 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1159
1160 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1161
1162 PR target/59784
1163 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1164 SFmode to DFmode case.
1165
1166 2014-01-29 DJ Delorie <dj@redhat.com>
1167
1168 * config/msp430/msp430.opt (-minrt): New.
1169 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1170 if -minrt given.
1171 (ENDFILE_SPEC): Likewise.
1172
1173 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1174
1175 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1176 (estimate_function_body_sizes): Use it.
1177
1178 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1179
1180 PR c++/58561
1181 * dwarf2out.c (is_cxx_auto): New.
1182 (is_base_type): Use it.
1183 (gen_type_die_with_usage): Likewise.
1184
1185 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1186
1187 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1188 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1189 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1190 -maltivec=be with LE targets.
1191 (vsx_mergeh_<mode>): Likewise.
1192 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1193 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1194 (altivec_vmrghb): Replace with define_expand and new
1195 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1196 (altivec_vmrghb_direct): New define_insn.
1197 (altivec_vmrghh): Replace with define_expand and new
1198 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1199 (altivec_vmrghh_direct): New define_insn.
1200 (altivec_vmrghw): Replace with define_expand and new
1201 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1202 (altivec_vmrghw_direct): New define_insn.
1203 (*altivec_vmrghsf): Adjust for endianness.
1204 (altivec_vmrglb): Replace with define_expand and new
1205 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1206 (altivec_vmrglb_direct): New define_insn.
1207 (altivec_vmrglh): Replace with define_expand and new
1208 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1209 (altivec_vmrglh_direct): New define_insn.
1210 (altivec_vmrglw): Replace with define_expand and new
1211 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1212 (altivec_vmrglw_direct): New define_insn.
1213 (*altivec_vmrglsf): Adjust for endianness.
1214 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1215 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1216 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1217 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1218 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1219 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1220 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1221 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1222
1223 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1224
1225 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1226 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1227 whitespace.
1228
1229 2014-01-29 Richard Biener <rguenther@suse.de>
1230
1231 PR tree-optimization/58742
1232 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1233 associate_pointerplus_align.
1234 (associate_pointerplus_diff): New function.
1235 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1236 and associate_pointerplus_diff.
1237
1238 2014-01-29 Richard Biener <rguenther@suse.de>
1239
1240 * lto-streamer.h (LTO_major_version): Bump to 3.
1241 (LTO_minor_version): Reset to 0.
1242
1243 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1244
1245 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1246 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1247 (arm_file_start): Generate correct asm header for armv7ve.
1248 * config/arm/bpabi.h: Add multilib support for armv7ve.
1249 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1250 and cortex-a15 to armv7ve.
1251 * config/arm/t-aprofile: Add multilib support for armv7ve.
1252 * doc/invoke.texi: Document -march=armv7ve.
1253
1254 2014-01-29 Richard Biener <rguenther@suse.de>
1255
1256 PR tree-optimization/58742
1257 * tree-ssa-forwprop.c (associate_plusminus): Return true
1258 if we changed sth, defer EH cleanup to ...
1259 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1260 (simplify_mult): New function.
1261
1262 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1263
1264 PR middle-end/59917
1265 PR tree-optimization/59920
1266 * tree.c (build_common_builtin_nodes): Remove
1267 __builtin_setjmp_dispatcher initialization.
1268 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1269 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1270 instead of gsi_after_labels + manually skipping debug stmts.
1271 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1272 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1273 * tree-inline.c (copy_edges_for_bb): Remove
1274 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1275 argument. Ignore computed_goto_p stmts. Don't call
1276 make_abnormal_goto_edges. If a call might need abnormal edges
1277 for non-local gotos, see if it already has an edge to
1278 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1279 with true argument, don't do anything then, otherwise add
1280 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1281 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1282 caller.
1283 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1284 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1285 (lower_stmt): Don't set data->calls_builtin_setjmp.
1286 (lower_builtin_setjmp): Adjust comment.
1287 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1288 * tree-cfg.c (found_computed_goto): Remove.
1289 (factor_computed_gotos): Remove.
1290 (make_goto_expr_edges): Return bool, true for computed gotos.
1291 Don't call make_abnormal_goto_edges.
1292 (build_gimple_cfg): Don't set found_computed_goto, don't call
1293 factor_computed_gotos.
1294 (computed_goto_p): No longer static.
1295 (make_blocks): Don't set found_computed_goto.
1296 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1297 (make_edges): If make_goto_expr_edges returns true, push bb
1298 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1299 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1300 vector. Record mapping between bbs and OpenMP regions if there
1301 are any, adjust make_gimple_omp_edges caller. Call
1302 handle_abnormal_edges.
1303 (make_abnormal_goto_edges): Remove.
1304 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1305 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1306 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1307 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1308 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1309 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1310 filling *region also set *region_idx to (*region)->entry->index.
1311
1312 PR other/58712
1313 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1314 For REGs set ORIGINAL_REGNO.
1315
1316 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1317
1318 * doc/md.texi: Mention that a target shouldn't implement
1319 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1320 than hi/lo pair.
1321
1322 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1323
1324 PR tree-optimization/59594
1325 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1326 a copy of the datarefs vector rather than the vector itself.
1327
1328 2014-01-28 Jason Merrill <jason@redhat.com>
1329
1330 PR c++/53756
1331 * dwarf2out.c (auto_die): New static.
1332 (gen_type_die_with_usage): Handle C++1y 'auto'.
1333 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1334 on definition.
1335
1336 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1337
1338 PR target/59672
1339 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1340 (SPEC_X32): Likewise.
1341 (SPEC_64): Likewise.
1342 * config/i386/i386.c (ix86_option_override_internal): Turn off
1343 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1344 for TARGET_16BIT.
1345 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1346 * config/i386/i386.h (TARGET_16BIT): New macro.
1347 (TARGET_16BIT_P): Likewise.
1348 * config/i386/i386.opt: Add m16.
1349 * doc/invoke.texi: Document -m16.
1350
1351 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1352
1353 PR preprocessor/59935
1354 * input.c (location_get_source_line): Bail out on when line number
1355 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1356
1357 2014-01-28 Richard Biener <rguenther@suse.de>
1358
1359 PR tree-optimization/58742
1360 * tree-ssa-forwprop.c (associate_plusminus): Handle
1361 pointer subtraction of the form (T)(P + A) - (T)P.
1362
1363 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1364
1365 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1366 at const_int_cost.
1367
1368 2014-01-28 Richard Biener <rguenther@suse.de>
1369
1370 Revert
1371 2014-01-28 Richard Biener <rguenther@suse.de>
1372
1373 PR rtl-optimization/45364
1374 PR rtl-optimization/59890
1375 * var-tracking.c (local_get_addr_clear_given_value): Handle
1376 already cleared slot.
1377 (val_reset): Handle not allocated local_get_addr_cache.
1378 (vt_find_locations): Use post-order on the inverted CFG.
1379
1380 2014-01-28 Richard Biener <rguenther@suse.de>
1381
1382 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1383
1384 2014-01-28 Richard Biener <rguenther@suse.de>
1385
1386 PR rtl-optimization/45364
1387 PR rtl-optimization/59890
1388 * var-tracking.c (local_get_addr_clear_given_value): Handle
1389 already cleared slot.
1390 (val_reset): Handle not allocated local_get_addr_cache.
1391 (vt_find_locations): Use post-order on the inverted CFG.
1392
1393 2014-01-28 Alan Modra <amodra@gmail.com>
1394
1395 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1396 * configure.ac <recursive call for build != host>: Define
1397 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1398 and LD_FOR_BUILD too.
1399 * configure: Regenerate.
1400
1401 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1402
1403 * config/i386/i386.c (get_builtin_code_for_version): Separate
1404 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1405 Broadwell from Haswell.
1406
1407 2014-01-27 Steve Ellcey <sellcey@mips.com>
1408
1409 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1410 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1411 * config/mips/mips.c (mips_option_override): Change setting
1412 of TARGET_DSP.
1413 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1414 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1415 Change from Mask to Var.
1416
1417 2014-01-27 Jeff Law <law@redhat.com>
1418
1419 * ipa-inline.c (inline_small_functions): Fix typo.
1420
1421 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1422
1423 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1424 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1425 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1426 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1427 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1428 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1429 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1430 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1431 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1432 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1433 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1434 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1435 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1436 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1437 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1438 (_mm512_storeu_epi64): Ditto.
