re PR target/59799 (aarch64_pass_by_reference never passes arrays by value, contrary...
[gcc.git] / gcc / ChangeLog
1 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2
3 PR target/59799
4 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
5 passing arrays in registers are the same as for structs, so remove the
6 special case for them.
7
8 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
9
10 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
11 destination type, extract only the valid bits if the source type is not
12 integral and has a different mode.
13
14 2014-02-19 Richard Biener <rguenther@suse.de>
15
16 PR ipa/60243
17 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
18 for all calls.
19
20 2014-02-19 Richard Biener <rguenther@suse.de>
21
22 PR ipa/60243
23 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
24 (ipa_modify_call_arguments): Emit an argument load explicitely and
25 preserve virtual SSA form there and for the replacement call.
26 Do not update SSA form nor free dominance info.
27
28 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
29
30 * ipa.c (function_and_variable_visibility): Also clear WEAK
31 flag when disolving COMDAT_GROUP.
32
33 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
34
35 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
36 * ipa-prop.c (ipa_set_jf_known_type): Return early when
37 not devirtualizing.
38 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
39 do more sanity checks.
40 (detect_type_change): Return true when giving up early.
41 (compute_complex_assign_jump_func): Fix type parameter of
42 ipa_set_ancestor_jf.
43 (compute_complex_ancestor_jump_func): Likewise.
44 (update_jump_functions_after_inlining): Fix updating of
45 ancestor function.
46 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type
47 to be NULL.
48
49 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
50
51 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
52 inline clones when edge disappears.
53
54 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
55
56 PR target/60203
57 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
58 Split 64-bit moves into 2 patterns. Do not allow the use of
59 direct move for TDmode in little endian, since the decimal value
60 has little endian bytes within a word, but the 64-bit pieces are
61 ordered in a big endian fashion, and normal subreg's of TDmode are
62 not allowed.
63 (mov<mode>_64bit_dm): Likewise.
64 (movtd_64bit_nodm): Likewise.
65
66 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
67
68 PR tree-optimization/60174
69 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
70 statement of an SSA_NAME that occurs in an abnormal PHI node.
71
72 2014-02-18 Jakub Jelinek <jakub@redhat.com>
73
74 PR sanitizer/60142
75 * final.c (SEEN_BB): Remove.
76 (SEEN_NOTE, SEEN_EMITTED): Renumber.
77 (final_scan_insn): Don't force_source_line on second
78 NOTE_INSN_BASIC_BLOCK.
79
80 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
81
82 PR target/60205
83 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
84 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
85 (type_natural_mode): Warn ABI change when %zmm register is not
86 available for AVX512F vector value passing.
87
88 2014-02-18 Kai Tietz <ktietz@redhat.com>
89
90 PR target/60193
91 * config/i386/i386.c (ix86_expand_prologue): Use value in
92 rax register as displacement when restoring %r10 or %rax.
93 Fix wrong offset when restoring both registers.
94
95 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
96
97 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
98 assertion with conditional return.
99
100 2014-02-18 Jakub Jelinek <jakub@redhat.com>
101 Uros Bizjak <ubizjak@gmail.com>
102
103 PR driver/60233
104 * config/i386/driver-i386.c (host_detect_local_cpu): If
105 YMM state is not saved by the OS, also clear has_f16c. Move
106 CPUID 0x80000001 handling before YMM state saving checking.
107
108 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
109
110 PR rtl-optimization/58960
111 * haifa-sched.c (alloc_global_sched_pressure_data): New,
112 factored out from ...
113 (sched_init): ... here.
114 (free_global_sched_pressure_data): New, factored out from ...
115 (sched_finish): ... here.
116 * sched-int.h (free_global_sched_pressure_data): Declare.
117 * sched-rgn.c (nr_regions_initial): New static global.
118 (haifa_find_rgns): Initialize it.
119 (schedule_region): Disable sched-pressure for the newly
120 generated regions.
121
122 2014-02-17 Richard Biener <rguenther@suse.de>
123
124 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
125 release SSA defs of pattern stmts.
126
127 2014-02-17 Richard Biener <rguenther@suse.de>
128
129 * tree-inline.c (expand_call_inline): Release the virtual
130 operand defined by the call we are about to inline.
131
132 2014-02-17 Richard Biener <rguenther@suse.de>
133
134 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
135
136 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
137 Ilya Tocar <ilya.tocar@intel.com>
138
139 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
140 arguments order in builtin.
141 (_mm512_permutexvar_epi64): Ditto.
142 (_mm512_mask_permutexvar_epi64): Ditto
143 (_mm512_maskz_permutexvar_epi32): Ditto
144 (_mm512_permutexvar_epi32): Ditto
145 (_mm512_mask_permutexvar_epi32): Ditto
146
147 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
148
149 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
150 (p8_vmrgow): Likewise.
151
152 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
153
154 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
155 endian targets.
156
157 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
158
159 PR target/60203
160 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
161 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
162 into 64-bit and 32-bit moves. On 64-bit moves, add support for
163 using direct move instructions on ISA 2.07. Also adjust
164 instruction length for 64-bit.
165 (mov<mode>_64bit, TFmode/TDmode): Likewise.
166 (mov<mode>_32bit, TFmode/TDmode): Likewise.
167
168 2014-02-15 Alan Modra <amodra@gmail.com>
169
170 PR target/58675
171 PR target/57935
172 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
173 find_replacement on parts of insn rtl that might be reloaded.
174
175 2014-02-15 Richard Biener <rguenther@suse.de>
176
177 PR tree-optimization/60183
178 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
179 (tree_ssa_phiprop): Calculate and free post-dominators.
180
181 2014-02-14 Jeff Law <law@redhat.com>
182
183 PR rtl-optimization/60131
184 * ree.c (get_extended_src_reg): New function.
185 (combine_reaching_defs): Use it rather than assuming location of REG.
186 (find_and_remove_re): Verify first operand of extension is
187 a REG before adding the insns to the copy list.
188
189 2014-02-14 Roland McGrath <mcgrathr@google.com>
190
191 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
192 * configure: Regenerated.
193 * config.in: Regenerated.
194 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
195 instead of ASM_SHORT.
196
197 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
198 Richard Earnshaw <rearnsha@arm.com>
199
200 PR rtl-optimization/59535
201 * lra-constraints.c (process_alt_operands): Encourage alternative
202 when unassigned pseudo class is superset of the alternative class.
203 (inherit_reload_reg): Don't inherit when optimizing for code size.
204 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
205 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
206 modes not less than 4 for Thumb1.
207
208 2014-02-14 Kyle McMartin <kyle@redhat.com>
209
210 PR pch/60010
211 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
212
213 2014-02-14 Richard Biener <rguenther@suse.de>
214
215 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
216 (get_frame_arg): Drop the assert with langhook types_compatible_p.
217 Do not strip INDIRECT_REFs.
218
219 2014-02-14 Richard Biener <rguenther@suse.de>
220
221 PR lto/60179
222 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
223 DECL_FUNCTION_SPECIFIC_TARGET.
224 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
225 * tree-streamer-out.c (pack_ts_target_option): Remove.
226 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
227 (write_ts_function_decl_tree_pointers): Do not stream
228 DECL_FUNCTION_SPECIFIC_TARGET.
229 * tree-streamer-in.c (unpack_ts_target_option): Remove.
230 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
231 (lto_input_ts_function_decl_tree_pointers): Do not stream
232 DECL_FUNCTION_SPECIFIC_TARGET.
233
234 2014-02-14 Jakub Jelinek <jakub@redhat.com>
235
236 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
237 (get_initial_def_for_induction, vectorizable_induction): Ignore
238 debug stmts when looking for exit_phi.
239 (vectorizable_live_operation): Fix up condition.
240
241 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
242
243 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
244 nreverse() because it changes the content of original tree list.
245
246 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
247
248 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
249 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
250
251 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
252
253 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
254 GNU coding standards.
255
256 2014-02-13 Jakub Jelinek <jakub@redhat.com>
257
258 PR debug/60152
259 * dwarf2out.c (gen_subprogram_die): Don't call
260 add_calling_convention_attribute if subr_die is old_die.
261
262 2014-02-13 Sharad Singhai <singhai@google.com>
263
264 * doc/optinfo.texi: Fix order of nodes.
265
266 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
267
268 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
269 operands[2], not operands[3].
270
271 2014-02-13 Richard Biener <rguenther@suse.de>
272
273 PR bootstrap/59878
274 * doc/install.texi (ISL): Update recommended version to 0.12.2,
275 mention the possibility of an in-tree build.
276 (CLooG): Update recommended version to 0.18.1, mention the
277 possibility of an in-tree build and clarify that the ISL
278 bundled with CLooG does not work.
279
280 2014-02-13 Jakub Jelinek <jakub@redhat.com>
281
282 PR target/43546
283 * expr.c (compress_float_constant): If x is a hard register,
284 extend into a pseudo and then move to x.
285
286 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
287
288 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
289 caused by bad second argument to warning_at() with -mhotpatch and
290 nested functions (e.g. with gfortran).
291
292 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
293
294 * opts.c (option_name): Remove "enabled by default" rider.
295
296 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
297
298 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
299
300 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
301 Uros Bizjak <ubizjak@gmail.com>
302
303 PR target/60151
304 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
305 * configure: Regenerated.
306
307 2014-02-12 Richard Biener <rguenther@suse.de>
308
309 * vec.c (vec_prefix::calculate_allocation): Move as
310 inline variant to vec.h.
311 (vec_prefix::calculate_allocation_1): New out-of-line version.
312 * vec.h (vec_prefix::calculate_allocation_1): Declare.
313 (vec_prefix::m_has_auto_buf): Rename to ...
314 (vec_prefix::m_using_auto_storage): ... this.
315 (vec_prefix::calculate_allocation): Inline the easy cases
316 and dispatch to calculate_allocation_1 which doesn't need the
317 prefix address.
318 (va_heap::reserve): Use gcc_checking_assert.
319 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
320 m_using_auto_storage.
321 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
322 member and adjust.
323 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
324 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
325 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
326
327 2014-02-12 Richard Biener <rguenther@suse.de>
328
329 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
330 when we found a dependence.
331
332 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
333
334 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
335 common code...
336 (maybe_fold_stmt): ... into this new function.
337 * omp-low.c (lower_omp): Update comment.
338
339 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
340 last use.
341
342 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
343 dereference.
344
345 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
346
347 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
348 identifiers in comments.
349 (cortexa53_extra_costs): Likewise.
350 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
351 (cortexa7_extra_costs): Likewise.
352 (cortexa12_extra_costs): Likewise.
353 (cortexa15_extra_costs): Likewise.
354 (v7m_extra_costs): Likewise.
355
356 2014-02-12 Richard Biener <rguenther@suse.de>
357
358 PR middle-end/60092
359 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
360 of posix_memalign being successful.
361 (lower_stmt): Restrict lowering of posix_memalign to when
362 -ftree-bit-ccp is enabled.
363
364 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
365
366 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
367 arg_loc.
368 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
369
370 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
371
372 PR rtl-optimization/60116
373 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
374 other_insn once the combination has been validated.
375
376 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
377
378 PR lto/59468
379 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
380 and wrapper.
381 * ipa-devirt.c: Include demangle.h
382 (odr_violation_reported): New static variable.
383 (add_type_duplicate): Update odr_violations.
384 (maybe_record_node): Add completep parameter; update it.
385 (record_target_from_binfo): Add COMPLETEP parameter;
386 update it as needed.
387 (possible_polymorphic_call_targets_1): Likewise.
388 (struct polymorphic_call_target_d): Add nonconstruction_targets;
389 rename FINAL to COMPLETE.
390 (record_targets_from_bases): Sanity check we found the binfo;
391 fix COMPLETEP updating.
392 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
393 parameter, fix computing of COMPLETEP.
394 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
395 at LTO time do demangling.
396 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
397 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
398 parameter.
399 (gimple_get_virt_method_for_binfo): Likewise.
400 * gimple-fold.h (gimple_get_virt_method_for_binfo,
401 gimple_get_virt_method_for_vtable): Update prototypes.
402
403 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
404
405 PR target/49008
406 * genautomata.c (add_presence_absence): Fix typo with
407 {final_}presence_list.
408
409 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
410
411 PR target/60137
412 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
413 for VSX/Altivec vectors that land in GPR registers.
414
415 2014-02-11 Richard Henderson <rth@redhat.com>
416 Jakub Jelinek <jakub@redhat.com>
417
418 PR debug/59776
419 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
420 around drhs if type conversion to lacc->type is not useless.
421
422 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
423
424 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
425 tuning struct.
426 (cortex-a57.cortex-a53): Likewise.
427 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
428
429 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
430
431 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
432 arm_restrict_it.
433
434 2014-02-11 Renlin Li <Renlin.Li@arm.com>
435
436 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
437 add_options_for_arm_vfp3.
438
439 2014-02-11 Jeff Law <law@redhat.com>
440
441 PR middle-end/54041
442 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
443 object with an undesirable mode.
444
445 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
446
447 PR libgomp/60107
448 * config/i386/sol2-9.h: New file.
449 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
450 *-*-solaris2.9*): Use it.
451
452 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
453
454 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
455 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
456
457 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
458
459 * config/microblaze/microblaze.c: Extend mcpu version format
460
461 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
462
463 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
464
465 2014-02-10 Richard Henderson <rth@redhat.com>
466
467 PR target/59927
468 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
469 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
470 ms-abi vs -mno-accumulate-outgoing-args.
471 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
472 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
473 respect to ms-abi.
474
475 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
476
477 PR middle-end/60080
478 * cfgexpand.c (expand_asm_operands): Attach source location to
479 ASM_INPUT rtx objects.
480 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
481
482 2014-02-10 Nick Clifton <nickc@redhat.com>
483
484 * config/mn10300/mn10300.c (popcount): New function.
