]> gcc.gnu.org Git - gcc.git/blob - gcc/ChangeLog
tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
[gcc.git] / gcc / ChangeLog
1 2014-02-17 Richard Biener <rguenther@suse.de>
2
3 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
4
5 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
6 Ilya Tocar <ilya.tocar@intel.com>
7
8 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
9 arguments order in builtin.
10 (_mm512_permutexvar_epi64): Ditto.
11 (_mm512_mask_permutexvar_epi64): Ditto
12 (_mm512_maskz_permutexvar_epi32): Ditto
13 (_mm512_permutexvar_epi32): Ditto
14 (_mm512_mask_permutexvar_epi32): Ditto
15
16 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17
18 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian
19 targets.
20 (p8_vmrgow): Likewise.
21
22 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23
24 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
25 endian targets.
26
27 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28
29 PR target/60203
30 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
31 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
32 into 64-bit and 32-bit moves. On 64-bit moves, add support for
33 using direct move instructions on ISA 2.07. Also adjust
34 instruction length for 64-bit.
35 (mov<mode>_64bit, TFmode/TDmode): Likewise.
36 (mov<mode>_32bit, TFmode/TDmode): Likewise.
37
38 2014-02-15 Alan Modra <amodra@gmail.com>
39
40 PR target/58675
41 PR target/57935
42 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
43 find_replacement on parts of insn rtl that might be reloaded.
44
45 2014-02-15 Richard Biener <rguenther@suse.de>
46
47 PR tree-optimization/60183
48 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating
49 loads.
50 (tree_ssa_phiprop): Calculate and free post-dominators.
51
52 2014-02-14 Jeff Law <law@redhat.com>
53
54 PR rtl-optimization/60131
55 * ree.c (get_extended_src_reg): New function.
56 (combine_reaching_defs): Use it rather than assuming location
57 of REG.
58 (find_and_remove_re): Verify first operand of extension is
59 a REG before adding the insns to the copy list.
60
61 2014-02-14 Roland McGrath <mcgrathr@google.com>
62
63 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
64 * configure: Regenerated.
65 * config.in: Regenerated.
66 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
67 instead of ASM_SHORT.
68
69 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
70 Richard Earnshaw <rearnsha@arm.com>
71
72 PR rtl-optimization/59535
73 * lra-constraints.c (process_alt_operands): Encourage alternative
74 when unassigned pseudo class is superset of the alternative class.
75 (inherit_reload_reg): Don't inherit when optimizing for code size.
76 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
77 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
78 modes not less than 4 for Thumb1.
79
80 2014-02-14 Kyle McMartin <kyle@redhat.com>
81
82 PR pch/60010
83 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
84
85 2014-02-14 Richard Biener <rguenther@suse.de>
86
87 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
88 (get_frame_arg): Drop the assert with langhook types_compatible_p.
89 Do not strip INDIRECT_REFs.
90
91 2014-02-14 Richard Biener <rguenther@suse.de>
92
93 PR lto/60179
94 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
95 DECL_FUNCTION_SPECIFIC_TARGET.
96 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
97 * tree-streamer-out.c (pack_ts_target_option): Remove.
98 (streamer_pack_tree_bitfields): Do not stream
99 TS_TARGET_OPTION.
100 (write_ts_function_decl_tree_pointers): Do not stream
101 DECL_FUNCTION_SPECIFIC_TARGET.
102 * tree-streamer-in.c (unpack_ts_target_option): Remove.
103 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
104 (lto_input_ts_function_decl_tree_pointers): Do not stream
105 DECL_FUNCTION_SPECIFIC_TARGET.
106
107 2014-02-14 Jakub Jelinek <jakub@redhat.com>
108
109 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
110 use_stmt twice.
111 (get_initial_def_for_induction, vectorizable_induction): Ignore
112 debug stmts when looking for exit_phi.
113 (vectorizable_live_operation): Fix up condition.
114
115 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
116
117 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
118 nreverse() because it changes the content of original tree list.
119
120 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
121
122 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
123 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
124
125 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
126
127 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
128 GNU coding standards.
129
130 2014-02-13 Jakub Jelinek <jakub@redhat.com>
131
132 PR debug/60152
133 * dwarf2out.c (gen_subprogram_die): Don't call
134 add_calling_convention_attribute if subr_die is old_die.
135
136 2014-02-13 Sharad Singhai <singhai@google.com>
137
138 * doc/optinfo.texi: Fix order of nodes.