1439 (_mm512_cmpge_epi32_mask): Ditto.
1440 (_mm512_cmpge_epu32_mask): Ditto.
1441 (_mm512_cmpge_epi64_mask): Ditto.
1442 (_mm512_cmpge_epu64_mask): Ditto.
1443 (_mm512_cmple_epi32_mask): Ditto.
1444 (_mm512_cmple_epu32_mask): Ditto.
1445 (_mm512_cmple_epi64_mask): Ditto.
1446 (_mm512_cmple_epu64_mask): Ditto.
1447 (_mm512_cmplt_epi32_mask): Ditto.
1448 (_mm512_cmplt_epu32_mask): Ditto.
1449 (_mm512_cmplt_epi64_mask): Ditto.
1450 (_mm512_cmplt_epu64_mask): Ditto.
1451 (_mm512_cmpneq_epi32_mask): Ditto.
1452 (_mm512_cmpneq_epu32_mask): Ditto.
1453 (_mm512_cmpneq_epi64_mask): Ditto.
1454 (_mm512_cmpneq_epu64_mask): Ditto.
1455 (_mm512_expand_pd): Ditto.
1456 (_mm512_expand_ps): Ditto.
1457 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1458 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1459 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1460 * config/i386/i386.c (ix86_builtins): Add
1461 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1462 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1463 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1464 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1465 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1466 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1467 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1468 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1469 IX86_BUILTIN_PMOVUSQW512_MEM.
1470 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1471 __builtin_ia32_pmovsqd512mem_mask,
1472 __builtin_ia32_pmovqd512mem_mask,
1473 __builtin_ia32_pmovusqw512mem_mask,
1474 __builtin_ia32_pmovsqw512mem_mask,
1475 __builtin_ia32_pmovqw512mem_mask,
1476 __builtin_ia32_pmovusdw512mem_mask,
1477 __builtin_ia32_pmovsdw512mem_mask,
1478 __builtin_ia32_pmovdw512mem_mask,
1479 __builtin_ia32_pmovqb512mem_mask,
1480 __builtin_ia32_pmovusqb512mem_mask,
1481 __builtin_ia32_pmovsqb512mem_mask,
1482 __builtin_ia32_pmovusdb512mem_mask,
1483 __builtin_ia32_pmovsdb512mem_mask,
1484 __builtin_ia32_pmovdb512mem_mask.
1485 (bdesc_args): Add __builtin_ia32_expanddf512,
1486 __builtin_ia32_expandsf512.
1487 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1488 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1489 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1490 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1491 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1492 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1493 (avx512f_<code>v8div16qi2_mask_store): This.
1494 (avx512f_expand<mode>): New.
1495
1496 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1497
1498 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1499 New.
1500 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1501 (_mm512_prefetch_i32scatter_pd): Ditto.
1502 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1503 (_mm512_prefetch_i64scatter_pd): Ditto.
1504 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1505 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1506 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1507 (_mm512_prefetch_i32scatter_ps): Ditto.
1508 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1509 (_mm512_prefetch_i64scatter_ps): Ditto.
1510 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1511 * config/i386/i386-builtin-types.def: Define
1512 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1513 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1514 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1515 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1516 IX86_BUILTIN_SCATTERPFQPD.
1517 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1518 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1519 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1520 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1521 __builtin_ia32_scatterpfqps.
1522 (ix86_expand_builtin): Expand new built-ins.
1523 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1524 fix memory access data type.
1525 (*avx512pf_gatherpf<mode>_mask): Ditto.
1526 (*avx512pf_gatherpf<mode>): Ditto.
1527 (avx512pf_scatterpf<mode>): Ditto.
1528 (*avx512pf_scatterpf<mode>_mask): Ditto.
1529 (*avx512pf_scatterpf<mode>): Ditto.
1530 (GATHER_SCATTER_SF_MEM_MODE): New.
1531 (avx512pf_gatherpf<mode>df): Ditto.
1532 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1533 (*avx512pf_scatterpf<mode>df): Ditto.
1534
1535 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1536
1537 PR bootstrap/59934
1538 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1539 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1540 reached.
1541
1542 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1543
1544 * common/config/arm/arm-common.c
1545 (arm_rewrite_mcpu): Handle multiple names.
1546 * config/arm/arm.h
1547 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1548
1549 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1550
1551 * gimple-builder.h (create_gimple_tmp): Delete.
1552
1553 2014-01-27 Christian Bruel <christian.bruel@st.com>
1554
1555 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1556 words comparisons.
1557
1558 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1559
1560 * config/pa/pa.md (call): Generate indirect long calls to non-local
1561 functions when outputing 32-bit code.
1562 (call_value): Likewise except for special call to buggy powf function.
1563
1564 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1565 portable runtime and PIC indirect calls.
1566 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1567 and PIC call sequences. Use ldo instead of blr to set return register
1568 in PIC call sequence.
1569
1570 2014-01-25 Walter Lee <walt@tilera.com>
1571
1572 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1573 avoid clobbering a live register.
1574
1575 2014-01-25 Walter Lee <walt@tilera.com>
1576
1577 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1578 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1579 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1580 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1581
1582 2014-01-25 Walter Lee <walt@tilera.com>
1583
1584 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1585 arguments on even registers.
1586 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1587 STACK_BOUNDARY.
1588 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1589 (BIGGEST_ALIGNMENT): Ditto.
1590 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1591
1592 2014-01-25 Walter Lee <walt@tilera.com>
1593
1594 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1595 insns before bundling.
1596 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1597
1598 2014-01-25 Walter Lee <walt@tilera.com>
1599
1600 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1601 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1602 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1603
1604 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1605
1606 * config/mips/constraints.md (kl): Delete.
1607 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1608 define expands, using...
1609 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1610 instructions for MIPS16.
1611 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1612 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1613
1614 2014-01-25 Walter Lee <walt@tilera.com>
1615
1616 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1617 (clzdi2): Ditto.
1618 (ffsdi2): Ditto.
1619
1620 2014-01-25 Walter Lee <walt@tilera.com>
1621
1622 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1623 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1624
1625 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1626
1627 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1628 Handle XOR.
1629
1630 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1631
1632 * print-rtl.c (in_call_function_usage): New var.
1633 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1634 EXPR_LIST mode as mode and not as reg note name.
1635
1636 PR middle-end/59561
1637 * cfgloopmanip.c (copy_loop_info): If
1638 loop->warned_aggressive_loop_optimizations, make sure
1639 the flag is set in target loop too.
1640
1641 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1642
1643 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1644 flag_cilkplus.
1645 * builtins.def: Likewise.
1646 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1647 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1648 * ira.c (ira_setup_eliminable_regset): Likewise.
1649 * omp-low.c (gate_expand_omp): Likewise.
1650 (execute_lower_omp): Likewise.
1651 (diagnose_sb_0): Likewise.
1652 (gate_diagnose_omp_blocks): Likewise.
1653 (simd_clone_clauses_extract): Likewise.
1654 (gate): Likewise.
1655
1656 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1657
1658 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1659 correction for little endian...
1660 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1661 here.
1662
1663 2014-01-24 Jeff Law <law@redhat.com>
1664
1665 PR tree-optimization/59919
1666 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1667 for non-returning calls.
1668
1669 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1670
1671 * common/config/aarch64/aarch64-common.c
1672 (aarch64_rewrite_mcpu): Handle multiple names.
1673 * config/aarch64/aarch64.h
1674 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1675
1676 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1677
1678 * input.c (add_file_to_cache_tab): Handle the case where fopen
1679 returns NULL.
1680
1681 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1682
1683 PR target/59929
1684 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1685 from push operand if code of push isn't PRE_DEC.
1686
1687 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1688
1689 PR target/59909
1690 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1691 -mquad-memory-atomic. Update -mquad-memory documentation to say
1692 it is only used for non-atomic loads/stores.
1693
1694 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1695 -mquad-memory or -mquad-memory-atomic switches.
1696
1697 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1698 -mquad-memory-atomic to ISA 2.07 support.
1699
1700 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1701 to separate support of normal quad word memory operations (ldq, stq)
1702 from the atomic quad word memory operations.