485 (mn10300_expand_prologue): Include saved registers in stack usage
486 count.
487
488 2014-02-10 Jeff Law <law@redhat.com>
489
490 PR middle-end/52306
491 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
492 when changing the SET_DEST of a prior insn to avoid an input reload.
493
494 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
495
496 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
497 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
498 -mcall-openbsd, or -mcall-linux.
499 (CC1_ENDIAN_BIG_SPEC): Remove.
500 (CC1_ENDIAN_LITTLE_SPEC): Remove.
501 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
502 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
503 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
504 and %cc1_endian_default.
505 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
506
507 2014-02-10 Richard Biener <rguenther@suse.de>
508
509 PR tree-optimization/60115
510 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
511 MEM_REF handling. Properly verify that the accesses are not
512 out of the objects bound.
513
514 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
515
516 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
517 coretex to cortex.
518
519 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
520
521 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
522 proper constants and fix formatting.
523 (possible_polymorphic_call_targets): Fix formatting.
524
525 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
526 Ilya Tocar <ilya.tocar@intel.com>
527
528 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
529 (_mm512_loadu_epi32): Renamed into...
530 (_mm512_loadu_si512): This.
531 (_mm512_storeu_epi32): Renamed into...
532 (_mm512_storeu_si512): This.
533 (_mm512_maskz_ceil_ps): Removed.
534 (_mm512_maskz_ceil_pd): Ditto.
535 (_mm512_maskz_floor_ps): Ditto.
536 (_mm512_maskz_floor_pd): Ditto.
537 (_mm512_floor_round_ps): Ditto.
538 (_mm512_floor_round_pd): Ditto.
539 (_mm512_ceil_round_ps): Ditto.
540 (_mm512_ceil_round_pd): Ditto.
541 (_mm512_mask_floor_round_ps): Ditto.
542 (_mm512_mask_floor_round_pd): Ditto.
543 (_mm512_mask_ceil_round_ps): Ditto.
544 (_mm512_mask_ceil_round_pd): Ditto.
545 (_mm512_maskz_floor_round_ps): Ditto.
546 (_mm512_maskz_floor_round_pd): Ditto.
547 (_mm512_maskz_ceil_round_ps): Ditto.
548 (_mm512_maskz_ceil_round_pd): Ditto.
549 (_mm512_expand_pd): Ditto.
550 (_mm512_expand_ps): Ditto.
551 * config/i386/i386.c (ix86_builtins): Remove
552 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
553 (bdesc_args): Ditto.
554 * config/i386/predicates.md (const1256_operand): New.
555 (const_1_to_2_operand): Ditto.
556 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
557 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
558 (*avx512pf_gatherpf<mode>sf): Ditto.
559 (avx512pf_gatherpf<mode>df): Ditto.
560 (*avx512pf_gatherpf<mode>df_mask): Ditto.
561 (*avx512pf_gatherpf<mode>df): Ditto.
562 (avx512pf_scatterpf<mode>sf): Ditto.
563 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
564 (*avx512pf_scatterpf<mode>sf): Ditto.
565 (avx512pf_scatterpf<mode>df): Ditto.
566 (*avx512pf_scatterpf<mode>df_mask): Ditto.
567 (*avx512pf_scatterpf<mode>df): Ditto.
568 (avx512f_expand<mode>): Removed.
569 (<shift_insn><mode>3<mask_name>): Change predicate type.
570
571 2014-02-08 Jakub Jelinek <jakub@redhat.com>
572
573 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
574 not at the end of datarefs vector use ordered_remove to avoid
575 reordering datarefs vector.
576
577 PR c/59984
578 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
579 mark local addressable non-static vars as GOVD_PRIVATE
580 instead of GOVD_LOCAL.
581 * omp-low.c (lower_omp_for): Move gimple_bind_vars
582 and BLOCK_VARS of gimple_bind_block to new_stmt rather
583 than copying them.
584
585 PR middle-end/60092
586 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
587 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
588 assume_aligned or alloc_align attributes.
589 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
590 arguments. Handle also assume_aligned and alloc_align attributes.
591 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
592 calls to functions with assume_aligned or alloc_align attributes.
593 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
594
595 2014-02-08 Terry Guo <terry.guo@arm.com>
596
597 * doc/invoke.texi: Document ARM -march=armv7e-m.
598
599 2014-02-08 Jakub Jelinek <jakub@redhat.com>
600
601 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
602 flag on __cilkrts_rethrow builtin.
603
604 PR ipa/60026
605 * ipa-cp.c (determine_versionability): Fail at -O0
606 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
607 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
608
609 Revert:
610 2014-02-04 Jakub Jelinek <jakub@redhat.com>
611
612 PR ipa/60026
613 * tree-inline.c (copy_forbidden): Fail for
614 __attribute__((optimize (0))) functions.
615
616 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
617
618 * varpool.c: Include pointer-set.h.
619 (varpool_remove_unreferenced_decls): Variables in other partitions
620 will not be output; be however careful to not lose information
621 about partitioning.
622
623 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
624
625 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
626 lookup in the vtable constructor.
627
628 2014-02-07 Jeff Law <law@redhat.com>
629
630 PR target/40977
631 * config/m68k/m68k.md (ashldi_extsi): Turn into a
632 define_insn_and_split.
633
634 * ipa-inline.c (inline_small_functions): Fix typos.
635
636 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
637
638 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
639 (s390_can_use_return_insn): Declare.
640 * config/s390/s390.h (EPILOGUE_USES): Define.
641 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
642 instructions.
643 (s390_chunkify_start): Handle return JUMP_LABELs.
644 (s390_early_mach): Emit a main_pool instruction on the entry edge.
645 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
646 (s390_can_use_return_insn): New functions.
647 (s390_fix_long_loop_prediction): Handle conditional returns.
648 (TARGET_SET_UP_BY_PROLOGUE): Define.
649 * config/s390/s390.md (ANY_RETURN): New code iterator.
650 (*creturn, *csimple_return, return, simple_return): New patterns.
651
652 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
653
654 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
655 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
656 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
657 REG_CFA_RESTORE list when deciding not to restore a register.
658
659 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
660
661 * config/s390/s390.c: Include tree-pass.h and context.h.
662 (s390_early_mach): New function, split out from...
663 (s390_emit_prologue): ...here.
664 (pass_data_s390_early_mach): New pass structure.
665 (pass_s390_early_mach): New class.
666 (s390_option_override): Create and register early_mach pass.
667 Move to end of file.
668
669 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
670
671 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
672 to match for the exit block.
673
674 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
675
676 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
677 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
678 Reject misaligned operands.
679
680 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
681
682 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
683
684 2014-02-07 Richard Biener <rguenther@suse.de>
685
686 PR middle-end/60092
687 * gimple-low.c (lower_builtin_posix_memalign): New function.
688 (lower_stmt): Call it to lower posix_memalign in a way
689 to make alignment info accessible.
690
691 2014-02-07 Jakub Jelinek <jakub@redhat.com>
692
693 PR c++/60082
694 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
695 __builtin_setjmp_receiver.
696
697 2014-02-07 Richard Biener <rguenther@suse.de>
698
699 PR middle-end/60092
700 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
701 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
702 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
703 Handle BUILT_IN_POSIX_MEMALIGN.
704 (find_func_clobbers): Likewise.
705 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
706 (call_may_clobber_ref_p_1): Likewise.
707
708 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
709
710 PR ipa/59918
711 * ipa-devirt.c (record_target_from_binfo): Remove overactive
712 sanity check.
713
714 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
715
716 PR ipa/59469
717 * lto-cgraph.c (lto_output_node): Use
718 symtab_get_symbol_partitioning_class.
719 (lto_output_varpool_node): likewise.
720 (symtab_get_symbol_partitioning_class): Move here from
721 lto/lto-partition.c
722 * cgraph.h (symbol_partitioning_class): Likewise.
723 (symtab_get_symbol_partitioning_class): Declare.
724
725 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
726
727 * ggc.h (ggc_internal_cleared_alloc): New macro.
728 * vec.h (vec_safe_copy): Handle memory stats.
729 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
730 * target-globals.c (save_target_globals): Likewise.
731
732 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
733
734 PR target/60077
735 * expr.c (emit_move_resolve_push): Export; be bit more selective
736 on when to clear alias set.
737 * expr.h (emit_move_resolve_push): Declare.
738 * function.h (struct function): Add tail_call_marked.
739 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
740 * config/i386/i386-protos.h (ix86_expand_push): Remove.
741 * config/i386/i386.md (TImode move expander): De not call
742 ix86_expand_push.
743 (FP push expanders): Preserve memory attributes.
744 * config/i386/sse.md (push<mode>1): Remove.
745 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
746 (ix86_expand_push): Remove.
747 * config/i386/mmx.md (push<mode>1): Remove.
748
749 2014-02-06 Jakub Jelinek <jakub@redhat.com>
750
751 PR rtl-optimization/60030
752 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
753 lopart with paradoxical subreg before shifting it up by hprec.
754
755 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
756
757 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
758 Remove extra newline at end of file.
759 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
760 (arm_issue_rate): Handle cortexa57.
761 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
762 (cortex-a57.cortex-a53): Likewise.
763
764 2014-02-06 Jakub Jelinek <jakub@redhat.com>
765
766 PR target/59575
767 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
768 don't record in REG_FRAME_RELATED_EXPR registers not set in that
769 bitmask.
770 (arm_expand_prologue): Adjust all callers.
771 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
772 info, registers also at the lowest numbered registers side. Use
773 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
774 XEXP.
775
776 PR debug/59992
777 * var-tracking.c (adjust_mems): Before adding a SET to
778 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
779
780 2014-02-06 Alan Modra <amodra@gmail.com>
781
782 PR target/60032
783 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
784 change SDmode to DDmode when lra_in_progress.
785
786 2014-02-06 Jakub Jelinek <jakub@redhat.com>
787
788 PR middle-end/59150
789 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
790 free_data_ref on the dr first, and before goto again also set dr
791 to the next dr. For simd_lane_access, free old datarefs[i] before
792 overwriting it. For get_vectype_for_scalar_type failure, don't
793 free_data_ref if simd_lane_access.
794
795 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
796
797 PR target/60062
798 * tree.h (opts_for_fn): New inline function.
799 (opt_for_fn): Define.
800 * config/i386/i386.c (ix86_function_regparm): Use
801 opt_for_fn (decl, optimize) instead of optimize.
802
803 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
804
805 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
806 for SYMBOL_REF in large memory model.
807
808 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
809
810 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
811 and crypto support.
812 (cortex-a57): Likewise.
813 (cortex-a57.cortex-a53): Likewise.
814
815 2014-02-06 Yury Gribov <y.gribov@samsung.com>
816 Kugan Vivekanandarajah <kuganv@linaro.org>
817
818 * config/arm/arm.c (arm_vector_alignment_reachable): Check
819 unaligned_access.
820 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
821
822 2014-02-06 Richard Biener <rguenther@suse.de>
823
824 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
825 set_loop_copy and initialize_original_copy_tables.
826
827 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
828
829 * config/aarch64/aarch64-simd.md
830 (aarch64_ashr_simddi): Change QI to SI.
831
832 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
833 Jakub Jelinek <jakub@redhat.com>
834
835 PR middle-end/60013
836 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
837 of the dataflow.
838
839 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
840
841 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
842 CODE_FOR_altivec_vpku[hw]um to
843 CODE_FOR_altivec_vpku[hw]um_direct.
844 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
845 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
846 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
847 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
848
849 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
850
851 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
852 generation for -maltivec=be.
853 (altivec_vsumsws): Simplify redundant test.
854
855 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
856
857 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
858 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
859 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
860 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
861 gen_altivec_vpkuwum.
862 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
863 BYTES_BIG_ENDIAN.
864 (altivec_vpks<VI_char>ss): Likewise.
865 (altivec_vpks<VI_char>us): Likewise.
866 (altivec_vpku<VI_char>us): Likewise.
867 (altivec_vpku<VI_char>um): Likewise.
868 (altivec_vpku<VI_char>um_direct): New (copy of
869 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
870 internal use).
871 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
872 target is little endian and -maltivec=be is not specified.
873 (*altivec_vupkhs<VU_char>_direct): New (copy of
874 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
875 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
876 target is little endian and -maltivec=be is not specified.
877 (*altivec_vupkls<VU_char>_direct): New (copy of
878 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
879 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
880 little endian and -maltivec=be is not specified.
881 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
882 little endian and -maltivec=be is not specified.
883
884 2014-02-05 Richard Henderson <rth@redhat.com>
885
886 PR debug/52727
887 * combine-stack-adj.c: Revert r206943.
888 * sched-int.h (struct deps_desc): Add last_args_size.
889 * sched-deps.c (init_deps): Initialize it.
890 (sched_analyze_insn): Add OUTPUT dependencies between insns that
891 contain REG_ARGS_SIZE notes.
892
893 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
894
895 * lto-cgraph.c (asm_nodes_output): Make global.
896 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
897 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
898 (driver_handle_option): Handle OPT_fwpa.
899
900 2014-02-05 Jakub Jelinek <jakub@redhat.com>
901
902 PR ipa/59947
903 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
904 a comment typo and formatting issue. If odr_hash hasn't been
905 created, return vNULL and set *completep to false.
906
907 PR middle-end/57499
908 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
909 bb with no successors.
910
911 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
912
913 PR target/59718
914 * doc/invoke.texi (-march): Clarify documentation for ARM.
915 (-mtune): Likewise.
916 (-mcpu): Likewise.
