re PR target/60203 (Support long double/_Decimal128 direct move on power8)
[gcc.git] / gcc / ChangeLog
1 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2
3 PR target/60203
4 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
5 Split 64-bit moves into 2 patterns. Do not allow the use of
6 direct move for TDmode in little endian, since the decimal value
7 has little endian bytes within a word, but the 64-bit pieces are
8 ordered in a big endian fashion, and normal subreg's of TDmode are
9 not allowed.
10 (mov<mode>_64bit_dm): Likewise.
11 (movtd_64bit_nodm): Likewise.
12
13 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
14
15 PR tree-optimization/60174
16 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
17 statement of an SSA_NAME that occurs in an abnormal PHI node.
18
19 2014-02-18 Jakub Jelinek <jakub@redhat.com>
20
21 PR sanitizer/60142
22 * final.c (SEEN_BB): Remove.
23 (SEEN_NOTE, SEEN_EMITTED): Renumber.
24 (final_scan_insn): Don't force_source_line on second
25 NOTE_INSN_BASIC_BLOCK.
26
27 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
28
29 PR target/60205
30 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
31 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
32 (type_natural_mode): Warn ABI change when %zmm register is not
33 available for AVX512F vector value passing.
34
35 2014-02-18 Kai Tietz <ktietz@redhat.com>
36
37 PR target/60193
38 * config/i386/i386.c (ix86_expand_prologue): Use value in
39 rax register as displacement when restoring %r10 or %rax.
40 Fix wrong offset when restoring both registers.
41
42 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
43
44 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
45 assertion with conditional return.
46
47 2014-02-18 Jakub Jelinek <jakub@redhat.com>
48 Uros Bizjak <ubizjak@gmail.com>
49
50 PR driver/60233
51 * config/i386/driver-i386.c (host_detect_local_cpu): If
52 YMM state is not saved by the OS, also clear has_f16c. Move
53 CPUID 0x80000001 handling before YMM state saving checking.
54
55 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
56
57 PR rtl-optimization/58960
58 * haifa-sched.c (alloc_global_sched_pressure_data): New,
59 factored out from ...
60 (sched_init): ... here.
61 (free_global_sched_pressure_data): New, factored out from ...
62 (sched_finish): ... here.
63 * sched-int.h (free_global_sched_pressure_data): Declare.
64 * sched-rgn.c (nr_regions_initial): New static global.
65 (haifa_find_rgns): Initialize it.
66 (schedule_region): Disable sched-pressure for the newly
67 generated regions.
68
69 2014-02-17 Richard Biener <rguenther@suse.de>
70
71 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
72 release SSA defs of pattern stmts.
73
74 2014-02-17 Richard Biener <rguenther@suse.de>
75
76 * tree-inline.c (expand_call_inline): Release the virtual
77 operand defined by the call we are about to inline.
78
79 2014-02-17 Richard Biener <rguenther@suse.de>
80
81 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
82
83 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
84 Ilya Tocar <ilya.tocar@intel.com>
85
86 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
87 arguments order in builtin.
88 (_mm512_permutexvar_epi64): Ditto.
89 (_mm512_mask_permutexvar_epi64): Ditto
90 (_mm512_maskz_permutexvar_epi32): Ditto
91 (_mm512_permutexvar_epi32): Ditto
92 (_mm512_mask_permutexvar_epi32): Ditto
93
94 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
95
96 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
97 (p8_vmrgow): Likewise.
98
99 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
100
101 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
102 endian targets.
103
104 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
105
106 PR target/60203
107 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
108 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
109 into 64-bit and 32-bit moves. On 64-bit moves, add support for
110 using direct move instructions on ISA 2.07. Also adjust
111 instruction length for 64-bit.
112 (mov<mode>_64bit, TFmode/TDmode): Likewise.
113 (mov<mode>_32bit, TFmode/TDmode): Likewise.
114
115 2014-02-15 Alan Modra <amodra@gmail.com>
116
117 PR target/58675
118 PR target/57935
119 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
120 find_replacement on parts of insn rtl that might be reloaded.
121
122 2014-02-15 Richard Biener <rguenther@suse.de>
123
124 PR tree-optimization/60183
125 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
126 (tree_ssa_phiprop): Calculate and free post-dominators.
127
128 2014-02-14 Jeff Law <law@redhat.com>
129
130 PR rtl-optimization/60131
131 * ree.c (get_extended_src_reg): New function.
132 (combine_reaching_defs): Use it rather than assuming location of REG.
133 (find_and_remove_re): Verify first operand of extension is
134 a REG before adding the insns to the copy list.
135
136 2014-02-14 Roland McGrath <mcgrathr@google.com>
137
138 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
139 * configure: Regenerated.
140 * config.in: Regenerated.
141 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
142 instead of ASM_SHORT.
143
144 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
145 Richard Earnshaw <rearnsha@arm.com>
146
147 PR rtl-optimization/59535
148 * lra-constraints.c (process_alt_operands): Encourage alternative
149 when unassigned pseudo class is superset of the alternative class.
150 (inherit_reload_reg): Don't inherit when optimizing for code size.
151 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
152 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
153 modes not less than 4 for Thumb1.
154
155 2014-02-14 Kyle McMartin <kyle@redhat.com>
156
157 PR pch/60010
158 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
159
160 2014-02-14 Richard Biener <rguenther@suse.de>
161
162 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
163 (get_frame_arg): Drop the assert with langhook types_compatible_p.
164 Do not strip INDIRECT_REFs.
165
166 2014-02-14 Richard Biener <rguenther@suse.de>
167
168 PR lto/60179
169 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
170 DECL_FUNCTION_SPECIFIC_TARGET.
171 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
172 * tree-streamer-out.c (pack_ts_target_option): Remove.
173 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
174 (write_ts_function_decl_tree_pointers): Do not stream
175 DECL_FUNCTION_SPECIFIC_TARGET.
176 * tree-streamer-in.c (unpack_ts_target_option): Remove.
177 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
178 (lto_input_ts_function_decl_tree_pointers): Do not stream
179 DECL_FUNCTION_SPECIFIC_TARGET.
180
181 2014-02-14 Jakub Jelinek <jakub@redhat.com>
182
183 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
184 (get_initial_def_for_induction, vectorizable_induction): Ignore
185 debug stmts when looking for exit_phi.
186 (vectorizable_live_operation): Fix up condition.
187
188 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
189
190 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
191 nreverse() because it changes the content of original tree list.
192
193 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
194
195 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
196 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
197
198 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
199
200 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
201 GNU coding standards.
202
203 2014-02-13 Jakub Jelinek <jakub@redhat.com>
204
205 PR debug/60152
206 * dwarf2out.c (gen_subprogram_die): Don't call
207 add_calling_convention_attribute if subr_die is old_die.
208
209 2014-02-13 Sharad Singhai <singhai@google.com>
210
211 * doc/optinfo.texi: Fix order of nodes.
212
213 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
214
215 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
216 operands[2], not operands[3].
217
218 2014-02-13 Richard Biener <rguenther@suse.de>
219
220 PR bootstrap/59878
221 * doc/install.texi (ISL): Update recommended version to 0.12.2,
222 mention the possibility of an in-tree build.
223 (CLooG): Update recommended version to 0.18.1, mention the
224 possibility of an in-tree build and clarify that the ISL
225 bundled with CLooG does not work.
226
227 2014-02-13 Jakub Jelinek <jakub@redhat.com>
228
229 PR target/43546
230 * expr.c (compress_float_constant): If x is a hard register,
231 extend into a pseudo and then move to x.
232
233 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
234
235 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
236 caused by bad second argument to warning_at() with -mhotpatch and
237 nested functions (e.g. with gfortran).
238
239 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
240
241 * opts.c (option_name): Remove "enabled by default" rider.
242
243 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
244
245 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
246
247 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
248 Uros Bizjak <ubizjak@gmail.com>
249
250 PR target/60151
251 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
252 * configure: Regenerated.
253
254 2014-02-12 Richard Biener <rguenther@suse.de>
255
256 * vec.c (vec_prefix::calculate_allocation): Move as
257 inline variant to vec.h.
258 (vec_prefix::calculate_allocation_1): New out-of-line version.
259 * vec.h (vec_prefix::calculate_allocation_1): Declare.
260 (vec_prefix::m_has_auto_buf): Rename to ...
261 (vec_prefix::m_using_auto_storage): ... this.
262 (vec_prefix::calculate_allocation): Inline the easy cases
263 and dispatch to calculate_allocation_1 which doesn't need the
264 prefix address.
265 (va_heap::reserve): Use gcc_checking_assert.
266 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
267 m_using_auto_storage.
268 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
269 member and adjust.
270 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
271 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
272 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
273
274 2014-02-12 Richard Biener <rguenther@suse.de>
275
276 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
277 when we found a dependence.
278
279 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
280
281 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
282 common code...
283 (maybe_fold_stmt): ... into this new function.
284 * omp-low.c (lower_omp): Update comment.
285
286 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
287 last use.
288
289 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
290 dereference.
291
292 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
293
294 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
295 identifiers in comments.
296 (cortexa53_extra_costs): Likewise.
297 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
298 (cortexa7_extra_costs): Likewise.
299 (cortexa12_extra_costs): Likewise.
300 (cortexa15_extra_costs): Likewise.
301 (v7m_extra_costs): Likewise.
302
303 2014-02-12 Richard Biener <rguenther@suse.de>
304
305 PR middle-end/60092
306 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
307 of posix_memalign being successful.
308 (lower_stmt): Restrict lowering of posix_memalign to when
309 -ftree-bit-ccp is enabled.
310
311 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
312
313 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
314 arg_loc.
315 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
316
317 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
318
319 PR rtl-optimization/60116
320 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
321 other_insn once the combination has been validated.
322
323 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
324
325 PR lto/59468
326 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
327 and wrapper.
328 * ipa-devirt.c: Include demangle.h
329 (odr_violation_reported): New static variable.
330 (add_type_duplicate): Update odr_violations.
331 (maybe_record_node): Add completep parameter; update it.
332 (record_target_from_binfo): Add COMPLETEP parameter;
333 update it as needed.
334 (possible_polymorphic_call_targets_1): Likewise.
335 (struct polymorphic_call_target_d): Add nonconstruction_targets;
336 rename FINAL to COMPLETE.
337 (record_targets_from_bases): Sanity check we found the binfo;
338 fix COMPLETEP updating.
339 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
340 parameter, fix computing of COMPLETEP.
341 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
342 at LTO time do demangling.
343 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
344 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
345 parameter.
346 (gimple_get_virt_method_for_binfo): Likewise.
347 * gimple-fold.h (gimple_get_virt_method_for_binfo,
348 gimple_get_virt_method_for_vtable): Update prototypes.
349
350 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
351
352 PR target/49008
353 * genautomata.c (add_presence_absence): Fix typo with
354 {final_}presence_list.
355
356 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
357
358 PR target/60137
359 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
360 for VSX/Altivec vectors that land in GPR registers.
361
362 2014-02-11 Richard Henderson <rth@redhat.com>
363 Jakub Jelinek <jakub@redhat.com>
364
365 PR debug/59776
366 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
367 around drhs if type conversion to lacc->type is not useless.
368
369 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
370
371 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
372 tuning struct.
373 (cortex-a57.cortex-a53): Likewise.
374 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
375
376 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
377
378 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
379 arm_restrict_it.
380
381 2014-02-11 Renlin Li <Renlin.Li@arm.com>
382
383 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
384 add_options_for_arm_vfp3.
385
386 2014-02-11 Jeff Law <law@redhat.com>
387
388 PR middle-end/54041
389 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
390 object with an undesirable mode.
391
392 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
393
394 PR libgomp/60107
395 * config/i386/sol2-9.h: New file.
396 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
397 *-*-solaris2.9*): Use it.
398
399 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
400
401 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
402 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
403
404 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
405
406 * config/microblaze/microblaze.c: Extend mcpu version format
407
408 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
409
410 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
411
412 2014-02-10 Richard Henderson <rth@redhat.com>
413
414 PR target/59927
415 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
416 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
417 ms-abi vs -mno-accumulate-outgoing-args.
418 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
419 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
420 respect to ms-abi.
421
422 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
423
424 PR middle-end/60080
425 * cfgexpand.c (expand_asm_operands): Attach source location to
426 ASM_INPUT rtx objects.
427 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
428
429 2014-02-10 Nick Clifton <nickc@redhat.com>
430
431 * config/mn10300/mn10300.c (popcount): New function.
432 (mn10300_expand_prologue): Include saved registers in stack usage
433 count.
434
435 2014-02-10 Jeff Law <law@redhat.com>
436
437 PR middle-end/52306
438 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
439 when changing the SET_DEST of a prior insn to avoid an input reload.
440
441 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
442
443 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
444 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
445 -mcall-openbsd, or -mcall-linux.
446 (CC1_ENDIAN_BIG_SPEC): Remove.
447 (CC1_ENDIAN_LITTLE_SPEC): Remove.
448 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
449 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
450 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
451 and %cc1_endian_default.
452 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
453
454 2014-02-10 Richard Biener <rguenther@suse.de>
455
456 PR tree-optimization/60115
457 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
458 MEM_REF handling. Properly verify that the accesses are not
459 out of the objects bound.
460
461 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
462
463 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
464 coretex to cortex.
465
466 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
467
468 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
469 proper constants and fix formatting.
470 (possible_polymorphic_call_targets): Fix formatting.
471
472 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
473 Ilya Tocar <ilya.tocar@intel.com>
474
475 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
476 (_mm512_loadu_epi32): Renamed into...