139
140 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
141
142 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
143 operands[2], not operands[3].
144
145 2014-02-13 Richard Biener <rguenther@suse.de>
146
147 PR bootstrap/59878
148 * doc/install.texi (ISL): Update recommended version to 0.12.2,
149 mention the possibility of an in-tree build.
150 (CLooG): Update recommended version to 0.18.1, mention the
151 possibility of an in-tree build and clarify that the ISL
152 bundled with CLooG does not work.
153
154 2014-02-13 Jakub Jelinek <jakub@redhat.com>
155
156 PR target/43546
157 * expr.c (compress_float_constant): If x is a hard register,
158 extend into a pseudo and then move to x.
159
160 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
161
162 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
163 caused by bad second argument to warning_at() with -mhotpatch and
164 nested functions (e.g. with gfortran).
165
166 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
167
168 * opts.c (option_name): Remove "enabled by default" rider.
169
170 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
171
172 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
173
174 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
175 Uros Bizjak <ubizjak@gmail.com>
176
177 PR target/60151
178 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
179 * configure: Regenerated.
180
181 2014-02-12 Richard Biener <rguenther@suse.de>
182
183 * vec.c (vec_prefix::calculate_allocation): Move as
184 inline variant to vec.h.
185 (vec_prefix::calculate_allocation_1): New out-of-line version.
186 * vec.h (vec_prefix::calculate_allocation_1): Declare.
187 (vec_prefix::m_has_auto_buf): Rename to ...
188 (vec_prefix::m_using_auto_storage): ... this.
189 (vec_prefix::calculate_allocation): Inline the easy cases
190 and dispatch to calculate_allocation_1 which doesn't need the
191 prefix address.
192 (va_heap::reserve): Use gcc_checking_assert.
193 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
194 m_using_auto_storage.
195 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
196 member and adjust.
197 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
198 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
199 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
200
201 2014-02-12 Richard Biener <rguenther@suse.de>
202
203 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
204 when we found a dependence.
205
206 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
207
208 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
209 common code...
210 (maybe_fold_stmt): ... into this new function.
211 * omp-low.c (lower_omp): Update comment.
212
213 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
214 last use.
215
216 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
217 dereference.
218
219 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
220
221 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
222 identifiers in comments.
223 (cortexa53_extra_costs): Likewise.
224 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
225 (cortexa7_extra_costs): Likewise.
226 (cortexa12_extra_costs): Likewise.
227 (cortexa15_extra_costs): Likewise.
228 (v7m_extra_costs): Likewise.
229
230 2014-02-12 Richard Biener <rguenther@suse.de>
231
232 PR middle-end/60092
233 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
234 of posix_memalign being successful.
235 (lower_stmt): Restrict lowering of posix_memalign to when
236 -ftree-bit-ccp is enabled.
237
238 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
239
240 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
241 arg_loc.
242 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
243
244 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
245
246 PR rtl-optimization/60116
247 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
248 other_insn once the combination has been validated.
249
250 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
251
252 PR lto/59468
253 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
254 and wrapper.
255 * ipa-devirt.c: Include demangle.h
256 (odr_violation_reported): New static variable.
257 (add_type_duplicate): Update odr_violations.
258 (maybe_record_node): Add completep parameter; update it.
259 (record_target_from_binfo): Add COMPLETEP parameter;
260 update it as needed.
261 (possible_polymorphic_call_targets_1): Likewise.
262 (struct polymorphic_call_target_d): Add nonconstruction_targets;
263 rename FINAL to COMPLETE.
264 (record_targets_from_bases): Sanity check we found the binfo;
265 fix COMPLETEP updating.
266 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
267 parameter, fix computing of COMPLETEP.
268 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
269 at LTO time do demangling.
270 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
271 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
272 parameter.
273 (gimple_get_virt_method_for_binfo): Likewise.
274 * gimple-fold.h (gimple_get_virt_method_for_binfo,
275 gimple_get_virt_method_for_vtable): Update prototypes.
276
277 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
278
279 PR target/49008
280 * genautomata.c (add_presence_absence): Fix typo with
281 {final_}presence_list.
282
283 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
284
285 PR target/60137
286 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
287 for VSX/Altivec vectors that land in GPR registers.
288
289 2014-02-11 Richard Henderson <rth@redhat.com>
290 Jakub Jelinek <jakub@redhat.com>
291
292 PR debug/59776
293 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
294 around drhs if type conversion to lacc->type is not useless.