917
918 2014-02-05 Richard Biener <rguenther@suse.de>
919
920 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
921 when not vectorizing because of too many alias checks.
922 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
923 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
924
925 2014-02-05 Nick Clifton <nickc@redhat.com>
926
927 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
928 accept extended registers in any mode when compiling for the MN10300.
929
930 2014-02-05 Yury Gribov <y.gribov@samsung.com>
931
932 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
933 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
934 sanitization attributes.
935 (can_inline_edge_p): Likewise.
936 (sanitize_attrs_match_for_inline_p): New function.
937
938 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
939
940 * ipa-prop.c (detect_type_change): Shor circuit testing of
941 type changes on THIS pointer.
942
943 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
944
945 PR target/59777
946 * config/pa/pa.c (legitimize_tls_address): Return original address
947 if not passed a SYMBOL_REF rtx.
948 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
949 addresses.
950 (pa_emit_move_sequence): Simplify TLS source operands.
951 (pa_legitimate_constant_p): Reject all TLS constants.
952 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
953 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
954
955 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
956
957 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
958 groups when we know they are controlled by LTO.
959 * varasm.c (default_binds_local_p_1): If object is in other partition,
960 it will be resolved locally.
961
962 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
963
964 * config/host-linux.c (linux_gt_pch_use_address): Don't
965 use SSIZE_MAX because it is not always defined.
966
967 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
968
969 PR bootstrap/59913
970 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
971 threshold for pseudo splitting.
972 (update_ebb_live_info): Process call argument hard registers and
973 hard registers from insn definition too.
974 (max_small_class_regs_num): New constant.
975 (inherit_in_ebb): Update live hard regs through EBBs. Update
976 reloads_num only for small register classes. Don't split for
977 outputs of jumps.
978
979 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
980
981 PR ipa/60058
982 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
983 is non-null.
984
985 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
986
987 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
988 visibility is safe.
989
990 2014-02-04 Marek Polacek <polacek@redhat.com>
991
992 * gdbinit.in (pel): Define.
993
994 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
995
996 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
997 behavior.
998
999 2014-02-04 Richard Biener <rguenther@suse.de>
1000
1001 PR lto/59723
1002 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1003 in function context local.
1004 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1005 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1006 similar to LTO_imported_decl_ref.
1007
1008 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1009
1010 PR tree-optimization/60002
1011 * cgraphclones.c (build_function_decl_skip_args): Clear
1012 DECL_LANG_SPECIFIC.
1013
1014 PR tree-optimization/60023
1015 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1016 false to gsi_replace.
1017 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1018 has been in some EH region and vec_stmt could throw, add
1019 vec_stmt into the same EH region.
1020 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1021 has no lhs, ignore it.
1022 * internal-fn.c (expand_MASK_LOAD): Likewise.
1023
1024 PR ipa/60026
1025 * tree-inline.c (copy_forbidden): Fail for
1026 __attribute__((optimize (0))) functions.
1027
1028 PR other/58712
1029 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1030 is set, copy one less argument.
1031 (expand_simd_clones): Don't subtract clone_info->inbranch
1032 from simd_clone_struct_alloc argument.
1033
1034 PR rtl-optimization/57915
1035 * recog.c (simplify_while_replacing): If all unary/binary/relational
1036 operation arguments are constant, attempt to simplify those.
1037
1038 PR middle-end/59261
1039 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1040 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1041
1042 2014-02-04 Richard Biener <rguenther@suse.de>
1043
1044 PR tree-optimization/60012
1045 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1046 TBAA disambiguation to all DDRs.
1047
1048 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1049
1050 PR target/59788
1051 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1052 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1053
1054 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1055
1056 PR ipa/59882
1057 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1058
1059 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1060
1061 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1062 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1063
1064 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1065
1066 PR ipa/59831
1067 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1068 to figure out targets of polymorphic calls with known decl.
1069 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1070 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1071 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1072 (get_polymorphic_call_info): ... here.
1073 (get_polymorphic_call_info_from_invariant): New function.
1074
1075 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1076
1077 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1078 lookup via vtable pointer; check for type consistency
1079 and turn inconsitent facts into UNREACHABLE.
1080 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1081 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1082 type inconsistent querries; return UNREACHABLE instead.
1083
1084 2014-02-03 Richard Henderson <rth@twiddle.net>
1085
1086 PR tree-opt/59924
1087 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1088 already processed this node.
1089 (normalize_one_pred_1): Pass along mark_set.
1090 (normalize_one_pred): Create and destroy a pointer_set_t.
1091 (normalize_one_pred_chain): Likewise.
1092
1093 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1094
1095 PR gcov-profile/58602
1096 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1097
1098 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1099
1100 PR ipa/59831
1101 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1102 -fno-devirtualize; try to devirtualize by the knowledge of
1103 virtual table pointer given by aggregate propagation.
1104 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1105 (ipa_print_node_jump_functions): Dump also offset that
1106 is relevant for polymorphic calls.
1107 (determine_known_aggregate_parts): Add arg_type parameter; use it
1108 instead of determining the type from pointer type.
1109 (ipa_compute_jump_functions_for_edge): Update call of
1110 determine_known_aggregate_parts.
1111 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1112 (gimple_get_virt_method_for_binfo): ... here; simplify using
1113 vtable_pointer_value_to_vtable.
1114 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1115 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1116 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1117 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1118 POINTER_PLUS_EXPR.
1119 (vtable_pointer_value_to_binfo): ... here.
1120 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1121
1122 2014-02-03 Teresa Johnson <tejohnson@google.com>
1123
1124 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1125 redef of outer loop index variable.
1126
1127 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1128
1129 PR c++/53017
1130 PR c++/59211
1131 * doc/extend.texi (Function Attributes): Typo.
1132
1133 2014-02-03 Cong Hou <congh@google.com>
1134
1135 PR tree-optimization/60000
1136 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1137 if the vectorized statement is a store. A store statement can only
1138 appear at the end of pattern statements.
1139
1140 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1141
1142 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1143 (ix86_option_override_internal): Default long double to 64-bit for
1144 32-bit Bionic and to 128-bit for 64-bit Bionic.
1145
1146 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1147 TARGET_LONG_DOUBLE_128 is true.
1148 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1149
1150 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1151 (mlong-double-64): Negate -mlong-double-128.
1152 (mlong-double-128): New option.
1153
1154 * config/i386/i386-c.c (ix86_target_macros): Define
1155 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1156
1157 * doc/invoke.texi: Document -mlong-double-128.
1158
1159 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1160
1161 PR rtl-optimization/60024
1162 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1163
1164 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1165
1166 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1167
1168 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1169
1170 PR rtl-optimization/57662
1171 * sel-sched.c (code_motion_path_driver): Do not mark already not
1172 existing blocks in the visiting bitmap.
1173
1174 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1175
1176 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1177 on the insn being emitted.
1178
1179 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1180 Will Deacon <will.deacon@arm.com>
1181
1182 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1183
1184 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1185
1186 * config/arm/arm-tables.opt: Regenerate.
1187
1188 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1189
1190 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1191 for vector types other than V16QImode.
1192 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1193 define_expand, and call altivec_expand_vec_perm_le when producing
1194 code with little endian element order.
1195 (*altivec_vperm_<mode>_internal): New insn having previous
1196 behavior of altivec_vperm_<mode>.
1197 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1198 altivec_expand_vec_perm_le when producing code with little endian
1199 element order.
1200 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1201 behavior of altivec_vperm_<mode>_uns.
1202
1203 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1204
1205 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1206 (altivec_vsumsws): Add handling for -maltivec=be with a little
1207 endian target.
1208 (altivec_vsumsws_direct): New.
1209 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1210 gen_altivec_vsumsws.
1211
1212 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1213
1214 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1215 vtable_pointer_value_to_binfo): New functions.
1216 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1217 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1218
1219 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1220
1221 * config/nios2/nios2.md (load_got_register): Initialize GOT
1222 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1223 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1224
1225 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1226
1227 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1228 preserverd by passthrough, do not propagate the type.
1229
1230 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1231
1232 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1233 (mips_atomic_assign_expand_fenv): New function.
1234 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1235
1236 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1237
1238 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1239 (__builtin_mips_set_fcsr): Likewise.
1240 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1241 MIPS_USI_FTYPE_VOID.
1242 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1243 (mips16_expand_set_fcsr): Likewise.
1244 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1245 (mips16_set_fcsr_stub): Likewise.
1246 (mips16_get_fcsr_one_only_stub): New class.
1247 (mips16_set_fcsr_one_only_stub): Likewise.
1248 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1249 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1250 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1251 (hard_float): New availability predicate.
1252 (mips_builtins): Add get_fcsr and set_fcsr.
1253 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1254 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1255 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1256 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1257 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1258 patterns.
1259
1260 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1261
1262 * config/mips/mips.c (mips_one_only_stub): New class.
1263 (mips_need_mips16_rdhwr_p): Replace with...
1264 (mips16_rdhwr_stub): ...this new variable.
1265 (mips16_stub_call_address): New function.
1266 (mips16_rdhwr_one_only_stub): New class.
1267 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1268 (mips_output_mips16_rdhwr): Delete.
1269 (mips_finish_stub): New function.
1270 (mips_code_end): Use it to handle rdhwr stubs.
1271
1272 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1273
1274 PR target/60017
1275 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1276 when calculating size of integer atomic types.
1277
1278 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1279
1280 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1281
1282 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1283
1284 PR tree-optimization/60003
1285 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1286 * profile.c (branch_prob): Use gimple_call_builtin_p
1287 to check for BUILT_IN_SETJMP_RECEIVER.
1288 * tree-inline.c (copy_bb): Call notice_special_calls.
1289
1290 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1291
1292 PR bootstrap/59985
1293 * lra-constraints.c (process_alt_operands): Update reload_sum only
1294 on the first pass.
1295
1296 2014-01-31 Richard Henderson <rth@redhat.com>
1297
1298 PR middle-end/60004
1299 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1300 until after else_eh is processed.
1301
1302 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1303
1304 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1305 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1306 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1307 in smmintrin.h, remove them.
1308 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1309 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1310 * config/i386/i386.md (ROUND_SAE): Fix value.
1311 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1312 (const48_operand): New.
1313 * config/i386/subst.md (round), (round_expand): Use
1314 const_4_or_8_to_11_operand.
1315 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1316
1317 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1318
1319 * config/i386/constraints.md (Yk): Swap meaning with k.
1320 * config/i386/i386.md (movhi_internal): Change Yk to k.
1321 (movqi_internal): Ditto.
1322 (*k<logic><mode>): Ditto.
1323 (*andhi_1): Ditto.
1324 (*andqi_1): Ditto.
1325 (kandn<mode>): Ditto.
1326 (*<code>hi_1): Ditto.
1327 (*<code>qi_1): Ditto.
1328 (kxnor<mode>): Ditto.
1329 (kortestzhi): Ditto.
1330 (kortestchi): Ditto.
1331 (kunpckhi): Ditto.
1332 (*one_cmplhi2_1): Ditto.
1333 (*one_cmplqi2_1): Ditto.
1334 * config/i386/sse.md (): Change k to Yk.
1335 (avx512f_load<mode>_mask): Ditto.
1336 (avx512f_blendm<mode>): Ditto.
1337 (avx512f_store<mode>_mask): Ditto.
1338 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1339 (avx512f_storedqu<mode>_mask): Ditto.
1340 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1341 Ditto.
1342 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1343 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1344 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1345 (avx512f_maskcmp<mode>3): Ditto.
1346 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1347 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1348 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1349 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1350 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1351 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1352 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1353 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1354 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1355 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1356 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1357 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1358 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1359 (vec_extract_lo_<mode>_maskm): Ditto.
1360 (vec_extract_hi_<mode>_maskm): Ditto.
1361 (avx512f_vternlog<mode>_mask): Ditto.
1362 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1363 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1364 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1365 (avx512f_<code>v8div16qi2_mask): Ditto.
1366 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1367 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1368 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1369 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1370 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1371 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1372 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1373 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1374 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1375 (avx512cd_maskb_vec_dupv8di): Ditto.
1376 (avx512cd_maskw_vec_dupv16si): Ditto.
1377 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1378 (avx512f_vpermi2var<mode>3_mask): Ditto.
1379 (avx512f_vpermi2var<mode>3_mask): Ditto.
1380 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1381 (*avx512f_gathersi<mode>): Ditto.
1382 (*avx512f_gathersi<mode>_2): Ditto.
1383 (*avx512f_gatherdi<mode>): Ditto.
1384 (*avx512f_gatherdi<mode>_2): Ditto.
1385 (*avx512f_scattersi<mode>): Ditto.
1386 (*avx512f_scatterdi<mode>): Ditto.
1387 (avx512f_compress<mode>_mask): Ditto.
1388 (avx512f_compressstore<mode>_mask): Ditto.
1389 (avx512f_expand<mode>_mask): Ditto.
1390 * config/i386/subst.md (mask): Change k to Yk.
1391 (mask_scalar_merge): Ditto.
1392 (sd): Ditto.
1393
1394 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1395
1396 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1397
1398 2014-01-31 Richard Biener <rguenther@suse.de>
1399
1400 PR middle-end/59990
1401 * builtins.c (fold_builtin_memory_op): Make sure to not
1402 use a floating-point mode or a boolean or enumeral type for
1403 the copy operation.
1404
1405 2014-01-30 DJ Delorie <dj@redhat.com>
1406
1407 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1408 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1409 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1410 whenever main() has an epilogue.
1411
1412 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1413
1414 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1415 unused variable "field".
1416 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1417 (vsx_mergeh_<mode>): Likewise.
1418 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1419 (altivec_vmrghh): Likewise.
1420 (altivec_vmrghw): Likewise.
1421 (altivec_vmrglb): Likewise.
