]> gcc.gnu.org Git - gcc.git/blob - gcc/ChangeLog
re PR target/58675 (ICE in rs6000_secondary_reload_inner:15460, type = store)
[gcc.git] / gcc / ChangeLog
1 2014-02-15 Alan Modra <amodra@gmail.com>
2
3 PR target/58675
4 PR target/57935
5 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
6 find_replacement on parts of insn rtl that might be reloaded.
7
8 2014-02-15 Richard Biener <rguenther@suse.de>
9
10 PR tree-optimization/60183
11 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating
12 loads.
13 (tree_ssa_phiprop): Calculate and free post-dominators.
14
15 2014-02-14 Jeff Law <law@redhat.com>
16
17 PR rtl-optimization/60131
18 * ree.c (get_extended_src_reg): New function.
19 (combine_reaching_defs): Use it rather than assuming location
20 of REG.
21 (find_and_remove_re): Verify first operand of extension is
22 a REG before adding the insns to the copy list.
23
24 2014-02-14 Roland McGrath <mcgrathr@google.com>
25
26 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
27 * configure: Regenerated.
28 * config.in: Regenerated.
29 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
30 instead of ASM_SHORT.
31
32 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
33 Richard Earnshaw <rearnsha@arm.com>
34
35 PR rtl-optimization/59535
36 * lra-constraints.c (process_alt_operands): Encourage alternative
37 when unassigned pseudo class is superset of the alternative class.
38 (inherit_reload_reg): Don't inherit when optimizing for code size.
39 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
40 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
41 modes not less than 4 for Thumb1.
42
43 2014-02-14 Kyle McMartin <kyle@redhat.com>
44
45 PR pch/60010
46 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
47
48 2014-02-14 Richard Biener <rguenther@suse.de>
49
50 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
51 (get_frame_arg): Drop the assert with langhook types_compatible_p.
52 Do not strip INDIRECT_REFs.
53
54 2014-02-14 Richard Biener <rguenther@suse.de>
55
56 PR lto/60179
57 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
58 DECL_FUNCTION_SPECIFIC_TARGET.
59 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
60 * tree-streamer-out.c (pack_ts_target_option): Remove.
61 (streamer_pack_tree_bitfields): Do not stream
62 TS_TARGET_OPTION.
63 (write_ts_function_decl_tree_pointers): Do not stream
64 DECL_FUNCTION_SPECIFIC_TARGET.
65 * tree-streamer-in.c (unpack_ts_target_option): Remove.
66 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
67 (lto_input_ts_function_decl_tree_pointers): Do not stream
68 DECL_FUNCTION_SPECIFIC_TARGET.
69
70 2014-02-14 Jakub Jelinek <jakub@redhat.com>
71
72 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
73 use_stmt twice.
74 (get_initial_def_for_induction, vectorizable_induction): Ignore
75 debug stmts when looking for exit_phi.
76 (vectorizable_live_operation): Fix up condition.
77
78 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
79
80 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
81 nreverse() because it changes the content of original tree list.
82
83 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
84
85 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
86 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
87
88 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
89
90 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
91 GNU coding standards.
92
93 2014-02-13 Jakub Jelinek <jakub@redhat.com>
94
95 PR debug/60152
96 * dwarf2out.c (gen_subprogram_die): Don't call
97 add_calling_convention_attribute if subr_die is old_die.
98
99 2014-02-13 Sharad Singhai <singhai@google.com>
100
101 * doc/optinfo.texi: Fix order of nodes.
102
103 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
104
105 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
106 operands[2], not operands[3].
107
108 2014-02-13 Richard Biener <rguenther@suse.de>
109
110 PR bootstrap/59878
111 * doc/install.texi (ISL): Update recommended version to 0.12.2,
112 mention the possibility of an in-tree build.
113 (CLooG): Update recommended version to 0.18.1, mention the
114 possibility of an in-tree build and clarify that the ISL
115 bundled with CLooG does not work.
116
117 2014-02-13 Jakub Jelinek <jakub@redhat.com>
118
119 PR target/43546
120 * expr.c (compress_float_constant): If x is a hard register,
121 extend into a pseudo and then move to x.
122
123 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
124
125 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
126 caused by bad second argument to warning_at() with -mhotpatch and
127 nested functions (e.g. with gfortran).
128
129 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
130
131 * opts.c (option_name): Remove "enabled by default" rider.
132
133 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
134
135 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
136
137 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
138 Uros Bizjak <ubizjak@gmail.com>
139
140 PR target/60151
141 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
142 * configure: Regenerated.
143
144 2014-02-12 Richard Biener <rguenther@suse.de>
145
146 * vec.c (vec_prefix::calculate_allocation): Move as
147 inline variant to vec.h.
148 (vec_prefix::calculate_allocation_1): New out-of-line version.
149 * vec.h (vec_prefix::calculate_allocation_1): Declare.
150 (vec_prefix::m_has_auto_buf): Rename to ...
151 (vec_prefix::m_using_auto_storage): ... this.
152 (vec_prefix::calculate_allocation): Inline the easy cases
153 and dispatch to calculate_allocation_1 which doesn't need the
154 prefix address.
155 (va_heap::reserve): Use gcc_checking_assert.
156 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
157 m_using_auto_storage.
158 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
159 member and adjust.
160 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
161 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
162 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
163
164 2014-02-12 Richard Biener <rguenther@suse.de>
165
166 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
167 when we found a dependence.
168
169 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
170
171 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
172 common code...
173 (maybe_fold_stmt): ... into this new function.
174 * omp-low.c (lower_omp): Update comment.
175
176 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
177 last use.
178
179 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
180 dereference.
181
182 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
183
184 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
185 identifiers in comments.
186 (cortexa53_extra_costs): Likewise.
187 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
188 (cortexa7_extra_costs): Likewise.
189 (cortexa12_extra_costs): Likewise.
190 (cortexa15_extra_costs): Likewise.
191 (v7m_extra_costs): Likewise.
192
193 2014-02-12 Richard Biener <rguenther@suse.de>
194
195 PR middle-end/60092
196 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
197 of posix_memalign being successful.
198 (lower_stmt): Restrict lowering of posix_memalign to when
199 -ftree-bit-ccp is enabled.
200
201 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
202
203 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
204 arg_loc.
205 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
206
207 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
208
209 PR rtl-optimization/60116
210 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
211 other_insn once the combination has been validated.
212
213 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
214
215 PR lto/59468
216 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
217 and wrapper.
218 * ipa-devirt.c: Include demangle.h
219 (odr_violation_reported): New static variable.
220 (add_type_duplicate): Update odr_violations.
221 (maybe_record_node): Add completep parameter; update it.
222 (record_target_from_binfo): Add COMPLETEP parameter;
223 update it as needed.
224 (possible_polymorphic_call_targets_1): Likewise.
225 (struct polymorphic_call_target_d): Add nonconstruction_targets;
226 rename FINAL to COMPLETE.
227 (record_targets_from_bases): Sanity check we found the binfo;
228 fix COMPLETEP updating.
229 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
230 parameter, fix computing of COMPLETEP.
231 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
232 at LTO time do demangling.
233 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
234 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
235 parameter.
236 (gimple_get_virt_method_for_binfo): Likewise.
237 * gimple-fold.h (gimple_get_virt_method_for_binfo,
238 gimple_get_virt_method_for_vtable): Update prototypes.
239
240 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
241
242 PR target/49008
243 * genautomata.c (add_presence_absence): Fix typo with
244 {final_}presence_list.
245
246 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
247
248 PR target/60137
249 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
250 for VSX/Altivec vectors that land in GPR registers.
251
252 2014-02-11 Richard Henderson <rth@redhat.com>
253 Jakub Jelinek <jakub@redhat.com>
254
255 PR debug/59776
256 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
257 around drhs if type conversion to lacc->type is not useless.
258
259 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
260
261 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
262 tuning struct.
263 (cortex-a57.cortex-a53): Likewise.
264 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
265
266 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
267
268 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
269 arm_restrict_it.
270
271 2014-02-11 Renlin Li <Renlin.Li@arm.com>
272
273 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
274 add_options_for_arm_vfp3.
275
276 2014-02-11 Jeff Law <law@redhat.com>
277
278 PR middle-end/54041
279 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
280 object with an undesirable mode.
281
282 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
283
284 PR libgomp/60107
285 * config/i386/sol2-9.h: New file.
286 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
287 *-*-solaris2.9*): Use it.
288
289 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
290
291 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
292 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
293
294 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
295
296 * config/microblaze/microblaze.c: Extend mcpu version format
297
298 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
299
300 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
301
302 2014-02-10 Richard Henderson <rth@redhat.com>
303
304 PR target/59927
305 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
306 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
307 ms-abi vs -mno-accumulate-outgoing-args.
308 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
309 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
310 respect to ms-abi.
311
312 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
313
314 PR middle-end/60080
315 * cfgexpand.c (expand_asm_operands): Attach source location to
316 ASM_INPUT rtx objects.
317 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
318
319 2014-02-10 Nick Clifton <nickc@redhat.com>
320
321 * config/mn10300/mn10300.c (popcount): New function.
322 (mn10300_expand_prologue): Include saved registers in stack usage
323 count.
324
325 2014-02-10 Jeff Law <law@redhat.com>
326
327 PR middle-end/52306
328 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
329 when changing the SET_DEST of a prior insn to avoid an input reload.
330
331 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
332
333 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
334 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
335 -mcall-openbsd, or -mcall-linux.
336 (CC1_ENDIAN_BIG_SPEC): Remove.
337 (CC1_ENDIAN_LITTLE_SPEC): Remove.
338 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
339 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
340 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
341 and %cc1_endian_default.
342 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
343
344 2014-02-10 Richard Biener <rguenther@suse.de>
345
346 PR tree-optimization/60115
347 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
348 MEM_REF handling. Properly verify that the accesses are not
349 out of the objects bound.
350
351 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
352
353 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
354 coretex to cortex.
355
356 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
357
358 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
359 proper constants and fix formatting.
360 (possible_polymorphic_call_targets): Fix formatting.
361
362 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
363 Ilya Tocar <ilya.tocar@intel.com>
364
365 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
366 (_mm512_loadu_epi32): Renamed into...
367 (_mm512_loadu_si512): This.
368 (_mm512_storeu_epi32): Renamed into...
369 (_mm512_storeu_si512): This.
370 (_mm512_maskz_ceil_ps): Removed.
371 (_mm512_maskz_ceil_pd): Ditto.
372 (_mm512_maskz_floor_ps): Ditto.
373 (_mm512_maskz_floor_pd): Ditto.
374 (_mm512_floor_round_ps): Ditto.
375 (_mm512_floor_round_pd): Ditto.
376 (_mm512_ceil_round_ps): Ditto.
377 (_mm512_ceil_round_pd): Ditto.
378 (_mm512_mask_floor_round_ps): Ditto.
379 (_mm512_mask_floor_round_pd): Ditto.
380 (_mm512_mask_ceil_round_ps): Ditto.
381 (_mm512_mask_ceil_round_pd): Ditto.
382 (_mm512_maskz_floor_round_ps): Ditto.
383 (_mm512_maskz_floor_round_pd): Ditto.
384 (_mm512_maskz_ceil_round_ps): Ditto.
385 (_mm512_maskz_ceil_round_pd): Ditto.
386 (_mm512_expand_pd): Ditto.
387 (_mm512_expand_ps): Ditto.
388 * config/i386/i386.c (ix86_builtins): Remove
389 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
390 (bdesc_args): Ditto.
391 * config/i386/predicates.md (const1256_operand): New.
392 (const_1_to_2_operand): Ditto.
393 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
394 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
395 (*avx512pf_gatherpf<mode>sf): Ditto.
396 (avx512pf_gatherpf<mode>df): Ditto.
397 (*avx512pf_gatherpf<mode>df_mask): Ditto.
398 (*avx512pf_gatherpf<mode>df): Ditto.
399 (avx512pf_scatterpf<mode>sf): Ditto.
400 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
401 (*avx512pf_scatterpf<mode>sf): Ditto.
402 (avx512pf_scatterpf<mode>df): Ditto.
403 (*avx512pf_scatterpf<mode>df_mask): Ditto.
404 (*avx512pf_scatterpf<mode>df): Ditto.
405 (avx512f_expand<mode>): Removed.
406 (<shift_insn><mode>3<mask_name>): Change predicate type.
407
408 2014-02-08 Jakub Jelinek <jakub@redhat.com>
409
410 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
411 not at the end of datarefs vector use ordered_remove to avoid
412 reordering datarefs vector.
413
414 PR c/59984
415 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
416 mark local addressable non-static vars as GOVD_PRIVATE
417 instead of GOVD_LOCAL.
418 * omp-low.c (lower_omp_for): Move gimple_bind_vars
419 and BLOCK_VARS of gimple_bind_block to new_stmt rather
420 than copying them.
421
422 PR middle-end/60092
423 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
424 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
425 assume_aligned or alloc_align attributes.
426 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
427 arguments. Handle also assume_aligned and alloc_align attributes.
428 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
429 calls to functions with assume_aligned or alloc_align attributes.