1703
1704 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1705 support to separate non-atomic quad word operations from atomic
1706 quad word operations. Disable non-atomic quad word operations in
1707 little endian mode so that we don't have to swap words after the
1708 load and before the store.
1709 (quad_load_store_p): Add comment about atomic quad word support.
1710 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1711 options printed with -mdebug=reg.
1712
1713 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1714 -mquad-memory-atomic as the test for whether we have quad word
1715 atomic instructions.
1716 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1717 or -mp8-vector are used, allow byte/half-word atomic operations.
1718
1719 * config/rs6000/sync.md (load_lockedti): Insure that the address
1720 is a proper indexed or indirect address for the lqarx instruction.
1721 On little endian systems, swap the hi/lo registers after the lqarx
1722 instruction.
1723 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1724 insure the address is valid for the lqarx instruction.
1725 (store_conditionalti): Insure that the address is a proper indexed
1726 or indirect address for the stqcrx. instruction. On little endian
1727 systems, swap the hi/lo registers before doing the stqcrx.
1728 instruction.
1729 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1730 insure the address is valid for the stqcrx. instruction.
1731
1732 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1733 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1734 type of quad memory support is available.
1735
1736 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1737
1738 PR regression/59915
1739 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1740 there is a danger of looping.
1741
1742 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1743
1744 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1745 force flag_ira_loop_pressure if set via command line.
1746
1747 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1748
1749 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1750 (ashr_simd): New builtin handling DI mode.
1751 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1752 (aarch64_sshr_simddi): New match pattern.
1753 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1754 (vshrd_n_s64): Likewise.
1755 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1756
1757 2014-01-23 Nick Clifton <nickc@redhat.com>
1758
1759 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1760 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1761 favour of mcu specific scripts.
1762 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1763 430x multilibs.
1764
1765 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1766 Alex Velenko <Alex.Velenko@arm.com>
1767
1768 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1769 (vaddv_s16): Likewise.
1770 (vaddv_s32): Likewise.
1771 (vaddv_u8): Likewise.
1772 (vaddv_u16): Likewise.
1773 (vaddv_u32): Likewise.
1774 (vaddvq_s8): Likewise.
1775 (vaddvq_s16): Likewise.
1776 (vaddvq_s32): Likewise.
1777 (vaddvq_s64): Likewise.
1778 (vaddvq_u8): Likewise.
1779 (vaddvq_u16): Likewise.
1780 (vaddvq_u32): Likewise.
1781 (vaddvq_u64): Likewise.
1782 (vaddv_f32): Likewise.
1783 (vaddvq_f32): Likewise.
1784 (vaddvq_f64): Likewise.
1785 (vmaxv_f32): Likewise.
1786 (vmaxv_s8): Likewise.
1787 (vmaxv_s16): Likewise.
1788 (vmaxv_s32): Likewise.
1789 (vmaxv_u8): Likewise.
1790 (vmaxv_u16): Likewise.
1791 (vmaxv_u32): Likewise.
1792 (vmaxvq_f32): Likewise.
1793 (vmaxvq_f64): Likewise.
1794 (vmaxvq_s8): Likewise.
1795 (vmaxvq_s16): Likewise.
1796 (vmaxvq_s32): Likewise.
1797 (vmaxvq_u8): Likewise.
1798 (vmaxvq_u16): Likewise.
1799 (vmaxvq_u32): Likewise.
1800 (vmaxnmv_f32): Likewise.
1801 (vmaxnmvq_f32): Likewise.
1802 (vmaxnmvq_f64): Likewise.
1803 (vminv_f32): Likewise.
1804 (vminv_s8): Likewise.
1805 (vminv_s16): Likewise.
1806 (vminv_s32): Likewise.
1807 (vminv_u8): Likewise.
1808 (vminv_u16): Likewise.
1809 (vminv_u32): Likewise.
1810 (vminvq_f32): Likewise.
1811 (vminvq_f64): Likewise.
1812 (vminvq_s8): Likewise.
1813 (vminvq_s16): Likewise.
1814 (vminvq_s32): Likewise.
1815 (vminvq_u8): Likewise.
1816 (vminvq_u16): Likewise.
1817 (vminvq_u32): Likewise.
1818 (vminnmv_f32): Likewise.
1819 (vminnmvq_f32): Likewise.
1820 (vminnmvq_f64): Likewise.
1821
1822 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1823
1824 * config/aarch64/aarch64-simd.md
1825 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1826 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1827 (*aarch64_mul3_elt<mode>): Likewise.
1828 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1829 (*aarch64_mul3_elt_to_64v2df): Likewise.
1830 (*aarch64_mla_elt<mode>): Likewise.
1831 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1832 (*aarch64_mls_elt<mode>): Likewise.
1833 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1834 (*aarch64_fma4_elt<mode>): Likewise.
1835 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1836 (*aarch64_fma4_elt_to_64v2df): Likewise.
1837 (*aarch64_fnma4_elt<mode>): Likewise.
1838 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1839 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1840 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1841 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1842 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1843 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1844 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1845 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1846 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1847
1848 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1849
1850 * config/aarch64/aarch64-simd.md
1851 (aarch64_be_checked_get_lane<mode>): New define_expand.
1852 * config/aarch64/aarch64-simd-builtins.def
1853 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1854 New builtin definition.
1855 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1856 Use new safe be builtin.
1857
1858 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1859
1860 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1861 New define_insn.
1862 (aarch64_be_st1<mode>): Likewise.
1863 (aarch_ld1<VALL:mode>): Define_expand modified.
1864 (aarch_st1<VALL:mode>): Likewise.
1865 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1866 (UNSPEC_ST1): Likewise.
1867
1868 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1869
1870 * config/microblaze/microblaze.md: Add trap insn and attribute
1871
1872 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1873
1874 PR preprocessor/58580
1875 * input.h (location_get_source_line): Take an additional line_size
1876 parameter.
1877 (void diagnostics_file_cache_fini): Declare new function.
1878 * input.c (struct fcache): New type.
1879 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1880 New static constants.
1881 (diagnostic_file_cache_init, total_lines_num)
1882 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1883 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1884 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1885 (get_next_line, read_next_line, goto_next_line, read_line_num):
1886 New static function definitions.
1887 (diagnostic_file_cache_fini): New function.
1888 (location_get_source_line): Take an additional output line_len
1889 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1890 read_line_num.
1891 * diagnostic.c (diagnostic_finish): Call
1892 diagnostic_file_cache_fini.
1893 (adjust_line): Take an additional input parameter for the length
1894 of the line, rather than calculating it with strlen.
1895 (diagnostic_show_locus): Adjust the use of
1896 location_get_source_line and adjust_line with respect to their new
1897 signature. While displaying a line now, do not stop at the first
1898 null byte. Rather, display the zero byte as a space and keep
1899 going until we reach the size of the line.
1900 * Makefile.in: Add vec.o to OBJS-libcommon
1901
1902 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1903 Ilya Tocar <ilya.tocar@intel.com>
1904
1905 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1906 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1907 (__builtin_ia32_kmov16): Ditto.
1908 * config/i386/i386.md (UNSPEC_KMOV): New.
1909 (kmovw): Ditto.
1910
1911 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1912
1913 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1914 (_mm512_storeu_si512): Ditto.
1915
1916 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1917
1918 PR target/52125
1919 * rtl.h (get_referenced_operands): Declare.
1920 * recog.c (get_referenced_operands): New function.
1921 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1922 operands have been referenced when recording LO_SUM references.
1923
1924 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1925
1926 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1927
1928 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1929
1930 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1931 Enable for generic and recent AMD targets.
1932
1933 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1934
1935 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1936 ARG_SIZE note when adjustment was eliminated.
1937
1938 2014-01-22 Jeff Law <law@redhat.com>
1939
1940 PR tree-optimization/59597
1941 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1942 in file. Accept new argument REGISTERING and use it to modify
1943 dump output appropriately.
1944 (register_jump_thread): Corresponding changes.
1945 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1946 thread paths involving joiner blocks. Add code to dump cancelled
1947 jump threading paths.
1948
1949 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1950
1951 PR rtl-optimization/59477
1952 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1953 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1954
1955 2014-01-22 Tom Tromey <tromey@redhat.com>
1956
1957 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1958 PARAMS.
1959 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1960
1961 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1962
1963 PR rtl-optimization/59896
1964 * lra-constraints.c (process_alt_operands): Check unused note for
1965 matched operands of insn with no output reloads.