477 (_mm512_loadu_si512): This.
478 (_mm512_storeu_epi32): Renamed into...
479 (_mm512_storeu_si512): This.
480 (_mm512_maskz_ceil_ps): Removed.
481 (_mm512_maskz_ceil_pd): Ditto.
482 (_mm512_maskz_floor_ps): Ditto.
483 (_mm512_maskz_floor_pd): Ditto.
484 (_mm512_floor_round_ps): Ditto.
485 (_mm512_floor_round_pd): Ditto.
486 (_mm512_ceil_round_ps): Ditto.
487 (_mm512_ceil_round_pd): Ditto.
488 (_mm512_mask_floor_round_ps): Ditto.
489 (_mm512_mask_floor_round_pd): Ditto.
490 (_mm512_mask_ceil_round_ps): Ditto.
491 (_mm512_mask_ceil_round_pd): Ditto.
492 (_mm512_maskz_floor_round_ps): Ditto.
493 (_mm512_maskz_floor_round_pd): Ditto.
494 (_mm512_maskz_ceil_round_ps): Ditto.
495 (_mm512_maskz_ceil_round_pd): Ditto.
496 (_mm512_expand_pd): Ditto.
497 (_mm512_expand_ps): Ditto.
498 * config/i386/i386.c (ix86_builtins): Remove
499 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
500 (bdesc_args): Ditto.
501 * config/i386/predicates.md (const1256_operand): New.
502 (const_1_to_2_operand): Ditto.
503 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
504 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
505 (*avx512pf_gatherpf<mode>sf): Ditto.
506 (avx512pf_gatherpf<mode>df): Ditto.
507 (*avx512pf_gatherpf<mode>df_mask): Ditto.
508 (*avx512pf_gatherpf<mode>df): Ditto.
509 (avx512pf_scatterpf<mode>sf): Ditto.
510 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
511 (*avx512pf_scatterpf<mode>sf): Ditto.
512 (avx512pf_scatterpf<mode>df): Ditto.
513 (*avx512pf_scatterpf<mode>df_mask): Ditto.
514 (*avx512pf_scatterpf<mode>df): Ditto.
515 (avx512f_expand<mode>): Removed.
516 (<shift_insn><mode>3<mask_name>): Change predicate type.
517
518 2014-02-08 Jakub Jelinek <jakub@redhat.com>
519
520 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
521 not at the end of datarefs vector use ordered_remove to avoid
522 reordering datarefs vector.
523
524 PR c/59984
525 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
526 mark local addressable non-static vars as GOVD_PRIVATE
527 instead of GOVD_LOCAL.
528 * omp-low.c (lower_omp_for): Move gimple_bind_vars
529 and BLOCK_VARS of gimple_bind_block to new_stmt rather
530 than copying them.
531
532 PR middle-end/60092
533 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
534 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
535 assume_aligned or alloc_align attributes.
536 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
537 arguments. Handle also assume_aligned and alloc_align attributes.
538 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
539 calls to functions with assume_aligned or alloc_align attributes.
540 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
541
542 2014-02-08 Terry Guo <terry.guo@arm.com>
543
544 * doc/invoke.texi: Document ARM -march=armv7e-m.
545
546 2014-02-08 Jakub Jelinek <jakub@redhat.com>
547
548 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
549 flag on __cilkrts_rethrow builtin.
550
551 PR ipa/60026
552 * ipa-cp.c (determine_versionability): Fail at -O0
553 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
554 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
555
556 Revert:
557 2014-02-04 Jakub Jelinek <jakub@redhat.com>
558
559 PR ipa/60026
560 * tree-inline.c (copy_forbidden): Fail for
561 __attribute__((optimize (0))) functions.
562
563 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
564
565 * varpool.c: Include pointer-set.h.
566 (varpool_remove_unreferenced_decls): Variables in other partitions
567 will not be output; be however careful to not lose information
568 about partitioning.
569
570 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
571
572 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
573 lookup in the vtable constructor.
574
575 2014-02-07 Jeff Law <law@redhat.com>
576
577 PR target/40977
578 * config/m68k/m68k.md (ashldi_extsi): Turn into a
579 define_insn_and_split.
580
581 * ipa-inline.c (inline_small_functions): Fix typos.
582
583 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
584
585 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
586 (s390_can_use_return_insn): Declare.
587 * config/s390/s390.h (EPILOGUE_USES): Define.
588 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
589 instructions.
590 (s390_chunkify_start): Handle return JUMP_LABELs.
591 (s390_early_mach): Emit a main_pool instruction on the entry edge.
592 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
593 (s390_can_use_return_insn): New functions.
594 (s390_fix_long_loop_prediction): Handle conditional returns.
595 (TARGET_SET_UP_BY_PROLOGUE): Define.
596 * config/s390/s390.md (ANY_RETURN): New code iterator.
597 (*creturn, *csimple_return, return, simple_return): New patterns.
598
599 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
600
601 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
602 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
603 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
604 REG_CFA_RESTORE list when deciding not to restore a register.
605
606 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
607
608 * config/s390/s390.c: Include tree-pass.h and context.h.
609 (s390_early_mach): New function, split out from...
610 (s390_emit_prologue): ...here.
611 (pass_data_s390_early_mach): New pass structure.
612 (pass_s390_early_mach): New class.
613 (s390_option_override): Create and register early_mach pass.
614 Move to end of file.
615
616 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
617
618 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
619 to match for the exit block.
620
621 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
622
623 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
624 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
625 Reject misaligned operands.
626
627 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
628
629 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
630
631 2014-02-07 Richard Biener <rguenther@suse.de>
632
633 PR middle-end/60092
634 * gimple-low.c (lower_builtin_posix_memalign): New function.
635 (lower_stmt): Call it to lower posix_memalign in a way
636 to make alignment info accessible.
637
638 2014-02-07 Jakub Jelinek <jakub@redhat.com>
639
640 PR c++/60082
641 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
642 __builtin_setjmp_receiver.
643
644 2014-02-07 Richard Biener <rguenther@suse.de>
645
646 PR middle-end/60092
647 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
648 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
649 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
650 Handle BUILT_IN_POSIX_MEMALIGN.
651 (find_func_clobbers): Likewise.
652 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
653 (call_may_clobber_ref_p_1): Likewise.
654
655 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
656
657 PR ipa/59918
658 * ipa-devirt.c (record_target_from_binfo): Remove overactive
659 sanity check.
660
661 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
662
663 PR ipa/59469
664 * lto-cgraph.c (lto_output_node): Use
665 symtab_get_symbol_partitioning_class.
666 (lto_output_varpool_node): likewise.
667 (symtab_get_symbol_partitioning_class): Move here from
668 lto/lto-partition.c
669 * cgraph.h (symbol_partitioning_class): Likewise.
670 (symtab_get_symbol_partitioning_class): Declare.
671
672 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
673
674 * ggc.h (ggc_internal_cleared_alloc): New macro.
675 * vec.h (vec_safe_copy): Handle memory stats.
676 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
677 * target-globals.c (save_target_globals): Likewise.
678
679 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
680
681 PR target/60077
682 * expr.c (emit_move_resolve_push): Export; be bit more selective
683 on when to clear alias set.
684 * expr.h (emit_move_resolve_push): Declare.
685 * function.h (struct function): Add tail_call_marked.
686 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
687 * config/i386/i386-protos.h (ix86_expand_push): Remove.
688 * config/i386/i386.md (TImode move expander): De not call
689 ix86_expand_push.
690 (FP push expanders): Preserve memory attributes.
691 * config/i386/sse.md (push<mode>1): Remove.
692 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
693 (ix86_expand_push): Remove.
694 * config/i386/mmx.md (push<mode>1): Remove.
695
696 2014-02-06 Jakub Jelinek <jakub@redhat.com>
697
698 PR rtl-optimization/60030
699 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
700 lopart with paradoxical subreg before shifting it up by hprec.
701
702 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
703
704 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
705 Remove extra newline at end of file.
706 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
707 (arm_issue_rate): Handle cortexa57.
708 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
709 (cortex-a57.cortex-a53): Likewise.
710
711 2014-02-06 Jakub Jelinek <jakub@redhat.com>
712
713 PR target/59575
714 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
715 don't record in REG_FRAME_RELATED_EXPR registers not set in that
716 bitmask.
717 (arm_expand_prologue): Adjust all callers.
718 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
719 info, registers also at the lowest numbered registers side. Use
720 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
721 XEXP.
722
723 PR debug/59992
724 * var-tracking.c (adjust_mems): Before adding a SET to
725 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
726
727 2014-02-06 Alan Modra <amodra@gmail.com>
728
729 PR target/60032
730 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
731 change SDmode to DDmode when lra_in_progress.
732
733 2014-02-06 Jakub Jelinek <jakub@redhat.com>
734
735 PR middle-end/59150
736 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
737 free_data_ref on the dr first, and before goto again also set dr
738 to the next dr. For simd_lane_access, free old datarefs[i] before
739 overwriting it. For get_vectype_for_scalar_type failure, don't
740 free_data_ref if simd_lane_access.
741
742 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
743
744 PR target/60062
745 * tree.h (opts_for_fn): New inline function.
746 (opt_for_fn): Define.
747 * config/i386/i386.c (ix86_function_regparm): Use
748 opt_for_fn (decl, optimize) instead of optimize.
749
750 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
751
752 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
753 for SYMBOL_REF in large memory model.
754
755 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
756
757 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
758 and crypto support.
759 (cortex-a57): Likewise.
760 (cortex-a57.cortex-a53): Likewise.
761
762 2014-02-06 Yury Gribov <y.gribov@samsung.com>
763 Kugan Vivekanandarajah <kuganv@linaro.org>
764
765 * config/arm/arm.c (arm_vector_alignment_reachable): Check
766 unaligned_access.
767 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
768
769 2014-02-06 Richard Biener <rguenther@suse.de>
770
771 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
772 set_loop_copy and initialize_original_copy_tables.
773
774 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
775
776 * config/aarch64/aarch64-simd.md
777 (aarch64_ashr_simddi): Change QI to SI.
778
779 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
780 Jakub Jelinek <jakub@redhat.com>
781
782 PR middle-end/60013
783 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
784 of the dataflow.
785
786 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
787
788 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
789 CODE_FOR_altivec_vpku[hw]um to
790 CODE_FOR_altivec_vpku[hw]um_direct.
791 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
792 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
793 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
794 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
795
796 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
797
798 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
799 generation for -maltivec=be.
800 (altivec_vsumsws): Simplify redundant test.
801
802 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
803
804 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
805 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
806 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
807 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
808 gen_altivec_vpkuwum.
809 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
810 BYTES_BIG_ENDIAN.
811 (altivec_vpks<VI_char>ss): Likewise.
812 (altivec_vpks<VI_char>us): Likewise.
813 (altivec_vpku<VI_char>us): Likewise.
814 (altivec_vpku<VI_char>um): Likewise.
815 (altivec_vpku<VI_char>um_direct): New (copy of
816 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
817 internal use).
818 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
819 target is little endian and -maltivec=be is not specified.
820 (*altivec_vupkhs<VU_char>_direct): New (copy of
821 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
822 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
823 target is little endian and -maltivec=be is not specified.
824 (*altivec_vupkls<VU_char>_direct): New (copy of
825 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
826 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
827 little endian and -maltivec=be is not specified.
828 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
829 little endian and -maltivec=be is not specified.
830
831 2014-02-05 Richard Henderson <rth@redhat.com>
832
833 PR debug/52727
834 * combine-stack-adj.c: Revert r206943.
835 * sched-int.h (struct deps_desc): Add last_args_size.
836 * sched-deps.c (init_deps): Initialize it.
837 (sched_analyze_insn): Add OUTPUT dependencies between insns that
838 contain REG_ARGS_SIZE notes.
839
840 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
841
842 * lto-cgraph.c (asm_nodes_output): Make global.
843 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
844 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
845 (driver_handle_option): Handle OPT_fwpa.
846
847 2014-02-05 Jakub Jelinek <jakub@redhat.com>
848
849 PR ipa/59947
850 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
851 a comment typo and formatting issue. If odr_hash hasn't been
852 created, return vNULL and set *completep to false.
853
854 PR middle-end/57499
855 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
856 bb with no successors.
857
858 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
859
860 PR target/59718
861 * doc/invoke.texi (-march): Clarify documentation for ARM.
862 (-mtune): Likewise.
863 (-mcpu): Likewise.
864
865 2014-02-05 Richard Biener <rguenther@suse.de>
866
867 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
868 when not vectorizing because of too many alias checks.
869 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
870 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
871
872 2014-02-05 Nick Clifton <nickc@redhat.com>
873
874 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
875 accept extended registers in any mode when compiling for the MN10300.
876
877 2014-02-05 Yury Gribov <y.gribov@samsung.com>
878
879 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
880 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
881 sanitization attributes.
882 (can_inline_edge_p): Likewise.
883 (sanitize_attrs_match_for_inline_p): New function.
884
885 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
886
887 * ipa-prop.c (detect_type_change): Shor circuit testing of
888 type changes on THIS pointer.
889
890 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
891
892 PR target/59777
893 * config/pa/pa.c (legitimize_tls_address): Return original address
894 if not passed a SYMBOL_REF rtx.
895 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
896 addresses.
897 (pa_emit_move_sequence): Simplify TLS source operands.
898 (pa_legitimate_constant_p): Reject all TLS constants.
899 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
900 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
901
902 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
903
904 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
905 groups when we know they are controlled by LTO.