1422 (altivec_vmrglh): Likewise.
1423 (altivec_vmrglw): Likewise.
1424 (altivec_vspltb): Add missing uses.
1425 (altivec_vsplth): Likewise.
1426 (altivec_vspltw): Likewise.
1427 (altivec_vspltsf): Likewise.
1428
1429 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1430
1431 PR target/59923
1432 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1433 frame related instructions.
1434
1435 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1436
1437 PR rtl-optimization/59959
1438 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1439 any reload of register whose subreg is invalid.
1440
1441 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1442
1443 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1444 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1445 Add missing return type - void.
1446
1447 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1448
1449 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1450 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1451 remove element index adjustment for endian (now handled in vsx.md
1452 and altivec.md).
1453 (altivec_expand_vec_perm_const): Use
1454 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1455 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1456 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1457 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1458 define_expand and a new define_insn *altivec_vspltb_internal;
1459 adjust for -maltivec=be on a little endian target.
1460 (altivec_vspltb_direct): New.
1461 (altivec_vsplth): Divide into a define_expand and a new
1462 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1463 little endian target.
1464 (altivec_vsplth_direct): New.
1465 (altivec_vspltw): Divide into a define_expand and a new
1466 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1467 little endian target.
1468 (altivec_vspltw_direct): New.
1469 (altivec_vspltsf): Divide into a define_expand and a new
1470 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1471 a little endian target.
1472
1473 2014-01-30 Richard Biener <rguenther@suse.de>
1474
1475 PR tree-optimization/59993
1476 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1477 can propagate form the earlier stmt and avoid the transform
1478 when the intermediate result is needed.
1479
1480 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1481
1482 * README.Portability: Fix typo.
1483
1484 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1485
1486 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1487 comparison_operator with ordered_comparison_operator.
1488
1489 2014-01-30 Nick Clifton <nickc@redhat.com>
1490
1491 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1492 Rename to mn10300_store_multiple_regs.
1493 * config/mn10300/mn10300.c: Likewise.
1494 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1495 store_multiple_regs.
1496 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1497 Call mn10300_store_multiple_regs.
1498
1499 2014-01-30 Nick Clifton <nickc@redhat.com>
1500 DJ Delorie <dj@redhat.com>
1501
1502 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1503 %fp 2 to keep registers after it properly word-aligned.
1504 (rl78_alloc_physical_registers_umul): Handle the case where both
1505 input operands are the same.
1506
1507 2014-01-30 Richard Biener <rguenther@suse.de>
1508
1509 PR tree-optimization/59903
1510 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1511 check properly.
1512
1513 2014-01-30 Jason Merrill <jason@redhat.com>
1514
1515 PR c++/59633
1516 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1517
1518 PR c++/59645
1519 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1520
1521 2014-01-30 Richard Biener <rguenther@suse.de>
1522
1523 PR tree-optimization/59951
1524 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1525
1526 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1527
1528 PR target/59784
1529 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1530 SFmode to DFmode case.
1531
1532 2014-01-29 DJ Delorie <dj@redhat.com>
1533
1534 * config/msp430/msp430.opt (-minrt): New.
1535 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1536 if -minrt given.
1537 (ENDFILE_SPEC): Likewise.
1538
1539 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1540
1541 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1542 (estimate_function_body_sizes): Use it.
1543
1544 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1545
1546 PR c++/58561
1547 * dwarf2out.c (is_cxx_auto): New.
1548 (is_base_type): Use it.
1549 (gen_type_die_with_usage): Likewise.
1550
1551 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1552
1553 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1554 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1555 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1556 -maltivec=be with LE targets.
1557 (vsx_mergeh_<mode>): Likewise.
1558 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1559 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1560 (altivec_vmrghb): Replace with define_expand and new
1561 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1562 (altivec_vmrghb_direct): New define_insn.
1563 (altivec_vmrghh): Replace with define_expand and new
1564 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1565 (altivec_vmrghh_direct): New define_insn.
1566 (altivec_vmrghw): Replace with define_expand and new
1567 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1568 (altivec_vmrghw_direct): New define_insn.
1569 (*altivec_vmrghsf): Adjust for endianness.
1570 (altivec_vmrglb): Replace with define_expand and new
1571 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1572 (altivec_vmrglb_direct): New define_insn.
1573 (altivec_vmrglh): Replace with define_expand and new
1574 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1575 (altivec_vmrglh_direct): New define_insn.
1576 (altivec_vmrglw): Replace with define_expand and new
1577 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1578 (altivec_vmrglw_direct): New define_insn.
1579 (*altivec_vmrglsf): Adjust for endianness.
1580 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1581 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1582 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1583 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1584 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1585 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1586 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1587 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1588
1589 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1590
1591 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1592 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1593 whitespace.
1594
1595 2014-01-29 Richard Biener <rguenther@suse.de>
1596
1597 PR tree-optimization/58742
1598 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1599 associate_pointerplus_align.
1600 (associate_pointerplus_diff): New function.
1601 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1602 and associate_pointerplus_diff.
1603
1604 2014-01-29 Richard Biener <rguenther@suse.de>
1605
1606 * lto-streamer.h (LTO_major_version): Bump to 3.
1607 (LTO_minor_version): Reset to 0.
1608
1609 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1610
1611 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1612 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1613 (arm_file_start): Generate correct asm header for armv7ve.
1614 * config/arm/bpabi.h: Add multilib support for armv7ve.
1615 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1616 and cortex-a15 to armv7ve.
1617 * config/arm/t-aprofile: Add multilib support for armv7ve.
1618 * doc/invoke.texi: Document -march=armv7ve.
1619
1620 2014-01-29 Richard Biener <rguenther@suse.de>
1621
1622 PR tree-optimization/58742
1623 * tree-ssa-forwprop.c (associate_plusminus): Return true
1624 if we changed sth, defer EH cleanup to ...
1625 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1626 (simplify_mult): New function.
1627
1628 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1629
1630 PR middle-end/59917
1631 PR tree-optimization/59920
1632 * tree.c (build_common_builtin_nodes): Remove
1633 __builtin_setjmp_dispatcher initialization.
1634 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1635 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1636 instead of gsi_after_labels + manually skipping debug stmts.
1637 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1638 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1639 * tree-inline.c (copy_edges_for_bb): Remove
1640 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1641 argument. Ignore computed_goto_p stmts. Don't call
1642 make_abnormal_goto_edges. If a call might need abnormal edges
1643 for non-local gotos, see if it already has an edge to
1644 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1645 with true argument, don't do anything then, otherwise add
1646 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1647 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1648 caller.
1649 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1650 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1651 (lower_stmt): Don't set data->calls_builtin_setjmp.
1652 (lower_builtin_setjmp): Adjust comment.
1653 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1654 * tree-cfg.c (found_computed_goto): Remove.
1655 (factor_computed_gotos): Remove.
1656 (make_goto_expr_edges): Return bool, true for computed gotos.
1657 Don't call make_abnormal_goto_edges.
1658 (build_gimple_cfg): Don't set found_computed_goto, don't call
1659 factor_computed_gotos.
1660 (computed_goto_p): No longer static.
1661 (make_blocks): Don't set found_computed_goto.
1662 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1663 (make_edges): If make_goto_expr_edges returns true, push bb
1664 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1665 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1666 vector. Record mapping between bbs and OpenMP regions if there
1667 are any, adjust make_gimple_omp_edges caller. Call
1668 handle_abnormal_edges.
1669 (make_abnormal_goto_edges): Remove.
1670 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1671 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1672 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1673 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1674 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1675 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1676 filling *region also set *region_idx to (*region)->entry->index.
1677
1678 PR other/58712
1679 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1680 For REGs set ORIGINAL_REGNO.
1681
1682 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1683
1684 * doc/md.texi: Mention that a target shouldn't implement
1685 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1686 than hi/lo pair.
1687
1688 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1689
1690 PR tree-optimization/59594
1691 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1692 a copy of the datarefs vector rather than the vector itself.
1693
1694 2014-01-28 Jason Merrill <jason@redhat.com>
1695
1696 PR c++/53756
1697 * dwarf2out.c (auto_die): New static.
1698 (gen_type_die_with_usage): Handle C++1y 'auto'.
1699 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1700 on definition.
1701
1702 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1703
1704 PR target/59672
1705 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1706 (SPEC_X32): Likewise.
1707 (SPEC_64): Likewise.
1708 * config/i386/i386.c (ix86_option_override_internal): Turn off
1709 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1710 for TARGET_16BIT.
1711 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1712 * config/i386/i386.h (TARGET_16BIT): New macro.
1713 (TARGET_16BIT_P): Likewise.
1714 * config/i386/i386.opt: Add m16.
1715 * doc/invoke.texi: Document -m16.
1716
1717 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1718
1719 PR preprocessor/59935
1720 * input.c (location_get_source_line): Bail out on when line number
1721 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1722
1723 2014-01-28 Richard Biener <rguenther@suse.de>
1724
1725 PR tree-optimization/58742
1726 * tree-ssa-forwprop.c (associate_plusminus): Handle
1727 pointer subtraction of the form (T)(P + A) - (T)P.
1728
1729 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1730
1731 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1732 at const_int_cost.
1733
1734 2014-01-28 Richard Biener <rguenther@suse.de>
1735
1736 Revert
1737 2014-01-28 Richard Biener <rguenther@suse.de>
1738
1739 PR rtl-optimization/45364
1740 PR rtl-optimization/59890
1741 * var-tracking.c (local_get_addr_clear_given_value): Handle
1742 already cleared slot.
1743 (val_reset): Handle not allocated local_get_addr_cache.
1744 (vt_find_locations): Use post-order on the inverted CFG.
1745
1746 2014-01-28 Richard Biener <rguenther@suse.de>
1747
1748 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1749
1750 2014-01-28 Richard Biener <rguenther@suse.de>
1751
1752 PR rtl-optimization/45364
1753 PR rtl-optimization/59890
1754 * var-tracking.c (local_get_addr_clear_given_value): Handle
1755 already cleared slot.
1756 (val_reset): Handle not allocated local_get_addr_cache.
1757 (vt_find_locations): Use post-order on the inverted CFG.
1758
1759 2014-01-28 Alan Modra <amodra@gmail.com>
1760
1761 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1762 * configure.ac <recursive call for build != host>: Define
1763 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1764 and LD_FOR_BUILD too.
1765 * configure: Regenerate.
1766
1767 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1768
1769 * config/i386/i386.c (get_builtin_code_for_version): Separate
1770 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1771 Broadwell from Haswell.
1772
1773 2014-01-27 Steve Ellcey <sellcey@mips.com>
1774
1775 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1776 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1777 * config/mips/mips.c (mips_option_override): Change setting
1778 of TARGET_DSP.
1779 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1780 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1781 Change from Mask to Var.
1782
1783 2014-01-27 Jeff Law <law@redhat.com>
1784
1785 * ipa-inline.c (inline_small_functions): Fix typo.
1786
1787 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1788
1789 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1790 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1791 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1792 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1793 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1794 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1795 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1796 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1797 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1798 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1799 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1800 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1801 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1802 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1803 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1804 (_mm512_storeu_epi64): Ditto.
1805 (_mm512_cmpge_epi32_mask): Ditto.
1806 (_mm512_cmpge_epu32_mask): Ditto.
1807 (_mm512_cmpge_epi64_mask): Ditto.
1808 (_mm512_cmpge_epu64_mask): Ditto.
1809 (_mm512_cmple_epi32_mask): Ditto.
1810 (_mm512_cmple_epu32_mask): Ditto.
1811 (_mm512_cmple_epi64_mask): Ditto.
1812 (_mm512_cmple_epu64_mask): Ditto.
1813 (_mm512_cmplt_epi32_mask): Ditto.
1814 (_mm512_cmplt_epu32_mask): Ditto.
1815 (_mm512_cmplt_epi64_mask): Ditto.
1816 (_mm512_cmplt_epu64_mask): Ditto.
1817 (_mm512_cmpneq_epi32_mask): Ditto.
1818 (_mm512_cmpneq_epu32_mask): Ditto.
1819 (_mm512_cmpneq_epi64_mask): Ditto.
1820 (_mm512_cmpneq_epu64_mask): Ditto.
1821 (_mm512_expand_pd): Ditto.
1822 (_mm512_expand_ps): Ditto.
1823 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1824 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1825 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1826 * config/i386/i386.c (ix86_builtins): Add
1827 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1828 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1829 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1830 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1831 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1832 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1833 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1834 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1835 IX86_BUILTIN_PMOVUSQW512_MEM.
1836 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1837 __builtin_ia32_pmovsqd512mem_mask,
1838 __builtin_ia32_pmovqd512mem_mask,
1839 __builtin_ia32_pmovusqw512mem_mask,
1840 __builtin_ia32_pmovsqw512mem_mask,
1841 __builtin_ia32_pmovqw512mem_mask,
1842 __builtin_ia32_pmovusdw512mem_mask,
1843 __builtin_ia32_pmovsdw512mem_mask,
1844 __builtin_ia32_pmovdw512mem_mask,
1845 __builtin_ia32_pmovqb512mem_mask,
1846 __builtin_ia32_pmovusqb512mem_mask,
1847 __builtin_ia32_pmovsqb512mem_mask,
1848 __builtin_ia32_pmovusdb512mem_mask,
1849 __builtin_ia32_pmovsdb512mem_mask,
1850 __builtin_ia32_pmovdb512mem_mask.
1851 (bdesc_args): Add __builtin_ia32_expanddf512,
1852 __builtin_ia32_expandsf512.
1853 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1854 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1855 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1856 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1857 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1858 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1859 (avx512f_<code>v8div16qi2_mask_store): This.
1860 (avx512f_expand<mode>): New.