1966
1967 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1968
1969 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1970 (mips_move_from_gpr_cost): Likewise.
1971
1972 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1973
1974 PR rtl-optimization/59858
1975 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1976 ira_class_hard_regs_num.
1977 (process_alt_operands): Increase reject for dying matched operand.
1978
1979 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1980
1981 PR target/59003
1982 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1983 smaller than size, perform several stores or loads and stores
1984 at dst + count - size to store or copy all of size bytes, rather
1985 than just last modesize bytes.
1986
1987 2014-01-20 DJ Delorie <dj@redhat.com>
1988
1989 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1990 that CLOBBERs are REGs before propogating their values.
1991
1992 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1993
1994 PR middle-end/59789
1995 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1996 (cgraph_inline_failed_type): New function.
1997 * cgraph.h (DEFCIFCODE): Add type.
1998 (cgraph_inline_failed_type_t): New enum.
1999 (cgraph_inline_failed_type): New prototype.
2000 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2001 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2002 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2003 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2004 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2005 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2006 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2007 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2008 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2009 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2010 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2011 OPTIMIZATION_MISMATCH.
2012 * tree-inline.c (expand_call_inline): Emit errors during
2013 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2014
2015 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2016
2017 PR target/59685
2018 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2019 mode attribute in insn output.
2020
2021 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2022
2023 * output.h (output_constant): Delete.
2024 * varasm.c (output_constant): Make private.
2025
2026 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2027
2028 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2029
2030 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2031
2032 PR middle-end/59860
2033 * tree.h (fold_builtin_strcat): New prototype.
2034 * builtins.c (fold_builtin_strcat): No longer static. Add len
2035 argument, if non-NULL, don't call c_strlen. Optimize
2036 directly into __builtin_memcpy instead of __builtin_strcpy.
2037 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2038 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2039
2040 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2041
2042 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2043 for SImode_address_operand operands, having only a REG argument.
2044
2045 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2046
2047 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2048 loader name using mbig-endian.
2049 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2050
2051 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2052
2053 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2054 (-mtune): Likewise.
2055 (-mcpu): Likewise.
2056
2057 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2058
2059 * config/aarch64/aarch64-protos.h
2060 (aarch64_cannot_change_mode_class_ptr): Declare.
2061 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2062 aarch64_cannot_change_mode_class_ptr): New.
2063 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2064 backend hook aarch64_cannot_change_mode_class.
2065
2066 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2067
2068 * common/config/aarch64/aarch64-common.c
2069 (aarch64_handle_option): Don't handle any option order logic here.
2070 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2071 selected_cpu, warn on architecture version mismatch.
2072 (aarch64_override_options): Fix parsing order for option strings.
2073
2074 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2075 Iain Sandoe <iain@codesourcery.com>
2076
2077 PR bootstrap/59496
2078 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2079 warning. Amend comment to reflect current functionality.
2080
2081 2014-01-20 Richard Biener <rguenther@suse.de>
2082
2083 PR middle-end/59860
2084 * builtins.c (fold_builtin_strcat): Remove case better handled
2085 by tree-ssa-strlen.c.
2086
2087 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2088
2089 * config/aarch64/aarch64.opt
2090 (mcpu, march, mtune): Make case-insensitive.
2091
2092 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2093
2094 PR target/59880
2095 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2096 if operands[1] is a REG or ZERO_EXTEND of a REG.
2097
2098 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2099
2100 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2101
2102 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2103
2104 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2105 long non-pic millicode calls.
2106
2107 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2108
2109 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2110
2111 2014-01-19 Kito Cheng <kito@0xlab.org>
2112
2113 * builtins.c (expand_movstr): Check movstr expand done or fail.
2114
2115 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2116 H.J. Lu <hongjiu.lu@intel.com>
2117
2118 PR target/59379
2119 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2120 to DImode for zero-extended addresses.
2121
2122 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2123
2124 PR rtl-optimization/57763
2125 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2126 on the new indirect jump_insn and increment LABEL_NUSES (label).
2127
2128 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2129
2130 PR bootstrap/59580
2131 PR bootstrap/59583
2132 * config.gcc (x86_archs): New variable.
2133 (x86_64_archs): Likewise.
2134 (x86_cpus): Likewise.
2135 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2136 --with-arch/--with-cpu= options.
2137 Support --with-arch=/--with-cpu={nehalem,westmere,
2138 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2139
2140 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2141
2142 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2143 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2144
2145 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2146
2147 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2148
2149 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2150
2151 PR target/58944
2152 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2153 clear cpp_get_options (parse_in)->warn_unused_macros for
2154 ix86_target_macros_internal with cpp_define.
2155
2156 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2157
2158 * jump.c (delete_related_insns): Keep (use (insn))s.
2159 * reorg.c (redundant_insn): Check for barriers too.
2160
2161 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2162
2163 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2164
2165 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2166
2167 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2168 call to $$dyncall when TARGET_LONG_CALLS is true.
2169
2170 2014-01-17 Jeff Law <law@redhat.com>
2171
2172 * ree.c (combine_set_extension): Temporarily disable test for
2173 changing number of hard registers.
2174
2175 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2176
2177 PR middle-end/58125
2178 * ipa-inline-analysis.c (inline_free_summary):
2179 Do not free summary of aliases.
2180
2181 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2182
2183 PR middle-end/59706
2184 * gimplify.c (gimplify_expr): Use create_tmp_var
2185 instead of create_tmp_var_raw. If cond doesn't have
2186 integral type, don't add the IFN_ANNOTATE builtin at all.
2187
2188 2014-01-17 Martin Jambor <mjambor@suse.cz>
2189
2190 PR ipa/59736
2191 * ipa-cp.c (prev_edge_clone): New variable.
2192 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2193 Also resize prev_edge_clone vector.
2194 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2195 (ipcp_edge_removal_hook): New function.
2196 (ipcp_driver): Register ipcp_edge_removal_hook.
2197
2198 2014-01-17 Andrew Pinski <apinski@cavium.com>
2199 Steve Ellcey <sellcey@mips.com>
2200
2201 PR target/59462
2202 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2203 of operator mode.
2204
2205 2014-01-17 Jeff Law <law@redhat.com>
2206
2207 PR middle-end/57904
2208 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2209 so that pass_ccp runs first.
2210
2211 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2212
2213 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2214 (ix86_adjust_cost): Use !TARGET_XXX.
2215 (do_reorder_for_imul): Likewise.
2216 (swap_top_of_ready_list): Likewise.
2217 (ix86_sched_reorder): Likewise.
2218
2219 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2220
2221 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2222 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2223 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2224 (intel_memset): New. Duplicate slm_memset.
2225 (intel_cost): New. Duplicate slm_cost.
2226 (m_INTEL): New macro.
2227 (processor_target_table): Add "intel".
2228 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2229 with PROCESSOR_INTEL for "intel".
2230 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2231 PROCESSOR_SILVERMONT.
2232 (ix86_issue_rate): Likewise.
2233 (ix86_adjust_cost): Likewise.
2234 (ia32_multipass_dfa_lookahead): Likewise.
2235 (swap_top_of_ready_list): Likewise.
2236 (ix86_sched_reorder): Likewise.
2237 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2238 instead of TARGET_OPT_AGU.
2239 * config/i386/i386.h (TARGET_INTEL): New.
2240 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2241 (processor_type): Add PROCESSOR_INTEL.
2242 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2243 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2244
2245 2014-01-17 Marek Polacek <polacek@redhat.com>
2246
2247 PR c/58346
2248 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2249 size is zero.
2250
2251 2014-01-17 Richard Biener <rguenther@suse.de>
2252
2253 PR tree-optimization/46590
2254 * opts.c (default_options_table): Add entries for
2255 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2256 all enabled at -O1 but not for -Og.
2257 * common.opt (fbranch-count-reg): Remove Init(1).
2258 (fmove-loop-invariants): Likewise.
2259 (ftree-pta): Likewise.
2260
2261 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2262
2263 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2264 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2265 decls to at least the GCC 4.8 used alignments.
2266
2267 PR fortran/59440
2268 * tree-nested.c (convert_nonlocal_reference_stmt,
2269 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2270 of GIMPLE_BIND stmts, adjust associated decls.