430 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
431
432 2014-02-08 Terry Guo <terry.guo@arm.com>
433
434 * doc/invoke.texi: Document ARM -march=armv7e-m.
435
436 2014-02-08 Jakub Jelinek <jakub@redhat.com>
437
438 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
439 flag on __cilkrts_rethrow builtin.
440
441 PR ipa/60026
442 * ipa-cp.c (determine_versionability): Fail at -O0
443 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
444 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
445
446 Revert:
447 2014-02-04 Jakub Jelinek <jakub@redhat.com>
448
449 PR ipa/60026
450 * tree-inline.c (copy_forbidden): Fail for
451 __attribute__((optimize (0))) functions.
452
453 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
454
455 * varpool.c: Include pointer-set.h.
456 (varpool_remove_unreferenced_decls): Variables in other partitions
457 will not be output; be however careful to not lose information
458 about partitioning.
459
460 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
461
462 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
463 lookup in the vtable constructor.
464
465 2014-02-07 Jeff Law <law@redhat.com>
466
467 PR target/40977
468 * config/m68k/m68k.md (ashldi_extsi): Turn into a
469 define_insn_and_split.
470
471 * ipa-inline.c (inline_small_functions): Fix typos.
472
473 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
474
475 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
476 (s390_can_use_return_insn): Declare.
477 * config/s390/s390.h (EPILOGUE_USES): Define.
478 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
479 instructions.
480 (s390_chunkify_start): Handle return JUMP_LABELs.
481 (s390_early_mach): Emit a main_pool instruction on the entry edge.
482 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
483 (s390_can_use_return_insn): New functions.
484 (s390_fix_long_loop_prediction): Handle conditional returns.
485 (TARGET_SET_UP_BY_PROLOGUE): Define.
486 * config/s390/s390.md (ANY_RETURN): New code iterator.
487 (*creturn, *csimple_return, return, simple_return): New patterns.
488
489 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
490
491 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
492 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
493 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
494 REG_CFA_RESTORE list when deciding not to restore a register.
495
496 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
497
498 * config/s390/s390.c: Include tree-pass.h and context.h.
499 (s390_early_mach): New function, split out from...
500 (s390_emit_prologue): ...here.
501 (pass_data_s390_early_mach): New pass structure.
502 (pass_s390_early_mach): New class.
503 (s390_option_override): Create and register early_mach pass.
504 Move to end of file.
505
506 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
507
508 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
509 to match for the exit block.
510
511 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
512
513 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
514 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
515 Reject misaligned operands.
516
517 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
518
519 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
520
521 2014-02-07 Richard Biener <rguenther@suse.de>
522
523 PR middle-end/60092
524 * gimple-low.c (lower_builtin_posix_memalign): New function.
525 (lower_stmt): Call it to lower posix_memalign in a way
526 to make alignment info accessible.
527
528 2014-02-07 Jakub Jelinek <jakub@redhat.com>
529
530 PR c++/60082
531 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
532 __builtin_setjmp_receiver.
533
534 2014-02-07 Richard Biener <rguenther@suse.de>
535
536 PR middle-end/60092
537 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
538 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
539 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
540 Handle BUILT_IN_POSIX_MEMALIGN.
541 (find_func_clobbers): Likewise.
542 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
543 (call_may_clobber_ref_p_1): Likewise.
544
545 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
546
547 PR ipa/59918
548 * ipa-devirt.c (record_target_from_binfo): Remove overactive
549 sanity check.
550
551 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
552
553 PR ipa/59469
554 * lto-cgraph.c (lto_output_node): Use
555 symtab_get_symbol_partitioning_class.
556 (lto_output_varpool_node): likewise.
557 (symtab_get_symbol_partitioning_class): Move here from
558 lto/lto-partition.c
559 * cgraph.h (symbol_partitioning_class): Likewise.
560 (symtab_get_symbol_partitioning_class): Declare.
561
562 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
563
564 * ggc.h (ggc_internal_cleared_alloc): New macro.
565 * vec.h (vec_safe_copy): Handle memory stats.
566 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
567 * target-globals.c (save_target_globals): Likewise.
568
569 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
570
571 PR target/60077
572 * expr.c (emit_move_resolve_push): Export; be bit more selective
573 on when to clear alias set.
574 * expr.h (emit_move_resolve_push): Declare.
575 * function.h (struct function): Add tail_call_marked.
576 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
577 * config/i386/i386-protos.h (ix86_expand_push): Remove.
578 * config/i386/i386.md (TImode move expander): De not call
579 ix86_expand_push.
580 (FP push expanders): Preserve memory attributes.
581 * config/i386/sse.md (push<mode>1): Remove.
582 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
583 (ix86_expand_push): Remove.
584 * config/i386/mmx.md (push<mode>1): Remove.
585
586 2014-02-06 Jakub Jelinek <jakub@redhat.com>
587
588 PR rtl-optimization/60030
589 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
590 lopart with paradoxical subreg before shifting it up by hprec.
591
592 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
593
594 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
595 Remove extra newline at end of file.
596 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
597 (arm_issue_rate): Handle cortexa57.
598 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
599 (cortex-a57.cortex-a53): Likewise.
600
601 2014-02-06 Jakub Jelinek <jakub@redhat.com>
602
603 PR target/59575
604 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
605 don't record in REG_FRAME_RELATED_EXPR registers not set in that
606 bitmask.
607 (arm_expand_prologue): Adjust all callers.
608 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
609 info, registers also at the lowest numbered registers side. Use
610 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
611 XEXP.
612
613 PR debug/59992
614 * var-tracking.c (adjust_mems): Before adding a SET to
615 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
616
617 2014-02-06 Alan Modra <amodra@gmail.com>
618
619 PR target/60032
620 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
621 change SDmode to DDmode when lra_in_progress.
622
623 2014-02-06 Jakub Jelinek <jakub@redhat.com>
624
625 PR middle-end/59150
626 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
627 free_data_ref on the dr first, and before goto again also set dr
628 to the next dr. For simd_lane_access, free old datarefs[i] before
629 overwriting it. For get_vectype_for_scalar_type failure, don't
630 free_data_ref if simd_lane_access.
631
632 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
633
634 PR target/60062
635 * tree.h (opts_for_fn): New inline function.
636 (opt_for_fn): Define.
637 * config/i386/i386.c (ix86_function_regparm): Use
638 opt_for_fn (decl, optimize) instead of optimize.
639
640 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
641
642 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
643 for SYMBOL_REF in large memory model.
644
645 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
646
647 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
648 and crypto support.
649 (cortex-a57): Likewise.
650 (cortex-a57.cortex-a53): Likewise.
651
652 2014-02-06 Yury Gribov <y.gribov@samsung.com>
653 Kugan Vivekanandarajah <kuganv@linaro.org>
654
655 * config/arm/arm.c (arm_vector_alignment_reachable): Check
656 unaligned_access.
657 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
658
659 2014-02-06 Richard Biener <rguenther@suse.de>
660
661 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
662 set_loop_copy and initialize_original_copy_tables.
663
664 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
665
666 * config/aarch64/aarch64-simd.md
667 (aarch64_ashr_simddi): Change QI to SI.
668
669 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
670 Jakub Jelinek <jakub@redhat.com>
671
672 PR middle-end/60013
673 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
674 of the dataflow.
675
676 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
677
678 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
679 CODE_FOR_altivec_vpku[hw]um to
680 CODE_FOR_altivec_vpku[hw]um_direct.
681 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
682 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
683 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
684 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
685
686 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
687
688 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
689 generation for -maltivec=be.
690 (altivec_vsumsws): Simplify redundant test.
691
692 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
693
694 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
695 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
696 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
697 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
698 gen_altivec_vpkuwum.
699 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
700 BYTES_BIG_ENDIAN.
701 (altivec_vpks<VI_char>ss): Likewise.
702 (altivec_vpks<VI_char>us): Likewise.
703 (altivec_vpku<VI_char>us): Likewise.
704 (altivec_vpku<VI_char>um): Likewise.
705 (altivec_vpku<VI_char>um_direct): New (copy of
706 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
707 internal use).
708 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
709 target is little endian and -maltivec=be is not specified.
710 (*altivec_vupkhs<VU_char>_direct): New (copy of
711 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
712 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
713 target is little endian and -maltivec=be is not specified.
714 (*altivec_vupkls<VU_char>_direct): New (copy of
715 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
716 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
717 little endian and -maltivec=be is not specified.
718 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
719 little endian and -maltivec=be is not specified.
720
721 2014-02-05 Richard Henderson <rth@redhat.com>
722
723 PR debug/52727
724 * combine-stack-adj.c: Revert r206943.
725 * sched-int.h (struct deps_desc): Add last_args_size.
726 * sched-deps.c (init_deps): Initialize it.
727 (sched_analyze_insn): Add OUTPUT dependencies between insns that
728 contain REG_ARGS_SIZE notes.
729
730 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
731
732 * lto-cgraph.c (asm_nodes_output): Make global.
733 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
734 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
735 (driver_handle_option): Handle OPT_fwpa.
736
737 2014-02-05 Jakub Jelinek <jakub@redhat.com>
738
739 PR ipa/59947
740 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
741 a comment typo and formatting issue. If odr_hash hasn't been
742 created, return vNULL and set *completep to false.
743
744 PR middle-end/57499
745 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
746 bb with no successors.
747
748 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
749
750 PR target/59718
751 * doc/invoke.texi (-march): Clarify documentation for ARM.
752 (-mtune): Likewise.
753 (-mcpu): Likewise.
754
755 2014-02-05 Richard Biener <rguenther@suse.de>
756
757 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
758 when not vectorizing because of too many alias checks.
759 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
760 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
761
762 2014-02-05 Nick Clifton <nickc@redhat.com>
763
764 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
765 accept extended registers in any mode when compiling for the MN10300.
766
767 2014-02-05 Yury Gribov <y.gribov@samsung.com>
768
769 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
770 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
771 sanitization attributes.
772 (can_inline_edge_p): Likewise.
773 (sanitize_attrs_match_for_inline_p): New function.
774
775 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
776
777 * ipa-prop.c (detect_type_change): Shor circuit testing of
778 type changes on THIS pointer.
779
780 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
781
782 PR target/59777
783 * config/pa/pa.c (legitimize_tls_address): Return original address
784 if not passed a SYMBOL_REF rtx.
785 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
786 addresses.
787 (pa_emit_move_sequence): Simplify TLS source operands.
788 (pa_legitimate_constant_p): Reject all TLS constants.
789 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
790 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
791
792 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
793
794 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
795 groups when we know they are controlled by LTO.
796 * varasm.c (default_binds_local_p_1): If object is in other partition,
797 it will be resolved locally.
798
799 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
800
801 * config/host-linux.c (linux_gt_pch_use_address): Don't
802 use SSIZE_MAX because it is not always defined.
803
804 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
805
806 PR bootstrap/59913
807 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
808 threshold for pseudo splitting.
809 (update_ebb_live_info): Process call argument hard registers and
810 hard registers from insn definition too.
811 (max_small_class_regs_num): New constant.
812 (inherit_in_ebb): Update live hard regs through EBBs. Update
813 reloads_num only for small register classes. Don't split for
814 outputs of jumps.
815
816 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
817
818 PR ipa/60058
819 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
820 is non-null.
821
822 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
823
824 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
825 visibility is safe.
826
827 2014-02-04 Marek Polacek <polacek@redhat.com>
828
829 * gdbinit.in (pel): Define.
830
831 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
832
833 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
834 behavior.
835
836 2014-02-04 Richard Biener <rguenther@suse.de>
837
838 PR lto/59723
839 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
840 in function context local.
841 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
842 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
843 similar to LTO_imported_decl_ref.
844
845 2014-02-04 Jakub Jelinek <jakub@redhat.com>
846
847 PR tree-optimization/60002
848 * cgraphclones.c (build_function_decl_skip_args): Clear
849 DECL_LANG_SPECIFIC.
850
851 PR tree-optimization/60023
852 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
853 false to gsi_replace.
854 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
855 has been in some EH region and vec_stmt could throw, add
856 vec_stmt into the same EH region.
857 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
858 has no lhs, ignore it.
859 * internal-fn.c (expand_MASK_LOAD): Likewise.
860
861 PR ipa/60026
862 * tree-inline.c (copy_forbidden): Fail for
863 __attribute__((optimize (0))) functions.
864
865 PR other/58712
866 * omp-low.c (simd_clone_struct_copy): If from->inbranch
867 is set, copy one less argument.
868 (expand_simd_clones): Don't subtract clone_info->inbranch
869 from simd_clone_struct_alloc argument.
870
871 PR rtl-optimization/57915
872 * recog.c (simplify_while_replacing): If all unary/binary/relational
873 operation arguments are constant, attempt to simplify those.
874
875 PR middle-end/59261
876 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
877 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
878
879 2014-02-04 Richard Biener <rguenther@suse.de>
880
881 PR tree-optimization/60012
882 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
883 TBAA disambiguation to all DDRs.
884
885 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
886
887 PR target/59788
888 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
889 (LINK_SPEC): Use it for -shared, -shared-libgcc.