1861
1862 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1863
1864 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1865 New.
1866 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1867 (_mm512_prefetch_i32scatter_pd): Ditto.
1868 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1869 (_mm512_prefetch_i64scatter_pd): Ditto.
1870 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1871 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1872 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1873 (_mm512_prefetch_i32scatter_ps): Ditto.
1874 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1875 (_mm512_prefetch_i64scatter_ps): Ditto.
1876 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1877 * config/i386/i386-builtin-types.def: Define
1878 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1879 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1880 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1881 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1882 IX86_BUILTIN_SCATTERPFQPD.
1883 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1884 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1885 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1886 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1887 __builtin_ia32_scatterpfqps.
1888 (ix86_expand_builtin): Expand new built-ins.
1889 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1890 fix memory access data type.
1891 (*avx512pf_gatherpf<mode>_mask): Ditto.
1892 (*avx512pf_gatherpf<mode>): Ditto.
1893 (avx512pf_scatterpf<mode>): Ditto.
1894 (*avx512pf_scatterpf<mode>_mask): Ditto.
1895 (*avx512pf_scatterpf<mode>): Ditto.
1896 (GATHER_SCATTER_SF_MEM_MODE): New.
1897 (avx512pf_gatherpf<mode>df): Ditto.
1898 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1899 (*avx512pf_scatterpf<mode>df): Ditto.
1900
1901 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1902
1903 PR bootstrap/59934
1904 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1905 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1906 reached.
1907
1908 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1909
1910 * common/config/arm/arm-common.c
1911 (arm_rewrite_mcpu): Handle multiple names.
1912 * config/arm/arm.h
1913 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1914
1915 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1916
1917 * gimple-builder.h (create_gimple_tmp): Delete.
1918
1919 2014-01-27 Christian Bruel <christian.bruel@st.com>
1920
1921 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1922 words comparisons.
1923
1924 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1925
1926 * config/pa/pa.md (call): Generate indirect long calls to non-local
1927 functions when outputing 32-bit code.
1928 (call_value): Likewise except for special call to buggy powf function.
1929
1930 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1931 portable runtime and PIC indirect calls.
1932 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1933 and PIC call sequences. Use ldo instead of blr to set return register
1934 in PIC call sequence.
1935
1936 2014-01-25 Walter Lee <walt@tilera.com>
1937
1938 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1939 avoid clobbering a live register.
1940
1941 2014-01-25 Walter Lee <walt@tilera.com>
1942
1943 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1944 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1945 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1946 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1947
1948 2014-01-25 Walter Lee <walt@tilera.com>
1949
1950 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1951 arguments on even registers.
1952 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1953 STACK_BOUNDARY.
1954 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1955 (BIGGEST_ALIGNMENT): Ditto.
1956 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1957
1958 2014-01-25 Walter Lee <walt@tilera.com>
1959
1960 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1961 insns before bundling.
1962 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1963
1964 2014-01-25 Walter Lee <walt@tilera.com>
1965
1966 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1967 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1968 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1969
1970 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1971
1972 * config/mips/constraints.md (kl): Delete.
1973 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1974 define expands, using...
1975 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1976 instructions for MIPS16.
1977 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1978 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1979
1980 2014-01-25 Walter Lee <walt@tilera.com>
1981
1982 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1983 (clzdi2): Ditto.
1984 (ffsdi2): Ditto.
1985
1986 2014-01-25 Walter Lee <walt@tilera.com>
1987
1988 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1989 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1990
1991 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1992
1993 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1994 Handle XOR.
1995
1996 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1997
1998 * print-rtl.c (in_call_function_usage): New var.
1999 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2000 EXPR_LIST mode as mode and not as reg note name.
2001
2002 PR middle-end/59561
2003 * cfgloopmanip.c (copy_loop_info): If
2004 loop->warned_aggressive_loop_optimizations, make sure
2005 the flag is set in target loop too.
2006
2007 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2008
2009 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2010 flag_cilkplus.
2011 * builtins.def: Likewise.
2012 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2013 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2014 * ira.c (ira_setup_eliminable_regset): Likewise.
2015 * omp-low.c (gate_expand_omp): Likewise.
2016 (execute_lower_omp): Likewise.
2017 (diagnose_sb_0): Likewise.
2018 (gate_diagnose_omp_blocks): Likewise.
2019 (simd_clone_clauses_extract): Likewise.
2020 (gate): Likewise.
2021
2022 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2023
2024 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2025 correction for little endian...
2026 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2027 here.
2028
2029 2014-01-24 Jeff Law <law@redhat.com>
2030
2031 PR tree-optimization/59919
2032 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2033 for non-returning calls.
2034
2035 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2036
2037 * common/config/aarch64/aarch64-common.c
2038 (aarch64_rewrite_mcpu): Handle multiple names.
2039 * config/aarch64/aarch64.h
2040 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2041
2042 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2043
2044 * input.c (add_file_to_cache_tab): Handle the case where fopen
2045 returns NULL.
2046
2047 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2048
2049 PR target/59929
2050 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2051 from push operand if code of push isn't PRE_DEC.
2052
2053 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2054
2055 PR target/59909
2056 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2057 -mquad-memory-atomic. Update -mquad-memory documentation to say
2058 it is only used for non-atomic loads/stores.
2059
2060 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2061 -mquad-memory or -mquad-memory-atomic switches.
2062
2063 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2064 -mquad-memory-atomic to ISA 2.07 support.
2065
2066 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2067 to separate support of normal quad word memory operations (ldq, stq)
2068 from the atomic quad word memory operations.
2069
2070 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2071 support to separate non-atomic quad word operations from atomic
2072 quad word operations. Disable non-atomic quad word operations in
2073 little endian mode so that we don't have to swap words after the
2074 load and before the store.
2075 (quad_load_store_p): Add comment about atomic quad word support.
2076 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2077 options printed with -mdebug=reg.
2078
2079 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2080 -mquad-memory-atomic as the test for whether we have quad word
2081 atomic instructions.
2082 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2083 or -mp8-vector are used, allow byte/half-word atomic operations.
2084
2085 * config/rs6000/sync.md (load_lockedti): Insure that the address
2086 is a proper indexed or indirect address for the lqarx instruction.
2087 On little endian systems, swap the hi/lo registers after the lqarx
2088 instruction.
2089 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2090 insure the address is valid for the lqarx instruction.
2091 (store_conditionalti): Insure that the address is a proper indexed
2092 or indirect address for the stqcrx. instruction. On little endian
2093 systems, swap the hi/lo registers before doing the stqcrx.
2094 instruction.
2095 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2096 insure the address is valid for the stqcrx. instruction.
2097
2098 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2099 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2100 type of quad memory support is available.
2101
2102 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2103
2104 PR regression/59915
2105 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2106 there is a danger of looping.
2107
2108 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2109
2110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2111 force flag_ira_loop_pressure if set via command line.
2112
2113 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2114
2115 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2116 (ashr_simd): New builtin handling DI mode.
2117 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2118 (aarch64_sshr_simddi): New match pattern.
2119 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2120 (vshrd_n_s64): Likewise.
2121 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2122
2123 2014-01-23 Nick Clifton <nickc@redhat.com>
2124
2125 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2126 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2127 favour of mcu specific scripts.
2128 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2129 430x multilibs.
2130
2131 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2132 Alex Velenko <Alex.Velenko@arm.com>
2133
2134 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2135 (vaddv_s16): Likewise.
2136 (vaddv_s32): Likewise.
2137 (vaddv_u8): Likewise.
2138 (vaddv_u16): Likewise.
2139 (vaddv_u32): Likewise.
2140 (vaddvq_s8): Likewise.
2141 (vaddvq_s16): Likewise.
2142 (vaddvq_s32): Likewise.
2143 (vaddvq_s64): Likewise.
2144 (vaddvq_u8): Likewise.
2145 (vaddvq_u16): Likewise.
2146 (vaddvq_u32): Likewise.
2147 (vaddvq_u64): Likewise.
2148 (vaddv_f32): Likewise.
2149 (vaddvq_f32): Likewise.
2150 (vaddvq_f64): Likewise.
2151 (vmaxv_f32): Likewise.
2152 (vmaxv_s8): Likewise.
2153 (vmaxv_s16): Likewise.
2154 (vmaxv_s32): Likewise.
2155 (vmaxv_u8): Likewise.
2156 (vmaxv_u16): Likewise.
2157 (vmaxv_u32): Likewise.
2158 (vmaxvq_f32): Likewise.
2159 (vmaxvq_f64): Likewise.
2160 (vmaxvq_s8): Likewise.
2161 (vmaxvq_s16): Likewise.
2162 (vmaxvq_s32): Likewise.
2163 (vmaxvq_u8): Likewise.
2164 (vmaxvq_u16): Likewise.
2165 (vmaxvq_u32): Likewise.
2166 (vmaxnmv_f32): Likewise.
2167 (vmaxnmvq_f32): Likewise.
2168 (vmaxnmvq_f64): Likewise.
2169 (vminv_f32): Likewise.
2170 (vminv_s8): Likewise.
2171 (vminv_s16): Likewise.
2172 (vminv_s32): Likewise.
2173 (vminv_u8): Likewise.
2174 (vminv_u16): Likewise.
2175 (vminv_u32): Likewise.
2176 (vminvq_f32): Likewise.
2177 (vminvq_f64): Likewise.
2178 (vminvq_s8): Likewise.
2179 (vminvq_s16): Likewise.
2180 (vminvq_s32): Likewise.
2181 (vminvq_u8): Likewise.
2182 (vminvq_u16): Likewise.
2183 (vminvq_u32): Likewise.
2184 (vminnmv_f32): Likewise.
2185 (vminnmvq_f32): Likewise.
2186 (vminnmvq_f64): Likewise.
2187
2188 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2189
2190 * config/aarch64/aarch64-simd.md
2191 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2192 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2193 (*aarch64_mul3_elt<mode>): Likewise.
2194 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2195 (*aarch64_mul3_elt_to_64v2df): Likewise.
2196 (*aarch64_mla_elt<mode>): Likewise.
2197 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2198 (*aarch64_mls_elt<mode>): Likewise.
2199 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2200 (*aarch64_fma4_elt<mode>): Likewise.
2201 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2202 (*aarch64_fma4_elt_to_64v2df): Likewise.
2203 (*aarch64_fnma4_elt<mode>): Likewise.
2204 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2205 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2206 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2207 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2208 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2209 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2210 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2211 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2212 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2213
2214 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2215
2216 * config/aarch64/aarch64-simd.md
2217 (aarch64_be_checked_get_lane<mode>): New define_expand.
2218 * config/aarch64/aarch64-simd-builtins.def
2219 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2220 New builtin definition.
2221 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2222 Use new safe be builtin.
2223
2224 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2225
2226 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2227 New define_insn.
2228 (aarch64_be_st1<mode>): Likewise.
2229 (aarch_ld1<VALL:mode>): Define_expand modified.
2230 (aarch_st1<VALL:mode>): Likewise.
2231 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2232 (UNSPEC_ST1): Likewise.
2233
2234 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2235
2236 * config/microblaze/microblaze.md: Add trap insn and attribute
2237
2238 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2239
2240 PR preprocessor/58580
2241 * input.h (location_get_source_line): Take an additional line_size
2242 parameter.
2243 (void diagnostics_file_cache_fini): Declare new function.
2244 * input.c (struct fcache): New type.
2245 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2246 New static constants.
2247 (diagnostic_file_cache_init, total_lines_num)
2248 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2249 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2250 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2251 (get_next_line, read_next_line, goto_next_line, read_line_num):
2252 New static function definitions.
2253 (diagnostic_file_cache_fini): New function.
2254 (location_get_source_line): Take an additional output line_len
2255 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2256 read_line_num.
2257 * diagnostic.c (diagnostic_finish): Call
2258 diagnostic_file_cache_fini.
2259 (adjust_line): Take an additional input parameter for the length
2260 of the line, rather than calculating it with strlen.
2261 (diagnostic_show_locus): Adjust the use of
2262 location_get_source_line and adjust_line with respect to their new
2263 signature. While displaying a line now, do not stop at the first
2264 null byte. Rather, display the zero byte as a space and keep
2265 going until we reach the size of the line.
2266 * Makefile.in: Add vec.o to OBJS-libcommon
2267
2268 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2269 Ilya Tocar <ilya.tocar@intel.com>
2270
2271 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2272 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2273 (__builtin_ia32_kmov16): Ditto.
2274 * config/i386/i386.md (UNSPEC_KMOV): New.
2275 (kmovw): Ditto.
2276
2277 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2278
2279 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2280 (_mm512_storeu_si512): Ditto.
2281
2282 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2283
2284 PR target/52125
2285 * rtl.h (get_referenced_operands): Declare.
2286 * recog.c (get_referenced_operands): New function.
2287 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2288 operands have been referenced when recording LO_SUM references.
2289
2290 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2291
2292 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2293
2294 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2295
2296 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2297 Enable for generic and recent AMD targets.
2298
2299 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2300
2301 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2302 ARG_SIZE note when adjustment was eliminated.
2303
2304 2014-01-22 Jeff Law <law@redhat.com>
2305
2306 PR tree-optimization/59597
2307 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2308 in file. Accept new argument REGISTERING and use it to modify
2309 dump output appropriately.
2310 (register_jump_thread): Corresponding changes.
2311 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2312 thread paths involving joiner blocks. Add code to dump cancelled
2313 jump threading paths.
2314
2315 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2316
2317 PR rtl-optimization/59477
2318 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2319 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2320
2321 2014-01-22 Tom Tromey <tromey@redhat.com>
2322
2323 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2324 PARAMS.