295
296 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
297
298 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
299 tuning struct.
300 (cortex-a57.cortex-a53): Likewise.
301 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
302
303 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
304
305 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
306 arm_restrict_it.
307
308 2014-02-11 Renlin Li <Renlin.Li@arm.com>
309
310 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
311 add_options_for_arm_vfp3.
312
313 2014-02-11 Jeff Law <law@redhat.com>
314
315 PR middle-end/54041
316 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
317 object with an undesirable mode.
318
319 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
320
321 PR libgomp/60107
322 * config/i386/sol2-9.h: New file.
323 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
324 *-*-solaris2.9*): Use it.
325
326 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
327
328 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
329 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
330
331 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
332
333 * config/microblaze/microblaze.c: Extend mcpu version format
334
335 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
336
337 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
338
339 2014-02-10 Richard Henderson <rth@redhat.com>
340
341 PR target/59927
342 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
343 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
344 ms-abi vs -mno-accumulate-outgoing-args.
345 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
346 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
347 respect to ms-abi.
348
349 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
350
351 PR middle-end/60080
352 * cfgexpand.c (expand_asm_operands): Attach source location to
353 ASM_INPUT rtx objects.
354 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
355
356 2014-02-10 Nick Clifton <nickc@redhat.com>
357
358 * config/mn10300/mn10300.c (popcount): New function.
359 (mn10300_expand_prologue): Include saved registers in stack usage
360 count.
361
362 2014-02-10 Jeff Law <law@redhat.com>
363
364 PR middle-end/52306
365 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
366 when changing the SET_DEST of a prior insn to avoid an input reload.
367
368 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
369
370 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
371 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
372 -mcall-openbsd, or -mcall-linux.
373 (CC1_ENDIAN_BIG_SPEC): Remove.
374 (CC1_ENDIAN_LITTLE_SPEC): Remove.
375 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
376 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
377 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
378 and %cc1_endian_default.
379 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
380
381 2014-02-10 Richard Biener <rguenther@suse.de>
382
383 PR tree-optimization/60115
384 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
385 MEM_REF handling. Properly verify that the accesses are not
386 out of the objects bound.
387
388 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
389
390 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
391 coretex to cortex.
392
393 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
394
395 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
396 proper constants and fix formatting.
397 (possible_polymorphic_call_targets): Fix formatting.
398
399 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
400 Ilya Tocar <ilya.tocar@intel.com>
401
402 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
403 (_mm512_loadu_epi32): Renamed into...
404 (_mm512_loadu_si512): This.
405 (_mm512_storeu_epi32): Renamed into...
406 (_mm512_storeu_si512): This.
407 (_mm512_maskz_ceil_ps): Removed.
408 (_mm512_maskz_ceil_pd): Ditto.
409 (_mm512_maskz_floor_ps): Ditto.
410 (_mm512_maskz_floor_pd): Ditto.
411 (_mm512_floor_round_ps): Ditto.
412 (_mm512_floor_round_pd): Ditto.
413 (_mm512_ceil_round_ps): Ditto.
414 (_mm512_ceil_round_pd): Ditto.
415 (_mm512_mask_floor_round_ps): Ditto.
416 (_mm512_mask_floor_round_pd): Ditto.
417 (_mm512_mask_ceil_round_ps): Ditto.
418 (_mm512_mask_ceil_round_pd): Ditto.
419 (_mm512_maskz_floor_round_ps): Ditto.
420 (_mm512_maskz_floor_round_pd): Ditto.
421 (_mm512_maskz_ceil_round_ps): Ditto.
422 (_mm512_maskz_ceil_round_pd): Ditto.
423 (_mm512_expand_pd): Ditto.
424 (_mm512_expand_ps): Ditto.
425 * config/i386/i386.c (ix86_builtins): Remove
426 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
427 (bdesc_args): Ditto.
428 * config/i386/predicates.md (const1256_operand): New.
429 (const_1_to_2_operand): Ditto.
430 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
431 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
432 (*avx512pf_gatherpf<mode>sf): Ditto.
433 (avx512pf_gatherpf<mode>df): Ditto.
434 (*avx512pf_gatherpf<mode>df_mask): Ditto.
435 (*avx512pf_gatherpf<mode>df): Ditto.
436 (avx512pf_scatterpf<mode>sf): Ditto.