2271
2272 2014-01-17 Richard Biener <rguenther@suse.de>
2273
2274 PR tree-optimization/46590
2275 * vec.h (vec<>::bseach): New member function implementing
2276 binary search according to C89 bsearch.
2277 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2278 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2279 bitmap pointer again. Make accesses_in_loop a flat array.
2280 (mem_ref_obstack): New global.
2281 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2282 (mark_ref_stored): Likewise.
2283 (ref_indep_loop_p_2): Likewise.
2284 (set_ref_stored_in_loop): New helper function.
2285 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2286 (memref_free): Adjust.
2287 (record_mem_ref_loc): Simplify.
2288 (gather_mem_refs_stmt): Adjust.
2289 (sort_locs_in_loop_postorder_cmp): New function.
2290 (analyze_memory_references): Sort accesses_in_loop after
2291 loop postorder number.
2292 (find_ref_loc_in_loop_cmp): New function.
2293 (for_all_locs_in_loop): Find relevant cluster of locs in
2294 accesses_in_loop and iterate without recursion.
2295 (execute_sm): Avoid uninit warning.
2296 (struct ref_always_accessed): Simplify.
2297 (ref_always_accessed::operator ()): Likewise.
2298 (ref_always_accessed_p): Likewise.
2299 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2300 loop postorder numbers here.
2301 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2302 numbers.
2303
2304 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2305
2306 PR c++/57945
2307 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2308 on decls for which assemble_alias has been called.
2309
2310 2014-01-17 Nick Clifton <nickc@redhat.com>
2311
2312 * config/msp430/msp430.opt: (mcpu): New option.
2313 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2314 (msp430_option_override): Parse target_cpu. If the MCU name
2315 matches a generic string, clear target_mcu.
2316 (msp430_attr): Allow numeric interrupt values up to 63.
2317 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2318 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2319 option.
2320 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2321 Add mcpu matches.
2322 * config/msp430/msp430.md (popm): Use %J rather than %I.
2323 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2324 (addhi_cy_i): Use immediate_operand for operand 2.
2325 * doc/invoke.texi: Document -mcpu option.
2326
2327 2014-01-17 Richard Biener <rguenther@suse.de>
2328
2329 PR rtl-optimization/38518
2330 * df.h (df_analyze_loop): Declare.
2331 * df-core.c: Include cfgloop.h.
2332 (df_analyze_1): Split out main part of df_analyze.
2333 (df_analyze): Adjust.
2334 (loop_inverted_post_order_compute): New function.
2335 (loop_post_order_compute): Likewise.
2336 (df_analyze_loop): New function avoiding whole-function
2337 postorder computes.
2338 * loop-invariant.c (find_defs): Use df_analyze_loop.
2339 (find_invariants): Adjust.
2340 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2341
2342 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2343
2344 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2345 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2346
2347 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2348
2349 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2350 traversal when removing references.
2351
2352 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2353
2354 PR ipa/59775
2355 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2356
2357 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2358
2359 PR middle-end/56791
2360 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2361 pushing a reload for an autoinc when we had previously reloaded an
2362 inner part of the address.
2363
2364 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2365
2366 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2367 field.
2368 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2369 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2370 when not giving up or versioning for alias only because of
2371 loop->safelen.
2372 (vect_analyze_data_ref_dependences): Set to true.
2373 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2374 is a GIMPLE_PHI.
2375 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2376 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2377 to the condition.
2378
2379 PR middle-end/58344
2380 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2381
2382 PR target/59839
2383 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2384 operand 0 predicate for gathers, use a new pseudo as subtarget.
2385
2386 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2387
2388 PR middle-end/59609
2389 * lra-constraints.c (process_alt_operands): Add printing debug info.
2390 Check absence of input/output reloads for matched operands too.
2391
2392 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2393
2394 PR rtl-optimization/59835
2395 * ira.c (ira_init_register_move_cost): Increase cost for
2396 impossible modes.
2397
2398 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2399
2400 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2401
2402 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2403
2404 PR target/59780
2405 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2406 non-register objects. Use gen_(high/low)part more consistently.
2407 Fix assertions.
2408
2409 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2410
2411 PR target/59844
2412 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2413 endian support, remove tests for WORDS_BIG_ENDIAN.
2414 (p8_mfvsrd_3_<mode>): Likewise.
2415 (reload_gpr_from_vsx<mode>): Likewise.
2416 (reload_gpr_from_vsxsf): Likewise.
2417 (p8_mfvsrd_4_disf): Likewise.
2418
2419 2014-01-16 Richard Biener <rguenther@suse.de>
2420
2421 PR rtl-optimization/46590
2422 * lcm.c (compute_antinout_edge): Use postorder iteration.
2423 (compute_laterin): Use inverted postorder iteration.
2424
2425 2014-01-16 Nick Clifton <nickc@redhat.com>
2426
2427 PR middle-end/28865
2428 * varasm.c (output_constant): Return the number of bytes actually
2429 emitted.
2430 (output_constructor_array_range): Update the field size with the
2431 number of bytes emitted by output_constant.
2432 (output_constructor_regular_field): Likewise. Also do not
2433 complain if the total number of bytes emitted is now greater
2434 than the expected fieldpos.
2435 * output.h (output_constant): Update prototype and descriptive comment.
2436
2437 2014-01-16 Marek Polacek <polacek@redhat.com>
2438
2439 PR middle-end/59827
2440 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2441 it is error_mark_node.
2442
2443 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2444
2445 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2446 VALID_AVX256_REG_OR_OI_MODE.
2447
2448 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2449
2450 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2451 current procedure should be profiled.
2452
2453 2014-01-15 Andrew Pinski <apinski@cavium.com>
2454
2455 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2456 of moving from/to the STACK_REG register class.
2457
2458 2014-01-15 Richard Henderson <rth@redhat.com>
2459
2460 PR debug/54694
2461 * reginfo.c (global_regs_decl): Globalize.
2462 * rtl.h (global_regs_decl): Declare.
2463 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2464 reserved via global_regs.
2465
2466 2014-01-15 Teresa Johnson <tejohnson@google.com>
2467
2468 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2469
2470 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2471
2472 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2473 and vmulosh rather than call gen_vec_widen_smult_*.
2474 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2475 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2476 (vec_widen_smult_even_v16qi): Likewise.
2477 (vec_widen_umult_even_v8hi): Likewise.
2478 (vec_widen_smult_even_v8hi): Likewise.
2479 (vec_widen_umult_odd_v16qi): Likewise.
2480 (vec_widen_smult_odd_v16qi): Likewise.
2481 (vec_widen_umult_odd_v8hi): Likewise.
2482 (vec_widen_smult_odd_v8hi): Likewise.
2483 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2484 vmuloub rather than call gen_vec_widen_umult_*.
2485 (vec_widen_umult_lo_v16qi): Likewise.
2486 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2487 vmulosb rather than call gen_vec_widen_smult_*.
2488 (vec_widen_smult_lo_v16qi): Likewise.
2489 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2490 rather than call gen_vec_widen_umult_*.
2491 (vec_widen_umult_lo_v8hi): Likewise.
2492 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2493 rather than call gen_vec_widen_smult_*.
2494 (vec_widen_smult_lo_v8hi): Likewise.
2495
2496 2014-01-15 Jeff Law <law@redhat.com>
2497
2498 PR tree-optimization/59747
2499 * ree.c (find_and_remove_re): Properly handle case where a second
2500 eliminated extension requires widening a copy created for elimination
2501 of a prior extension.
2502 (combine_set_extension): Ensure that the number of hard regs needed
2503 for a destination register does not change when we widen it.
2504
2505 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2506
2507 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2508 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2509 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2510 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2511 (avr-*-rtems*): Likewise.
2512 (bfin*-rtems*): Likewise.
2513 (moxie-*-rtems*): Likewise.
2514 (h8300-*-rtems*): Likewise.
2515 (i[34567]86-*-rtems*): Likewise.
2516 (lm32-*-rtems*): Likewise.
2517 (m32r-*-rtems*): Likewise.
2518 (m68k-*-rtems*): Likewise.
2519 (microblaze*-*-rtems*): Likewise.
2520 (mips*-*-rtems*): Likewise.
2521 (powerpc-*-rtems*): Likewise.
2522 (sh-*-rtems*): Likewise.
2523 (sparc-*-rtems*): Likewise.