2325 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2326
2327 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2328
2329 PR rtl-optimization/59896
2330 * lra-constraints.c (process_alt_operands): Check unused note for
2331 matched operands of insn with no output reloads.
2332
2333 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2334
2335 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2336 (mips_move_from_gpr_cost): Likewise.
2337
2338 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2339
2340 PR rtl-optimization/59858
2341 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2342 ira_class_hard_regs_num.
2343 (process_alt_operands): Increase reject for dying matched operand.
2344
2345 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2346
2347 PR target/59003
2348 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2349 smaller than size, perform several stores or loads and stores
2350 at dst + count - size to store or copy all of size bytes, rather
2351 than just last modesize bytes.
2352
2353 2014-01-20 DJ Delorie <dj@redhat.com>
2354
2355 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2356 that CLOBBERs are REGs before propogating their values.
2357
2358 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2359
2360 PR middle-end/59789
2361 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2362 (cgraph_inline_failed_type): New function.
2363 * cgraph.h (DEFCIFCODE): Add type.
2364 (cgraph_inline_failed_type_t): New enum.
2365 (cgraph_inline_failed_type): New prototype.
2366 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2367 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2368 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2369 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2370 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2371 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2372 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2373 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2374 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2375 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2376 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2377 OPTIMIZATION_MISMATCH.
2378 * tree-inline.c (expand_call_inline): Emit errors during
2379 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2380
2381 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2382
2383 PR target/59685
2384 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2385 mode attribute in insn output.
2386
2387 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2388
2389 * output.h (output_constant): Delete.
2390 * varasm.c (output_constant): Make private.
2391
2392 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2393
2394 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2395
2396 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2397
2398 PR middle-end/59860
2399 * tree.h (fold_builtin_strcat): New prototype.
2400 * builtins.c (fold_builtin_strcat): No longer static. Add len
2401 argument, if non-NULL, don't call c_strlen. Optimize
2402 directly into __builtin_memcpy instead of __builtin_strcpy.
2403 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2404 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2405
2406 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2407
2408 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2409 for SImode_address_operand operands, having only a REG argument.
2410
2411 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2412
2413 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2414 loader name using mbig-endian.
2415 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2416
2417 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2418
2419 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2420 (-mtune): Likewise.
2421 (-mcpu): Likewise.
2422
2423 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2424
2425 * config/aarch64/aarch64-protos.h
2426 (aarch64_cannot_change_mode_class_ptr): Declare.
2427 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2428 aarch64_cannot_change_mode_class_ptr): New.
2429 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2430 backend hook aarch64_cannot_change_mode_class.
2431
2432 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2433
2434 * common/config/aarch64/aarch64-common.c
2435 (aarch64_handle_option): Don't handle any option order logic here.
2436 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2437 selected_cpu, warn on architecture version mismatch.
2438 (aarch64_override_options): Fix parsing order for option strings.
2439
2440 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2441 Iain Sandoe <iain@codesourcery.com>
2442
2443 PR bootstrap/59496
2444 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2445 warning. Amend comment to reflect current functionality.
2446
2447 2014-01-20 Richard Biener <rguenther@suse.de>
2448
2449 PR middle-end/59860
2450 * builtins.c (fold_builtin_strcat): Remove case better handled
2451 by tree-ssa-strlen.c.
2452
2453 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2454
2455 * config/aarch64/aarch64.opt
2456 (mcpu, march, mtune): Make case-insensitive.
2457
2458 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2459
2460 PR target/59880
2461 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2462 if operands[1] is a REG or ZERO_EXTEND of a REG.
2463
2464 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2465
2466 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2467
2468 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2469
2470 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2471 long non-pic millicode calls.
2472
2473 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2474
2475 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2476
2477 2014-01-19 Kito Cheng <kito@0xlab.org>
2478
2479 * builtins.c (expand_movstr): Check movstr expand done or fail.
2480
2481 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2482 H.J. Lu <hongjiu.lu@intel.com>
2483
2484 PR target/59379
2485 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2486 to DImode for zero-extended addresses.
2487
2488 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2489
2490 PR rtl-optimization/57763
2491 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2492 on the new indirect jump_insn and increment LABEL_NUSES (label).
2493
2494 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2495
2496 PR bootstrap/59580
2497 PR bootstrap/59583
2498 * config.gcc (x86_archs): New variable.
2499 (x86_64_archs): Likewise.
2500 (x86_cpus): Likewise.
2501 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2502 --with-arch/--with-cpu= options.
2503 Support --with-arch=/--with-cpu={nehalem,westmere,
2504 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2505
2506 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2507
2508 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2509 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2510
2511 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2512
2513 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2514
2515 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2516
2517 PR target/58944
2518 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2519 clear cpp_get_options (parse_in)->warn_unused_macros for
2520 ix86_target_macros_internal with cpp_define.
2521
2522 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2523
2524 * jump.c (delete_related_insns): Keep (use (insn))s.
2525 * reorg.c (redundant_insn): Check for barriers too.
2526
2527 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2528
2529 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2530
2531 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2532
2533 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2534 call to $$dyncall when TARGET_LONG_CALLS is true.
2535
2536 2014-01-17 Jeff Law <law@redhat.com>
2537
2538 * ree.c (combine_set_extension): Temporarily disable test for
2539 changing number of hard registers.
2540
2541 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2542
2543 PR middle-end/58125
2544 * ipa-inline-analysis.c (inline_free_summary):
2545 Do not free summary of aliases.
2546
2547 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2548
2549 PR middle-end/59706
2550 * gimplify.c (gimplify_expr): Use create_tmp_var
2551 instead of create_tmp_var_raw. If cond doesn't have
2552 integral type, don't add the IFN_ANNOTATE builtin at all.
2553
2554 2014-01-17 Martin Jambor <mjambor@suse.cz>
2555
2556 PR ipa/59736
2557 * ipa-cp.c (prev_edge_clone): New variable.
2558 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2559 Also resize prev_edge_clone vector.
2560 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2561 (ipcp_edge_removal_hook): New function.
2562 (ipcp_driver): Register ipcp_edge_removal_hook.
2563
2564 2014-01-17 Andrew Pinski <apinski@cavium.com>
2565 Steve Ellcey <sellcey@mips.com>
2566
2567 PR target/59462
2568 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2569 of operator mode.
2570
2571 2014-01-17 Jeff Law <law@redhat.com>
2572
2573 PR middle-end/57904
2574 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2575 so that pass_ccp runs first.
2576
2577 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2578
2579 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2580 (ix86_adjust_cost): Use !TARGET_XXX.
2581 (do_reorder_for_imul): Likewise.
2582 (swap_top_of_ready_list): Likewise.
2583 (ix86_sched_reorder): Likewise.
2584
2585 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2586
2587 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2588 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2589 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2590 (intel_memset): New. Duplicate slm_memset.
2591 (intel_cost): New. Duplicate slm_cost.
2592 (m_INTEL): New macro.
2593 (processor_target_table): Add "intel".
2594 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2595 with PROCESSOR_INTEL for "intel".
2596 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2597 PROCESSOR_SILVERMONT.
2598 (ix86_issue_rate): Likewise.
2599 (ix86_adjust_cost): Likewise.
2600 (ia32_multipass_dfa_lookahead): Likewise.
2601 (swap_top_of_ready_list): Likewise.
2602 (ix86_sched_reorder): Likewise.
2603 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2604 instead of TARGET_OPT_AGU.
2605 * config/i386/i386.h (TARGET_INTEL): New.
2606 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2607 (processor_type): Add PROCESSOR_INTEL.
2608 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2609 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2610
2611 2014-01-17 Marek Polacek <polacek@redhat.com>
2612
2613 PR c/58346
2614 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2615 size is zero.
2616
2617 2014-01-17 Richard Biener <rguenther@suse.de>
2618
2619 PR tree-optimization/46590
2620 * opts.c (default_options_table): Add entries for
2621 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2622 all enabled at -O1 but not for -Og.
2623 * common.opt (fbranch-count-reg): Remove Init(1).
2624 (fmove-loop-invariants): Likewise.
2625 (ftree-pta): Likewise.
2626
2627 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2628
2629 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2630 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2631 decls to at least the GCC 4.8 used alignments.
2632
2633 PR fortran/59440
2634 * tree-nested.c (convert_nonlocal_reference_stmt,
2635 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2636 of GIMPLE_BIND stmts, adjust associated decls.
2637
2638 2014-01-17 Richard Biener <rguenther@suse.de>
2639
2640 PR tree-optimization/46590
2641 * vec.h (vec<>::bseach): New member function implementing
2642 binary search according to C89 bsearch.
2643 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2644 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2645 bitmap pointer again. Make accesses_in_loop a flat array.
2646 (mem_ref_obstack): New global.
2647 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2648 (mark_ref_stored): Likewise.
2649 (ref_indep_loop_p_2): Likewise.
2650 (set_ref_stored_in_loop): New helper function.
2651 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2652 (memref_free): Adjust.
2653 (record_mem_ref_loc): Simplify.
2654 (gather_mem_refs_stmt): Adjust.
2655 (sort_locs_in_loop_postorder_cmp): New function.
2656 (analyze_memory_references): Sort accesses_in_loop after
2657 loop postorder number.
2658 (find_ref_loc_in_loop_cmp): New function.
2659 (for_all_locs_in_loop): Find relevant cluster of locs in
2660 accesses_in_loop and iterate without recursion.
2661 (execute_sm): Avoid uninit warning.
2662 (struct ref_always_accessed): Simplify.
2663 (ref_always_accessed::operator ()): Likewise.
2664 (ref_always_accessed_p): Likewise.
2665 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2666 loop postorder numbers here.
2667 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2668 numbers.
2669
2670 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2671
2672 PR c++/57945
2673 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2674 on decls for which assemble_alias has been called.
2675
2676 2014-01-17 Nick Clifton <nickc@redhat.com>
2677
2678 * config/msp430/msp430.opt: (mcpu): New option.
2679 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2680 (msp430_option_override): Parse target_cpu. If the MCU name
2681 matches a generic string, clear target_mcu.
2682 (msp430_attr): Allow numeric interrupt values up to 63.
2683 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2684 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2685 option.
2686 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2687 Add mcpu matches.
2688 * config/msp430/msp430.md (popm): Use %J rather than %I.
2689 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2690 (addhi_cy_i): Use immediate_operand for operand 2.
2691 * doc/invoke.texi: Document -mcpu option.
2692
2693 2014-01-17 Richard Biener <rguenther@suse.de>
2694
2695 PR rtl-optimization/38518
2696 * df.h (df_analyze_loop): Declare.
2697 * df-core.c: Include cfgloop.h.
2698 (df_analyze_1): Split out main part of df_analyze.
2699 (df_analyze): Adjust.
2700 (loop_inverted_post_order_compute): New function.
2701 (loop_post_order_compute): Likewise.
2702 (df_analyze_loop): New function avoiding whole-function
2703 postorder computes.
2704 * loop-invariant.c (find_defs): Use df_analyze_loop.
2705 (find_invariants): Adjust.
2706 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2707
2708 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2709
2710 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2711 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2712
2713 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2714
2715 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2716 traversal when removing references.
2717
2718 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2719
2720 PR ipa/59775
2721 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2722
2723 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2724
2725 PR middle-end/56791
2726 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2727 pushing a reload for an autoinc when we had previously reloaded an
2728 inner part of the address.
2729
2730 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2731
2732 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2733 field.
2734 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2735 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2736 when not giving up or versioning for alias only because of
2737 loop->safelen.
2738 (vect_analyze_data_ref_dependences): Set to true.
2739 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2740 is a GIMPLE_PHI.
2741 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2742 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2743 to the condition.
2744
2745 PR middle-end/58344
2746 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2747
2748 PR target/59839
2749 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2750 operand 0 predicate for gathers, use a new pseudo as subtarget.
2751
2752 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2753
2754 PR middle-end/59609
2755 * lra-constraints.c (process_alt_operands): Add printing debug info.
2756 Check absence of input/output reloads for matched operands too.
2757
2758 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2759
2760 PR rtl-optimization/59835
2761 * ira.c (ira_init_register_move_cost): Increase cost for
2762 impossible modes.
2763
2764 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2765
2766 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2767
2768 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2769
2770 PR target/59780
2771 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2772 non-register objects. Use gen_(high/low)part more consistently.
2773 Fix assertions.
2774
2775 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2776
2777 PR target/59844
2778 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2779 endian support, remove tests for WORDS_BIG_ENDIAN.
2780 (p8_mfvsrd_3_<mode>): Likewise.
2781 (reload_gpr_from_vsx<mode>): Likewise.
2782 (reload_gpr_from_vsxsf): Likewise.
2783 (p8_mfvsrd_4_disf): Likewise.
2784
2785 2014-01-16 Richard Biener <rguenther@suse.de>
2786
2787 PR rtl-optimization/46590
2788 * lcm.c (compute_antinout_edge): Use postorder iteration.
2789 (compute_laterin): Use inverted postorder iteration.
2790
2791 2014-01-16 Nick Clifton <nickc@redhat.com>
2792
2793 PR middle-end/28865
2794 * varasm.c (output_constant): Return the number of bytes actually
2795 emitted.
2796 (output_constructor_array_range): Update the field size with the
2797 number of bytes emitted by output_constant.
2798 (output_constructor_regular_field): Likewise. Also do not
2799 complain if the total number of bytes emitted is now greater
2800 than the expected fieldpos.
2801 * output.h (output_constant): Update prototype and descriptive comment.
2802
2803 2014-01-16 Marek Polacek <polacek@redhat.com>
2804
2805 PR middle-end/59827
2806 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2807 it is error_mark_node.
2808
2809 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2810
2811 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2812 VALID_AVX256_REG_OR_OI_MODE.