890
891 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
892
893 PR ipa/59882
894 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
895
896 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
897
898 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
899 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
900
901 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
902
903 PR ipa/59831
904 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
905 to figure out targets of polymorphic calls with known decl.
906 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
907 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
908 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
909 (get_polymorphic_call_info): ... here.
910 (get_polymorphic_call_info_from_invariant): New function.
911
912 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
913
914 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
915 lookup via vtable pointer; check for type consistency
916 and turn inconsitent facts into UNREACHABLE.
917 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
918 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
919 type inconsistent querries; return UNREACHABLE instead.
920
921 2014-02-03 Richard Henderson <rth@twiddle.net>
922
923 PR tree-opt/59924
924 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
925 already processed this node.
926 (normalize_one_pred_1): Pass along mark_set.
927 (normalize_one_pred): Create and destroy a pointer_set_t.
928 (normalize_one_pred_chain): Likewise.
929
930 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
931
932 PR gcov-profile/58602
933 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
934
935 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
936
937 PR ipa/59831
938 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
939 -fno-devirtualize; try to devirtualize by the knowledge of
940 virtual table pointer given by aggregate propagation.
941 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
942 (ipa_print_node_jump_functions): Dump also offset that
943 is relevant for polymorphic calls.
944 (determine_known_aggregate_parts): Add arg_type parameter; use it
945 instead of determining the type from pointer type.
946 (ipa_compute_jump_functions_for_edge): Update call of
947 determine_known_aggregate_parts.
948 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
949 (gimple_get_virt_method_for_binfo): ... here; simplify using
950 vtable_pointer_value_to_vtable.
951 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
952 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
953 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
954 (vtable_pointer_value_to_vtable): Break out from ...; handle also
955 POINTER_PLUS_EXPR.
956 (vtable_pointer_value_to_binfo): ... here.
957 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
958
959 2014-02-03 Teresa Johnson <tejohnson@google.com>
960
961 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
962 redef of outer loop index variable.
963
964 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
965
966 PR c++/53017
967 PR c++/59211
968 * doc/extend.texi (Function Attributes): Typo.
969
970 2014-02-03 Cong Hou <congh@google.com>
971
972 PR tree-optimization/60000
973 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
974 if the vectorized statement is a store. A store statement can only
975 appear at the end of pattern statements.
976
977 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
978
979 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
980 (ix86_option_override_internal): Default long double to 64-bit for
981 32-bit Bionic and to 128-bit for 64-bit Bionic.
982
983 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
984 TARGET_LONG_DOUBLE_128 is true.
985 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
986
987 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
988 (mlong-double-64): Negate -mlong-double-128.
989 (mlong-double-128): New option.
990
991 * config/i386/i386-c.c (ix86_target_macros): Define
992 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
993
994 * doc/invoke.texi: Document -mlong-double-128.
995
996 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
997
998 PR rtl-optimization/60024
999 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1000
1001 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1002
1003 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1004
1005 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1006
1007 PR rtl-optimization/57662
1008 * sel-sched.c (code_motion_path_driver): Do not mark already not
1009 existing blocks in the visiting bitmap.
1010
1011 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1012
1013 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1014 on the insn being emitted.
1015
1016 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1017 Will Deacon <will.deacon@arm.com>
1018
1019 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1020
1021 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1022
1023 * config/arm/arm-tables.opt: Regenerate.
1024
1025 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1026
1027 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1028 for vector types other than V16QImode.
1029 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1030 define_expand, and call altivec_expand_vec_perm_le when producing
1031 code with little endian element order.
1032 (*altivec_vperm_<mode>_internal): New insn having previous
1033 behavior of altivec_vperm_<mode>.
1034 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1035 altivec_expand_vec_perm_le when producing code with little endian
1036 element order.
1037 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1038 behavior of altivec_vperm_<mode>_uns.
1039
1040 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1041
1042 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1043 (altivec_vsumsws): Add handling for -maltivec=be with a little
1044 endian target.
1045 (altivec_vsumsws_direct): New.
1046 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1047 gen_altivec_vsumsws.
1048
1049 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1050
1051 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1052 vtable_pointer_value_to_binfo): New functions.
1053 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1054 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1055
1056 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1057
1058 * config/nios2/nios2.md (load_got_register): Initialize GOT
1059 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1060 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1061
1062 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1063
1064 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1065 preserverd by passthrough, do not propagate the type.
1066
1067 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1068
1069 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1070 (mips_atomic_assign_expand_fenv): New function.
1071 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1072
1073 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1074
1075 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1076 (__builtin_mips_set_fcsr): Likewise.
1077 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1078 MIPS_USI_FTYPE_VOID.
1079 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1080 (mips16_expand_set_fcsr): Likewise.
1081 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1082 (mips16_set_fcsr_stub): Likewise.
1083 (mips16_get_fcsr_one_only_stub): New class.
1084 (mips16_set_fcsr_one_only_stub): Likewise.
1085 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1086 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1087 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1088 (hard_float): New availability predicate.
1089 (mips_builtins): Add get_fcsr and set_fcsr.
1090 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1091 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1092 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1093 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1094 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1095 patterns.
1096
1097 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1098
1099 * config/mips/mips.c (mips_one_only_stub): New class.
1100 (mips_need_mips16_rdhwr_p): Replace with...
1101 (mips16_rdhwr_stub): ...this new variable.
1102 (mips16_stub_call_address): New function.
1103 (mips16_rdhwr_one_only_stub): New class.
1104 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1105 (mips_output_mips16_rdhwr): Delete.
1106 (mips_finish_stub): New function.
1107 (mips_code_end): Use it to handle rdhwr stubs.
1108
1109 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1110
1111 PR target/60017
1112 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1113 when calculating size of integer atomic types.
1114
1115 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1116
1117 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1118
1119 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1120
1121 PR tree-optimization/60003
1122 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1123 * profile.c (branch_prob): Use gimple_call_builtin_p
1124 to check for BUILT_IN_SETJMP_RECEIVER.
1125 * tree-inline.c (copy_bb): Call notice_special_calls.
1126
1127 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1128
1129 PR bootstrap/59985
1130 * lra-constraints.c (process_alt_operands): Update reload_sum only
1131 on the first pass.
1132
1133 2014-01-31 Richard Henderson <rth@redhat.com>
1134
1135 PR middle-end/60004
1136 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1137 until after else_eh is processed.
1138
1139 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1140
1141 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1142 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1143 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1144 in smmintrin.h, remove them.
1145 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1146 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1147 * config/i386/i386.md (ROUND_SAE): Fix value.
1148 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1149 (const48_operand): New.
1150 * config/i386/subst.md (round), (round_expand): Use
1151 const_4_or_8_to_11_operand.
1152 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1153
1154 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1155
1156 * config/i386/constraints.md (Yk): Swap meaning with k.
1157 * config/i386/i386.md (movhi_internal): Change Yk to k.
1158 (movqi_internal): Ditto.
1159 (*k<logic><mode>): Ditto.
1160 (*andhi_1): Ditto.
1161 (*andqi_1): Ditto.
1162 (kandn<mode>): Ditto.
1163 (*<code>hi_1): Ditto.
1164 (*<code>qi_1): Ditto.
1165 (kxnor<mode>): Ditto.
1166 (kortestzhi): Ditto.
1167 (kortestchi): Ditto.
1168 (kunpckhi): Ditto.
1169 (*one_cmplhi2_1): Ditto.
1170 (*one_cmplqi2_1): Ditto.
1171 * config/i386/sse.md (): Change k to Yk.
1172 (avx512f_load<mode>_mask): Ditto.
1173 (avx512f_blendm<mode>): Ditto.
1174 (avx512f_store<mode>_mask): Ditto.
1175 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1176 (avx512f_storedqu<mode>_mask): Ditto.
1177 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1178 Ditto.
1179 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1180 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1181 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1182 (avx512f_maskcmp<mode>3): Ditto.
1183 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1184 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1185 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1186 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1187 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1188 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1189 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1190 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1191 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1192 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1193 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1194 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1195 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1196 (vec_extract_lo_<mode>_maskm): Ditto.
1197 (vec_extract_hi_<mode>_maskm): Ditto.
1198 (avx512f_vternlog<mode>_mask): Ditto.
1199 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1200 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1201 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1202 (avx512f_<code>v8div16qi2_mask): Ditto.
1203 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1204 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1205 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1206 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1207 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1208 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1209 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1210 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1211 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1212 (avx512cd_maskb_vec_dupv8di): Ditto.
1213 (avx512cd_maskw_vec_dupv16si): Ditto.
1214 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1215 (avx512f_vpermi2var<mode>3_mask): Ditto.
1216 (avx512f_vpermi2var<mode>3_mask): Ditto.
1217 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1218 (*avx512f_gathersi<mode>): Ditto.
1219 (*avx512f_gathersi<mode>_2): Ditto.
1220 (*avx512f_gatherdi<mode>): Ditto.
1221 (*avx512f_gatherdi<mode>_2): Ditto.
1222 (*avx512f_scattersi<mode>): Ditto.
1223 (*avx512f_scatterdi<mode>): Ditto.
1224 (avx512f_compress<mode>_mask): Ditto.
1225 (avx512f_compressstore<mode>_mask): Ditto.
1226 (avx512f_expand<mode>_mask): Ditto.
1227 * config/i386/subst.md (mask): Change k to Yk.
1228 (mask_scalar_merge): Ditto.
1229 (sd): Ditto.
1230
1231 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1232
1233 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1234
1235 2014-01-31 Richard Biener <rguenther@suse.de>
1236
1237 PR middle-end/59990
1238 * builtins.c (fold_builtin_memory_op): Make sure to not
1239 use a floating-point mode or a boolean or enumeral type for
1240 the copy operation.
1241
1242 2014-01-30 DJ Delorie <dj@redhat.com>
1243
1244 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1245 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1246 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1247 whenever main() has an epilogue.
1248
1249 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1250
1251 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1252 unused variable "field".
1253 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1254 (vsx_mergeh_<mode>): Likewise.
1255 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1256 (altivec_vmrghh): Likewise.
1257 (altivec_vmrghw): Likewise.
1258 (altivec_vmrglb): Likewise.
1259 (altivec_vmrglh): Likewise.
1260 (altivec_vmrglw): Likewise.
1261 (altivec_vspltb): Add missing uses.
1262 (altivec_vsplth): Likewise.
1263 (altivec_vspltw): Likewise.
1264 (altivec_vspltsf): Likewise.
1265
1266 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1267
1268 PR target/59923
1269 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1270 frame related instructions.
1271
1272 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1273
1274 PR rtl-optimization/59959
1275 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1276 any reload of register whose subreg is invalid.
1277
1278 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1279
1280 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1281 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1282 Add missing return type - void.
1283
1284 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1285
1286 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1287 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1288 remove element index adjustment for endian (now handled in vsx.md
1289 and altivec.md).
1290 (altivec_expand_vec_perm_const): Use
1291 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1292 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1293 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1294 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1295 define_expand and a new define_insn *altivec_vspltb_internal;
1296 adjust for -maltivec=be on a little endian target.
1297 (altivec_vspltb_direct): New.
1298 (altivec_vsplth): Divide into a define_expand and a new
1299 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1300 little endian target.
1301 (altivec_vsplth_direct): New.
1302 (altivec_vspltw): Divide into a define_expand and a new
1303 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1304 little endian target.
1305 (altivec_vspltw_direct): New.
1306 (altivec_vspltsf): Divide into a define_expand and a new
1307 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1308 a little endian target.
1309
1310 2014-01-30 Richard Biener <rguenther@suse.de>
1311
1312 PR tree-optimization/59993
1313 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1314 can propagate form the earlier stmt and avoid the transform
1315 when the intermediate result is needed.
1316
1317 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1318
1319 * README.Portability: Fix typo.
1320
1321 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1322
1323 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1324 comparison_operator with ordered_comparison_operator.
1325
1326 2014-01-30 Nick Clifton <nickc@redhat.com>
1327
1328 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1329 Rename to mn10300_store_multiple_regs.
1330 * config/mn10300/mn10300.c: Likewise.
1331 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1332 store_multiple_regs.
1333 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1334 Call mn10300_store_multiple_regs.
1335
1336 2014-01-30 Nick Clifton <nickc@redhat.com>
1337 DJ Delorie <dj@redhat.com>
1338
1339 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1340 %fp 2 to keep registers after it properly word-aligned.
1341 (rl78_alloc_physical_registers_umul): Handle the case where both
1342 input operands are the same.
1343
1344 2014-01-30 Richard Biener <rguenther@suse.de>
1345
1346 PR tree-optimization/59903
1347 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1348 check properly.
1349
1350 2014-01-30 Jason Merrill <jason@redhat.com>
1351
1352 PR c++/59633
1353 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1354
1355 PR c++/59645
1356 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1357
1358 2014-01-30 Richard Biener <rguenther@suse.de>
1359
1360 PR tree-optimization/59951
1361 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1362
1363 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1364
1365 PR target/59784
1366 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1367 SFmode to DFmode case.
1368
1369 2014-01-29 DJ Delorie <dj@redhat.com>
1370
1371 * config/msp430/msp430.opt (-minrt): New.