2524 (sparc64-*-rtems*): Likewise.
2525 (v850-*-rtems*): Likewise.
2526 (m32c-*-rtems*): Likewise.
2527
2528 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2529
2530 PR rtl-optimization/59511
2531 * ira.c (ira_init_register_move_cost): Use memory costs for some
2532 cases of register move cost calculations.
2533 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2534 instead of BB frequency.
2535 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2536 * lra-assigns.c (find_hard_regno_for): Ditto.
2537
2538 2014-01-15 Richard Biener <rguenther@suse.de>
2539
2540 PR tree-optimization/59822
2541 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2542 (vectorizable_load): Use it to hoist defs of uses of invariant
2543 loads out of the loop.
2544
2545 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2546 Kugan Vivekanandarajah <kuganv@linaro.org>
2547
2548 PR target/59695
2549 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2550 truncation.
2551
2552 2014-01-15 Richard Biener <rguenther@suse.de>
2553
2554 PR rtl-optimization/59802
2555 * lcm.c (compute_available): Use inverted postorder to seed
2556 the initial worklist.
2557
2558 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2559
2560 PR target/59803
2561 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2562 ADDR_REGS for invalid symrefs in non-PIC code.
2563
2564 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2565
2566 PR other/58712
2567 * builtins.c (determine_block_size): Initialize *probable_max_size
2568 even if len_rtx is CONST_INT.
2569
2570 2014-01-14 Andrew Pinski <apinski@cavium.com>
2571
2572 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2573 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2574 (cortexa53_tunings): Likewise.
2575 (aarch64_sched_issue_rate): New function.
2576 (TARGET_SCHED_ISSUE_RATE): Define.
2577
2578 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2579
2580 * ira-costs.c (find_costs_and_classes): Add missed
2581 ira_init_register_move_cost_if_necessary.
2582
2583 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2584
2585 PR target/59787
2586 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2587
2588 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2589
2590 PR target/59794
2591 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2592 to indicate if type is used for function return value. Warn ABI
2593 change if the vector mode isn't available for function return value.
2594 (ix86_function_arg_advance): Pass false to type_natural_mode.
2595 (ix86_function_arg): Likewise.
2596 (ix86_gimplify_va_arg): Likewise.
2597 (function_arg_32): Don't warn ABI change.
2598 (ix86_function_value): Pass true to type_natural_mode.
2599 (ix86_return_in_memory): Likewise.
2600 (ix86_struct_value_rtx): Removed.
2601 (TARGET_STRUCT_VALUE_RTX): Likewise.
2602
2603 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2604
2605 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2606 converting a conditional jump into a conditional return.
2607
2608 2014-01-14 Richard Biener <rguenther@suse.de>
2609
2610 PR tree-optimization/58921
2611 PR tree-optimization/59006
2612 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2613 hoisting invariant stmts.
2614 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2615 invariant loads on the preheader edge if possible.
2616
2617 2014-01-14 Joey Ye <joey.ye@arm.com>
2618
2619 * doc/plugin.texi (Building GCC plugins): Update to C++.
2620
2621 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2622
2623 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2624 (_mm_rcp28_round_ss): Ditto.
2625 (_mm_rsqrt28_round_sd): Ditto.
2626 (_mm_rsqrt28_round_ss): Ditto.
2627 (_mm_rcp28_sd): Ditto.
2628 (_mm_rcp28_ss): Ditto.
2629 (_mm_rsqrt28_sd): Ditto.
2630 (_mm_rsqrt28_ss): Ditto.
2631 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2632 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2633 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2634 (IX86_BUILTIN_RCP28SD): Ditto.
2635 (IX86_BUILTIN_RCP28SS): Ditto.
2636 (IX86_BUILTIN_RSQRT28SD): Ditto.
2637 (IX86_BUILTIN_RSQRT28SS): Ditto.
2638 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2639 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2640 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2641 (ix86_expand_special_args_builtin): Expand new FTYPE.
2642 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2643 (srcp14<mode>): Make insn unary.
2644 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2645 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2646 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2647 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2648 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2649 Fix rounding: make it SAE only.
2650 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2651 Ditto.
2652 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2653 Ditto.
2654 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2655 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2656 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2657 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2658 (round_saeonly_mask_scalar_operand4): Ditto.
2659 (round_saeonly_mask_scalar_op3): Ditto.
2660 (round_saeonly_mask_scalar_op4): Ditto.
2661
2662 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2663
2664 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2665 Implement -maltivec=be for vec_insert and vec_extract.
2666
2667 2014-01-10 DJ Delorie <dj@redhat.com>
2668
2669 * config/msp430/msp430.md (call_internal): Don't allow memory
2670 references with SP as the base register.
2671 (call_value_internal): Likewise.
2672 * config/msp430/constraints.md (Yc): New. For memory references
2673 that don't use SP as a base register.
2674
2675 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2676 "an integer without a # prefix"
2677 * config/msp430/msp430.md (epilogue_helper): Use it.
2678
2679 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2680
2681 PR target/59617
2682 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2683 AVX512F gather builtins.
2684 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2685 on gather decls with INTEGER_TYPE masktype.
2686 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2687 directly into the builtin rather than hoisting it before loop.
2688
2689 PR tree-optimization/59387
2690 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2691 (scev_const_prop): If folded_casts and type has undefined overflow,
2692 use force_gimple_operand instead of force_gimple_operand_gsi and
2693 for each added stmt if it is assign with
2694 arith_code_with_undefined_signed_overflow, call
2695 rewrite_to_defined_overflow.
2696 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2697 gimple-fold.h instead.
2698 (arith_code_with_undefined_signed_overflow,
2699 rewrite_to_defined_overflow): Moved to ...
2700 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2701 rewrite_to_defined_overflow): ... here. No longer static.
2702 Include gimplify-me.h.
2703 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2704 rewrite_to_defined_overflow): New prototypes.
2705
2706 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2707
2708 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2709
2710 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2711
2712 * builtins.c (get_object_alignment_2): Minor tweak.
2713 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2714
2715 2014-01-13 Christian Bruel <christian.bruel@st.com>
2716
2717 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2718 optimized non constant lengths.
2719
2720 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2721
2722 PR libgomp/59194
2723 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2724 load as __atomic_load_N if possible.
2725
2726 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2727
2728 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2729 target parameter.
2730 (rs6000_expand_builtin): Adjust call.
2731
2732 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2733
2734 PR target/58115
2735 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2736 * config/rs6000/rs6000.c: Include target-globals.h.
2737 (rs6000_set_current_function): Instead of doing target_reinit
2738 unconditionally, use save_target_globals_default_opts and
2739 restore_target_globals.
2740
2741 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2742 FPSCR.
2743 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2744 (rs6000_expand_builtin): Handle mffs and mtfsf.
2745 (rs6000_init_builtins): Define mffs and mtfsf.
2746 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2747 (rs6000_mffs): New pattern.
2748 (rs6000_mtfsf): New pattern.
2749
2750 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2751
2752 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2753 Start narrowing with START. Apply candidate-use pair
2754 and check overall cost in narrowing.
2755 (iv_ca_prune): Pass new argument.
2756
2757 2014-01-10 Jeff Law <law@redhat.com>
2758
2759 PR middle-end/59743
2760 * ree.c (combine_reaching_defs): Ensure the defining statement
2761 occurs before the extension when optimizing extensions with
2762 different source and destination hard registers.
2763
2764 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2765
2766 PR ipa/58585
2767 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2768 vtables into the type inheritance graph.
2769
2770 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2771
2772 PR rtl-optimization/59754
2773 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2774 modes in the REGNO != REGNO case.
2775
2776 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2777
2778 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2779
2780 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2781
2782 PR tree-optimization/59745
2783 * tree-predcom.c (tree_predictive_commoning_loop): Call
2784 free_affine_expand_cache if giving up because components is NULL.
2785
2786 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2787 GC in payload of target_globals struct instead of allocating them on
2788 the heap and the larger structs separately using GC.
2789 * target-globals.h (struct target_globals): Make regs, hard_regs,
2790 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2791 of GTY((skip)) and change type to void *.
2792 (reset_target_globals): Cast loads from those fields to corresponding
2793 types.