2813
2814 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2815
2816 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2817 current procedure should be profiled.
2818
2819 2014-01-15 Andrew Pinski <apinski@cavium.com>
2820
2821 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2822 of moving from/to the STACK_REG register class.
2823
2824 2014-01-15 Richard Henderson <rth@redhat.com>
2825
2826 PR debug/54694
2827 * reginfo.c (global_regs_decl): Globalize.
2828 * rtl.h (global_regs_decl): Declare.
2829 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2830 reserved via global_regs.
2831
2832 2014-01-15 Teresa Johnson <tejohnson@google.com>
2833
2834 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2835
2836 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2837
2838 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2839 and vmulosh rather than call gen_vec_widen_smult_*.
2840 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2841 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2842 (vec_widen_smult_even_v16qi): Likewise.
2843 (vec_widen_umult_even_v8hi): Likewise.
2844 (vec_widen_smult_even_v8hi): Likewise.
2845 (vec_widen_umult_odd_v16qi): Likewise.
2846 (vec_widen_smult_odd_v16qi): Likewise.
2847 (vec_widen_umult_odd_v8hi): Likewise.
2848 (vec_widen_smult_odd_v8hi): Likewise.
2849 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2850 vmuloub rather than call gen_vec_widen_umult_*.
2851 (vec_widen_umult_lo_v16qi): Likewise.
2852 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2853 vmulosb rather than call gen_vec_widen_smult_*.
2854 (vec_widen_smult_lo_v16qi): Likewise.
2855 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2856 rather than call gen_vec_widen_umult_*.
2857 (vec_widen_umult_lo_v8hi): Likewise.
2858 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2859 rather than call gen_vec_widen_smult_*.
2860 (vec_widen_smult_lo_v8hi): Likewise.
2861
2862 2014-01-15 Jeff Law <law@redhat.com>
2863
2864 PR tree-optimization/59747
2865 * ree.c (find_and_remove_re): Properly handle case where a second
2866 eliminated extension requires widening a copy created for elimination
2867 of a prior extension.
2868 (combine_set_extension): Ensure that the number of hard regs needed
2869 for a destination register does not change when we widen it.
2870
2871 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2872
2873 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2874 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2875 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2876 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2877 (avr-*-rtems*): Likewise.
2878 (bfin*-rtems*): Likewise.
2879 (moxie-*-rtems*): Likewise.
2880 (h8300-*-rtems*): Likewise.
2881 (i[34567]86-*-rtems*): Likewise.
2882 (lm32-*-rtems*): Likewise.
2883 (m32r-*-rtems*): Likewise.
2884 (m68k-*-rtems*): Likewise.
2885 (microblaze*-*-rtems*): Likewise.
2886 (mips*-*-rtems*): Likewise.
2887 (powerpc-*-rtems*): Likewise.
2888 (sh-*-rtems*): Likewise.
2889 (sparc-*-rtems*): Likewise.
2890 (sparc64-*-rtems*): Likewise.
2891 (v850-*-rtems*): Likewise.
2892 (m32c-*-rtems*): Likewise.
2893
2894 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2895
2896 PR rtl-optimization/59511
2897 * ira.c (ira_init_register_move_cost): Use memory costs for some
2898 cases of register move cost calculations.
2899 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2900 instead of BB frequency.
2901 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2902 * lra-assigns.c (find_hard_regno_for): Ditto.
2903
2904 2014-01-15 Richard Biener <rguenther@suse.de>
2905
2906 PR tree-optimization/59822
2907 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2908 (vectorizable_load): Use it to hoist defs of uses of invariant
2909 loads out of the loop.
2910
2911 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2912 Kugan Vivekanandarajah <kuganv@linaro.org>
2913
2914 PR target/59695
2915 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2916 truncation.
2917
2918 2014-01-15 Richard Biener <rguenther@suse.de>
2919
2920 PR rtl-optimization/59802
2921 * lcm.c (compute_available): Use inverted postorder to seed
2922 the initial worklist.
2923
2924 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2925
2926 PR target/59803
2927 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2928 ADDR_REGS for invalid symrefs in non-PIC code.
2929
2930 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2931
2932 PR other/58712
2933 * builtins.c (determine_block_size): Initialize *probable_max_size
2934 even if len_rtx is CONST_INT.
2935
2936 2014-01-14 Andrew Pinski <apinski@cavium.com>
2937
2938 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2939 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2940 (cortexa53_tunings): Likewise.
2941 (aarch64_sched_issue_rate): New function.
2942 (TARGET_SCHED_ISSUE_RATE): Define.
2943
2944 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2945
2946 * ira-costs.c (find_costs_and_classes): Add missed
2947 ira_init_register_move_cost_if_necessary.
2948
2949 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2950
2951 PR target/59787
2952 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2953
2954 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2955
2956 PR target/59794
2957 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2958 to indicate if type is used for function return value. Warn ABI
2959 change if the vector mode isn't available for function return value.
2960 (ix86_function_arg_advance): Pass false to type_natural_mode.
2961 (ix86_function_arg): Likewise.
2962 (ix86_gimplify_va_arg): Likewise.
2963 (function_arg_32): Don't warn ABI change.
2964 (ix86_function_value): Pass true to type_natural_mode.
2965 (ix86_return_in_memory): Likewise.
2966 (ix86_struct_value_rtx): Removed.
2967 (TARGET_STRUCT_VALUE_RTX): Likewise.
2968
2969 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2970
2971 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2972 converting a conditional jump into a conditional return.
2973
2974 2014-01-14 Richard Biener <rguenther@suse.de>
2975
2976 PR tree-optimization/58921
2977 PR tree-optimization/59006
2978 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2979 hoisting invariant stmts.
2980 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2981 invariant loads on the preheader edge if possible.
2982
2983 2014-01-14 Joey Ye <joey.ye@arm.com>
2984
2985 * doc/plugin.texi (Building GCC plugins): Update to C++.
2986
2987 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2988
2989 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2990 (_mm_rcp28_round_ss): Ditto.
2991 (_mm_rsqrt28_round_sd): Ditto.
2992 (_mm_rsqrt28_round_ss): Ditto.
2993 (_mm_rcp28_sd): Ditto.
2994 (_mm_rcp28_ss): Ditto.
2995 (_mm_rsqrt28_sd): Ditto.
2996 (_mm_rsqrt28_ss): Ditto.
2997 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2998 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2999 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3000 (IX86_BUILTIN_RCP28SD): Ditto.
3001 (IX86_BUILTIN_RCP28SS): Ditto.
3002 (IX86_BUILTIN_RSQRT28SD): Ditto.
3003 (IX86_BUILTIN_RSQRT28SS): Ditto.
3004 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3005 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3006 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3007 (ix86_expand_special_args_builtin): Expand new FTYPE.
3008 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3009 (srcp14<mode>): Make insn unary.
3010 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3011 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3012 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3013 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3014 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3015 Fix rounding: make it SAE only.
3016 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3017 Ditto.
3018 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3019 Ditto.
3020 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3021 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3022 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3023 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3024 (round_saeonly_mask_scalar_operand4): Ditto.
3025 (round_saeonly_mask_scalar_op3): Ditto.
3026 (round_saeonly_mask_scalar_op4): Ditto.
3027
3028 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3029
3030 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3031 Implement -maltivec=be for vec_insert and vec_extract.
3032
3033 2014-01-10 DJ Delorie <dj@redhat.com>
3034
3035 * config/msp430/msp430.md (call_internal): Don't allow memory
3036 references with SP as the base register.
3037 (call_value_internal): Likewise.
3038 * config/msp430/constraints.md (Yc): New. For memory references
3039 that don't use SP as a base register.
3040
3041 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3042 "an integer without a # prefix"
3043 * config/msp430/msp430.md (epilogue_helper): Use it.
3044
3045 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3046
3047 PR target/59617
3048 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3049 AVX512F gather builtins.
3050 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3051 on gather decls with INTEGER_TYPE masktype.
3052 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3053 directly into the builtin rather than hoisting it before loop.
3054
3055 PR tree-optimization/59387
3056 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3057 (scev_const_prop): If folded_casts and type has undefined overflow,
3058 use force_gimple_operand instead of force_gimple_operand_gsi and
3059 for each added stmt if it is assign with
3060 arith_code_with_undefined_signed_overflow, call
3061 rewrite_to_defined_overflow.
3062 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3063 gimple-fold.h instead.
3064 (arith_code_with_undefined_signed_overflow,
3065 rewrite_to_defined_overflow): Moved to ...
3066 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3067 rewrite_to_defined_overflow): ... here. No longer static.
3068 Include gimplify-me.h.
3069 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3070 rewrite_to_defined_overflow): New prototypes.
3071
3072 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3073
3074 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3075
3076 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3077
3078 * builtins.c (get_object_alignment_2): Minor tweak.
3079 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3080
3081 2014-01-13 Christian Bruel <christian.bruel@st.com>
3082
3083 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3084 optimized non constant lengths.
3085
3086 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3087
3088 PR libgomp/59194
3089 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3090 load as __atomic_load_N if possible.
3091
3092 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3093
3094 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3095 target parameter.
3096 (rs6000_expand_builtin): Adjust call.
3097
3098 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3099
3100 PR target/58115
3101 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3102 * config/rs6000/rs6000.c: Include target-globals.h.
3103 (rs6000_set_current_function): Instead of doing target_reinit
3104 unconditionally, use save_target_globals_default_opts and
3105 restore_target_globals.
3106
3107 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3108 FPSCR.
3109 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3110 (rs6000_expand_builtin): Handle mffs and mtfsf.
3111 (rs6000_init_builtins): Define mffs and mtfsf.
3112 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3113 (rs6000_mffs): New pattern.
3114 (rs6000_mtfsf): New pattern.
3115
3116 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3117
3118 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3119 Start narrowing with START. Apply candidate-use pair
3120 and check overall cost in narrowing.
3121 (iv_ca_prune): Pass new argument.
3122
3123 2014-01-10 Jeff Law <law@redhat.com>
3124
3125 PR middle-end/59743
3126 * ree.c (combine_reaching_defs): Ensure the defining statement
3127 occurs before the extension when optimizing extensions with
3128 different source and destination hard registers.
3129
3130 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3131
3132 PR ipa/58585
3133 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3134 vtables into the type inheritance graph.
3135
3136 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3137
3138 PR rtl-optimization/59754
3139 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3140 modes in the REGNO != REGNO case.
3141
3142 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3143
3144 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3145
3146 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3147
3148 PR tree-optimization/59745
3149 * tree-predcom.c (tree_predictive_commoning_loop): Call
3150 free_affine_expand_cache if giving up because components is NULL.
3151
3152 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3153 GC in payload of target_globals struct instead of allocating them on
3154 the heap and the larger structs separately using GC.
3155 * target-globals.h (struct target_globals): Make regs, hard_regs,
3156 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3157 of GTY((skip)) and change type to void *.
3158 (reset_target_globals): Cast loads from those fields to corresponding
3159 types.
3160
3161 2014-01-10 Steve Ellcey <sellcey@mips.com>
3162
3163 PR plugins/59335
3164 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3165 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3166 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3167
3168 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3169
3170 PR target/59744
3171 * aarch64-modes.def (CC_Zmode): New flags mode.
3172 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3173 represents an equality.
3174 (aarch64_get_condition_code): Handle CC_Zmode.
3175 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3176
3177 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3178
3179 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3180 extraction in good case.
3181
3182 2014-01-10 Richard Biener <rguenther@suse.de>
3183
3184 PR tree-optimization/59374
3185 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3186 checking after SLP discovery. Mark stmts not participating
3187 in any SLP instance properly.
3188
3189 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3190
3191 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3192 when handling a SET rtx.
3193
3194 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3195
3196 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3197 (cortex-a57): Likewise.
3198 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3199
3200 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3201
3202 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3203 non-iwmmxt builtins.
3204
3205 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3206
3207 PR ipa/58252
3208 PR ipa/59226
3209 * ipa-devirt.c record_target_from_binfo): Take as argument
3210 stack of binfos and lookup matching one for virtual inheritance.
3211 (possible_polymorphic_call_targets_1): Update.
3212
3213 2014-01-10 Huacai Chen <chenhc@lemote.com>
3214
3215 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3216 kernel strings for Loongson-2E/2F/3A.
3217
3218 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3219
3220 PR middle-end/59670
3221 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3222 is_gimple_call before calling gimple_call_internal_p.
3223
3224 2014-01-09 Steve Ellcey <sellcey@mips.com>
3225
3226 * Makefile.in (TREE_FLOW_H): Remove.
3227 (TREE_SSA_H): Add file names from tree-flow.h.
3228 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3229 * tree.h: Remove tree-flow.h reference.
3230 * hash-table.h: Remove tree-flow.h reference.
3231 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3232 reference with tree-ssa-loop.h.
3233
3234 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3235
3236 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3237 default element-order behavior for -maltivec.
3238 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3239 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3240 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3241 when targeting big endian, at least for now.
3242 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3243
3244 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3245
3246 PR middle-end/47735
3247 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3248 var satisfies use_register_for_decl, just take into account type
3249 alignment, rather than decl alignment.
3250
3251 PR tree-optimization/59622
3252 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3253 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3254 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3255 Don't devirtualize for inplace at all. For targets.length () == 1,
3256 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3257
3258 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3259
3260 * config/i386/i386.md (cpu): Remove the unused btver1.
3261
3262 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3263
3264 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3265
3266 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3267
3268 PR target/58115
3269 * tree-core.h (struct target_globals): New forward declaration.
3270 (struct tree_target_option): Add globals field.
3271 * tree.h (TREE_TARGET_GLOBALS): Define.
3272 (prepare_target_option_nodes_for_pch): New prototype.