906 * varasm.c (default_binds_local_p_1): If object is in other partition,
907 it will be resolved locally.
908
909 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
910
911 * config/host-linux.c (linux_gt_pch_use_address): Don't
912 use SSIZE_MAX because it is not always defined.
913
914 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
915
916 PR bootstrap/59913
917 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
918 threshold for pseudo splitting.
919 (update_ebb_live_info): Process call argument hard registers and
920 hard registers from insn definition too.
921 (max_small_class_regs_num): New constant.
922 (inherit_in_ebb): Update live hard regs through EBBs. Update
923 reloads_num only for small register classes. Don't split for
924 outputs of jumps.
925
926 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
927
928 PR ipa/60058
929 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
930 is non-null.
931
932 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
933
934 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
935 visibility is safe.
936
937 2014-02-04 Marek Polacek <polacek@redhat.com>
938
939 * gdbinit.in (pel): Define.
940
941 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
942
943 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
944 behavior.
945
946 2014-02-04 Richard Biener <rguenther@suse.de>
947
948 PR lto/59723
949 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
950 in function context local.
951 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
952 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
953 similar to LTO_imported_decl_ref.
954
955 2014-02-04 Jakub Jelinek <jakub@redhat.com>
956
957 PR tree-optimization/60002
958 * cgraphclones.c (build_function_decl_skip_args): Clear
959 DECL_LANG_SPECIFIC.
960
961 PR tree-optimization/60023
962 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
963 false to gsi_replace.
964 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
965 has been in some EH region and vec_stmt could throw, add
966 vec_stmt into the same EH region.
967 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
968 has no lhs, ignore it.
969 * internal-fn.c (expand_MASK_LOAD): Likewise.
970
971 PR ipa/60026
972 * tree-inline.c (copy_forbidden): Fail for
973 __attribute__((optimize (0))) functions.
974
975 PR other/58712
976 * omp-low.c (simd_clone_struct_copy): If from->inbranch
977 is set, copy one less argument.
978 (expand_simd_clones): Don't subtract clone_info->inbranch
979 from simd_clone_struct_alloc argument.
980
981 PR rtl-optimization/57915
982 * recog.c (simplify_while_replacing): If all unary/binary/relational
983 operation arguments are constant, attempt to simplify those.
984
985 PR middle-end/59261
986 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
987 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
988
989 2014-02-04 Richard Biener <rguenther@suse.de>
990
991 PR tree-optimization/60012
992 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
993 TBAA disambiguation to all DDRs.
994
995 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
996
997 PR target/59788
998 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
999 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1000
1001 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1002
1003 PR ipa/59882
1004 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1005
1006 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1007
1008 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1009 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1010
1011 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1012
1013 PR ipa/59831
1014 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1015 to figure out targets of polymorphic calls with known decl.
1016 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1017 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1018 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1019 (get_polymorphic_call_info): ... here.
1020 (get_polymorphic_call_info_from_invariant): New function.
1021
1022 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1023
1024 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1025 lookup via vtable pointer; check for type consistency
1026 and turn inconsitent facts into UNREACHABLE.
1027 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1028 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1029 type inconsistent querries; return UNREACHABLE instead.
1030
1031 2014-02-03 Richard Henderson <rth@twiddle.net>
1032
1033 PR tree-opt/59924
1034 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1035 already processed this node.
1036 (normalize_one_pred_1): Pass along mark_set.
1037 (normalize_one_pred): Create and destroy a pointer_set_t.
1038 (normalize_one_pred_chain): Likewise.
1039
1040 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1041
1042 PR gcov-profile/58602
1043 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1044
1045 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1046
1047 PR ipa/59831
1048 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1049 -fno-devirtualize; try to devirtualize by the knowledge of
1050 virtual table pointer given by aggregate propagation.
1051 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1052 (ipa_print_node_jump_functions): Dump also offset that
1053 is relevant for polymorphic calls.
1054 (determine_known_aggregate_parts): Add arg_type parameter; use it
1055 instead of determining the type from pointer type.
1056 (ipa_compute_jump_functions_for_edge): Update call of
1057 determine_known_aggregate_parts.
1058 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1059 (gimple_get_virt_method_for_binfo): ... here; simplify using
1060 vtable_pointer_value_to_vtable.
1061 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1062 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1063 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1064 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1065 POINTER_PLUS_EXPR.
1066 (vtable_pointer_value_to_binfo): ... here.
1067 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1068
1069 2014-02-03 Teresa Johnson <tejohnson@google.com>
1070
1071 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1072 redef of outer loop index variable.
1073
1074 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1075
1076 PR c++/53017
1077 PR c++/59211
1078 * doc/extend.texi (Function Attributes): Typo.
1079
1080 2014-02-03 Cong Hou <congh@google.com>
1081
1082 PR tree-optimization/60000
1083 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1084 if the vectorized statement is a store. A store statement can only
1085 appear at the end of pattern statements.
1086
1087 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1088
1089 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1090 (ix86_option_override_internal): Default long double to 64-bit for
1091 32-bit Bionic and to 128-bit for 64-bit Bionic.
1092
1093 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1094 TARGET_LONG_DOUBLE_128 is true.
1095 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1096
1097 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1098 (mlong-double-64): Negate -mlong-double-128.
1099 (mlong-double-128): New option.
1100
1101 * config/i386/i386-c.c (ix86_target_macros): Define
1102 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1103
1104 * doc/invoke.texi: Document -mlong-double-128.
1105
1106 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1107
1108 PR rtl-optimization/60024
1109 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1110
1111 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1112
1113 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1114
1115 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1116
1117 PR rtl-optimization/57662
1118 * sel-sched.c (code_motion_path_driver): Do not mark already not
1119 existing blocks in the visiting bitmap.
1120
1121 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1122
1123 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1124 on the insn being emitted.
1125
1126 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1127 Will Deacon <will.deacon@arm.com>
1128
1129 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1130
1131 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1132
1133 * config/arm/arm-tables.opt: Regenerate.
1134
1135 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1136
1137 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1138 for vector types other than V16QImode.
1139 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1140 define_expand, and call altivec_expand_vec_perm_le when producing
1141 code with little endian element order.
1142 (*altivec_vperm_<mode>_internal): New insn having previous
1143 behavior of altivec_vperm_<mode>.
1144 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1145 altivec_expand_vec_perm_le when producing code with little endian
1146 element order.
1147 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1148 behavior of altivec_vperm_<mode>_uns.
1149
1150 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1151
1152 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1153 (altivec_vsumsws): Add handling for -maltivec=be with a little
1154 endian target.
1155 (altivec_vsumsws_direct): New.
1156 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1157 gen_altivec_vsumsws.
1158
1159 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1160
1161 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1162 vtable_pointer_value_to_binfo): New functions.
1163 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1164 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1165
1166 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1167
1168 * config/nios2/nios2.md (load_got_register): Initialize GOT
1169 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1170 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1171
1172 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1173
1174 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1175 preserverd by passthrough, do not propagate the type.
1176
1177 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1178
1179 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1180 (mips_atomic_assign_expand_fenv): New function.
1181 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1182
1183 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1184
1185 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1186 (__builtin_mips_set_fcsr): Likewise.
1187 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1188 MIPS_USI_FTYPE_VOID.
1189 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1190 (mips16_expand_set_fcsr): Likewise.
1191 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1192 (mips16_set_fcsr_stub): Likewise.
1193 (mips16_get_fcsr_one_only_stub): New class.
1194 (mips16_set_fcsr_one_only_stub): Likewise.
1195 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1196 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1197 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1198 (hard_float): New availability predicate.
1199 (mips_builtins): Add get_fcsr and set_fcsr.
1200 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1201 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1202 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1203 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1204 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1205 patterns.
1206
1207 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1208
1209 * config/mips/mips.c (mips_one_only_stub): New class.
1210 (mips_need_mips16_rdhwr_p): Replace with...
1211 (mips16_rdhwr_stub): ...this new variable.
1212 (mips16_stub_call_address): New function.
1213 (mips16_rdhwr_one_only_stub): New class.
1214 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1215 (mips_output_mips16_rdhwr): Delete.
1216 (mips_finish_stub): New function.
1217 (mips_code_end): Use it to handle rdhwr stubs.
1218
1219 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1220
1221 PR target/60017
1222 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1223 when calculating size of integer atomic types.
1224
1225 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1226
1227 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1228
1229 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1230
1231 PR tree-optimization/60003
1232 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1233 * profile.c (branch_prob): Use gimple_call_builtin_p
1234 to check for BUILT_IN_SETJMP_RECEIVER.
1235 * tree-inline.c (copy_bb): Call notice_special_calls.
1236
1237 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1238
1239 PR bootstrap/59985
1240 * lra-constraints.c (process_alt_operands): Update reload_sum only
1241 on the first pass.
1242
1243 2014-01-31 Richard Henderson <rth@redhat.com>
1244
1245 PR middle-end/60004
1246 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1247 until after else_eh is processed.
1248
1249 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1250
1251 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1252 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1253 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1254 in smmintrin.h, remove them.
1255 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1256 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1257 * config/i386/i386.md (ROUND_SAE): Fix value.
1258 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1259 (const48_operand): New.
1260 * config/i386/subst.md (round), (round_expand): Use
1261 const_4_or_8_to_11_operand.
1262 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1263
1264 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1265
1266 * config/i386/constraints.md (Yk): Swap meaning with k.
1267 * config/i386/i386.md (movhi_internal): Change Yk to k.
1268 (movqi_internal): Ditto.
1269 (*k<logic><mode>): Ditto.
1270 (*andhi_1): Ditto.
1271 (*andqi_1): Ditto.
1272 (kandn<mode>): Ditto.
1273 (*<code>hi_1): Ditto.
1274 (*<code>qi_1): Ditto.
1275 (kxnor<mode>): Ditto.
1276 (kortestzhi): Ditto.
1277 (kortestchi): Ditto.
1278 (kunpckhi): Ditto.
1279 (*one_cmplhi2_1): Ditto.
1280 (*one_cmplqi2_1): Ditto.
1281 * config/i386/sse.md (): Change k to Yk.
1282 (avx512f_load<mode>_mask): Ditto.
1283 (avx512f_blendm<mode>): Ditto.
1284 (avx512f_store<mode>_mask): Ditto.
1285 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1286 (avx512f_storedqu<mode>_mask): Ditto.
1287 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1288 Ditto.
1289 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1290 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1291 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1292 (avx512f_maskcmp<mode>3): Ditto.
1293 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1294 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1295 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1296 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1297 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1298 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1299 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1300 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1301 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1302 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1303 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1304 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1305 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1306 (vec_extract_lo_<mode>_maskm): Ditto.
1307 (vec_extract_hi_<mode>_maskm): Ditto.
1308 (avx512f_vternlog<mode>_mask): Ditto.
1309 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1310 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1311 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1312 (avx512f_<code>v8div16qi2_mask): Ditto.
1313 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1314 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1315 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1316 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1317 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1318 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1319 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1320 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1321 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1322 (avx512cd_maskb_vec_dupv8di): Ditto.
1323 (avx512cd_maskw_vec_dupv16si): Ditto.
1324 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1325 (avx512f_vpermi2var<mode>3_mask): Ditto.
1326 (avx512f_vpermi2var<mode>3_mask): Ditto.
1327 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1328 (*avx512f_gathersi<mode>): Ditto.
1329 (*avx512f_gathersi<mode>_2): Ditto.
1330 (*avx512f_gatherdi<mode>): Ditto.
1331 (*avx512f_gatherdi<mode>_2): Ditto.
1332 (*avx512f_scattersi<mode>): Ditto.
1333 (*avx512f_scatterdi<mode>): Ditto.
1334 (avx512f_compress<mode>_mask): Ditto.
1335 (avx512f_compressstore<mode>_mask): Ditto.
1336 (avx512f_expand<mode>_mask): Ditto.
1337 * config/i386/subst.md (mask): Change k to Yk.
1338 (mask_scalar_merge): Ditto.
1339 (sd): Ditto.
1340
1341 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1342
1343 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1344
1345 2014-01-31 Richard Biener <rguenther@suse.de>
1346
1347 PR middle-end/59990
1348 * builtins.c (fold_builtin_memory_op): Make sure to not
1349 use a floating-point mode or a boolean or enumeral type for
1350 the copy operation.
1351
1352 2014-01-30 DJ Delorie <dj@redhat.com>
1353
1354 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1355 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1356 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1357 whenever main() has an epilogue.
1358
1359 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1360
1361 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1362 unused variable "field".
1363 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1364 (vsx_mergeh_<mode>): Likewise.
1365 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1366 (altivec_vmrghh): Likewise.
1367 (altivec_vmrghw): Likewise.
1368 (altivec_vmrglb): Likewise.
1369 (altivec_vmrglh): Likewise.
1370 (altivec_vmrglw): Likewise.
1371 (altivec_vspltb): Add missing uses.
1372 (altivec_vsplth): Likewise.
1373 (altivec_vspltw): Likewise.
1374 (altivec_vspltsf): Likewise.
1375
1376 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1377
1378 PR target/59923
1379 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1380 frame related instructions.
1381
1382 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1383
1384 PR rtl-optimization/59959
1385 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1386 any reload of register whose subreg is invalid.
1387
1388 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1389
1390 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1391 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1392 Add missing return type - void.
1393
1394 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1395
1396 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1397 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1398 remove element index adjustment for endian (now handled in vsx.md
1399 and altivec.md).
1400 (altivec_expand_vec_perm_const): Use
1401 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1402 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1403 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1404 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1405 define_expand and a new define_insn *altivec_vspltb_internal;
1406 adjust for -maltivec=be on a little endian target.