437 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
438 (*avx512pf_scatterpf<mode>sf): Ditto.
439 (avx512pf_scatterpf<mode>df): Ditto.
440 (*avx512pf_scatterpf<mode>df_mask): Ditto.
441 (*avx512pf_scatterpf<mode>df): Ditto.
442 (avx512f_expand<mode>): Removed.
443 (<shift_insn><mode>3<mask_name>): Change predicate type.
444
445 2014-02-08 Jakub Jelinek <jakub@redhat.com>
446
447 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
448 not at the end of datarefs vector use ordered_remove to avoid
449 reordering datarefs vector.
450
451 PR c/59984
452 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
453 mark local addressable non-static vars as GOVD_PRIVATE
454 instead of GOVD_LOCAL.
455 * omp-low.c (lower_omp_for): Move gimple_bind_vars
456 and BLOCK_VARS of gimple_bind_block to new_stmt rather
457 than copying them.
458
459 PR middle-end/60092
460 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
461 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
462 assume_aligned or alloc_align attributes.
463 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
464 arguments. Handle also assume_aligned and alloc_align attributes.
465 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
466 calls to functions with assume_aligned or alloc_align attributes.
467 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
468
469 2014-02-08 Terry Guo <terry.guo@arm.com>
470
471 * doc/invoke.texi: Document ARM -march=armv7e-m.
472
473 2014-02-08 Jakub Jelinek <jakub@redhat.com>
474
475 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
476 flag on __cilkrts_rethrow builtin.
477
478 PR ipa/60026
479 * ipa-cp.c (determine_versionability): Fail at -O0
480 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
481 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
482
483 Revert:
484 2014-02-04 Jakub Jelinek <jakub@redhat.com>
485
486 PR ipa/60026
487 * tree-inline.c (copy_forbidden): Fail for
488 __attribute__((optimize (0))) functions.
489
490 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
491
492 * varpool.c: Include pointer-set.h.
493 (varpool_remove_unreferenced_decls): Variables in other partitions
494 will not be output; be however careful to not lose information
495 about partitioning.
496
497 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
498
499 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
500 lookup in the vtable constructor.
501
502 2014-02-07 Jeff Law <law@redhat.com>
503
504 PR target/40977
505 * config/m68k/m68k.md (ashldi_extsi): Turn into a
506 define_insn_and_split.
507
508 * ipa-inline.c (inline_small_functions): Fix typos.
509
510 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
511
512 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
513 (s390_can_use_return_insn): Declare.
514 * config/s390/s390.h (EPILOGUE_USES): Define.
515 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
516 instructions.
517 (s390_chunkify_start): Handle return JUMP_LABELs.
518 (s390_early_mach): Emit a main_pool instruction on the entry edge.
519 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
520 (s390_can_use_return_insn): New functions.
521 (s390_fix_long_loop_prediction): Handle conditional returns.
522 (TARGET_SET_UP_BY_PROLOGUE): Define.
523 * config/s390/s390.md (ANY_RETURN): New code iterator.
524 (*creturn, *csimple_return, return, simple_return): New patterns.
525
526 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
527
528 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
529 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
530 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
531 REG_CFA_RESTORE list when deciding not to restore a register.
532
533 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
534
535 * config/s390/s390.c: Include tree-pass.h and context.h.
536 (s390_early_mach): New function, split out from...
537 (s390_emit_prologue): ...here.
538 (pass_data_s390_early_mach): New pass structure.
539 (pass_s390_early_mach): New class.
540 (s390_option_override): Create and register early_mach pass.
541 Move to end of file.
542
543 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
544
545 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
546 to match for the exit block.
547
548 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
549
550 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
551 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
552 Reject misaligned operands.
553
554 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
555
556 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
557
558 2014-02-07 Richard Biener <rguenther@suse.de>
559
560 PR middle-end/60092
561 * gimple-low.c (lower_builtin_posix_memalign): New function.
562 (lower_stmt): Call it to lower posix_memalign in a way
563 to make alignment info accessible.
564
565 2014-02-07 Jakub Jelinek <jakub@redhat.com>
566
567 PR c++/60082
568 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
569 __builtin_setjmp_receiver.
570
571 2014-02-07 Richard Biener <rguenther@suse.de>
572
573 PR middle-end/60092
574 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
575 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
576 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
577 Handle BUILT_IN_POSIX_MEMALIGN.
578 (find_func_clobbers): Likewise.