2794
2795 2014-01-10 Steve Ellcey <sellcey@mips.com>
2796
2797 PR plugins/59335
2798 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2799 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2800 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2801
2802 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2803
2804 PR target/59744
2805 * aarch64-modes.def (CC_Zmode): New flags mode.
2806 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2807 represents an equality.
2808 (aarch64_get_condition_code): Handle CC_Zmode.
2809 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2810
2811 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2812
2813 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2814 extraction in good case.
2815
2816 2014-01-10 Richard Biener <rguenther@suse.de>
2817
2818 PR tree-optimization/59374
2819 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2820 checking after SLP discovery. Mark stmts not participating
2821 in any SLP instance properly.
2822
2823 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2824
2825 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2826 when handling a SET rtx.
2827
2828 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2829
2830 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2831 (cortex-a57): Likewise.
2832 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2833
2834 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2835
2836 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2837 non-iwmmxt builtins.
2838
2839 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2840
2841 PR ipa/58252
2842 PR ipa/59226
2843 * ipa-devirt.c record_target_from_binfo): Take as argument
2844 stack of binfos and lookup matching one for virtual inheritance.
2845 (possible_polymorphic_call_targets_1): Update.
2846
2847 2014-01-10 Huacai Chen <chenhc@lemote.com>
2848
2849 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2850 kernel strings for Loongson-2E/2F/3A.
2851
2852 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2853
2854 PR middle-end/59670
2855 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2856 is_gimple_call before calling gimple_call_internal_p.
2857
2858 2014-01-09 Steve Ellcey <sellcey@mips.com>
2859
2860 * Makefile.in (TREE_FLOW_H): Remove.
2861 (TREE_SSA_H): Add file names from tree-flow.h.
2862 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2863 * tree.h: Remove tree-flow.h reference.
2864 * hash-table.h: Remove tree-flow.h reference.
2865 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2866 reference with tree-ssa-loop.h.
2867
2868 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2869
2870 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2871 default element-order behavior for -maltivec.
2872 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2873 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2874 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2875 when targeting big endian, at least for now.
2876 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2877
2878 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2879
2880 PR middle-end/47735
2881 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2882 var satisfies use_register_for_decl, just take into account type
2883 alignment, rather than decl alignment.
2884
2885 PR tree-optimization/59622
2886 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2887 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2888 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2889 Don't devirtualize for inplace at all. For targets.length () == 1,
2890 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2891
2892 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2893
2894 * config/i386/i386.md (cpu): Remove the unused btver1.
2895
2896 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2897
2898 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2899
2900 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2901
2902 PR target/58115
2903 * tree-core.h (struct target_globals): New forward declaration.
2904 (struct tree_target_option): Add globals field.
2905 * tree.h (TREE_TARGET_GLOBALS): Define.
2906 (prepare_target_option_nodes_for_pch): New prototype.
2907 * target-globals.h (struct target_globals): Define even if
2908 !SWITCHABLE_TARGET.
2909 * tree.c (prepare_target_option_node_for_pch,
2910 prepare_target_option_nodes_for_pch): New functions.
2911 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2912 * config/i386/i386.c: Include target-globals.h.
2913 (ix86_set_current_function): Instead of doing target_reinit
2914 unconditionally, use save_target_globals_default_opts and
2915 restore_target_globals.
2916
2917 2014-01-09 Richard Biener <rguenther@suse.de>
2918
2919 PR tree-optimization/59715
2920 * tree-cfg.h (split_critical_edges): Declare.
2921 * tree-cfg.c (split_critical_edges): Export.
2922 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2923
2924 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2925
2926 * cfgexpand.c (expand_stack_vars): Optionally disable
2927 asan stack protection.
2928 (expand_used_vars): Likewise.
2929 (partition_stack_vars): Likewise.
2930 * asan.c (asan_emit_stack_protection): Optionally disable
2931 after return stack usage.
2932 (instrument_derefs): Optionally disable memory access instrumentation.
2933 (instrument_builtin_call): Likewise.
2934 (instrument_strlen_call): Likewise.
2935 (asan_protect_global): Optionally disable global variables protection.
2936 * doc/invoke.texi: Added doc for new options.
2937 * params.def: Added new options.
2938 * params.h: Likewise.
2939
2940 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2941
2942 PR rtl-optimization/59724
2943 * ifcvt.c (cond_exec_process_if_block): Don't call
2944 flow_find_head_matching_sequence with 0 longest_match.
2945 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2946 non-active insns if !stop_after.
2947 (try_head_merge_bb): Revert 2014-01-07 changes.
2948
2949 2014-01-08 Jeff Law <law@redhat.com>
2950
2951 * ree.c (get_sub_rtx): New function, extracted from...
2952 (merge_def_and_ext): Here.
2953 (combine_reaching_defs): Use get_sub_rtx.
2954
2955 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2956
2957 * cgraph.h (varpool_variable_node): Do not choke on null node.
2958
2959 2014-01-08 Catherine Moore <clm@codesourcery.com>
2960
2961 * config/mips/mips.md (simple_return): Attempt to use JRC
2962 for microMIPS.
2963 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2964
2965 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2966
2967 PR rtl-optimization/59137
2968 * reorg.c (steal_delay_list_from_target): Call update_block for
2969 elided insns.
2970 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2971
2972 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2973
2974 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2975 two duplicate entries.
2976
2977 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2978
2979 Revert:
2980 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2981
2982 * config/mips/mips.c (mips_truncated_op_cost): New function.
2983 (mips_rtx_costs): Adjust test for BADDU.
2984 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2985
2986 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2987
2988 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2989 (*baddu_si): ...this new pattern.
2990
2991 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2992
2993 PR ipa/59722
2994 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2995
2996 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2997
2998 PR middle-end/57748
2999 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3000 inner_reference_p.
3001 (expand_expr, expand_normal): Adjust.
3002 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3003 inner_reference_p. Use inner_reference_p to expand inner references.
3004 (store_expr): Adjust.
3005 * cfgexpand.c (expand_call_stmt): Adjust.
3006
3007 2014-01-08 Rong Xu <xur@google.com>
3008
3009 * gcov-io.c (gcov_var): Move from gcov-io.h.
3010 (gcov_position): Ditto.
3011 (gcov_is_error): Ditto.
3012 (gcov_rewrite): Ditto.
3013 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3014 only part to libgcc/libgcov.h.
3015
3016 2014-01-08 Marek Polacek <polacek@redhat.com>
3017
3018 PR middle-end/59669
3019 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3020
3021 2014-01-08 Marek Polacek <polacek@redhat.com>
3022
3023 PR sanitizer/59667
3024 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3025
3026 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3027
3028 PR rtl-optimization/59649
3029 * stor-layout.c (get_mode_bounds): For BImode return
3030 0 and STORE_FLAG_VALUE.
3031
3032 2014-01-08 Richard Biener <rguenther@suse.de>
3033
3034 PR middle-end/59630
3035 * gimple.h (is_gimple_builtin_call): Remove.
3036 (gimple_builtin_call_types_compatible_p): New.
3037 (gimple_call_builtin_p): New overload.
3038 * gimple.c (is_gimple_builtin_call): Remove.
3039 (validate_call): Rename to ...
3040 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3041 check return types.
3042 (validate_type): New static function.
3043 (gimple_call_builtin_p): New overload and adjust.
3044 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3045 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3046 (gimple_fold_stmt_to_constant_1): Likewise.
3047 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3048
3049 2014-01-08 Richard Biener <rguenther@suse.de>
3050
3051 PR middle-end/59471
3052 * gimplify.c (gimplify_expr): Gimplify register-register type
3053 VIEW_CONVERT_EXPRs to separate stmts.
3054
3055 2014-01-07 Jeff Law <law@redhat.com>
3056
3057 PR middle-end/53623
3058 * ree.c (combine_set_extension): Handle case where source
3059 and destination registers in an extension insn are different.
3060 (combine_reaching_defs): Allow source and destination registers
3061 in extension to be different under limited circumstances.
3062 (add_removable_extension): Remove restriction that the
3063 source and destination registers in the extension are the same.
3064 (find_and_remove_re): Emit a copy from the extension's
3065 destination to its source after the defining insn if
3066 the source and destination registers are different.
3067
3068 PR middle-end/59285
3069 * ifcvt.c (merge_if_block): If we are merging a block with more than
3070 one successor with a block with no successors, remove any BARRIER
3071 after the second block.