1407 (altivec_vspltb_direct): New.
1408 (altivec_vsplth): Divide into a define_expand and a new
1409 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1410 little endian target.
1411 (altivec_vsplth_direct): New.
1412 (altivec_vspltw): Divide into a define_expand and a new
1413 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1414 little endian target.
1415 (altivec_vspltw_direct): New.
1416 (altivec_vspltsf): Divide into a define_expand and a new
1417 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1418 a little endian target.
1419
1420 2014-01-30 Richard Biener <rguenther@suse.de>
1421
1422 PR tree-optimization/59993
1423 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1424 can propagate form the earlier stmt and avoid the transform
1425 when the intermediate result is needed.
1426
1427 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1428
1429 * README.Portability: Fix typo.
1430
1431 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1432
1433 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1434 comparison_operator with ordered_comparison_operator.
1435
1436 2014-01-30 Nick Clifton <nickc@redhat.com>
1437
1438 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1439 Rename to mn10300_store_multiple_regs.
1440 * config/mn10300/mn10300.c: Likewise.
1441 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1442 store_multiple_regs.
1443 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1444 Call mn10300_store_multiple_regs.
1445
1446 2014-01-30 Nick Clifton <nickc@redhat.com>
1447 DJ Delorie <dj@redhat.com>
1448
1449 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1450 %fp 2 to keep registers after it properly word-aligned.
1451 (rl78_alloc_physical_registers_umul): Handle the case where both
1452 input operands are the same.
1453
1454 2014-01-30 Richard Biener <rguenther@suse.de>
1455
1456 PR tree-optimization/59903
1457 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1458 check properly.
1459
1460 2014-01-30 Jason Merrill <jason@redhat.com>
1461
1462 PR c++/59633
1463 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1464
1465 PR c++/59645
1466 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1467
1468 2014-01-30 Richard Biener <rguenther@suse.de>
1469
1470 PR tree-optimization/59951
1471 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1472
1473 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1474
1475 PR target/59784
1476 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1477 SFmode to DFmode case.
1478
1479 2014-01-29 DJ Delorie <dj@redhat.com>
1480
1481 * config/msp430/msp430.opt (-minrt): New.
1482 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1483 if -minrt given.
1484 (ENDFILE_SPEC): Likewise.
1485
1486 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1487
1488 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1489 (estimate_function_body_sizes): Use it.
1490
1491 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1492
1493 PR c++/58561
1494 * dwarf2out.c (is_cxx_auto): New.
1495 (is_base_type): Use it.
1496 (gen_type_die_with_usage): Likewise.
1497
1498 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1499
1500 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1501 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1502 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1503 -maltivec=be with LE targets.
1504 (vsx_mergeh_<mode>): Likewise.
1505 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1506 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1507 (altivec_vmrghb): Replace with define_expand and new
1508 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1509 (altivec_vmrghb_direct): New define_insn.
1510 (altivec_vmrghh): Replace with define_expand and new
1511 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1512 (altivec_vmrghh_direct): New define_insn.
1513 (altivec_vmrghw): Replace with define_expand and new
1514 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1515 (altivec_vmrghw_direct): New define_insn.
1516 (*altivec_vmrghsf): Adjust for endianness.
1517 (altivec_vmrglb): Replace with define_expand and new
1518 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1519 (altivec_vmrglb_direct): New define_insn.
1520 (altivec_vmrglh): Replace with define_expand and new
1521 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1522 (altivec_vmrglh_direct): New define_insn.
1523 (altivec_vmrglw): Replace with define_expand and new
1524 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1525 (altivec_vmrglw_direct): New define_insn.
1526 (*altivec_vmrglsf): Adjust for endianness.
1527 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1528 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1529 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1530 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1531 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1532 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1533 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1534 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1535
1536 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1537
1538 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1539 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1540 whitespace.
1541
1542 2014-01-29 Richard Biener <rguenther@suse.de>
1543
1544 PR tree-optimization/58742
1545 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1546 associate_pointerplus_align.
1547 (associate_pointerplus_diff): New function.
1548 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1549 and associate_pointerplus_diff.
1550
1551 2014-01-29 Richard Biener <rguenther@suse.de>
1552
1553 * lto-streamer.h (LTO_major_version): Bump to 3.
1554 (LTO_minor_version): Reset to 0.
1555
1556 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1557
1558 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1559 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1560 (arm_file_start): Generate correct asm header for armv7ve.
1561 * config/arm/bpabi.h: Add multilib support for armv7ve.
1562 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1563 and cortex-a15 to armv7ve.
1564 * config/arm/t-aprofile: Add multilib support for armv7ve.
1565 * doc/invoke.texi: Document -march=armv7ve.
1566
1567 2014-01-29 Richard Biener <rguenther@suse.de>
1568
1569 PR tree-optimization/58742
1570 * tree-ssa-forwprop.c (associate_plusminus): Return true
1571 if we changed sth, defer EH cleanup to ...
1572 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1573 (simplify_mult): New function.
1574
1575 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1576
1577 PR middle-end/59917
1578 PR tree-optimization/59920
1579 * tree.c (build_common_builtin_nodes): Remove
1580 __builtin_setjmp_dispatcher initialization.
1581 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1582 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1583 instead of gsi_after_labels + manually skipping debug stmts.
1584 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1585 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1586 * tree-inline.c (copy_edges_for_bb): Remove
1587 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1588 argument. Ignore computed_goto_p stmts. Don't call
1589 make_abnormal_goto_edges. If a call might need abnormal edges
1590 for non-local gotos, see if it already has an edge to
1591 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1592 with true argument, don't do anything then, otherwise add
1593 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1594 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1595 caller.
1596 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1597 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1598 (lower_stmt): Don't set data->calls_builtin_setjmp.
1599 (lower_builtin_setjmp): Adjust comment.
1600 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1601 * tree-cfg.c (found_computed_goto): Remove.
1602 (factor_computed_gotos): Remove.
1603 (make_goto_expr_edges): Return bool, true for computed gotos.
1604 Don't call make_abnormal_goto_edges.
1605 (build_gimple_cfg): Don't set found_computed_goto, don't call
1606 factor_computed_gotos.
1607 (computed_goto_p): No longer static.
1608 (make_blocks): Don't set found_computed_goto.
1609 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1610 (make_edges): If make_goto_expr_edges returns true, push bb
1611 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1612 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1613 vector. Record mapping between bbs and OpenMP regions if there
1614 are any, adjust make_gimple_omp_edges caller. Call
1615 handle_abnormal_edges.
1616 (make_abnormal_goto_edges): Remove.
1617 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1618 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1619 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1620 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1621 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1622 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1623 filling *region also set *region_idx to (*region)->entry->index.
1624
1625 PR other/58712
1626 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1627 For REGs set ORIGINAL_REGNO.
1628
1629 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1630
1631 * doc/md.texi: Mention that a target shouldn't implement
1632 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1633 than hi/lo pair.
1634
1635 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1636
1637 PR tree-optimization/59594
1638 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1639 a copy of the datarefs vector rather than the vector itself.
1640
1641 2014-01-28 Jason Merrill <jason@redhat.com>
1642
1643 PR c++/53756
1644 * dwarf2out.c (auto_die): New static.
1645 (gen_type_die_with_usage): Handle C++1y 'auto'.
1646 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1647 on definition.
1648
1649 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1650
1651 PR target/59672
1652 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1653 (SPEC_X32): Likewise.
1654 (SPEC_64): Likewise.
1655 * config/i386/i386.c (ix86_option_override_internal): Turn off
1656 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1657 for TARGET_16BIT.
1658 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1659 * config/i386/i386.h (TARGET_16BIT): New macro.
1660 (TARGET_16BIT_P): Likewise.
1661 * config/i386/i386.opt: Add m16.
1662 * doc/invoke.texi: Document -m16.
1663
1664 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1665
1666 PR preprocessor/59935
1667 * input.c (location_get_source_line): Bail out on when line number
1668 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1669
1670 2014-01-28 Richard Biener <rguenther@suse.de>
1671
1672 PR tree-optimization/58742
1673 * tree-ssa-forwprop.c (associate_plusminus): Handle
1674 pointer subtraction of the form (T)(P + A) - (T)P.
1675
1676 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1677
1678 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1679 at const_int_cost.
1680
1681 2014-01-28 Richard Biener <rguenther@suse.de>
1682
1683 Revert
1684 2014-01-28 Richard Biener <rguenther@suse.de>
1685
1686 PR rtl-optimization/45364
1687 PR rtl-optimization/59890
1688 * var-tracking.c (local_get_addr_clear_given_value): Handle
1689 already cleared slot.
1690 (val_reset): Handle not allocated local_get_addr_cache.
1691 (vt_find_locations): Use post-order on the inverted CFG.
1692
1693 2014-01-28 Richard Biener <rguenther@suse.de>
1694
1695 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1696
1697 2014-01-28 Richard Biener <rguenther@suse.de>
1698
1699 PR rtl-optimization/45364
1700 PR rtl-optimization/59890
1701 * var-tracking.c (local_get_addr_clear_given_value): Handle
1702 already cleared slot.
1703 (val_reset): Handle not allocated local_get_addr_cache.
1704 (vt_find_locations): Use post-order on the inverted CFG.
1705
1706 2014-01-28 Alan Modra <amodra@gmail.com>
1707
1708 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1709 * configure.ac <recursive call for build != host>: Define
1710 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1711 and LD_FOR_BUILD too.
1712 * configure: Regenerate.
1713
1714 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1715
1716 * config/i386/i386.c (get_builtin_code_for_version): Separate
1717 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1718 Broadwell from Haswell.
1719
1720 2014-01-27 Steve Ellcey <sellcey@mips.com>
1721
1722 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1723 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1724 * config/mips/mips.c (mips_option_override): Change setting
1725 of TARGET_DSP.
1726 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1727 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1728 Change from Mask to Var.
1729
1730 2014-01-27 Jeff Law <law@redhat.com>
1731
1732 * ipa-inline.c (inline_small_functions): Fix typo.
1733
1734 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1735
1736 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1737 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1738 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1739 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1740 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1741 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1742 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1743 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1744 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1745 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1746 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1747 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1748 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1749 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1750 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1751 (_mm512_storeu_epi64): Ditto.
1752 (_mm512_cmpge_epi32_mask): Ditto.
1753 (_mm512_cmpge_epu32_mask): Ditto.
1754 (_mm512_cmpge_epi64_mask): Ditto.
1755 (_mm512_cmpge_epu64_mask): Ditto.
1756 (_mm512_cmple_epi32_mask): Ditto.
1757 (_mm512_cmple_epu32_mask): Ditto.
1758 (_mm512_cmple_epi64_mask): Ditto.
1759 (_mm512_cmple_epu64_mask): Ditto.
1760 (_mm512_cmplt_epi32_mask): Ditto.
1761 (_mm512_cmplt_epu32_mask): Ditto.
1762 (_mm512_cmplt_epi64_mask): Ditto.
1763 (_mm512_cmplt_epu64_mask): Ditto.
1764 (_mm512_cmpneq_epi32_mask): Ditto.
1765 (_mm512_cmpneq_epu32_mask): Ditto.
1766 (_mm512_cmpneq_epi64_mask): Ditto.
1767 (_mm512_cmpneq_epu64_mask): Ditto.
1768 (_mm512_expand_pd): Ditto.
1769 (_mm512_expand_ps): Ditto.
1770 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1771 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1772 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1773 * config/i386/i386.c (ix86_builtins): Add
1774 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1775 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1776 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1777 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1778 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1779 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1780 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1781 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1782 IX86_BUILTIN_PMOVUSQW512_MEM.
1783 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1784 __builtin_ia32_pmovsqd512mem_mask,
1785 __builtin_ia32_pmovqd512mem_mask,
1786 __builtin_ia32_pmovusqw512mem_mask,
1787 __builtin_ia32_pmovsqw512mem_mask,
1788 __builtin_ia32_pmovqw512mem_mask,
1789 __builtin_ia32_pmovusdw512mem_mask,
1790 __builtin_ia32_pmovsdw512mem_mask,
1791 __builtin_ia32_pmovdw512mem_mask,
1792 __builtin_ia32_pmovqb512mem_mask,
1793 __builtin_ia32_pmovusqb512mem_mask,
1794 __builtin_ia32_pmovsqb512mem_mask,
1795 __builtin_ia32_pmovusdb512mem_mask,
1796 __builtin_ia32_pmovsdb512mem_mask,
1797 __builtin_ia32_pmovdb512mem_mask.
1798 (bdesc_args): Add __builtin_ia32_expanddf512,
1799 __builtin_ia32_expandsf512.
1800 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1801 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1802 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1803 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1804 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1805 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1806 (avx512f_<code>v8div16qi2_mask_store): This.
1807 (avx512f_expand<mode>): New.
1808
1809 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1810
1811 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1812 New.
1813 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1814 (_mm512_prefetch_i32scatter_pd): Ditto.
1815 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1816 (_mm512_prefetch_i64scatter_pd): Ditto.
1817 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1818 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1819 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1820 (_mm512_prefetch_i32scatter_ps): Ditto.
1821 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1822 (_mm512_prefetch_i64scatter_ps): Ditto.
1823 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1824 * config/i386/i386-builtin-types.def: Define
1825 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1826 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1827 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1828 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1829 IX86_BUILTIN_SCATTERPFQPD.
1830 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1831 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1832 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1833 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1834 __builtin_ia32_scatterpfqps.