1372 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1373 if -minrt given.
1374 (ENDFILE_SPEC): Likewise.
1375
1376 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1377
1378 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1379 (estimate_function_body_sizes): Use it.
1380
1381 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1382
1383 PR c++/58561
1384 * dwarf2out.c (is_cxx_auto): New.
1385 (is_base_type): Use it.
1386 (gen_type_die_with_usage): Likewise.
1387
1388 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1389
1390 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1391 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1392 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1393 -maltivec=be with LE targets.
1394 (vsx_mergeh_<mode>): Likewise.
1395 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1396 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1397 (altivec_vmrghb): Replace with define_expand and new
1398 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1399 (altivec_vmrghb_direct): New define_insn.
1400 (altivec_vmrghh): Replace with define_expand and new
1401 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1402 (altivec_vmrghh_direct): New define_insn.
1403 (altivec_vmrghw): Replace with define_expand and new
1404 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1405 (altivec_vmrghw_direct): New define_insn.
1406 (*altivec_vmrghsf): Adjust for endianness.
1407 (altivec_vmrglb): Replace with define_expand and new
1408 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1409 (altivec_vmrglb_direct): New define_insn.
1410 (altivec_vmrglh): Replace with define_expand and new
1411 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1412 (altivec_vmrglh_direct): New define_insn.
1413 (altivec_vmrglw): Replace with define_expand and new
1414 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1415 (altivec_vmrglw_direct): New define_insn.
1416 (*altivec_vmrglsf): Adjust for endianness.
1417 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1418 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1419 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1420 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1421 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1422 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1423 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1424 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1425
1426 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1427
1428 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1429 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1430 whitespace.
1431
1432 2014-01-29 Richard Biener <rguenther@suse.de>
1433
1434 PR tree-optimization/58742
1435 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1436 associate_pointerplus_align.
1437 (associate_pointerplus_diff): New function.
1438 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1439 and associate_pointerplus_diff.
1440
1441 2014-01-29 Richard Biener <rguenther@suse.de>
1442
1443 * lto-streamer.h (LTO_major_version): Bump to 3.
1444 (LTO_minor_version): Reset to 0.
1445
1446 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1447
1448 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1449 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1450 (arm_file_start): Generate correct asm header for armv7ve.
1451 * config/arm/bpabi.h: Add multilib support for armv7ve.
1452 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1453 and cortex-a15 to armv7ve.
1454 * config/arm/t-aprofile: Add multilib support for armv7ve.
1455 * doc/invoke.texi: Document -march=armv7ve.
1456
1457 2014-01-29 Richard Biener <rguenther@suse.de>
1458
1459 PR tree-optimization/58742
1460 * tree-ssa-forwprop.c (associate_plusminus): Return true
1461 if we changed sth, defer EH cleanup to ...
1462 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1463 (simplify_mult): New function.
1464
1465 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1466
1467 PR middle-end/59917
1468 PR tree-optimization/59920
1469 * tree.c (build_common_builtin_nodes): Remove
1470 __builtin_setjmp_dispatcher initialization.
1471 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1472 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1473 instead of gsi_after_labels + manually skipping debug stmts.
1474 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1475 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1476 * tree-inline.c (copy_edges_for_bb): Remove
1477 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1478 argument. Ignore computed_goto_p stmts. Don't call
1479 make_abnormal_goto_edges. If a call might need abnormal edges
1480 for non-local gotos, see if it already has an edge to
1481 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1482 with true argument, don't do anything then, otherwise add
1483 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1484 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1485 caller.
1486 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1487 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1488 (lower_stmt): Don't set data->calls_builtin_setjmp.
1489 (lower_builtin_setjmp): Adjust comment.
1490 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1491 * tree-cfg.c (found_computed_goto): Remove.
1492 (factor_computed_gotos): Remove.
1493 (make_goto_expr_edges): Return bool, true for computed gotos.
1494 Don't call make_abnormal_goto_edges.
1495 (build_gimple_cfg): Don't set found_computed_goto, don't call
1496 factor_computed_gotos.
1497 (computed_goto_p): No longer static.
1498 (make_blocks): Don't set found_computed_goto.
1499 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1500 (make_edges): If make_goto_expr_edges returns true, push bb
1501 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1502 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1503 vector. Record mapping between bbs and OpenMP regions if there
1504 are any, adjust make_gimple_omp_edges caller. Call
1505 handle_abnormal_edges.
1506 (make_abnormal_goto_edges): Remove.
1507 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1508 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1509 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1510 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1511 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1512 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1513 filling *region also set *region_idx to (*region)->entry->index.
1514
1515 PR other/58712
1516 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1517 For REGs set ORIGINAL_REGNO.
1518
1519 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1520
1521 * doc/md.texi: Mention that a target shouldn't implement
1522 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1523 than hi/lo pair.
1524
1525 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1526
1527 PR tree-optimization/59594
1528 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1529 a copy of the datarefs vector rather than the vector itself.
1530
1531 2014-01-28 Jason Merrill <jason@redhat.com>
1532
1533 PR c++/53756
1534 * dwarf2out.c (auto_die): New static.
1535 (gen_type_die_with_usage): Handle C++1y 'auto'.
1536 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1537 on definition.
1538
1539 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1540
1541 PR target/59672
1542 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1543 (SPEC_X32): Likewise.
1544 (SPEC_64): Likewise.
1545 * config/i386/i386.c (ix86_option_override_internal): Turn off
1546 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1547 for TARGET_16BIT.
1548 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1549 * config/i386/i386.h (TARGET_16BIT): New macro.
1550 (TARGET_16BIT_P): Likewise.
1551 * config/i386/i386.opt: Add m16.
1552 * doc/invoke.texi: Document -m16.
1553
1554 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1555
1556 PR preprocessor/59935
1557 * input.c (location_get_source_line): Bail out on when line number
1558 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1559
1560 2014-01-28 Richard Biener <rguenther@suse.de>
1561
1562 PR tree-optimization/58742
1563 * tree-ssa-forwprop.c (associate_plusminus): Handle
1564 pointer subtraction of the form (T)(P + A) - (T)P.
1565
1566 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1567
1568 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1569 at const_int_cost.
1570
1571 2014-01-28 Richard Biener <rguenther@suse.de>
1572
1573 Revert
1574 2014-01-28 Richard Biener <rguenther@suse.de>
1575
1576 PR rtl-optimization/45364
1577 PR rtl-optimization/59890
1578 * var-tracking.c (local_get_addr_clear_given_value): Handle
1579 already cleared slot.
1580 (val_reset): Handle not allocated local_get_addr_cache.
1581 (vt_find_locations): Use post-order on the inverted CFG.
1582
1583 2014-01-28 Richard Biener <rguenther@suse.de>
1584
1585 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1586
1587 2014-01-28 Richard Biener <rguenther@suse.de>
1588
1589 PR rtl-optimization/45364
1590 PR rtl-optimization/59890
1591 * var-tracking.c (local_get_addr_clear_given_value): Handle
1592 already cleared slot.
1593 (val_reset): Handle not allocated local_get_addr_cache.
1594 (vt_find_locations): Use post-order on the inverted CFG.
1595
1596 2014-01-28 Alan Modra <amodra@gmail.com>
1597
1598 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1599 * configure.ac <recursive call for build != host>: Define
1600 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1601 and LD_FOR_BUILD too.
1602 * configure: Regenerate.
1603
1604 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1605
1606 * config/i386/i386.c (get_builtin_code_for_version): Separate
1607 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1608 Broadwell from Haswell.
1609
1610 2014-01-27 Steve Ellcey <sellcey@mips.com>
1611
1612 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1613 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1614 * config/mips/mips.c (mips_option_override): Change setting
1615 of TARGET_DSP.
1616 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1617 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1618 Change from Mask to Var.
1619
1620 2014-01-27 Jeff Law <law@redhat.com>
1621
1622 * ipa-inline.c (inline_small_functions): Fix typo.
1623
1624 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1625
1626 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1627 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1628 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1629 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1630 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1631 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1632 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1633 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1634 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1635 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1636 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1637 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1638 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1639 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1640 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1641 (_mm512_storeu_epi64): Ditto.
1642 (_mm512_cmpge_epi32_mask): Ditto.
1643 (_mm512_cmpge_epu32_mask): Ditto.
1644 (_mm512_cmpge_epi64_mask): Ditto.
1645 (_mm512_cmpge_epu64_mask): Ditto.
1646 (_mm512_cmple_epi32_mask): Ditto.
1647 (_mm512_cmple_epu32_mask): Ditto.
1648 (_mm512_cmple_epi64_mask): Ditto.
1649 (_mm512_cmple_epu64_mask): Ditto.
1650 (_mm512_cmplt_epi32_mask): Ditto.
1651 (_mm512_cmplt_epu32_mask): Ditto.
1652 (_mm512_cmplt_epi64_mask): Ditto.
1653 (_mm512_cmplt_epu64_mask): Ditto.
1654 (_mm512_cmpneq_epi32_mask): Ditto.
1655 (_mm512_cmpneq_epu32_mask): Ditto.
1656 (_mm512_cmpneq_epi64_mask): Ditto.
1657 (_mm512_cmpneq_epu64_mask): Ditto.
1658 (_mm512_expand_pd): Ditto.
1659 (_mm512_expand_ps): Ditto.
1660 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1661 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1662 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1663 * config/i386/i386.c (ix86_builtins): Add
1664 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1665 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1666 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1667 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1668 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1669 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1670 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1671 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1672 IX86_BUILTIN_PMOVUSQW512_MEM.
1673 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1674 __builtin_ia32_pmovsqd512mem_mask,
1675 __builtin_ia32_pmovqd512mem_mask,
1676 __builtin_ia32_pmovusqw512mem_mask,
1677 __builtin_ia32_pmovsqw512mem_mask,
1678 __builtin_ia32_pmovqw512mem_mask,
1679 __builtin_ia32_pmovusdw512mem_mask,
1680 __builtin_ia32_pmovsdw512mem_mask,
1681 __builtin_ia32_pmovdw512mem_mask,
1682 __builtin_ia32_pmovqb512mem_mask,
1683 __builtin_ia32_pmovusqb512mem_mask,
1684 __builtin_ia32_pmovsqb512mem_mask,
1685 __builtin_ia32_pmovusdb512mem_mask,
1686 __builtin_ia32_pmovsdb512mem_mask,
1687 __builtin_ia32_pmovdb512mem_mask.
1688 (bdesc_args): Add __builtin_ia32_expanddf512,
1689 __builtin_ia32_expandsf512.
1690 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1691 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1692 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1693 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1694 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1695 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1696 (avx512f_<code>v8div16qi2_mask_store): This.
1697 (avx512f_expand<mode>): New.
1698
1699 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1700
1701 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1702 New.
1703 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1704 (_mm512_prefetch_i32scatter_pd): Ditto.
1705 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1706 (_mm512_prefetch_i64scatter_pd): Ditto.
1707 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1708 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1709 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1710 (_mm512_prefetch_i32scatter_ps): Ditto.
1711 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1712 (_mm512_prefetch_i64scatter_ps): Ditto.
1713 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1714 * config/i386/i386-builtin-types.def: Define
1715 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1716 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1717 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1718 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1719 IX86_BUILTIN_SCATTERPFQPD.
1720 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1721 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1722 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1723 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1724 __builtin_ia32_scatterpfqps.
1725 (ix86_expand_builtin): Expand new built-ins.
1726 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1727 fix memory access data type.
1728 (*avx512pf_gatherpf<mode>_mask): Ditto.
1729 (*avx512pf_gatherpf<mode>): Ditto.
1730 (avx512pf_scatterpf<mode>): Ditto.
1731 (*avx512pf_scatterpf<mode>_mask): Ditto.
1732 (*avx512pf_scatterpf<mode>): Ditto.
1733 (GATHER_SCATTER_SF_MEM_MODE): New.
1734 (avx512pf_gatherpf<mode>df): Ditto.
1735 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1736 (*avx512pf_scatterpf<mode>df): Ditto.
1737
1738 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1739
1740 PR bootstrap/59934
1741 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1742 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1743 reached.
1744
1745 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1746
1747 * common/config/arm/arm-common.c
1748 (arm_rewrite_mcpu): Handle multiple names.
1749 * config/arm/arm.h
1750 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1751
1752 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1753
1754 * gimple-builder.h (create_gimple_tmp): Delete.
1755
1756 2014-01-27 Christian Bruel <christian.bruel@st.com>
1757
1758 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1759 words comparisons.
1760
1761 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1762
1763 * config/pa/pa.md (call): Generate indirect long calls to non-local
1764 functions when outputing 32-bit code.
1765 (call_value): Likewise except for special call to buggy powf function.
1766
1767 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1768 portable runtime and PIC indirect calls.
1769 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1770 and PIC call sequences. Use ldo instead of blr to set return register
1771 in PIC call sequence.
1772
1773 2014-01-25 Walter Lee <walt@tilera.com>
1774
1775 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1776 avoid clobbering a live register.