3273 * target-globals.h (struct target_globals): Define even if
3274 !SWITCHABLE_TARGET.
3275 * tree.c (prepare_target_option_node_for_pch,
3276 prepare_target_option_nodes_for_pch): New functions.
3277 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3278 * config/i386/i386.c: Include target-globals.h.
3279 (ix86_set_current_function): Instead of doing target_reinit
3280 unconditionally, use save_target_globals_default_opts and
3281 restore_target_globals.
3282
3283 2014-01-09 Richard Biener <rguenther@suse.de>
3284
3285 PR tree-optimization/59715
3286 * tree-cfg.h (split_critical_edges): Declare.
3287 * tree-cfg.c (split_critical_edges): Export.
3288 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3289
3290 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3291
3292 * cfgexpand.c (expand_stack_vars): Optionally disable
3293 asan stack protection.
3294 (expand_used_vars): Likewise.
3295 (partition_stack_vars): Likewise.
3296 * asan.c (asan_emit_stack_protection): Optionally disable
3297 after return stack usage.
3298 (instrument_derefs): Optionally disable memory access instrumentation.
3299 (instrument_builtin_call): Likewise.
3300 (instrument_strlen_call): Likewise.
3301 (asan_protect_global): Optionally disable global variables protection.
3302 * doc/invoke.texi: Added doc for new options.
3303 * params.def: Added new options.
3304 * params.h: Likewise.
3305
3306 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3307
3308 PR rtl-optimization/59724
3309 * ifcvt.c (cond_exec_process_if_block): Don't call
3310 flow_find_head_matching_sequence with 0 longest_match.
3311 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3312 non-active insns if !stop_after.
3313 (try_head_merge_bb): Revert 2014-01-07 changes.
3314
3315 2014-01-08 Jeff Law <law@redhat.com>
3316
3317 * ree.c (get_sub_rtx): New function, extracted from...
3318 (merge_def_and_ext): Here.
3319 (combine_reaching_defs): Use get_sub_rtx.
3320
3321 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3322
3323 * cgraph.h (varpool_variable_node): Do not choke on null node.
3324
3325 2014-01-08 Catherine Moore <clm@codesourcery.com>
3326
3327 * config/mips/mips.md (simple_return): Attempt to use JRC
3328 for microMIPS.
3329 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3330
3331 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3332
3333 PR rtl-optimization/59137
3334 * reorg.c (steal_delay_list_from_target): Call update_block for
3335 elided insns.
3336 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3337
3338 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3339
3340 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3341 two duplicate entries.
3342
3343 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3344
3345 Revert:
3346 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3347
3348 * config/mips/mips.c (mips_truncated_op_cost): New function.
3349 (mips_rtx_costs): Adjust test for BADDU.
3350 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3351
3352 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3353
3354 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3355 (*baddu_si): ...this new pattern.
3356
3357 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3358
3359 PR ipa/59722
3360 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3361
3362 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3363
3364 PR middle-end/57748
3365 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3366 inner_reference_p.
3367 (expand_expr, expand_normal): Adjust.
3368 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3369 inner_reference_p. Use inner_reference_p to expand inner references.
3370 (store_expr): Adjust.
3371 * cfgexpand.c (expand_call_stmt): Adjust.
3372
3373 2014-01-08 Rong Xu <xur@google.com>
3374
3375 * gcov-io.c (gcov_var): Move from gcov-io.h.
3376 (gcov_position): Ditto.
3377 (gcov_is_error): Ditto.
3378 (gcov_rewrite): Ditto.
3379 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3380 only part to libgcc/libgcov.h.
3381
3382 2014-01-08 Marek Polacek <polacek@redhat.com>
3383
3384 PR middle-end/59669
3385 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3386
3387 2014-01-08 Marek Polacek <polacek@redhat.com>
3388
3389 PR sanitizer/59667
3390 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3391
3392 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3393
3394 PR rtl-optimization/59649
3395 * stor-layout.c (get_mode_bounds): For BImode return
3396 0 and STORE_FLAG_VALUE.
3397
3398 2014-01-08 Richard Biener <rguenther@suse.de>
3399
3400 PR middle-end/59630
3401 * gimple.h (is_gimple_builtin_call): Remove.
3402 (gimple_builtin_call_types_compatible_p): New.
3403 (gimple_call_builtin_p): New overload.
3404 * gimple.c (is_gimple_builtin_call): Remove.
3405 (validate_call): Rename to ...
3406 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3407 check return types.
3408 (validate_type): New static function.
3409 (gimple_call_builtin_p): New overload and adjust.
3410 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3411 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3412 (gimple_fold_stmt_to_constant_1): Likewise.
3413 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3414
3415 2014-01-08 Richard Biener <rguenther@suse.de>
3416
3417 PR middle-end/59471
3418 * gimplify.c (gimplify_expr): Gimplify register-register type
3419 VIEW_CONVERT_EXPRs to separate stmts.
3420
3421 2014-01-07 Jeff Law <law@redhat.com>
3422
3423 PR middle-end/53623
3424 * ree.c (combine_set_extension): Handle case where source
3425 and destination registers in an extension insn are different.
3426 (combine_reaching_defs): Allow source and destination registers
3427 in extension to be different under limited circumstances.
3428 (add_removable_extension): Remove restriction that the
3429 source and destination registers in the extension are the same.
3430 (find_and_remove_re): Emit a copy from the extension's
3431 destination to its source after the defining insn if
3432 the source and destination registers are different.
3433
3434 PR middle-end/59285
3435 * ifcvt.c (merge_if_block): If we are merging a block with more than
3436 one successor with a block with no successors, remove any BARRIER
3437 after the second block.
3438
3439 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3440
3441 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3442
3443 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3444
3445 PR target/59652
3446 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3447 for 14-bit register offsets when INT14_OK_STRICT is false.
3448
3449 2014-01-07 Roland Stigge <stigge@antcom.de>
3450 Michael Meissner <meissner@linux.vnet.ibm.com>
3451
3452 PR 57386/target
3453 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3454 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3455
3456 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3457
3458 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3459 -mcpu.
3460
3461 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3462
3463 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3464 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3465 rtx is const0_rtx or not.
3466
3467 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3468
3469 PR target/58115
3470 * target-globals.c (save_target_globals): Remove this_fn_optab
3471 handling.
3472 * toplev.c: Include optabs.h.
3473 (target_reinit): Temporarily restore the global options if another
3474 set of options are in force.
3475
3476 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3477
3478 PR rtl-optimization/58668
3479 * cfgcleanup.c (flow_find_cross_jump): Don't count
3480 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3481 to determine what is counted.
3482 (flow_find_head_matching_sequence): Use active_insn_p to determine
3483 what is counted.
3484 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3485 counting change.
3486 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3487 determine what is counted.
3488
3489 PR tree-optimization/59643
3490 * tree-predcom.c (split_data_refs_to_components): If one dr is
3491 read and one write, determine_offset fails and the write isn't
3492 in the bad component, just put the read into the bad component.
3493
3494 2014-01-07 Mike Stump <mikestump@comcast.net>
3495 Jakub Jelinek <jakub@redhat.com>
3496
3497 PR pch/59436
3498 * tree-core.h (struct tree_optimization_option): Change optabs
3499 type from unsigned char * to void *.
3500 * optabs.c (init_tree_optimization_optabs): Adjust
3501 TREE_OPTIMIZATION_OPTABS initialization.
3502
3503 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3504
3505 PR target/59644
3506 * config/i386/i386.h (struct machine_function): Add
3507 no_drap_save_restore field.
3508 * config/i386/i386.c (ix86_save_reg): Use
3509 !cfun->machine->no_drap_save_restore instead of
3510 crtl->stack_realign_needed.
3511 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3512 this function clears frame_pointer_needed. Set
3513 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3514 and DRAP reg is needed.
3515
3516 2014-01-06 Marek Polacek <polacek@redhat.com>
3517
3518 PR c/57773
3519 * doc/implement-c.texi: Mention that other integer types are
3520 permitted as bit-field types in strictly conforming mode.
3521
3522 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3523
3524 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3525 is newly allocated.
3526
3527 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3528
3529 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3530
3531 2014-01-06 Martin Jambor <mjambor@suse.cz>
3532
3533 PR ipa/59008
3534 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3535 to int.
3536 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3537
3538 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3539
3540 PR debug/59350
3541 PR debug/59510
3542 * var-tracking.c (add_stores): Preserve the value of the source even if
3543 we don't record the store.
3544
3545 2014-01-06 Terry Guo <terry.guo@arm.com>
3546
3547 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3548
3549 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3550
3551 PR bootstrap/59541
3552 * config/darwin.c (darwin_function_section): Adjust return values to
3553 correspond to optimisation changes made in r206070.
3554
3555 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3556
3557 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3558 from prefetch_block tune setting.
3559 (nocona_cost): Correct size of prefetch block to 64.
3560
3561 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3562
3563 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3564 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3565 used to save the static chain register in the computation of the offset
3566 from which the FP registers need to be restored.
3567
3568 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3569
3570 PR tree-optimization/59519
3571 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3572 ICE if get_current_def (current_new_name) is already non-NULL, as long
3573 as it is a phi result of some other phi in *new_exit_bb that has
3574 the same argument.
3575
3576 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3577 or vmovdqu* for misaligned_operand.
3578 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3579 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3580 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3581 aligned_mem for AVX512F masked aligned load and store builtins and for
3582 non-temporal moves.
3583
3584 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3585
3586 PR tree-optimization/59651
3587 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3588 Address range for negative step should be added by TYPE_SIZE_UNIT.
3589
3590 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3591
3592 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3593 overlapping registers also for registers other than the stack pointer.
3594
3595 2014-01-03 Marek Polacek <polacek@redhat.com>
3596
3597 PR other/59661
3598 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3599 __builtin_FILE.
3600
3601 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3602
3603 PR target/59625
3604 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3605 asm goto as jump.
3606
3607 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3608 (push splitter): Use <P:MODE_SIZE> instead of
3609 GET_MODE_SIZE (<P:MODE>mode).
3610 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3611 (mov -1, reg peephole2): Likewise.
3612 * config/i386/sse.md (*mov<mode>_internal,
3613 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3614 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3615 *<code><mode>3, *andnot<mode>3<mask_name>,
3616 <mask_codefor><code><mode>3<mask_name>): Likewise.
3617 * config/i386/subst.md (mask_mode512bit_condition,
3618 sd_mask_mode512bit_condition): Likewise.
3619
3620 2014-01-02 Xinliang David Li <davidxl@google.com>
3621
3622 PR tree-optimization/59303
3623 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3624 (dump_predicates): Better output format.
3625 (pred_equal_p): New function.
3626 (is_neq_relop_p): Ditto.
3627 (is_neq_zero_form_p): Ditto.
3628 (pred_expr_equal_p): Ditto.
3629 (pred_neg_p): Ditto.
3630 (simplify_pred): Ditto.
3631 (simplify_preds_2): Ditto.
3632 (simplify_preds_3): Ditto.
3633 (simplify_preds_4): Ditto.
3634 (simplify_preds): Ditto.
3635 (push_pred): Ditto.
3636 (push_to_worklist): Ditto.
3637 (get_pred_info_from_cmp): Ditto.
3638 (is_degenerated_phi): Ditto.
3639 (normalize_one_pred_1): Ditto.
3640 (normalize_one_pred): Ditto.
3641 (normalize_one_pred_chain): Ditto.
3642 (normalize_preds): Ditto.
3643 (normalize_cond_1): Remove function.
3644 (normalize_cond): Ditto.
3645 (is_gcond_subset_of): Ditto.
3646 (is_subset_of_any): Ditto.
3647 (is_or_set_subset_of): Ditto.
3648 (is_and_set_subset_of): Ditto.
3649 (is_norm_cond_subset_of): Ditto.
3650 (pred_chain_length_cmp): Ditto.
3651 (convert_control_dep_chain_into_preds): Type change.
3652 (find_predicates): Ditto.
3653 (find_def_preds): Ditto.
3654 (destroy_predicates_vecs): Ditto.
3655 (find_matching_predicates_in_rest_chains): Ditto.
3656 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3657 (is_pred_expr_subset): Ditto.
3658 (is_pred_chain_subset_of): Ditto.
3659 (is_included_in): Ditto.
3660 (is_superset_of): Ditto.
3661
3662 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3663
3664 Update copyright years.
3665
3666 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3667
3668 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3669 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3670 config/arc/arc.md, config/arc/arc.opt,
3671 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3672 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3673 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3674 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3675 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3676 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3677 vtable-verify.c, vtable-verify.h: Use the standard form for the
3678 copyright notice.
3679
3680 2014-01-02 Tobias Burnus <burnus@net-b.de>
3681
3682 * gcc.c (process_command): Update copyright notice dates.
3683 * gcov-dump.c: Ditto.
3684 * gcov.c: Ditto.
3685 * doc/cpp.texi: Bump @copying's copyright year.
3686 * doc/cppinternals.texi: Ditto.
3687 * doc/gcc.texi: Ditto.
3688 * doc/gccint.texi: Ditto.
3689 * doc/gcov.texi: Ditto.
3690 * doc/install.texi: Ditto.
3691 * doc/invoke.texi: Ditto.
3692
3693 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3694
3695 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3696
3697 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3698
3699 * config/i386/sse.md (*mov<mode>_internal): Guard
3700 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3701
3702 PR rtl-optimization/59647
3703 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3704 new_rtx into UNSIGNED_FLOAT rtxes.
3705 \f
3706 Copyright (C) 2014 Free Software Foundation, Inc.
3707
3708 Copying and distribution of this file, with or without modification,
3709 are permitted in any medium without royalty provided the copyright
3710 notice and this notice are preserved.