1835 (ix86_expand_builtin): Expand new built-ins.
1836 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1837 fix memory access data type.
1838 (*avx512pf_gatherpf<mode>_mask): Ditto.
1839 (*avx512pf_gatherpf<mode>): Ditto.
1840 (avx512pf_scatterpf<mode>): Ditto.
1841 (*avx512pf_scatterpf<mode>_mask): Ditto.
1842 (*avx512pf_scatterpf<mode>): Ditto.
1843 (GATHER_SCATTER_SF_MEM_MODE): New.
1844 (avx512pf_gatherpf<mode>df): Ditto.
1845 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1846 (*avx512pf_scatterpf<mode>df): Ditto.
1847
1848 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1849
1850 PR bootstrap/59934
1851 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1852 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1853 reached.
1854
1855 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1856
1857 * common/config/arm/arm-common.c
1858 (arm_rewrite_mcpu): Handle multiple names.
1859 * config/arm/arm.h
1860 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1861
1862 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1863
1864 * gimple-builder.h (create_gimple_tmp): Delete.
1865
1866 2014-01-27 Christian Bruel <christian.bruel@st.com>
1867
1868 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1869 words comparisons.
1870
1871 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1872
1873 * config/pa/pa.md (call): Generate indirect long calls to non-local
1874 functions when outputing 32-bit code.
1875 (call_value): Likewise except for special call to buggy powf function.
1876
1877 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1878 portable runtime and PIC indirect calls.
1879 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1880 and PIC call sequences. Use ldo instead of blr to set return register
1881 in PIC call sequence.
1882
1883 2014-01-25 Walter Lee <walt@tilera.com>
1884
1885 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1886 avoid clobbering a live register.
1887
1888 2014-01-25 Walter Lee <walt@tilera.com>
1889
1890 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1891 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1892 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1893 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1894
1895 2014-01-25 Walter Lee <walt@tilera.com>
1896
1897 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1898 arguments on even registers.
1899 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1900 STACK_BOUNDARY.
1901 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1902 (BIGGEST_ALIGNMENT): Ditto.
1903 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1904
1905 2014-01-25 Walter Lee <walt@tilera.com>
1906
1907 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1908 insns before bundling.
1909 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1910
1911 2014-01-25 Walter Lee <walt@tilera.com>
1912
1913 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1914 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1915 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1916
1917 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1918
1919 * config/mips/constraints.md (kl): Delete.
1920 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1921 define expands, using...
1922 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1923 instructions for MIPS16.
1924 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1925 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1926
1927 2014-01-25 Walter Lee <walt@tilera.com>
1928
1929 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1930 (clzdi2): Ditto.
1931 (ffsdi2): Ditto.
1932
1933 2014-01-25 Walter Lee <walt@tilera.com>
1934
1935 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1936 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1937
1938 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1939
1940 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1941 Handle XOR.
1942
1943 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1944
1945 * print-rtl.c (in_call_function_usage): New var.
1946 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1947 EXPR_LIST mode as mode and not as reg note name.
1948
1949 PR middle-end/59561
1950 * cfgloopmanip.c (copy_loop_info): If
1951 loop->warned_aggressive_loop_optimizations, make sure
1952 the flag is set in target loop too.
1953
1954 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1955
1956 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1957 flag_cilkplus.
1958 * builtins.def: Likewise.
1959 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1960 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1961 * ira.c (ira_setup_eliminable_regset): Likewise.
1962 * omp-low.c (gate_expand_omp): Likewise.
1963 (execute_lower_omp): Likewise.
1964 (diagnose_sb_0): Likewise.
1965 (gate_diagnose_omp_blocks): Likewise.
1966 (simd_clone_clauses_extract): Likewise.
1967 (gate): Likewise.
1968
1969 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1970
1971 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1972 correction for little endian...
1973 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1974 here.
1975
1976 2014-01-24 Jeff Law <law@redhat.com>
1977
1978 PR tree-optimization/59919
1979 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1980 for non-returning calls.
1981
1982 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1983
1984 * common/config/aarch64/aarch64-common.c
1985 (aarch64_rewrite_mcpu): Handle multiple names.
1986 * config/aarch64/aarch64.h
1987 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1988
1989 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1990
1991 * input.c (add_file_to_cache_tab): Handle the case where fopen
1992 returns NULL.
1993
1994 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1995
1996 PR target/59929
1997 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1998 from push operand if code of push isn't PRE_DEC.
1999
2000 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2001
2002 PR target/59909
2003 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2004 -mquad-memory-atomic. Update -mquad-memory documentation to say
2005 it is only used for non-atomic loads/stores.
2006
2007 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2008 -mquad-memory or -mquad-memory-atomic switches.
2009
2010 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2011 -mquad-memory-atomic to ISA 2.07 support.
2012
2013 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2014 to separate support of normal quad word memory operations (ldq, stq)
2015 from the atomic quad word memory operations.
2016
2017 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2018 support to separate non-atomic quad word operations from atomic
2019 quad word operations. Disable non-atomic quad word operations in
2020 little endian mode so that we don't have to swap words after the
2021 load and before the store.
2022 (quad_load_store_p): Add comment about atomic quad word support.
2023 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2024 options printed with -mdebug=reg.
2025
2026 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2027 -mquad-memory-atomic as the test for whether we have quad word
2028 atomic instructions.
2029 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2030 or -mp8-vector are used, allow byte/half-word atomic operations.
2031
2032 * config/rs6000/sync.md (load_lockedti): Insure that the address
2033 is a proper indexed or indirect address for the lqarx instruction.
2034 On little endian systems, swap the hi/lo registers after the lqarx
2035 instruction.
2036 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2037 insure the address is valid for the lqarx instruction.
2038 (store_conditionalti): Insure that the address is a proper indexed
2039 or indirect address for the stqcrx. instruction. On little endian
2040 systems, swap the hi/lo registers before doing the stqcrx.
2041 instruction.
2042 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2043 insure the address is valid for the stqcrx. instruction.
2044
2045 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2046 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2047 type of quad memory support is available.
2048
2049 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2050
2051 PR regression/59915
2052 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2053 there is a danger of looping.
2054
2055 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2056
2057 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2058 force flag_ira_loop_pressure if set via command line.
2059
2060 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2061
2062 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2063 (ashr_simd): New builtin handling DI mode.
2064 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2065 (aarch64_sshr_simddi): New match pattern.
2066 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2067 (vshrd_n_s64): Likewise.
2068 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2069
2070 2014-01-23 Nick Clifton <nickc@redhat.com>
2071
2072 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2073 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2074 favour of mcu specific scripts.
2075 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2076 430x multilibs.
2077
2078 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2079 Alex Velenko <Alex.Velenko@arm.com>
2080
2081 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2082 (vaddv_s16): Likewise.
2083 (vaddv_s32): Likewise.
2084 (vaddv_u8): Likewise.
2085 (vaddv_u16): Likewise.
2086 (vaddv_u32): Likewise.
2087 (vaddvq_s8): Likewise.
2088 (vaddvq_s16): Likewise.
2089 (vaddvq_s32): Likewise.
2090 (vaddvq_s64): Likewise.
2091 (vaddvq_u8): Likewise.
2092 (vaddvq_u16): Likewise.
2093 (vaddvq_u32): Likewise.
2094 (vaddvq_u64): Likewise.
2095 (vaddv_f32): Likewise.
2096 (vaddvq_f32): Likewise.
2097 (vaddvq_f64): Likewise.
2098 (vmaxv_f32): Likewise.
2099 (vmaxv_s8): Likewise.
2100 (vmaxv_s16): Likewise.
2101 (vmaxv_s32): Likewise.
2102 (vmaxv_u8): Likewise.
2103 (vmaxv_u16): Likewise.
2104 (vmaxv_u32): Likewise.
2105 (vmaxvq_f32): Likewise.
2106 (vmaxvq_f64): Likewise.
2107 (vmaxvq_s8): Likewise.
2108 (vmaxvq_s16): Likewise.
2109 (vmaxvq_s32): Likewise.
2110 (vmaxvq_u8): Likewise.
2111 (vmaxvq_u16): Likewise.
2112 (vmaxvq_u32): Likewise.
2113 (vmaxnmv_f32): Likewise.
2114 (vmaxnmvq_f32): Likewise.
2115 (vmaxnmvq_f64): Likewise.
2116 (vminv_f32): Likewise.
2117 (vminv_s8): Likewise.
2118 (vminv_s16): Likewise.
2119 (vminv_s32): Likewise.
2120 (vminv_u8): Likewise.
2121 (vminv_u16): Likewise.
2122 (vminv_u32): Likewise.
2123 (vminvq_f32): Likewise.
2124 (vminvq_f64): Likewise.
2125 (vminvq_s8): Likewise.
2126 (vminvq_s16): Likewise.
2127 (vminvq_s32): Likewise.
2128 (vminvq_u8): Likewise.
2129 (vminvq_u16): Likewise.
2130 (vminvq_u32): Likewise.
2131 (vminnmv_f32): Likewise.
2132 (vminnmvq_f32): Likewise.
2133 (vminnmvq_f64): Likewise.
2134
2135 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2136
2137 * config/aarch64/aarch64-simd.md
2138 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2139 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2140 (*aarch64_mul3_elt<mode>): Likewise.
2141 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2142 (*aarch64_mul3_elt_to_64v2df): Likewise.
2143 (*aarch64_mla_elt<mode>): Likewise.
2144 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2145 (*aarch64_mls_elt<mode>): Likewise.
2146 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2147 (*aarch64_fma4_elt<mode>): Likewise.
2148 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2149 (*aarch64_fma4_elt_to_64v2df): Likewise.
2150 (*aarch64_fnma4_elt<mode>): Likewise.
2151 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2152 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2153 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2154 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2155 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2156 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2157 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2158 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2159 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2160
2161 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2162
2163 * config/aarch64/aarch64-simd.md
2164 (aarch64_be_checked_get_lane<mode>): New define_expand.
2165 * config/aarch64/aarch64-simd-builtins.def
2166 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2167 New builtin definition.
2168 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2169 Use new safe be builtin.
2170
2171 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2172
2173 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2174 New define_insn.
2175 (aarch64_be_st1<mode>): Likewise.
2176 (aarch_ld1<VALL:mode>): Define_expand modified.
2177 (aarch_st1<VALL:mode>): Likewise.
2178 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2179 (UNSPEC_ST1): Likewise.
2180
2181 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2182
2183 * config/microblaze/microblaze.md: Add trap insn and attribute
2184
2185 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2186
2187 PR preprocessor/58580
2188 * input.h (location_get_source_line): Take an additional line_size
2189 parameter.
2190 (void diagnostics_file_cache_fini): Declare new function.
2191 * input.c (struct fcache): New type.
2192 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2193 New static constants.
2194 (diagnostic_file_cache_init, total_lines_num)
2195 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2196 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2197 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2198 (get_next_line, read_next_line, goto_next_line, read_line_num):
2199 New static function definitions.
2200 (diagnostic_file_cache_fini): New function.
2201 (location_get_source_line): Take an additional output line_len
2202 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2203 read_line_num.
2204 * diagnostic.c (diagnostic_finish): Call
2205 diagnostic_file_cache_fini.
2206 (adjust_line): Take an additional input parameter for the length
2207 of the line, rather than calculating it with strlen.
2208 (diagnostic_show_locus): Adjust the use of
2209 location_get_source_line and adjust_line with respect to their new
2210 signature. While displaying a line now, do not stop at the first
2211 null byte. Rather, display the zero byte as a space and keep
2212 going until we reach the size of the line.
2213 * Makefile.in: Add vec.o to OBJS-libcommon
2214
2215 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2216 Ilya Tocar <ilya.tocar@intel.com>
2217
2218 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2219 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2220 (__builtin_ia32_kmov16): Ditto.
2221 * config/i386/i386.md (UNSPEC_KMOV): New.
2222 (kmovw): Ditto.
2223
2224 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2225
2226 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2227 (_mm512_storeu_si512): Ditto.
2228
2229 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2230
2231 PR target/52125
2232 * rtl.h (get_referenced_operands): Declare.
2233 * recog.c (get_referenced_operands): New function.
2234 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2235 operands have been referenced when recording LO_SUM references.
2236
2237 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2238
2239 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2240
2241 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2242
2243 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2244 Enable for generic and recent AMD targets.
2245
2246 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2247
2248 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2249 ARG_SIZE note when adjustment was eliminated.
2250
2251 2014-01-22 Jeff Law <law@redhat.com>
2252
2253 PR tree-optimization/59597
2254 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2255 in file. Accept new argument REGISTERING and use it to modify
2256 dump output appropriately.
2257 (register_jump_thread): Corresponding changes.
2258 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2259 thread paths involving joiner blocks. Add code to dump cancelled
2260 jump threading paths.
2261
2262 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2263
2264 PR rtl-optimization/59477
2265 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2266 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2267
2268 2014-01-22 Tom Tromey <tromey@redhat.com>
2269
2270 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2271 PARAMS.
2272 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2273
2274 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2275
2276 PR rtl-optimization/59896
2277 * lra-constraints.c (process_alt_operands): Check unused note for
2278 matched operands of insn with no output reloads.
2279
2280 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2281
2282 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2283 (mips_move_from_gpr_cost): Likewise.
2284
2285 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2286
2287 PR rtl-optimization/59858
2288 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2289 ira_class_hard_regs_num.
2290 (process_alt_operands): Increase reject for dying matched operand.