1777
1778 2014-01-25 Walter Lee <walt@tilera.com>
1779
1780 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1781 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1782 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1783 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1784
1785 2014-01-25 Walter Lee <walt@tilera.com>
1786
1787 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1788 arguments on even registers.
1789 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1790 STACK_BOUNDARY.
1791 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1792 (BIGGEST_ALIGNMENT): Ditto.
1793 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1794
1795 2014-01-25 Walter Lee <walt@tilera.com>
1796
1797 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1798 insns before bundling.
1799 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1800
1801 2014-01-25 Walter Lee <walt@tilera.com>
1802
1803 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1804 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1805 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1806
1807 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1808
1809 * config/mips/constraints.md (kl): Delete.
1810 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1811 define expands, using...
1812 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1813 instructions for MIPS16.
1814 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1815 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1816
1817 2014-01-25 Walter Lee <walt@tilera.com>
1818
1819 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1820 (clzdi2): Ditto.
1821 (ffsdi2): Ditto.
1822
1823 2014-01-25 Walter Lee <walt@tilera.com>
1824
1825 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1826 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1827
1828 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1829
1830 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1831 Handle XOR.
1832
1833 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1834
1835 * print-rtl.c (in_call_function_usage): New var.
1836 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1837 EXPR_LIST mode as mode and not as reg note name.
1838
1839 PR middle-end/59561
1840 * cfgloopmanip.c (copy_loop_info): If
1841 loop->warned_aggressive_loop_optimizations, make sure
1842 the flag is set in target loop too.
1843
1844 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1845
1846 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1847 flag_cilkplus.
1848 * builtins.def: Likewise.
1849 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1850 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1851 * ira.c (ira_setup_eliminable_regset): Likewise.
1852 * omp-low.c (gate_expand_omp): Likewise.
1853 (execute_lower_omp): Likewise.
1854 (diagnose_sb_0): Likewise.
1855 (gate_diagnose_omp_blocks): Likewise.
1856 (simd_clone_clauses_extract): Likewise.
1857 (gate): Likewise.
1858
1859 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1860
1861 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1862 correction for little endian...
1863 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1864 here.
1865
1866 2014-01-24 Jeff Law <law@redhat.com>
1867
1868 PR tree-optimization/59919
1869 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1870 for non-returning calls.
1871
1872 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1873
1874 * common/config/aarch64/aarch64-common.c
1875 (aarch64_rewrite_mcpu): Handle multiple names.
1876 * config/aarch64/aarch64.h
1877 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1878
1879 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1880
1881 * input.c (add_file_to_cache_tab): Handle the case where fopen
1882 returns NULL.
1883
1884 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1885
1886 PR target/59929
1887 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1888 from push operand if code of push isn't PRE_DEC.
1889
1890 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1891
1892 PR target/59909
1893 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1894 -mquad-memory-atomic. Update -mquad-memory documentation to say
1895 it is only used for non-atomic loads/stores.
1896
1897 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1898 -mquad-memory or -mquad-memory-atomic switches.
1899
1900 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1901 -mquad-memory-atomic to ISA 2.07 support.
1902
1903 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1904 to separate support of normal quad word memory operations (ldq, stq)
1905 from the atomic quad word memory operations.
1906
1907 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1908 support to separate non-atomic quad word operations from atomic
1909 quad word operations. Disable non-atomic quad word operations in
1910 little endian mode so that we don't have to swap words after the
1911 load and before the store.
1912 (quad_load_store_p): Add comment about atomic quad word support.
1913 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1914 options printed with -mdebug=reg.
1915
1916 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1917 -mquad-memory-atomic as the test for whether we have quad word
1918 atomic instructions.
1919 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1920 or -mp8-vector are used, allow byte/half-word atomic operations.
1921
1922 * config/rs6000/sync.md (load_lockedti): Insure that the address
1923 is a proper indexed or indirect address for the lqarx instruction.
1924 On little endian systems, swap the hi/lo registers after the lqarx
1925 instruction.
1926 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1927 insure the address is valid for the lqarx instruction.
1928 (store_conditionalti): Insure that the address is a proper indexed
1929 or indirect address for the stqcrx. instruction. On little endian
1930 systems, swap the hi/lo registers before doing the stqcrx.
1931 instruction.
1932 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1933 insure the address is valid for the stqcrx. instruction.
1934
1935 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1936 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1937 type of quad memory support is available.
1938
1939 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1940
1941 PR regression/59915
1942 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1943 there is a danger of looping.
1944
1945 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1946
1947 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1948 force flag_ira_loop_pressure if set via command line.
1949
1950 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1951
1952 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1953 (ashr_simd): New builtin handling DI mode.
1954 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1955 (aarch64_sshr_simddi): New match pattern.
1956 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1957 (vshrd_n_s64): Likewise.
1958 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1959
1960 2014-01-23 Nick Clifton <nickc@redhat.com>
1961
1962 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1963 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1964 favour of mcu specific scripts.
1965 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1966 430x multilibs.
1967
1968 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1969 Alex Velenko <Alex.Velenko@arm.com>
1970
1971 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1972 (vaddv_s16): Likewise.
1973 (vaddv_s32): Likewise.
1974 (vaddv_u8): Likewise.
1975 (vaddv_u16): Likewise.
1976 (vaddv_u32): Likewise.
1977 (vaddvq_s8): Likewise.
1978 (vaddvq_s16): Likewise.
1979 (vaddvq_s32): Likewise.
1980 (vaddvq_s64): Likewise.
1981 (vaddvq_u8): Likewise.
1982 (vaddvq_u16): Likewise.
1983 (vaddvq_u32): Likewise.
1984 (vaddvq_u64): Likewise.
1985 (vaddv_f32): Likewise.
1986 (vaddvq_f32): Likewise.
1987 (vaddvq_f64): Likewise.
1988 (vmaxv_f32): Likewise.
1989 (vmaxv_s8): Likewise.
1990 (vmaxv_s16): Likewise.
1991 (vmaxv_s32): Likewise.
1992 (vmaxv_u8): Likewise.
1993 (vmaxv_u16): Likewise.
1994 (vmaxv_u32): Likewise.
1995 (vmaxvq_f32): Likewise.
1996 (vmaxvq_f64): Likewise.
1997 (vmaxvq_s8): Likewise.
1998 (vmaxvq_s16): Likewise.
1999 (vmaxvq_s32): Likewise.
2000 (vmaxvq_u8): Likewise.
2001 (vmaxvq_u16): Likewise.
2002 (vmaxvq_u32): Likewise.
2003 (vmaxnmv_f32): Likewise.
2004 (vmaxnmvq_f32): Likewise.
2005 (vmaxnmvq_f64): Likewise.
2006 (vminv_f32): Likewise.
2007 (vminv_s8): Likewise.
2008 (vminv_s16): Likewise.
2009 (vminv_s32): Likewise.
2010 (vminv_u8): Likewise.
2011 (vminv_u16): Likewise.
2012 (vminv_u32): Likewise.
2013 (vminvq_f32): Likewise.
2014 (vminvq_f64): Likewise.
2015 (vminvq_s8): Likewise.
2016 (vminvq_s16): Likewise.
2017 (vminvq_s32): Likewise.
2018 (vminvq_u8): Likewise.
2019 (vminvq_u16): Likewise.
2020 (vminvq_u32): Likewise.
2021 (vminnmv_f32): Likewise.
2022 (vminnmvq_f32): Likewise.
2023 (vminnmvq_f64): Likewise.
2024
2025 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2026
2027 * config/aarch64/aarch64-simd.md
2028 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2029 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2030 (*aarch64_mul3_elt<mode>): Likewise.
2031 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2032 (*aarch64_mul3_elt_to_64v2df): Likewise.
2033 (*aarch64_mla_elt<mode>): Likewise.
2034 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2035 (*aarch64_mls_elt<mode>): Likewise.
2036 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2037 (*aarch64_fma4_elt<mode>): Likewise.
2038 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2039 (*aarch64_fma4_elt_to_64v2df): Likewise.
2040 (*aarch64_fnma4_elt<mode>): Likewise.
2041 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2042 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2043 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2044 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2045 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2046 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2047 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2048 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2049 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2050
2051 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2052
2053 * config/aarch64/aarch64-simd.md
2054 (aarch64_be_checked_get_lane<mode>): New define_expand.
2055 * config/aarch64/aarch64-simd-builtins.def
2056 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2057 New builtin definition.
2058 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2059 Use new safe be builtin.
2060
2061 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2062
2063 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2064 New define_insn.
2065 (aarch64_be_st1<mode>): Likewise.
2066 (aarch_ld1<VALL:mode>): Define_expand modified.
2067 (aarch_st1<VALL:mode>): Likewise.
2068 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2069 (UNSPEC_ST1): Likewise.
2070
2071 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2072
2073 * config/microblaze/microblaze.md: Add trap insn and attribute
2074
2075 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2076
2077 PR preprocessor/58580
2078 * input.h (location_get_source_line): Take an additional line_size
2079 parameter.
2080 (void diagnostics_file_cache_fini): Declare new function.
2081 * input.c (struct fcache): New type.
2082 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2083 New static constants.
2084 (diagnostic_file_cache_init, total_lines_num)
2085 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2086 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2087 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2088 (get_next_line, read_next_line, goto_next_line, read_line_num):
2089 New static function definitions.
2090 (diagnostic_file_cache_fini): New function.
2091 (location_get_source_line): Take an additional output line_len
2092 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2093 read_line_num.
2094 * diagnostic.c (diagnostic_finish): Call
2095 diagnostic_file_cache_fini.
2096 (adjust_line): Take an additional input parameter for the length
2097 of the line, rather than calculating it with strlen.
2098 (diagnostic_show_locus): Adjust the use of
2099 location_get_source_line and adjust_line with respect to their new
2100 signature. While displaying a line now, do not stop at the first
2101 null byte. Rather, display the zero byte as a space and keep
2102 going until we reach the size of the line.
2103 * Makefile.in: Add vec.o to OBJS-libcommon
2104
2105 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2106 Ilya Tocar <ilya.tocar@intel.com>
2107
2108 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2109 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2110 (__builtin_ia32_kmov16): Ditto.
2111 * config/i386/i386.md (UNSPEC_KMOV): New.
2112 (kmovw): Ditto.
2113
2114 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2115
2116 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2117 (_mm512_storeu_si512): Ditto.
2118
2119 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2120
2121 PR target/52125
2122 * rtl.h (get_referenced_operands): Declare.
2123 * recog.c (get_referenced_operands): New function.
2124 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2125 operands have been referenced when recording LO_SUM references.
2126
2127 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2128
2129 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2130
2131 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2132
2133 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2134 Enable for generic and recent AMD targets.
2135
2136 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2137
2138 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2139 ARG_SIZE note when adjustment was eliminated.
2140
2141 2014-01-22 Jeff Law <law@redhat.com>
2142
2143 PR tree-optimization/59597
2144 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2145 in file. Accept new argument REGISTERING and use it to modify
2146 dump output appropriately.
2147 (register_jump_thread): Corresponding changes.
2148 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2149 thread paths involving joiner blocks. Add code to dump cancelled
2150 jump threading paths.
2151
2152 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2153
2154 PR rtl-optimization/59477
2155 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2156 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2157
2158 2014-01-22 Tom Tromey <tromey@redhat.com>
2159
2160 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2161 PARAMS.
2162 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2163
2164 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2165
2166 PR rtl-optimization/59896
2167 * lra-constraints.c (process_alt_operands): Check unused note for
2168 matched operands of insn with no output reloads.
2169
2170 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2171
2172 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2173 (mips_move_from_gpr_cost): Likewise.
2174
2175 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2176
2177 PR rtl-optimization/59858
2178 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2179 ira_class_hard_regs_num.
2180 (process_alt_operands): Increase reject for dying matched operand.
2181
2182 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2183
2184 PR target/59003
2185 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2186 smaller than size, perform several stores or loads and stores
2187 at dst + count - size to store or copy all of size bytes, rather
2188 than just last modesize bytes.
2189
2190 2014-01-20 DJ Delorie <dj@redhat.com>
2191
2192 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2193 that CLOBBERs are REGs before propogating their values.
2194
2195 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2196
2197 PR middle-end/59789
2198 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2199 (cgraph_inline_failed_type): New function.
2200 * cgraph.h (DEFCIFCODE): Add type.
2201 (cgraph_inline_failed_type_t): New enum.
2202 (cgraph_inline_failed_type): New prototype.
2203 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2204 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2205 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2206 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2207 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2208 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2209 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2210 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2211 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2212 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2213 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2214 OPTIMIZATION_MISMATCH.
2215 * tree-inline.c (expand_call_inline): Emit errors during
2216 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2217
2218 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2219
2220 PR target/59685
2221 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2222 mode attribute in insn output.
2223
2224 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2225
2226 * output.h (output_constant): Delete.
2227 * varasm.c (output_constant): Make private.
2228
2229 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2230
2231 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2232
2233 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2234
2235 PR middle-end/59860
2236 * tree.h (fold_builtin_strcat): New prototype.
2237 * builtins.c (fold_builtin_strcat): No longer static. Add len
2238 argument, if non-NULL, don't call c_strlen. Optimize
2239 directly into __builtin_memcpy instead of __builtin_strcpy.