3072
3073 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3074
3075 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3076
3077 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3078
3079 PR target/59652
3080 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3081 for 14-bit register offsets when INT14_OK_STRICT is false.
3082
3083 2014-01-07 Roland Stigge <stigge@antcom.de>
3084 Michael Meissner <meissner@linux.vnet.ibm.com>
3085
3086 PR 57386/target
3087 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3088 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3089
3090 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3091
3092 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3093 -mcpu.
3094
3095 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3096
3097 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3098 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3099 rtx is const0_rtx or not.
3100
3101 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3102
3103 PR target/58115
3104 * target-globals.c (save_target_globals): Remove this_fn_optab
3105 handling.
3106 * toplev.c: Include optabs.h.
3107 (target_reinit): Temporarily restore the global options if another
3108 set of options are in force.
3109
3110 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3111
3112 PR rtl-optimization/58668
3113 * cfgcleanup.c (flow_find_cross_jump): Don't count
3114 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3115 to determine what is counted.
3116 (flow_find_head_matching_sequence): Use active_insn_p to determine
3117 what is counted.
3118 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3119 counting change.
3120 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3121 determine what is counted.
3122
3123 PR tree-optimization/59643
3124 * tree-predcom.c (split_data_refs_to_components): If one dr is
3125 read and one write, determine_offset fails and the write isn't
3126 in the bad component, just put the read into the bad component.
3127
3128 2014-01-07 Mike Stump <mikestump@comcast.net>
3129 Jakub Jelinek <jakub@redhat.com>
3130
3131 PR pch/59436
3132 * tree-core.h (struct tree_optimization_option): Change optabs
3133 type from unsigned char * to void *.
3134 * optabs.c (init_tree_optimization_optabs): Adjust
3135 TREE_OPTIMIZATION_OPTABS initialization.
3136
3137 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3138
3139 PR target/59644
3140 * config/i386/i386.h (struct machine_function): Add
3141 no_drap_save_restore field.
3142 * config/i386/i386.c (ix86_save_reg): Use
3143 !cfun->machine->no_drap_save_restore instead of
3144 crtl->stack_realign_needed.
3145 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3146 this function clears frame_pointer_needed. Set
3147 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3148 and DRAP reg is needed.
3149
3150 2014-01-06 Marek Polacek <polacek@redhat.com>
3151
3152 PR c/57773
3153 * doc/implement-c.texi: Mention that other integer types are
3154 permitted as bit-field types in strictly conforming mode.
3155
3156 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3157
3158 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3159 is newly allocated.
3160
3161 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3162
3163 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3164
3165 2014-01-06 Martin Jambor <mjambor@suse.cz>
3166
3167 PR ipa/59008
3168 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3169 to int.
3170 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3171
3172 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3173
3174 PR debug/59350
3175 PR debug/59510
3176 * var-tracking.c (add_stores): Preserve the value of the source even if
3177 we don't record the store.
3178
3179 2014-01-06 Terry Guo <terry.guo@arm.com>
3180
3181 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3182
3183 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3184
3185 PR bootstrap/59541
3186 * config/darwin.c (darwin_function_section): Adjust return values to
3187 correspond to optimisation changes made in r206070.
3188
3189 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3190
3191 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3192 from prefetch_block tune setting.
3193 (nocona_cost): Correct size of prefetch block to 64.
3194
3195 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3196
3197 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3198 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3199 used to save the static chain register in the computation of the offset
3200 from which the FP registers need to be restored.
3201
3202 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3203
3204 PR tree-optimization/59519
3205 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3206 ICE if get_current_def (current_new_name) is already non-NULL, as long
3207 as it is a phi result of some other phi in *new_exit_bb that has
3208 the same argument.
3209
3210 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3211 or vmovdqu* for misaligned_operand.
3212 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3213 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3214 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3215 aligned_mem for AVX512F masked aligned load and store builtins and for
3216 non-temporal moves.
3217
3218 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3219
3220 PR tree-optimization/59651
3221 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3222 Address range for negative step should be added by TYPE_SIZE_UNIT.
3223
3224 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3225
3226 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3227 overlapping registers also for registers other than the stack pointer.
3228
3229 2014-01-03 Marek Polacek <polacek@redhat.com>
3230
3231 PR other/59661
3232 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3233 __builtin_FILE.
3234
3235 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3236
3237 PR target/59625
3238 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3239 asm goto as jump.
3240
3241 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3242 (push splitter): Use <P:MODE_SIZE> instead of
3243 GET_MODE_SIZE (<P:MODE>mode).
3244 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3245 (mov -1, reg peephole2): Likewise.
3246 * config/i386/sse.md (*mov<mode>_internal,
3247 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3248 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3249 *<code><mode>3, *andnot<mode>3<mask_name>,
3250 <mask_codefor><code><mode>3<mask_name>): Likewise.
3251 * config/i386/subst.md (mask_mode512bit_condition,
3252 sd_mask_mode512bit_condition): Likewise.
3253
3254 2014-01-02 Xinliang David Li <davidxl@google.com>
3255
3256 PR tree-optimization/59303
3257 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3258 (dump_predicates): Better output format.
3259 (pred_equal_p): New function.
3260 (is_neq_relop_p): Ditto.
3261 (is_neq_zero_form_p): Ditto.
3262 (pred_expr_equal_p): Ditto.
3263 (pred_neg_p): Ditto.
3264 (simplify_pred): Ditto.
3265 (simplify_preds_2): Ditto.
3266 (simplify_preds_3): Ditto.
3267 (simplify_preds_4): Ditto.
3268 (simplify_preds): Ditto.
3269 (push_pred): Ditto.
3270 (push_to_worklist): Ditto.
3271 (get_pred_info_from_cmp): Ditto.
3272 (is_degenerated_phi): Ditto.
3273 (normalize_one_pred_1): Ditto.
3274 (normalize_one_pred): Ditto.
3275 (normalize_one_pred_chain): Ditto.
3276 (normalize_preds): Ditto.
3277 (normalize_cond_1): Remove function.
3278 (normalize_cond): Ditto.
3279 (is_gcond_subset_of): Ditto.
3280 (is_subset_of_any): Ditto.
3281 (is_or_set_subset_of): Ditto.
3282 (is_and_set_subset_of): Ditto.
3283 (is_norm_cond_subset_of): Ditto.
3284 (pred_chain_length_cmp): Ditto.
3285 (convert_control_dep_chain_into_preds): Type change.
3286 (find_predicates): Ditto.
3287 (find_def_preds): Ditto.
3288 (destroy_predicates_vecs): Ditto.
3289 (find_matching_predicates_in_rest_chains): Ditto.
3290 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3291 (is_pred_expr_subset): Ditto.
3292 (is_pred_chain_subset_of): Ditto.
3293 (is_included_in): Ditto.
3294 (is_superset_of): Ditto.
3295
3296 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3297
3298 Update copyright years.
3299
3300 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3301
3302 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3303 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3304 config/arc/arc.md, config/arc/arc.opt,
3305 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3306 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3307 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3308 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3309 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3310 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3311 vtable-verify.c, vtable-verify.h: Use the standard form for the
3312 copyright notice.
3313
3314 2014-01-02 Tobias Burnus <burnus@net-b.de>
3315
3316 * gcc.c (process_command): Update copyright notice dates.
3317 * gcov-dump.c: Ditto.
3318 * gcov.c: Ditto.
3319 * doc/cpp.texi: Bump @copying's copyright year.
3320 * doc/cppinternals.texi: Ditto.
3321 * doc/gcc.texi: Ditto.
3322 * doc/gccint.texi: Ditto.
3323 * doc/gcov.texi: Ditto.
3324 * doc/install.texi: Ditto.
3325 * doc/invoke.texi: Ditto.
3326
3327 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3328
3329 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3330
3331 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3332
3333 * config/i386/sse.md (*mov<mode>_internal): Guard
3334 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3335
3336 PR rtl-optimization/59647
3337 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3338 new_rtx into UNSIGNED_FLOAT rtxes.
3339 \f
3340 Copyright (C) 2014 Free Software Foundation, Inc.
3341
3342 Copying and distribution of this file, with or without modification,
3343 are permitted in any medium without royalty provided the copyright
3344 notice and this notice are preserved.
This page took 0.163329 seconds and 6 git commands to generate.