2291
2292 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2293
2294 PR target/59003
2295 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2296 smaller than size, perform several stores or loads and stores
2297 at dst + count - size to store or copy all of size bytes, rather
2298 than just last modesize bytes.
2299
2300 2014-01-20 DJ Delorie <dj@redhat.com>
2301
2302 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2303 that CLOBBERs are REGs before propogating their values.
2304
2305 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2306
2307 PR middle-end/59789
2308 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2309 (cgraph_inline_failed_type): New function.
2310 * cgraph.h (DEFCIFCODE): Add type.
2311 (cgraph_inline_failed_type_t): New enum.
2312 (cgraph_inline_failed_type): New prototype.
2313 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2314 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2315 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2316 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2317 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2318 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2319 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2320 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2321 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2322 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2323 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2324 OPTIMIZATION_MISMATCH.
2325 * tree-inline.c (expand_call_inline): Emit errors during
2326 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2327
2328 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2329
2330 PR target/59685
2331 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2332 mode attribute in insn output.
2333
2334 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2335
2336 * output.h (output_constant): Delete.
2337 * varasm.c (output_constant): Make private.
2338
2339 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2340
2341 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2342
2343 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2344
2345 PR middle-end/59860
2346 * tree.h (fold_builtin_strcat): New prototype.
2347 * builtins.c (fold_builtin_strcat): No longer static. Add len
2348 argument, if non-NULL, don't call c_strlen. Optimize
2349 directly into __builtin_memcpy instead of __builtin_strcpy.
2350 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2351 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2352
2353 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2354
2355 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2356 for SImode_address_operand operands, having only a REG argument.
2357
2358 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2359
2360 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2361 loader name using mbig-endian.
2362 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2363
2364 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2365
2366 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2367 (-mtune): Likewise.
2368 (-mcpu): Likewise.
2369
2370 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2371
2372 * config/aarch64/aarch64-protos.h
2373 (aarch64_cannot_change_mode_class_ptr): Declare.
2374 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2375 aarch64_cannot_change_mode_class_ptr): New.
2376 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2377 backend hook aarch64_cannot_change_mode_class.
2378
2379 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2380
2381 * common/config/aarch64/aarch64-common.c
2382 (aarch64_handle_option): Don't handle any option order logic here.
2383 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2384 selected_cpu, warn on architecture version mismatch.
2385 (aarch64_override_options): Fix parsing order for option strings.
2386
2387 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2388 Iain Sandoe <iain@codesourcery.com>
2389
2390 PR bootstrap/59496
2391 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2392 warning. Amend comment to reflect current functionality.
2393
2394 2014-01-20 Richard Biener <rguenther@suse.de>
2395
2396 PR middle-end/59860
2397 * builtins.c (fold_builtin_strcat): Remove case better handled
2398 by tree-ssa-strlen.c.
2399
2400 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2401
2402 * config/aarch64/aarch64.opt
2403 (mcpu, march, mtune): Make case-insensitive.
2404
2405 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2406
2407 PR target/59880
2408 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2409 if operands[1] is a REG or ZERO_EXTEND of a REG.
2410
2411 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2412
2413 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2414
2415 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2416
2417 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2418 long non-pic millicode calls.
2419
2420 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2421
2422 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2423
2424 2014-01-19 Kito Cheng <kito@0xlab.org>
2425
2426 * builtins.c (expand_movstr): Check movstr expand done or fail.
2427
2428 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2429 H.J. Lu <hongjiu.lu@intel.com>
2430
2431 PR target/59379
2432 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2433 to DImode for zero-extended addresses.
2434
2435 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2436
2437 PR rtl-optimization/57763
2438 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2439 on the new indirect jump_insn and increment LABEL_NUSES (label).
2440
2441 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2442
2443 PR bootstrap/59580
2444 PR bootstrap/59583
2445 * config.gcc (x86_archs): New variable.
2446 (x86_64_archs): Likewise.
2447 (x86_cpus): Likewise.
2448 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2449 --with-arch/--with-cpu= options.
2450 Support --with-arch=/--with-cpu={nehalem,westmere,
2451 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2452
2453 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2454
2455 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2456 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2457
2458 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2459
2460 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2461
2462 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2463
2464 PR target/58944
2465 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2466 clear cpp_get_options (parse_in)->warn_unused_macros for
2467 ix86_target_macros_internal with cpp_define.
2468
2469 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2470
2471 * jump.c (delete_related_insns): Keep (use (insn))s.
2472 * reorg.c (redundant_insn): Check for barriers too.
2473
2474 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2475
2476 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2477
2478 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2479
2480 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2481 call to $$dyncall when TARGET_LONG_CALLS is true.
2482
2483 2014-01-17 Jeff Law <law@redhat.com>
2484
2485 * ree.c (combine_set_extension): Temporarily disable test for
2486 changing number of hard registers.
2487
2488 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2489
2490 PR middle-end/58125
2491 * ipa-inline-analysis.c (inline_free_summary):
2492 Do not free summary of aliases.
2493
2494 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2495
2496 PR middle-end/59706
2497 * gimplify.c (gimplify_expr): Use create_tmp_var
2498 instead of create_tmp_var_raw. If cond doesn't have
2499 integral type, don't add the IFN_ANNOTATE builtin at all.
2500
2501 2014-01-17 Martin Jambor <mjambor@suse.cz>
2502
2503 PR ipa/59736
2504 * ipa-cp.c (prev_edge_clone): New variable.
2505 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2506 Also resize prev_edge_clone vector.
2507 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2508 (ipcp_edge_removal_hook): New function.
2509 (ipcp_driver): Register ipcp_edge_removal_hook.
2510
2511 2014-01-17 Andrew Pinski <apinski@cavium.com>
2512 Steve Ellcey <sellcey@mips.com>
2513
2514 PR target/59462
2515 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2516 of operator mode.
2517
2518 2014-01-17 Jeff Law <law@redhat.com>
2519
2520 PR middle-end/57904
2521 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2522 so that pass_ccp runs first.
2523
2524 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2525
2526 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2527 (ix86_adjust_cost): Use !TARGET_XXX.
2528 (do_reorder_for_imul): Likewise.
2529 (swap_top_of_ready_list): Likewise.
2530 (ix86_sched_reorder): Likewise.
2531
2532 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2533
2534 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2535 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2536 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2537 (intel_memset): New. Duplicate slm_memset.
2538 (intel_cost): New. Duplicate slm_cost.
2539 (m_INTEL): New macro.
2540 (processor_target_table): Add "intel".
2541 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2542 with PROCESSOR_INTEL for "intel".
2543 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2544 PROCESSOR_SILVERMONT.
2545 (ix86_issue_rate): Likewise.
2546 (ix86_adjust_cost): Likewise.
2547 (ia32_multipass_dfa_lookahead): Likewise.
2548 (swap_top_of_ready_list): Likewise.
2549 (ix86_sched_reorder): Likewise.
2550 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2551 instead of TARGET_OPT_AGU.
2552 * config/i386/i386.h (TARGET_INTEL): New.
2553 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2554 (processor_type): Add PROCESSOR_INTEL.
2555 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2556 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2557
2558 2014-01-17 Marek Polacek <polacek@redhat.com>
2559
2560 PR c/58346
2561 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2562 size is zero.
2563
2564 2014-01-17 Richard Biener <rguenther@suse.de>
2565
2566 PR tree-optimization/46590
2567 * opts.c (default_options_table): Add entries for
2568 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2569 all enabled at -O1 but not for -Og.
2570 * common.opt (fbranch-count-reg): Remove Init(1).
2571 (fmove-loop-invariants): Likewise.
2572 (ftree-pta): Likewise.
2573
2574 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2575
2576 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2577 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2578 decls to at least the GCC 4.8 used alignments.
2579
2580 PR fortran/59440
2581 * tree-nested.c (convert_nonlocal_reference_stmt,
2582 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2583 of GIMPLE_BIND stmts, adjust associated decls.
2584
2585 2014-01-17 Richard Biener <rguenther@suse.de>
2586
2587 PR tree-optimization/46590
2588 * vec.h (vec<>::bseach): New member function implementing
2589 binary search according to C89 bsearch.
2590 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2591 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2592 bitmap pointer again. Make accesses_in_loop a flat array.
2593 (mem_ref_obstack): New global.
2594 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2595 (mark_ref_stored): Likewise.
2596 (ref_indep_loop_p_2): Likewise.
2597 (set_ref_stored_in_loop): New helper function.
2598 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2599 (memref_free): Adjust.
2600 (record_mem_ref_loc): Simplify.
2601 (gather_mem_refs_stmt): Adjust.
2602 (sort_locs_in_loop_postorder_cmp): New function.
2603 (analyze_memory_references): Sort accesses_in_loop after
2604 loop postorder number.
2605 (find_ref_loc_in_loop_cmp): New function.
2606 (for_all_locs_in_loop): Find relevant cluster of locs in
2607 accesses_in_loop and iterate without recursion.
2608 (execute_sm): Avoid uninit warning.
2609 (struct ref_always_accessed): Simplify.
2610 (ref_always_accessed::operator ()): Likewise.
2611 (ref_always_accessed_p): Likewise.
2612 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2613 loop postorder numbers here.
2614 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2615 numbers.
2616
2617 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2618
2619 PR c++/57945
2620 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2621 on decls for which assemble_alias has been called.
2622
2623 2014-01-17 Nick Clifton <nickc@redhat.com>
2624
2625 * config/msp430/msp430.opt: (mcpu): New option.
2626 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2627 (msp430_option_override): Parse target_cpu. If the MCU name
2628 matches a generic string, clear target_mcu.
2629 (msp430_attr): Allow numeric interrupt values up to 63.
2630 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2631 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2632 option.
2633 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2634 Add mcpu matches.
2635 * config/msp430/msp430.md (popm): Use %J rather than %I.
2636 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2637 (addhi_cy_i): Use immediate_operand for operand 2.
2638 * doc/invoke.texi: Document -mcpu option.
2639
2640 2014-01-17 Richard Biener <rguenther@suse.de>
2641
2642 PR rtl-optimization/38518
2643 * df.h (df_analyze_loop): Declare.
2644 * df-core.c: Include cfgloop.h.
2645 (df_analyze_1): Split out main part of df_analyze.
2646 (df_analyze): Adjust.
2647 (loop_inverted_post_order_compute): New function.
2648 (loop_post_order_compute): Likewise.
2649 (df_analyze_loop): New function avoiding whole-function
2650 postorder computes.
2651 * loop-invariant.c (find_defs): Use df_analyze_loop.
2652 (find_invariants): Adjust.
2653 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2654
2655 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2656
2657 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2658 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2659
2660 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2661
2662 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2663 traversal when removing references.
2664
2665 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2666
2667 PR ipa/59775
2668 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2669
2670 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2671
2672 PR middle-end/56791
2673 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2674 pushing a reload for an autoinc when we had previously reloaded an
2675 inner part of the address.
2676
2677 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2678
2679 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2680 field.
2681 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2682 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2683 when not giving up or versioning for alias only because of
2684 loop->safelen.
2685 (vect_analyze_data_ref_dependences): Set to true.
2686 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2687 is a GIMPLE_PHI.
2688 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2689 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2690 to the condition.
2691
2692 PR middle-end/58344
2693 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2694
2695 PR target/59839
2696 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2697 operand 0 predicate for gathers, use a new pseudo as subtarget.
2698
2699 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2700
2701 PR middle-end/59609
2702 * lra-constraints.c (process_alt_operands): Add printing debug info.
2703 Check absence of input/output reloads for matched operands too.
2704
2705 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2706
2707 PR rtl-optimization/59835
2708 * ira.c (ira_init_register_move_cost): Increase cost for
2709 impossible modes.
2710
2711 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2712
2713 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2714
2715 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2716
2717 PR target/59780
2718 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2719 non-register objects. Use gen_(high/low)part more consistently.
2720 Fix assertions.
2721
2722 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2723
2724 PR target/59844
2725 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2726 endian support, remove tests for WORDS_BIG_ENDIAN.
2727 (p8_mfvsrd_3_<mode>): Likewise.
2728 (reload_gpr_from_vsx<mode>): Likewise.
2729 (reload_gpr_from_vsxsf): Likewise.
2730 (p8_mfvsrd_4_disf): Likewise.
2731
2732 2014-01-16 Richard Biener <rguenther@suse.de>
2733
2734 PR rtl-optimization/46590
2735 * lcm.c (compute_antinout_edge): Use postorder iteration.
2736 (compute_laterin): Use inverted postorder iteration.
2737
2738 2014-01-16 Nick Clifton <nickc@redhat.com>
2739
2740 PR middle-end/28865
2741 * varasm.c (output_constant): Return the number of bytes actually
2742 emitted.
2743 (output_constructor_array_range): Update the field size with the
2744 number of bytes emitted by output_constant.
2745 (output_constructor_regular_field): Likewise. Also do not
2746 complain if the total number of bytes emitted is now greater
2747 than the expected fieldpos.
2748 * output.h (output_constant): Update prototype and descriptive comment.
2749
2750 2014-01-16 Marek Polacek <polacek@redhat.com>
2751
2752 PR middle-end/59827
2753 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2754 it is error_mark_node.
2755
2756 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2757
2758 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2759 VALID_AVX256_REG_OR_OI_MODE.
2760
2761 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2762
2763 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2764 current procedure should be profiled.
2765
2766 2014-01-15 Andrew Pinski <apinski@cavium.com>
2767
2768 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2769 of moving from/to the STACK_REG register class.
2770
2771 2014-01-15 Richard Henderson <rth@redhat.com>
2772
2773 PR debug/54694
2774 * reginfo.c (global_regs_decl): Globalize.