2240 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2241 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2242
2243 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2244
2245 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2246 for SImode_address_operand operands, having only a REG argument.
2247
2248 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2249
2250 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2251 loader name using mbig-endian.
2252 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2253
2254 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2255
2256 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2257 (-mtune): Likewise.
2258 (-mcpu): Likewise.
2259
2260 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2261
2262 * config/aarch64/aarch64-protos.h
2263 (aarch64_cannot_change_mode_class_ptr): Declare.
2264 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2265 aarch64_cannot_change_mode_class_ptr): New.
2266 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2267 backend hook aarch64_cannot_change_mode_class.
2268
2269 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2270
2271 * common/config/aarch64/aarch64-common.c
2272 (aarch64_handle_option): Don't handle any option order logic here.
2273 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2274 selected_cpu, warn on architecture version mismatch.
2275 (aarch64_override_options): Fix parsing order for option strings.
2276
2277 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2278 Iain Sandoe <iain@codesourcery.com>
2279
2280 PR bootstrap/59496
2281 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2282 warning. Amend comment to reflect current functionality.
2283
2284 2014-01-20 Richard Biener <rguenther@suse.de>
2285
2286 PR middle-end/59860
2287 * builtins.c (fold_builtin_strcat): Remove case better handled
2288 by tree-ssa-strlen.c.
2289
2290 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2291
2292 * config/aarch64/aarch64.opt
2293 (mcpu, march, mtune): Make case-insensitive.
2294
2295 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2296
2297 PR target/59880
2298 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2299 if operands[1] is a REG or ZERO_EXTEND of a REG.
2300
2301 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2302
2303 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2304
2305 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2306
2307 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2308 long non-pic millicode calls.
2309
2310 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2311
2312 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2313
2314 2014-01-19 Kito Cheng <kito@0xlab.org>
2315
2316 * builtins.c (expand_movstr): Check movstr expand done or fail.
2317
2318 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2319 H.J. Lu <hongjiu.lu@intel.com>
2320
2321 PR target/59379
2322 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2323 to DImode for zero-extended addresses.
2324
2325 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2326
2327 PR rtl-optimization/57763
2328 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2329 on the new indirect jump_insn and increment LABEL_NUSES (label).
2330
2331 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2332
2333 PR bootstrap/59580
2334 PR bootstrap/59583
2335 * config.gcc (x86_archs): New variable.
2336 (x86_64_archs): Likewise.
2337 (x86_cpus): Likewise.
2338 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2339 --with-arch/--with-cpu= options.
2340 Support --with-arch=/--with-cpu={nehalem,westmere,
2341 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2342
2343 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2344
2345 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2346 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2347
2348 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2349
2350 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2351
2352 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2353
2354 PR target/58944
2355 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2356 clear cpp_get_options (parse_in)->warn_unused_macros for
2357 ix86_target_macros_internal with cpp_define.
2358
2359 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2360
2361 * jump.c (delete_related_insns): Keep (use (insn))s.
2362 * reorg.c (redundant_insn): Check for barriers too.
2363
2364 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2365
2366 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2367
2368 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2369
2370 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2371 call to $$dyncall when TARGET_LONG_CALLS is true.
2372
2373 2014-01-17 Jeff Law <law@redhat.com>
2374
2375 * ree.c (combine_set_extension): Temporarily disable test for
2376 changing number of hard registers.
2377
2378 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2379
2380 PR middle-end/58125
2381 * ipa-inline-analysis.c (inline_free_summary):
2382 Do not free summary of aliases.
2383
2384 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2385
2386 PR middle-end/59706
2387 * gimplify.c (gimplify_expr): Use create_tmp_var
2388 instead of create_tmp_var_raw. If cond doesn't have
2389 integral type, don't add the IFN_ANNOTATE builtin at all.
2390
2391 2014-01-17 Martin Jambor <mjambor@suse.cz>
2392
2393 PR ipa/59736
2394 * ipa-cp.c (prev_edge_clone): New variable.
2395 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2396 Also resize prev_edge_clone vector.
2397 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2398 (ipcp_edge_removal_hook): New function.
2399 (ipcp_driver): Register ipcp_edge_removal_hook.
2400
2401 2014-01-17 Andrew Pinski <apinski@cavium.com>
2402 Steve Ellcey <sellcey@mips.com>
2403
2404 PR target/59462
2405 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2406 of operator mode.
2407
2408 2014-01-17 Jeff Law <law@redhat.com>
2409
2410 PR middle-end/57904
2411 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2412 so that pass_ccp runs first.
2413
2414 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2415
2416 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2417 (ix86_adjust_cost): Use !TARGET_XXX.
2418 (do_reorder_for_imul): Likewise.
2419 (swap_top_of_ready_list): Likewise.
2420 (ix86_sched_reorder): Likewise.
2421
2422 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2423
2424 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2425 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2426 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2427 (intel_memset): New. Duplicate slm_memset.
2428 (intel_cost): New. Duplicate slm_cost.
2429 (m_INTEL): New macro.
2430 (processor_target_table): Add "intel".
2431 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2432 with PROCESSOR_INTEL for "intel".
2433 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2434 PROCESSOR_SILVERMONT.
2435 (ix86_issue_rate): Likewise.
2436 (ix86_adjust_cost): Likewise.
2437 (ia32_multipass_dfa_lookahead): Likewise.
2438 (swap_top_of_ready_list): Likewise.
2439 (ix86_sched_reorder): Likewise.
2440 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2441 instead of TARGET_OPT_AGU.
2442 * config/i386/i386.h (TARGET_INTEL): New.
2443 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2444 (processor_type): Add PROCESSOR_INTEL.
2445 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2446 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2447
2448 2014-01-17 Marek Polacek <polacek@redhat.com>
2449
2450 PR c/58346
2451 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2452 size is zero.
2453
2454 2014-01-17 Richard Biener <rguenther@suse.de>
2455
2456 PR tree-optimization/46590
2457 * opts.c (default_options_table): Add entries for
2458 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2459 all enabled at -O1 but not for -Og.
2460 * common.opt (fbranch-count-reg): Remove Init(1).
2461 (fmove-loop-invariants): Likewise.
2462 (ftree-pta): Likewise.
2463
2464 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2465
2466 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2467 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2468 decls to at least the GCC 4.8 used alignments.
2469
2470 PR fortran/59440
2471 * tree-nested.c (convert_nonlocal_reference_stmt,
2472 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2473 of GIMPLE_BIND stmts, adjust associated decls.
2474
2475 2014-01-17 Richard Biener <rguenther@suse.de>
2476
2477 PR tree-optimization/46590
2478 * vec.h (vec<>::bseach): New member function implementing
2479 binary search according to C89 bsearch.
2480 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2481 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2482 bitmap pointer again. Make accesses_in_loop a flat array.
2483 (mem_ref_obstack): New global.
2484 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2485 (mark_ref_stored): Likewise.
2486 (ref_indep_loop_p_2): Likewise.
2487 (set_ref_stored_in_loop): New helper function.
2488 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2489 (memref_free): Adjust.
2490 (record_mem_ref_loc): Simplify.
2491 (gather_mem_refs_stmt): Adjust.
2492 (sort_locs_in_loop_postorder_cmp): New function.
2493 (analyze_memory_references): Sort accesses_in_loop after
2494 loop postorder number.
2495 (find_ref_loc_in_loop_cmp): New function.
2496 (for_all_locs_in_loop): Find relevant cluster of locs in
2497 accesses_in_loop and iterate without recursion.
2498 (execute_sm): Avoid uninit warning.
2499 (struct ref_always_accessed): Simplify.
2500 (ref_always_accessed::operator ()): Likewise.
2501 (ref_always_accessed_p): Likewise.
2502 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2503 loop postorder numbers here.
2504 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2505 numbers.
2506
2507 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2508
2509 PR c++/57945
2510 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2511 on decls for which assemble_alias has been called.
2512
2513 2014-01-17 Nick Clifton <nickc@redhat.com>
2514
2515 * config/msp430/msp430.opt: (mcpu): New option.
2516 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2517 (msp430_option_override): Parse target_cpu. If the MCU name
2518 matches a generic string, clear target_mcu.
2519 (msp430_attr): Allow numeric interrupt values up to 63.
2520 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2521 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2522 option.
2523 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2524 Add mcpu matches.
2525 * config/msp430/msp430.md (popm): Use %J rather than %I.
2526 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2527 (addhi_cy_i): Use immediate_operand for operand 2.
2528 * doc/invoke.texi: Document -mcpu option.
2529
2530 2014-01-17 Richard Biener <rguenther@suse.de>
2531
2532 PR rtl-optimization/38518
2533 * df.h (df_analyze_loop): Declare.
2534 * df-core.c: Include cfgloop.h.
2535 (df_analyze_1): Split out main part of df_analyze.
2536 (df_analyze): Adjust.
2537 (loop_inverted_post_order_compute): New function.
2538 (loop_post_order_compute): Likewise.
2539 (df_analyze_loop): New function avoiding whole-function
2540 postorder computes.
2541 * loop-invariant.c (find_defs): Use df_analyze_loop.
2542 (find_invariants): Adjust.
2543 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2544
2545 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2546
2547 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2548 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2549
2550 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2551
2552 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2553 traversal when removing references.
2554
2555 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2556
2557 PR ipa/59775
2558 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2559
2560 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2561
2562 PR middle-end/56791
2563 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2564 pushing a reload for an autoinc when we had previously reloaded an
2565 inner part of the address.
2566
2567 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2568
2569 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2570 field.
2571 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2572 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2573 when not giving up or versioning for alias only because of
2574 loop->safelen.
2575 (vect_analyze_data_ref_dependences): Set to true.
2576 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2577 is a GIMPLE_PHI.
2578 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2579 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2580 to the condition.
2581
2582 PR middle-end/58344
2583 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2584
2585 PR target/59839
2586 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2587 operand 0 predicate for gathers, use a new pseudo as subtarget.
2588
2589 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2590
2591 PR middle-end/59609
2592 * lra-constraints.c (process_alt_operands): Add printing debug info.
2593 Check absence of input/output reloads for matched operands too.
2594
2595 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2596
2597 PR rtl-optimization/59835
2598 * ira.c (ira_init_register_move_cost): Increase cost for
2599 impossible modes.
2600
2601 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2602
2603 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2604
2605 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2606
2607 PR target/59780
2608 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2609 non-register objects. Use gen_(high/low)part more consistently.
2610 Fix assertions.
2611
2612 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2613
2614 PR target/59844
2615 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2616 endian support, remove tests for WORDS_BIG_ENDIAN.
2617 (p8_mfvsrd_3_<mode>): Likewise.
2618 (reload_gpr_from_vsx<mode>): Likewise.
2619 (reload_gpr_from_vsxsf): Likewise.
2620 (p8_mfvsrd_4_disf): Likewise.
2621
2622 2014-01-16 Richard Biener <rguenther@suse.de>
2623
2624 PR rtl-optimization/46590
2625 * lcm.c (compute_antinout_edge): Use postorder iteration.
2626 (compute_laterin): Use inverted postorder iteration.
2627
2628 2014-01-16 Nick Clifton <nickc@redhat.com>
2629
2630 PR middle-end/28865
2631 * varasm.c (output_constant): Return the number of bytes actually
2632 emitted.
2633 (output_constructor_array_range): Update the field size with the
2634 number of bytes emitted by output_constant.
2635 (output_constructor_regular_field): Likewise. Also do not
2636 complain if the total number of bytes emitted is now greater
2637 than the expected fieldpos.
2638 * output.h (output_constant): Update prototype and descriptive comment.
2639
2640 2014-01-16 Marek Polacek <polacek@redhat.com>
2641
2642 PR middle-end/59827
2643 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2644 it is error_mark_node.
2645
2646 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2647
2648 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2649 VALID_AVX256_REG_OR_OI_MODE.
2650
2651 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2652
2653 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2654 current procedure should be profiled.
2655
2656 2014-01-15 Andrew Pinski <apinski@cavium.com>
2657
2658 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2659 of moving from/to the STACK_REG register class.
2660
2661 2014-01-15 Richard Henderson <rth@redhat.com>
2662
2663 PR debug/54694
2664 * reginfo.c (global_regs_decl): Globalize.
2665 * rtl.h (global_regs_decl): Declare.
2666 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2667 reserved via global_regs.
2668
2669 2014-01-15 Teresa Johnson <tejohnson@google.com>
2670
2671 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2672
2673 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2674
2675 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2676 and vmulosh rather than call gen_vec_widen_smult_*.
2677 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2678 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2679 (vec_widen_smult_even_v16qi): Likewise.
2680 (vec_widen_umult_even_v8hi): Likewise.
2681 (vec_widen_smult_even_v8hi): Likewise.
2682 (vec_widen_umult_odd_v16qi): Likewise.
2683 (vec_widen_smult_odd_v16qi): Likewise.
2684 (vec_widen_umult_odd_v8hi): Likewise.
2685 (vec_widen_smult_odd_v8hi): Likewise.
2686 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2687 vmuloub rather than call gen_vec_widen_umult_*.