2775 * rtl.h (global_regs_decl): Declare.
2776 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2777 reserved via global_regs.
2778
2779 2014-01-15 Teresa Johnson <tejohnson@google.com>
2780
2781 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2782
2783 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2784
2785 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2786 and vmulosh rather than call gen_vec_widen_smult_*.
2787 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2788 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2789 (vec_widen_smult_even_v16qi): Likewise.
2790 (vec_widen_umult_even_v8hi): Likewise.
2791 (vec_widen_smult_even_v8hi): Likewise.
2792 (vec_widen_umult_odd_v16qi): Likewise.
2793 (vec_widen_smult_odd_v16qi): Likewise.
2794 (vec_widen_umult_odd_v8hi): Likewise.
2795 (vec_widen_smult_odd_v8hi): Likewise.
2796 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2797 vmuloub rather than call gen_vec_widen_umult_*.
2798 (vec_widen_umult_lo_v16qi): Likewise.
2799 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2800 vmulosb rather than call gen_vec_widen_smult_*.
2801 (vec_widen_smult_lo_v16qi): Likewise.
2802 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2803 rather than call gen_vec_widen_umult_*.
2804 (vec_widen_umult_lo_v8hi): Likewise.
2805 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2806 rather than call gen_vec_widen_smult_*.
2807 (vec_widen_smult_lo_v8hi): Likewise.
2808
2809 2014-01-15 Jeff Law <law@redhat.com>
2810
2811 PR tree-optimization/59747
2812 * ree.c (find_and_remove_re): Properly handle case where a second
2813 eliminated extension requires widening a copy created for elimination
2814 of a prior extension.
2815 (combine_set_extension): Ensure that the number of hard regs needed
2816 for a destination register does not change when we widen it.
2817
2818 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2819
2820 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2821 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2822 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2823 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2824 (avr-*-rtems*): Likewise.
2825 (bfin*-rtems*): Likewise.
2826 (moxie-*-rtems*): Likewise.
2827 (h8300-*-rtems*): Likewise.
2828 (i[34567]86-*-rtems*): Likewise.
2829 (lm32-*-rtems*): Likewise.
2830 (m32r-*-rtems*): Likewise.
2831 (m68k-*-rtems*): Likewise.
2832 (microblaze*-*-rtems*): Likewise.
2833 (mips*-*-rtems*): Likewise.
2834 (powerpc-*-rtems*): Likewise.
2835 (sh-*-rtems*): Likewise.
2836 (sparc-*-rtems*): Likewise.
2837 (sparc64-*-rtems*): Likewise.
2838 (v850-*-rtems*): Likewise.
2839 (m32c-*-rtems*): Likewise.
2840
2841 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2842
2843 PR rtl-optimization/59511
2844 * ira.c (ira_init_register_move_cost): Use memory costs for some
2845 cases of register move cost calculations.
2846 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2847 instead of BB frequency.
2848 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2849 * lra-assigns.c (find_hard_regno_for): Ditto.
2850
2851 2014-01-15 Richard Biener <rguenther@suse.de>
2852
2853 PR tree-optimization/59822
2854 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2855 (vectorizable_load): Use it to hoist defs of uses of invariant
2856 loads out of the loop.
2857
2858 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2859 Kugan Vivekanandarajah <kuganv@linaro.org>
2860
2861 PR target/59695
2862 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2863 truncation.
2864
2865 2014-01-15 Richard Biener <rguenther@suse.de>
2866
2867 PR rtl-optimization/59802
2868 * lcm.c (compute_available): Use inverted postorder to seed
2869 the initial worklist.
2870
2871 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2872
2873 PR target/59803
2874 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2875 ADDR_REGS for invalid symrefs in non-PIC code.
2876
2877 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2878
2879 PR other/58712
2880 * builtins.c (determine_block_size): Initialize *probable_max_size
2881 even if len_rtx is CONST_INT.
2882
2883 2014-01-14 Andrew Pinski <apinski@cavium.com>
2884
2885 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2886 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2887 (cortexa53_tunings): Likewise.
2888 (aarch64_sched_issue_rate): New function.
2889 (TARGET_SCHED_ISSUE_RATE): Define.
2890
2891 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2892
2893 * ira-costs.c (find_costs_and_classes): Add missed
2894 ira_init_register_move_cost_if_necessary.
2895
2896 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2897
2898 PR target/59787
2899 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2900
2901 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2902
2903 PR target/59794
2904 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2905 to indicate if type is used for function return value. Warn ABI
2906 change if the vector mode isn't available for function return value.
2907 (ix86_function_arg_advance): Pass false to type_natural_mode.
2908 (ix86_function_arg): Likewise.
2909 (ix86_gimplify_va_arg): Likewise.
2910 (function_arg_32): Don't warn ABI change.
2911 (ix86_function_value): Pass true to type_natural_mode.
2912 (ix86_return_in_memory): Likewise.
2913 (ix86_struct_value_rtx): Removed.
2914 (TARGET_STRUCT_VALUE_RTX): Likewise.
2915
2916 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2917
2918 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2919 converting a conditional jump into a conditional return.
2920
2921 2014-01-14 Richard Biener <rguenther@suse.de>
2922
2923 PR tree-optimization/58921
2924 PR tree-optimization/59006
2925 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2926 hoisting invariant stmts.
2927 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2928 invariant loads on the preheader edge if possible.
2929
2930 2014-01-14 Joey Ye <joey.ye@arm.com>
2931
2932 * doc/plugin.texi (Building GCC plugins): Update to C++.
2933
2934 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2935
2936 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2937 (_mm_rcp28_round_ss): Ditto.
2938 (_mm_rsqrt28_round_sd): Ditto.
2939 (_mm_rsqrt28_round_ss): Ditto.
2940 (_mm_rcp28_sd): Ditto.
2941 (_mm_rcp28_ss): Ditto.
2942 (_mm_rsqrt28_sd): Ditto.
2943 (_mm_rsqrt28_ss): Ditto.
2944 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2945 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2946 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2947 (IX86_BUILTIN_RCP28SD): Ditto.
2948 (IX86_BUILTIN_RCP28SS): Ditto.
2949 (IX86_BUILTIN_RSQRT28SD): Ditto.
2950 (IX86_BUILTIN_RSQRT28SS): Ditto.
2951 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2952 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2953 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2954 (ix86_expand_special_args_builtin): Expand new FTYPE.
2955 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2956 (srcp14<mode>): Make insn unary.
2957 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2958 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2959 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2960 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2961 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2962 Fix rounding: make it SAE only.
2963 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2964 Ditto.
2965 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2966 Ditto.
2967 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2968 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2969 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2970 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2971 (round_saeonly_mask_scalar_operand4): Ditto.
2972 (round_saeonly_mask_scalar_op3): Ditto.
2973 (round_saeonly_mask_scalar_op4): Ditto.
2974
2975 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2976
2977 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2978 Implement -maltivec=be for vec_insert and vec_extract.
2979
2980 2014-01-10 DJ Delorie <dj@redhat.com>
2981
2982 * config/msp430/msp430.md (call_internal): Don't allow memory
2983 references with SP as the base register.
2984 (call_value_internal): Likewise.
2985 * config/msp430/constraints.md (Yc): New. For memory references
2986 that don't use SP as a base register.
2987
2988 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2989 "an integer without a # prefix"
2990 * config/msp430/msp430.md (epilogue_helper): Use it.
2991
2992 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2993
2994 PR target/59617
2995 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2996 AVX512F gather builtins.
2997 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2998 on gather decls with INTEGER_TYPE masktype.
2999 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3000 directly into the builtin rather than hoisting it before loop.
3001
3002 PR tree-optimization/59387
3003 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3004 (scev_const_prop): If folded_casts and type has undefined overflow,
3005 use force_gimple_operand instead of force_gimple_operand_gsi and
3006 for each added stmt if it is assign with
3007 arith_code_with_undefined_signed_overflow, call
3008 rewrite_to_defined_overflow.
3009 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3010 gimple-fold.h instead.
3011 (arith_code_with_undefined_signed_overflow,
3012 rewrite_to_defined_overflow): Moved to ...
3013 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3014 rewrite_to_defined_overflow): ... here. No longer static.
3015 Include gimplify-me.h.
3016 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3017 rewrite_to_defined_overflow): New prototypes.
3018
3019 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3020
3021 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3022
3023 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3024
3025 * builtins.c (get_object_alignment_2): Minor tweak.
3026 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3027
3028 2014-01-13 Christian Bruel <christian.bruel@st.com>
3029
3030 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3031 optimized non constant lengths.
3032
3033 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3034
3035 PR libgomp/59194
3036 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3037 load as __atomic_load_N if possible.
3038
3039 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3040
3041 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3042 target parameter.
3043 (rs6000_expand_builtin): Adjust call.
3044
3045 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3046
3047 PR target/58115
3048 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3049 * config/rs6000/rs6000.c: Include target-globals.h.
3050 (rs6000_set_current_function): Instead of doing target_reinit
3051 unconditionally, use save_target_globals_default_opts and
3052 restore_target_globals.
3053
3054 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3055 FPSCR.
3056 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3057 (rs6000_expand_builtin): Handle mffs and mtfsf.
3058 (rs6000_init_builtins): Define mffs and mtfsf.
3059 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3060 (rs6000_mffs): New pattern.
3061 (rs6000_mtfsf): New pattern.
3062
3063 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3064
3065 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3066 Start narrowing with START. Apply candidate-use pair
3067 and check overall cost in narrowing.
3068 (iv_ca_prune): Pass new argument.
3069
3070 2014-01-10 Jeff Law <law@redhat.com>
3071
3072 PR middle-end/59743
3073 * ree.c (combine_reaching_defs): Ensure the defining statement
3074 occurs before the extension when optimizing extensions with
3075 different source and destination hard registers.
3076
3077 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3078
3079 PR ipa/58585
3080 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3081 vtables into the type inheritance graph.
3082
3083 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3084
3085 PR rtl-optimization/59754
3086 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3087 modes in the REGNO != REGNO case.
3088
3089 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3090
3091 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3092
3093 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3094
3095 PR tree-optimization/59745
3096 * tree-predcom.c (tree_predictive_commoning_loop): Call
3097 free_affine_expand_cache if giving up because components is NULL.
3098
3099 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3100 GC in payload of target_globals struct instead of allocating them on
3101 the heap and the larger structs separately using GC.
3102 * target-globals.h (struct target_globals): Make regs, hard_regs,
3103 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3104 of GTY((skip)) and change type to void *.
3105 (reset_target_globals): Cast loads from those fields to corresponding
3106 types.
3107
3108 2014-01-10 Steve Ellcey <sellcey@mips.com>
3109
3110 PR plugins/59335
3111 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3112 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3113 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3114
3115 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3116
3117 PR target/59744
3118 * aarch64-modes.def (CC_Zmode): New flags mode.
3119 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3120 represents an equality.
3121 (aarch64_get_condition_code): Handle CC_Zmode.
3122 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3123
3124 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3125
3126 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3127 extraction in good case.
3128
3129 2014-01-10 Richard Biener <rguenther@suse.de>
3130
3131 PR tree-optimization/59374
3132 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3133 checking after SLP discovery. Mark stmts not participating
3134 in any SLP instance properly.
3135
3136 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3137
3138 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3139 when handling a SET rtx.
3140
3141 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3142
3143 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3144 (cortex-a57): Likewise.
3145 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3146
3147 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3148
3149 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3150 non-iwmmxt builtins.
3151
3152 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3153
3154 PR ipa/58252
3155 PR ipa/59226
3156 * ipa-devirt.c record_target_from_binfo): Take as argument
3157 stack of binfos and lookup matching one for virtual inheritance.
3158 (possible_polymorphic_call_targets_1): Update.
3159
3160 2014-01-10 Huacai Chen <chenhc@lemote.com>
3161
3162 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3163 kernel strings for Loongson-2E/2F/3A.
3164
3165 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3166
3167 PR middle-end/59670
3168 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3169 is_gimple_call before calling gimple_call_internal_p.
3170
3171 2014-01-09 Steve Ellcey <sellcey@mips.com>
3172
3173 * Makefile.in (TREE_FLOW_H): Remove.
3174 (TREE_SSA_H): Add file names from tree-flow.h.
3175 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3176 * tree.h: Remove tree-flow.h reference.
3177 * hash-table.h: Remove tree-flow.h reference.
3178 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3179 reference with tree-ssa-loop.h.
3180
3181 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3182
3183 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3184 default element-order behavior for -maltivec.
3185 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3186 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3187 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3188 when targeting big endian, at least for now.
3189 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3190
3191 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3192
3193 PR middle-end/47735
3194 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3195 var satisfies use_register_for_decl, just take into account type
3196 alignment, rather than decl alignment.
3197
3198 PR tree-optimization/59622
3199 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3200 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3201 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3202 Don't devirtualize for inplace at all. For targets.length () == 1,
3203 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3204
3205 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3206
3207 * config/i386/i386.md (cpu): Remove the unused btver1.
3208
3209 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3210
3211 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3212
3213 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3214
3215 PR target/58115
3216 * tree-core.h (struct target_globals): New forward declaration.
3217 (struct tree_target_option): Add globals field.
3218 * tree.h (TREE_TARGET_GLOBALS): Define.
3219 (prepare_target_option_nodes_for_pch): New prototype.
3220 * target-globals.h (struct target_globals): Define even if
3221 !SWITCHABLE_TARGET.
3222 * tree.c (prepare_target_option_node_for_pch,
3223 prepare_target_option_nodes_for_pch): New functions.
3224 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3225 * config/i386/i386.c: Include target-globals.h.
3226 (ix86_set_current_function): Instead of doing target_reinit
3227 unconditionally, use save_target_globals_default_opts and
3228 restore_target_globals.
3229
3230 2014-01-09 Richard Biener <rguenther@suse.de>
3231
3232 PR tree-optimization/59715
3233 * tree-cfg.h (split_critical_edges): Declare.
3234 * tree-cfg.c (split_critical_edges): Export.
3235 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3236
3237 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3238
3239 * cfgexpand.c (expand_stack_vars): Optionally disable
3240 asan stack protection.
3241 (expand_used_vars): Likewise.
3242 (partition_stack_vars): Likewise.
3243 * asan.c (asan_emit_stack_protection): Optionally disable
3244 after return stack usage.
3245 (instrument_derefs): Optionally disable memory access instrumentation.
3246 (instrument_builtin_call): Likewise.
3247 (instrument_strlen_call): Likewise.
3248 (asan_protect_global): Optionally disable global variables protection.
3249 * doc/invoke.texi: Added doc for new options.
3250 * params.def: Added new options.
3251 * params.h: Likewise.
3252
3253 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3254
3255 PR rtl-optimization/59724
3256 * ifcvt.c (cond_exec_process_if_block): Don't call
3257 flow_find_head_matching_sequence with 0 longest_match.
3258 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3259 non-active insns if !stop_after.
3260 (try_head_merge_bb): Revert 2014-01-07 changes.
3261
3262 2014-01-08 Jeff Law <law@redhat.com>
3263
3264 * ree.c (get_sub_rtx): New function, extracted from...
3265 (merge_def_and_ext): Here.
3266 (combine_reaching_defs): Use get_sub_rtx.
3267
3268 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3269
3270 * cgraph.h (varpool_variable_node): Do not choke on null node.
3271
3272 2014-01-08 Catherine Moore <clm@codesourcery.com>
3273
3274 * config/mips/mips.md (simple_return): Attempt to use JRC
3275 for microMIPS.
3276 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3277
3278 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3279
3280 PR rtl-optimization/59137
3281 * reorg.c (steal_delay_list_from_target): Call update_block for
3282 elided insns.
3283 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3284
3285 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3286
3287 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3288 two duplicate entries.
3289
3290 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3291
3292 Revert:
3293 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3294
3295 * config/mips/mips.c (mips_truncated_op_cost): New function.
3296 (mips_rtx_costs): Adjust test for BADDU.
3297 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3298
3299 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3300
3301 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3302 (*baddu_si): ...this new pattern.
3303
3304 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3305
3306 PR ipa/59722
3307 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3308
3309 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3310
3311 PR middle-end/57748
3312 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3313 inner_reference_p.
3314 (expand_expr, expand_normal): Adjust.
3315 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3316 inner_reference_p. Use inner_reference_p to expand inner references.
3317 (store_expr): Adjust.
3318 * cfgexpand.c (expand_call_stmt): Adjust.
3319
3320 2014-01-08 Rong Xu <xur@google.com>
3321
3322 * gcov-io.c (gcov_var): Move from gcov-io.h.
3323 (gcov_position): Ditto.
3324 (gcov_is_error): Ditto.
3325 (gcov_rewrite): Ditto.
3326 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3327 only part to libgcc/libgcov.h.
3328
3329 2014-01-08 Marek Polacek <polacek@redhat.com>
3330
3331 PR middle-end/59669
3332 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3333
3334 2014-01-08 Marek Polacek <polacek@redhat.com>
3335
3336 PR sanitizer/59667
3337 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3338
3339 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3340
3341 PR rtl-optimization/59649
3342 * stor-layout.c (get_mode_bounds): For BImode return
3343 0 and STORE_FLAG_VALUE.
3344
3345 2014-01-08 Richard Biener <rguenther@suse.de>
3346
3347 PR middle-end/59630
3348 * gimple.h (is_gimple_builtin_call): Remove.
3349 (gimple_builtin_call_types_compatible_p): New.
3350 (gimple_call_builtin_p): New overload.
3351 * gimple.c (is_gimple_builtin_call): Remove.
3352 (validate_call): Rename to ...
3353 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3354 check return types.
3355 (validate_type): New static function.
3356 (gimple_call_builtin_p): New overload and adjust.
3357 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3358 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3359 (gimple_fold_stmt_to_constant_1): Likewise.
3360 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3361
3362 2014-01-08 Richard Biener <rguenther@suse.de>
3363
3364 PR middle-end/59471
3365 * gimplify.c (gimplify_expr): Gimplify register-register type
3366 VIEW_CONVERT_EXPRs to separate stmts.
3367
3368 2014-01-07 Jeff Law <law@redhat.com>
3369
3370 PR middle-end/53623
3371 * ree.c (combine_set_extension): Handle case where source
3372 and destination registers in an extension insn are different.
3373 (combine_reaching_defs): Allow source and destination registers
3374 in extension to be different under limited circumstances.
3375 (add_removable_extension): Remove restriction that the
3376 source and destination registers in the extension are the same.
3377 (find_and_remove_re): Emit a copy from the extension's
3378 destination to its source after the defining insn if
3379 the source and destination registers are different.
3380
3381 PR middle-end/59285
3382 * ifcvt.c (merge_if_block): If we are merging a block with more than
3383 one successor with a block with no successors, remove any BARRIER
3384 after the second block.
3385
3386 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3387
3388 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3389
3390 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3391
3392 PR target/59652
3393 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3394 for 14-bit register offsets when INT14_OK_STRICT is false.
3395
3396 2014-01-07 Roland Stigge <stigge@antcom.de>
3397 Michael Meissner <meissner@linux.vnet.ibm.com>
3398
3399 PR 57386/target
3400 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3401 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3402
3403 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3404
3405 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3406 -mcpu.
3407
3408 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3409
3410 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3411 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3412 rtx is const0_rtx or not.
3413
3414 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3415
3416 PR target/58115
3417 * target-globals.c (save_target_globals): Remove this_fn_optab
3418 handling.
3419 * toplev.c: Include optabs.h.
3420 (target_reinit): Temporarily restore the global options if another
3421 set of options are in force.
3422
3423 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3424
3425 PR rtl-optimization/58668
3426 * cfgcleanup.c (flow_find_cross_jump): Don't count
3427 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3428 to determine what is counted.
3429 (flow_find_head_matching_sequence): Use active_insn_p to determine
3430 what is counted.
3431 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3432 counting change.
3433 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3434 determine what is counted.
3435
3436 PR tree-optimization/59643
3437 * tree-predcom.c (split_data_refs_to_components): If one dr is
3438 read and one write, determine_offset fails and the write isn't
3439 in the bad component, just put the read into the bad component.
3440
3441 2014-01-07 Mike Stump <mikestump@comcast.net>
3442 Jakub Jelinek <jakub@redhat.com>
3443
3444 PR pch/59436
3445 * tree-core.h (struct tree_optimization_option): Change optabs
3446 type from unsigned char * to void *.
3447 * optabs.c (init_tree_optimization_optabs): Adjust
3448 TREE_OPTIMIZATION_OPTABS initialization.
3449
3450 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3451
3452 PR target/59644
3453 * config/i386/i386.h (struct machine_function): Add
3454 no_drap_save_restore field.
3455 * config/i386/i386.c (ix86_save_reg): Use
3456 !cfun->machine->no_drap_save_restore instead of
3457 crtl->stack_realign_needed.
3458 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3459 this function clears frame_pointer_needed. Set
3460 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3461 and DRAP reg is needed.
3462
3463 2014-01-06 Marek Polacek <polacek@redhat.com>
3464
3465 PR c/57773
3466 * doc/implement-c.texi: Mention that other integer types are
3467 permitted as bit-field types in strictly conforming mode.
3468
3469 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3470
3471 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3472 is newly allocated.
3473
3474 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3475
3476 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3477
3478 2014-01-06 Martin Jambor <mjambor@suse.cz>
3479
3480 PR ipa/59008
3481 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3482 to int.
3483 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3484
3485 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3486
3487 PR debug/59350
3488 PR debug/59510
3489 * var-tracking.c (add_stores): Preserve the value of the source even if
3490 we don't record the store.
3491
3492 2014-01-06 Terry Guo <terry.guo@arm.com>
3493
3494 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3495
3496 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3497
3498 PR bootstrap/59541
3499 * config/darwin.c (darwin_function_section): Adjust return values to
3500 correspond to optimisation changes made in r206070.
3501
3502 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3503
3504 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3505 from prefetch_block tune setting.
3506 (nocona_cost): Correct size of prefetch block to 64.
3507
3508 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3509
3510 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3511 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3512 used to save the static chain register in the computation of the offset
3513 from which the FP registers need to be restored.
3514
3515 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3516
3517 PR tree-optimization/59519
3518 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3519 ICE if get_current_def (current_new_name) is already non-NULL, as long
3520 as it is a phi result of some other phi in *new_exit_bb that has
3521 the same argument.
3522
3523 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3524 or vmovdqu* for misaligned_operand.
3525 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3526 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3527 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3528 aligned_mem for AVX512F masked aligned load and store builtins and for
3529 non-temporal moves.
3530
3531 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3532
3533 PR tree-optimization/59651
3534 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3535 Address range for negative step should be added by TYPE_SIZE_UNIT.
3536
3537 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3538
3539 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3540 overlapping registers also for registers other than the stack pointer.
3541
3542 2014-01-03 Marek Polacek <polacek@redhat.com>
3543
3544 PR other/59661
3545 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3546 __builtin_FILE.
3547
3548 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3549
3550 PR target/59625
3551 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3552 asm goto as jump.
3553
3554 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3555 (push splitter): Use <P:MODE_SIZE> instead of
3556 GET_MODE_SIZE (<P:MODE>mode).
3557 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3558 (mov -1, reg peephole2): Likewise.
3559 * config/i386/sse.md (*mov<mode>_internal,
3560 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3561 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3562 *<code><mode>3, *andnot<mode>3<mask_name>,
3563 <mask_codefor><code><mode>3<mask_name>): Likewise.
3564 * config/i386/subst.md (mask_mode512bit_condition,
3565 sd_mask_mode512bit_condition): Likewise.
3566
3567 2014-01-02 Xinliang David Li <davidxl@google.com>
3568
3569 PR tree-optimization/59303
3570 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3571 (dump_predicates): Better output format.
3572 (pred_equal_p): New function.
3573 (is_neq_relop_p): Ditto.
3574 (is_neq_zero_form_p): Ditto.
3575 (pred_expr_equal_p): Ditto.
3576 (pred_neg_p): Ditto.
3577 (simplify_pred): Ditto.
3578 (simplify_preds_2): Ditto.
3579 (simplify_preds_3): Ditto.
3580 (simplify_preds_4): Ditto.
3581 (simplify_preds): Ditto.
3582 (push_pred): Ditto.
3583 (push_to_worklist): Ditto.
3584 (get_pred_info_from_cmp): Ditto.
3585 (is_degenerated_phi): Ditto.
3586 (normalize_one_pred_1): Ditto.
3587 (normalize_one_pred): Ditto.
3588 (normalize_one_pred_chain): Ditto.
3589 (normalize_preds): Ditto.
3590 (normalize_cond_1): Remove function.
3591 (normalize_cond): Ditto.
3592 (is_gcond_subset_of): Ditto.
3593 (is_subset_of_any): Ditto.
3594 (is_or_set_subset_of): Ditto.
3595 (is_and_set_subset_of): Ditto.
3596 (is_norm_cond_subset_of): Ditto.
3597 (pred_chain_length_cmp): Ditto.
3598 (convert_control_dep_chain_into_preds): Type change.
3599 (find_predicates): Ditto.
3600 (find_def_preds): Ditto.
3601 (destroy_predicates_vecs): Ditto.
3602 (find_matching_predicates_in_rest_chains): Ditto.
3603 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3604 (is_pred_expr_subset): Ditto.
3605 (is_pred_chain_subset_of): Ditto.
3606 (is_included_in): Ditto.
3607 (is_superset_of): Ditto.
3608
3609 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3610
3611 Update copyright years.
3612
3613 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3614
3615 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3616 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3617 config/arc/arc.md, config/arc/arc.opt,
3618 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3619 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3620 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3621 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3622 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3623 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3624 vtable-verify.c, vtable-verify.h: Use the standard form for the
3625 copyright notice.
3626
3627 2014-01-02 Tobias Burnus <burnus@net-b.de>
3628
3629 * gcc.c (process_command): Update copyright notice dates.
3630 * gcov-dump.c: Ditto.
3631 * gcov.c: Ditto.
3632 * doc/cpp.texi: Bump @copying's copyright year.
3633 * doc/cppinternals.texi: Ditto.
3634 * doc/gcc.texi: Ditto.
3635 * doc/gccint.texi: Ditto.
3636 * doc/gcov.texi: Ditto.
3637 * doc/install.texi: Ditto.
3638 * doc/invoke.texi: Ditto.
3639
3640 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3641
3642 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3643
3644 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3645
3646 * config/i386/sse.md (*mov<mode>_internal): Guard
3647 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3648
3649 PR rtl-optimization/59647
3650 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3651 new_rtx into UNSIGNED_FLOAT rtxes.
3652 \f
3653 Copyright (C) 2014 Free Software Foundation, Inc.
3654
3655 Copying and distribution of this file, with or without modification,
3656 are permitted in any medium without royalty provided the copyright
3657 notice and this notice are preserved.