2688 (vec_widen_umult_lo_v16qi): Likewise.
2689 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2690 vmulosb rather than call gen_vec_widen_smult_*.
2691 (vec_widen_smult_lo_v16qi): Likewise.
2692 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2693 rather than call gen_vec_widen_umult_*.
2694 (vec_widen_umult_lo_v8hi): Likewise.
2695 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2696 rather than call gen_vec_widen_smult_*.
2697 (vec_widen_smult_lo_v8hi): Likewise.
2698
2699 2014-01-15 Jeff Law <law@redhat.com>
2700
2701 PR tree-optimization/59747
2702 * ree.c (find_and_remove_re): Properly handle case where a second
2703 eliminated extension requires widening a copy created for elimination
2704 of a prior extension.
2705 (combine_set_extension): Ensure that the number of hard regs needed
2706 for a destination register does not change when we widen it.
2707
2708 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2709
2710 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2711 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2712 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2713 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2714 (avr-*-rtems*): Likewise.
2715 (bfin*-rtems*): Likewise.
2716 (moxie-*-rtems*): Likewise.
2717 (h8300-*-rtems*): Likewise.
2718 (i[34567]86-*-rtems*): Likewise.
2719 (lm32-*-rtems*): Likewise.
2720 (m32r-*-rtems*): Likewise.
2721 (m68k-*-rtems*): Likewise.
2722 (microblaze*-*-rtems*): Likewise.
2723 (mips*-*-rtems*): Likewise.
2724 (powerpc-*-rtems*): Likewise.
2725 (sh-*-rtems*): Likewise.
2726 (sparc-*-rtems*): Likewise.
2727 (sparc64-*-rtems*): Likewise.
2728 (v850-*-rtems*): Likewise.
2729 (m32c-*-rtems*): Likewise.
2730
2731 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2732
2733 PR rtl-optimization/59511
2734 * ira.c (ira_init_register_move_cost): Use memory costs for some
2735 cases of register move cost calculations.
2736 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2737 instead of BB frequency.
2738 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2739 * lra-assigns.c (find_hard_regno_for): Ditto.
2740
2741 2014-01-15 Richard Biener <rguenther@suse.de>
2742
2743 PR tree-optimization/59822
2744 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2745 (vectorizable_load): Use it to hoist defs of uses of invariant
2746 loads out of the loop.
2747
2748 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2749 Kugan Vivekanandarajah <kuganv@linaro.org>
2750
2751 PR target/59695
2752 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2753 truncation.
2754
2755 2014-01-15 Richard Biener <rguenther@suse.de>
2756
2757 PR rtl-optimization/59802
2758 * lcm.c (compute_available): Use inverted postorder to seed
2759 the initial worklist.
2760
2761 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2762
2763 PR target/59803
2764 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2765 ADDR_REGS for invalid symrefs in non-PIC code.
2766
2767 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2768
2769 PR other/58712
2770 * builtins.c (determine_block_size): Initialize *probable_max_size
2771 even if len_rtx is CONST_INT.
2772
2773 2014-01-14 Andrew Pinski <apinski@cavium.com>
2774
2775 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2776 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2777 (cortexa53_tunings): Likewise.
2778 (aarch64_sched_issue_rate): New function.
2779 (TARGET_SCHED_ISSUE_RATE): Define.
2780
2781 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2782
2783 * ira-costs.c (find_costs_and_classes): Add missed
2784 ira_init_register_move_cost_if_necessary.
2785
2786 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2787
2788 PR target/59787
2789 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2790
2791 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2792
2793 PR target/59794
2794 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2795 to indicate if type is used for function return value. Warn ABI
2796 change if the vector mode isn't available for function return value.
2797 (ix86_function_arg_advance): Pass false to type_natural_mode.
2798 (ix86_function_arg): Likewise.
2799 (ix86_gimplify_va_arg): Likewise.
2800 (function_arg_32): Don't warn ABI change.
2801 (ix86_function_value): Pass true to type_natural_mode.
2802 (ix86_return_in_memory): Likewise.
2803 (ix86_struct_value_rtx): Removed.
2804 (TARGET_STRUCT_VALUE_RTX): Likewise.
2805
2806 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2807
2808 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2809 converting a conditional jump into a conditional return.
2810
2811 2014-01-14 Richard Biener <rguenther@suse.de>
2812
2813 PR tree-optimization/58921
2814 PR tree-optimization/59006
2815 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2816 hoisting invariant stmts.
2817 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2818 invariant loads on the preheader edge if possible.
2819
2820 2014-01-14 Joey Ye <joey.ye@arm.com>
2821
2822 * doc/plugin.texi (Building GCC plugins): Update to C++.
2823
2824 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2825
2826 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2827 (_mm_rcp28_round_ss): Ditto.
2828 (_mm_rsqrt28_round_sd): Ditto.
2829 (_mm_rsqrt28_round_ss): Ditto.
2830 (_mm_rcp28_sd): Ditto.
2831 (_mm_rcp28_ss): Ditto.
2832 (_mm_rsqrt28_sd): Ditto.
2833 (_mm_rsqrt28_ss): Ditto.
2834 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2835 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2836 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2837 (IX86_BUILTIN_RCP28SD): Ditto.
2838 (IX86_BUILTIN_RCP28SS): Ditto.
2839 (IX86_BUILTIN_RSQRT28SD): Ditto.
2840 (IX86_BUILTIN_RSQRT28SS): Ditto.
2841 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2842 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2843 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2844 (ix86_expand_special_args_builtin): Expand new FTYPE.
2845 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2846 (srcp14<mode>): Make insn unary.
2847 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2848 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2849 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2850 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2851 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2852 Fix rounding: make it SAE only.
2853 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2854 Ditto.
2855 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2856 Ditto.
2857 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2858 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2859 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2860 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2861 (round_saeonly_mask_scalar_operand4): Ditto.
2862 (round_saeonly_mask_scalar_op3): Ditto.
2863 (round_saeonly_mask_scalar_op4): Ditto.
2864
2865 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2866
2867 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2868 Implement -maltivec=be for vec_insert and vec_extract.
2869
2870 2014-01-10 DJ Delorie <dj@redhat.com>
2871
2872 * config/msp430/msp430.md (call_internal): Don't allow memory
2873 references with SP as the base register.
2874 (call_value_internal): Likewise.
2875 * config/msp430/constraints.md (Yc): New. For memory references
2876 that don't use SP as a base register.
2877
2878 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2879 "an integer without a # prefix"
2880 * config/msp430/msp430.md (epilogue_helper): Use it.
2881
2882 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2883
2884 PR target/59617
2885 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2886 AVX512F gather builtins.
2887 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2888 on gather decls with INTEGER_TYPE masktype.
2889 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2890 directly into the builtin rather than hoisting it before loop.
2891
2892 PR tree-optimization/59387
2893 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2894 (scev_const_prop): If folded_casts and type has undefined overflow,
2895 use force_gimple_operand instead of force_gimple_operand_gsi and
2896 for each added stmt if it is assign with
2897 arith_code_with_undefined_signed_overflow, call
2898 rewrite_to_defined_overflow.
2899 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2900 gimple-fold.h instead.
2901 (arith_code_with_undefined_signed_overflow,
2902 rewrite_to_defined_overflow): Moved to ...
2903 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2904 rewrite_to_defined_overflow): ... here. No longer static.
2905 Include gimplify-me.h.
2906 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2907 rewrite_to_defined_overflow): New prototypes.
2908
2909 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2910
2911 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2912
2913 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2914
2915 * builtins.c (get_object_alignment_2): Minor tweak.
2916 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2917
2918 2014-01-13 Christian Bruel <christian.bruel@st.com>
2919
2920 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2921 optimized non constant lengths.
2922
2923 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2924
2925 PR libgomp/59194
2926 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2927 load as __atomic_load_N if possible.
2928
2929 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2930
2931 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2932 target parameter.
2933 (rs6000_expand_builtin): Adjust call.
2934
2935 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2936
2937 PR target/58115
2938 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2939 * config/rs6000/rs6000.c: Include target-globals.h.
2940 (rs6000_set_current_function): Instead of doing target_reinit
2941 unconditionally, use save_target_globals_default_opts and
2942 restore_target_globals.
2943
2944 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2945 FPSCR.
2946 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2947 (rs6000_expand_builtin): Handle mffs and mtfsf.
2948 (rs6000_init_builtins): Define mffs and mtfsf.
2949 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2950 (rs6000_mffs): New pattern.
2951 (rs6000_mtfsf): New pattern.
2952
2953 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2954
2955 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2956 Start narrowing with START. Apply candidate-use pair
2957 and check overall cost in narrowing.
2958 (iv_ca_prune): Pass new argument.
2959
2960 2014-01-10 Jeff Law <law@redhat.com>
2961
2962 PR middle-end/59743
2963 * ree.c (combine_reaching_defs): Ensure the defining statement
2964 occurs before the extension when optimizing extensions with
2965 different source and destination hard registers.
2966
2967 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2968
2969 PR ipa/58585
2970 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2971 vtables into the type inheritance graph.
2972
2973 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2974
2975 PR rtl-optimization/59754
2976 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2977 modes in the REGNO != REGNO case.
2978
2979 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2980
2981 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2982
2983 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2984
2985 PR tree-optimization/59745
2986 * tree-predcom.c (tree_predictive_commoning_loop): Call
2987 free_affine_expand_cache if giving up because components is NULL.
2988
2989 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2990 GC in payload of target_globals struct instead of allocating them on
2991 the heap and the larger structs separately using GC.
2992 * target-globals.h (struct target_globals): Make regs, hard_regs,
2993 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2994 of GTY((skip)) and change type to void *.
2995 (reset_target_globals): Cast loads from those fields to corresponding
2996 types.
2997
2998 2014-01-10 Steve Ellcey <sellcey@mips.com>
2999
3000 PR plugins/59335
3001 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3002 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3003 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3004
3005 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3006
3007 PR target/59744
3008 * aarch64-modes.def (CC_Zmode): New flags mode.
3009 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3010 represents an equality.
3011 (aarch64_get_condition_code): Handle CC_Zmode.
3012 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3013
3014 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3015
3016 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3017 extraction in good case.
3018
3019 2014-01-10 Richard Biener <rguenther@suse.de>
3020
3021 PR tree-optimization/59374
3022 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3023 checking after SLP discovery. Mark stmts not participating
3024 in any SLP instance properly.
3025
3026 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3027
3028 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3029 when handling a SET rtx.
3030
3031 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3032
3033 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3034 (cortex-a57): Likewise.
3035 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3036
3037 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3038
3039 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3040 non-iwmmxt builtins.
3041
3042 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3043
3044 PR ipa/58252
3045 PR ipa/59226
3046 * ipa-devirt.c record_target_from_binfo): Take as argument
3047 stack of binfos and lookup matching one for virtual inheritance.
3048 (possible_polymorphic_call_targets_1): Update.
3049
3050 2014-01-10 Huacai Chen <chenhc@lemote.com>
3051
3052 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3053 kernel strings for Loongson-2E/2F/3A.
3054
3055 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3056
3057 PR middle-end/59670
3058 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3059 is_gimple_call before calling gimple_call_internal_p.
3060
3061 2014-01-09 Steve Ellcey <sellcey@mips.com>
3062
3063 * Makefile.in (TREE_FLOW_H): Remove.
3064 (TREE_SSA_H): Add file names from tree-flow.h.
3065 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3066 * tree.h: Remove tree-flow.h reference.
3067 * hash-table.h: Remove tree-flow.h reference.
3068 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3069 reference with tree-ssa-loop.h.
3070
3071 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3072
3073 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3074 default element-order behavior for -maltivec.
3075 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3076 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3077 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3078 when targeting big endian, at least for now.
3079 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3080
3081 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3082
3083 PR middle-end/47735
3084 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3085 var satisfies use_register_for_decl, just take into account type
3086 alignment, rather than decl alignment.
3087
3088 PR tree-optimization/59622
3089 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3090 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3091 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3092 Don't devirtualize for inplace at all. For targets.length () == 1,
3093 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3094
3095 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3096
3097 * config/i386/i386.md (cpu): Remove the unused btver1.
3098
3099 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3100
3101 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3102
3103 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3104
3105 PR target/58115
3106 * tree-core.h (struct target_globals): New forward declaration.
3107 (struct tree_target_option): Add globals field.
3108 * tree.h (TREE_TARGET_GLOBALS): Define.
3109 (prepare_target_option_nodes_for_pch): New prototype.
3110 * target-globals.h (struct target_globals): Define even if
3111 !SWITCHABLE_TARGET.
3112 * tree.c (prepare_target_option_node_for_pch,
3113 prepare_target_option_nodes_for_pch): New functions.
3114 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3115 * config/i386/i386.c: Include target-globals.h.
3116 (ix86_set_current_function): Instead of doing target_reinit
3117 unconditionally, use save_target_globals_default_opts and
3118 restore_target_globals.
3119
3120 2014-01-09 Richard Biener <rguenther@suse.de>
3121
3122 PR tree-optimization/59715
3123 * tree-cfg.h (split_critical_edges): Declare.
3124 * tree-cfg.c (split_critical_edges): Export.
3125 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3126
3127 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3128
3129 * cfgexpand.c (expand_stack_vars): Optionally disable
3130 asan stack protection.
3131 (expand_used_vars): Likewise.
3132 (partition_stack_vars): Likewise.
3133 * asan.c (asan_emit_stack_protection): Optionally disable
3134 after return stack usage.
3135 (instrument_derefs): Optionally disable memory access instrumentation.
3136 (instrument_builtin_call): Likewise.
3137 (instrument_strlen_call): Likewise.
3138 (asan_protect_global): Optionally disable global variables protection.
3139 * doc/invoke.texi: Added doc for new options.
3140 * params.def: Added new options.
3141 * params.h: Likewise.
3142
3143 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3144
3145 PR rtl-optimization/59724
3146 * ifcvt.c (cond_exec_process_if_block): Don't call
3147 flow_find_head_matching_sequence with 0 longest_match.
3148 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3149 non-active insns if !stop_after.
3150 (try_head_merge_bb): Revert 2014-01-07 changes.
3151
3152 2014-01-08 Jeff Law <law@redhat.com>
3153
3154 * ree.c (get_sub_rtx): New function, extracted from...
3155 (merge_def_and_ext): Here.
3156 (combine_reaching_defs): Use get_sub_rtx.
3157
3158 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3159
3160 * cgraph.h (varpool_variable_node): Do not choke on null node.
3161
3162 2014-01-08 Catherine Moore <clm@codesourcery.com>
3163
3164 * config/mips/mips.md (simple_return): Attempt to use JRC
3165 for microMIPS.
3166 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3167
3168 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3169
3170 PR rtl-optimization/59137
3171 * reorg.c (steal_delay_list_from_target): Call update_block for
3172 elided insns.
3173 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3174
3175 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3176
3177 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3178 two duplicate entries.
3179
3180 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3181
3182 Revert:
3183 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3184
3185 * config/mips/mips.c (mips_truncated_op_cost): New function.
3186 (mips_rtx_costs): Adjust test for BADDU.
3187 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3188
3189 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3190
3191 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3192 (*baddu_si): ...this new pattern.
3193
3194 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3195
3196 PR ipa/59722
3197 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3198
3199 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3200
3201 PR middle-end/57748
3202 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3203 inner_reference_p.
3204 (expand_expr, expand_normal): Adjust.
3205 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3206 inner_reference_p. Use inner_reference_p to expand inner references.
3207 (store_expr): Adjust.
3208 * cfgexpand.c (expand_call_stmt): Adjust.
3209
3210 2014-01-08 Rong Xu <xur@google.com>
3211
3212 * gcov-io.c (gcov_var): Move from gcov-io.h.
3213 (gcov_position): Ditto.
3214 (gcov_is_error): Ditto.
3215 (gcov_rewrite): Ditto.
3216 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3217 only part to libgcc/libgcov.h.
3218
3219 2014-01-08 Marek Polacek <polacek@redhat.com>
3220
3221 PR middle-end/59669
3222 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3223
3224 2014-01-08 Marek Polacek <polacek@redhat.com>
3225
3226 PR sanitizer/59667
3227 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3228
3229 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3230
3231 PR rtl-optimization/59649
3232 * stor-layout.c (get_mode_bounds): For BImode return
3233 0 and STORE_FLAG_VALUE.
3234
3235 2014-01-08 Richard Biener <rguenther@suse.de>
3236
3237 PR middle-end/59630
3238 * gimple.h (is_gimple_builtin_call): Remove.
3239 (gimple_builtin_call_types_compatible_p): New.
3240 (gimple_call_builtin_p): New overload.
3241 * gimple.c (is_gimple_builtin_call): Remove.
3242 (validate_call): Rename to ...
3243 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3244 check return types.
3245 (validate_type): New static function.
3246 (gimple_call_builtin_p): New overload and adjust.
3247 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3248 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3249 (gimple_fold_stmt_to_constant_1): Likewise.
3250 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3251
3252 2014-01-08 Richard Biener <rguenther@suse.de>
3253
3254 PR middle-end/59471
3255 * gimplify.c (gimplify_expr): Gimplify register-register type
3256 VIEW_CONVERT_EXPRs to separate stmts.
3257
3258 2014-01-07 Jeff Law <law@redhat.com>
3259
3260 PR middle-end/53623
3261 * ree.c (combine_set_extension): Handle case where source
3262 and destination registers in an extension insn are different.
3263 (combine_reaching_defs): Allow source and destination registers
3264 in extension to be different under limited circumstances.
3265 (add_removable_extension): Remove restriction that the
3266 source and destination registers in the extension are the same.
3267 (find_and_remove_re): Emit a copy from the extension's
3268 destination to its source after the defining insn if
3269 the source and destination registers are different.
3270
3271 PR middle-end/59285
3272 * ifcvt.c (merge_if_block): If we are merging a block with more than
3273 one successor with a block with no successors, remove any BARRIER
3274 after the second block.
3275
3276 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3277
3278 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3279
3280 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3281
3282 PR target/59652
3283 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3284 for 14-bit register offsets when INT14_OK_STRICT is false.
3285
3286 2014-01-07 Roland Stigge <stigge@antcom.de>
3287 Michael Meissner <meissner@linux.vnet.ibm.com>
3288
3289 PR 57386/target
3290 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3291 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3292
3293 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3294
3295 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3296 -mcpu.
3297
3298 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3299
3300 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3301 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3302 rtx is const0_rtx or not.
3303
3304 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3305
3306 PR target/58115
3307 * target-globals.c (save_target_globals): Remove this_fn_optab
3308 handling.
3309 * toplev.c: Include optabs.h.
3310 (target_reinit): Temporarily restore the global options if another
3311 set of options are in force.
3312
3313 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3314
3315 PR rtl-optimization/58668
3316 * cfgcleanup.c (flow_find_cross_jump): Don't count
3317 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3318 to determine what is counted.
3319 (flow_find_head_matching_sequence): Use active_insn_p to determine
3320 what is counted.
3321 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3322 counting change.
3323 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3324 determine what is counted.
3325
3326 PR tree-optimization/59643
3327 * tree-predcom.c (split_data_refs_to_components): If one dr is
3328 read and one write, determine_offset fails and the write isn't
3329 in the bad component, just put the read into the bad component.
3330
3331 2014-01-07 Mike Stump <mikestump@comcast.net>
3332 Jakub Jelinek <jakub@redhat.com>
3333
3334 PR pch/59436
3335 * tree-core.h (struct tree_optimization_option): Change optabs
3336 type from unsigned char * to void *.
3337 * optabs.c (init_tree_optimization_optabs): Adjust
3338 TREE_OPTIMIZATION_OPTABS initialization.
3339
3340 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3341
3342 PR target/59644
3343 * config/i386/i386.h (struct machine_function): Add
3344 no_drap_save_restore field.
3345 * config/i386/i386.c (ix86_save_reg): Use
3346 !cfun->machine->no_drap_save_restore instead of
3347 crtl->stack_realign_needed.
3348 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3349 this function clears frame_pointer_needed. Set
3350 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3351 and DRAP reg is needed.
3352
3353 2014-01-06 Marek Polacek <polacek@redhat.com>
3354
3355 PR c/57773
3356 * doc/implement-c.texi: Mention that other integer types are
3357 permitted as bit-field types in strictly conforming mode.
3358
3359 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3360
3361 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3362 is newly allocated.
3363
3364 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3365
3366 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3367
3368 2014-01-06 Martin Jambor <mjambor@suse.cz>
3369
3370 PR ipa/59008
3371 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3372 to int.
3373 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3374
3375 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3376
3377 PR debug/59350
3378 PR debug/59510
3379 * var-tracking.c (add_stores): Preserve the value of the source even if
3380 we don't record the store.
3381
3382 2014-01-06 Terry Guo <terry.guo@arm.com>
3383
3384 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3385
3386 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3387
3388 PR bootstrap/59541
3389 * config/darwin.c (darwin_function_section): Adjust return values to
3390 correspond to optimisation changes made in r206070.
3391
3392 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3393
3394 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3395 from prefetch_block tune setting.
3396 (nocona_cost): Correct size of prefetch block to 64.
3397
3398 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3399
3400 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3401 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3402 used to save the static chain register in the computation of the offset
3403 from which the FP registers need to be restored.
3404
3405 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3406
3407 PR tree-optimization/59519
3408 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3409 ICE if get_current_def (current_new_name) is already non-NULL, as long
3410 as it is a phi result of some other phi in *new_exit_bb that has
3411 the same argument.
3412
3413 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3414 or vmovdqu* for misaligned_operand.
3415 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3416 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3417 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3418 aligned_mem for AVX512F masked aligned load and store builtins and for
3419 non-temporal moves.
3420
3421 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3422
3423 PR tree-optimization/59651
3424 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3425 Address range for negative step should be added by TYPE_SIZE_UNIT.
3426
3427 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3428
3429 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3430 overlapping registers also for registers other than the stack pointer.
3431
3432 2014-01-03 Marek Polacek <polacek@redhat.com>
3433
3434 PR other/59661
3435 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3436 __builtin_FILE.
3437
3438 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3439
3440 PR target/59625
3441 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3442 asm goto as jump.
3443
3444 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3445 (push splitter): Use <P:MODE_SIZE> instead of
3446 GET_MODE_SIZE (<P:MODE>mode).
3447 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3448 (mov -1, reg peephole2): Likewise.
3449 * config/i386/sse.md (*mov<mode>_internal,
3450 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3451 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3452 *<code><mode>3, *andnot<mode>3<mask_name>,
3453 <mask_codefor><code><mode>3<mask_name>): Likewise.
3454 * config/i386/subst.md (mask_mode512bit_condition,
3455 sd_mask_mode512bit_condition): Likewise.
3456
3457 2014-01-02 Xinliang David Li <davidxl@google.com>
3458
3459 PR tree-optimization/59303
3460 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3461 (dump_predicates): Better output format.
3462 (pred_equal_p): New function.
3463 (is_neq_relop_p): Ditto.
3464 (is_neq_zero_form_p): Ditto.
3465 (pred_expr_equal_p): Ditto.
3466 (pred_neg_p): Ditto.
3467 (simplify_pred): Ditto.
3468 (simplify_preds_2): Ditto.
3469 (simplify_preds_3): Ditto.
3470 (simplify_preds_4): Ditto.
3471 (simplify_preds): Ditto.
3472 (push_pred): Ditto.
3473 (push_to_worklist): Ditto.
3474 (get_pred_info_from_cmp): Ditto.
3475 (is_degenerated_phi): Ditto.
3476 (normalize_one_pred_1): Ditto.
3477 (normalize_one_pred): Ditto.
3478 (normalize_one_pred_chain): Ditto.
3479 (normalize_preds): Ditto.
3480 (normalize_cond_1): Remove function.
3481 (normalize_cond): Ditto.
3482 (is_gcond_subset_of): Ditto.
3483 (is_subset_of_any): Ditto.
3484 (is_or_set_subset_of): Ditto.
3485 (is_and_set_subset_of): Ditto.
3486 (is_norm_cond_subset_of): Ditto.
3487 (pred_chain_length_cmp): Ditto.
3488 (convert_control_dep_chain_into_preds): Type change.
3489 (find_predicates): Ditto.
3490 (find_def_preds): Ditto.
3491 (destroy_predicates_vecs): Ditto.
3492 (find_matching_predicates_in_rest_chains): Ditto.
3493 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3494 (is_pred_expr_subset): Ditto.
3495 (is_pred_chain_subset_of): Ditto.
3496 (is_included_in): Ditto.
3497 (is_superset_of): Ditto.
3498
3499 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3500
3501 Update copyright years.
3502
3503 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3504
3505 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3506 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3507 config/arc/arc.md, config/arc/arc.opt,
3508 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3509 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3510 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3511 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3512 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3513 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3514 vtable-verify.c, vtable-verify.h: Use the standard form for the
3515 copyright notice.
3516
3517 2014-01-02 Tobias Burnus <burnus@net-b.de>
3518
3519 * gcc.c (process_command): Update copyright notice dates.
3520 * gcov-dump.c: Ditto.
3521 * gcov.c: Ditto.
3522 * doc/cpp.texi: Bump @copying's copyright year.
3523 * doc/cppinternals.texi: Ditto.
3524 * doc/gcc.texi: Ditto.
3525 * doc/gccint.texi: Ditto.
3526 * doc/gcov.texi: Ditto.
3527 * doc/install.texi: Ditto.
3528 * doc/invoke.texi: Ditto.
3529
3530 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3531
3532 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3533
3534 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3535
3536 * config/i386/sse.md (*mov<mode>_internal): Guard
3537 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3538
3539 PR rtl-optimization/59647
3540 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3541 new_rtx into UNSIGNED_FLOAT rtxes.
3542 \f
3543 Copyright (C) 2014 Free Software Foundation, Inc.
3544
3545 Copying and distribution of this file, with or without modification,
3546 are permitted in any medium without royalty provided the copyright
3547 notice and this notice are preserved.
This page took 0.180778 seconds and 6 git commands to generate.