579 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
580 (call_may_clobber_ref_p_1): Likewise.
581
582 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
583
584 PR ipa/59918
585 * ipa-devirt.c (record_target_from_binfo): Remove overactive
586 sanity check.
587
588 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
589
590 PR ipa/59469
591 * lto-cgraph.c (lto_output_node): Use
592 symtab_get_symbol_partitioning_class.
593 (lto_output_varpool_node): likewise.
594 (symtab_get_symbol_partitioning_class): Move here from
595 lto/lto-partition.c
596 * cgraph.h (symbol_partitioning_class): Likewise.
597 (symtab_get_symbol_partitioning_class): Declare.
598
599 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
600
601 * ggc.h (ggc_internal_cleared_alloc): New macro.
602 * vec.h (vec_safe_copy): Handle memory stats.
603 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
604 * target-globals.c (save_target_globals): Likewise.
605
606 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
607
608 PR target/60077
609 * expr.c (emit_move_resolve_push): Export; be bit more selective
610 on when to clear alias set.
611 * expr.h (emit_move_resolve_push): Declare.
612 * function.h (struct function): Add tail_call_marked.
613 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
614 * config/i386/i386-protos.h (ix86_expand_push): Remove.
615 * config/i386/i386.md (TImode move expander): De not call
616 ix86_expand_push.
617 (FP push expanders): Preserve memory attributes.
618 * config/i386/sse.md (push<mode>1): Remove.
619 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
620 (ix86_expand_push): Remove.
621 * config/i386/mmx.md (push<mode>1): Remove.
622
623 2014-02-06 Jakub Jelinek <jakub@redhat.com>
624
625 PR rtl-optimization/60030
626 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
627 lopart with paradoxical subreg before shifting it up by hprec.
628
629 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
630
631 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
632 Remove extra newline at end of file.
633 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
634 (arm_issue_rate): Handle cortexa57.
635 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
636 (cortex-a57.cortex-a53): Likewise.
637
638 2014-02-06 Jakub Jelinek <jakub@redhat.com>
639
640 PR target/59575
641 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
642 don't record in REG_FRAME_RELATED_EXPR registers not set in that
643 bitmask.
644 (arm_expand_prologue): Adjust all callers.
645 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
646 info, registers also at the lowest numbered registers side. Use
647 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
648 XEXP.
649
650 PR debug/59992
651 * var-tracking.c (adjust_mems): Before adding a SET to
652 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
653
654 2014-02-06 Alan Modra <amodra@gmail.com>
655
656 PR target/60032
657 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
658 change SDmode to DDmode when lra_in_progress.
659
660 2014-02-06 Jakub Jelinek <jakub@redhat.com>
661
662 PR middle-end/59150
663 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
664 free_data_ref on the dr first, and before goto again also set dr
665 to the next dr. For simd_lane_access, free old datarefs[i] before
666 overwriting it. For get_vectype_for_scalar_type failure, don't
667 free_data_ref if simd_lane_access.
668
669 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
670
671 PR target/60062
672 * tree.h (opts_for_fn): New inline function.
673 (opt_for_fn): Define.
674 * config/i386/i386.c (ix86_function_regparm): Use
675 opt_for_fn (decl, optimize) instead of optimize.
676
677 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
678
679 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
680 for SYMBOL_REF in large memory model.
681
682 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
683
684 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
685 and crypto support.
686 (cortex-a57): Likewise.
687 (cortex-a57.cortex-a53): Likewise.
688
689 2014-02-06 Yury Gribov <y.gribov@samsung.com>
690 Kugan Vivekanandarajah <kuganv@linaro.org>
691
692 * config/arm/arm.c (arm_vector_alignment_reachable): Check
693 unaligned_access.
694 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
695
696 2014-02-06 Richard Biener <rguenther@suse.de>
697
698 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
699 set_loop_copy and initialize_original_copy_tables.
700
701 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
702
703 * config/aarch64/aarch64-simd.md
704 (aarch64_ashr_simddi): Change QI to SI.
705
706 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
707 Jakub Jelinek <jakub@redhat.com>
708
709 PR middle-end/60013
710 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
711 of the dataflow.
712
713 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
714
715 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
716 CODE_FOR_altivec_vpku[hw]um to
717 CODE_FOR_altivec_vpku[hw]um_direct.
718 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
719 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
720 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
721 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
722
723 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
724
725 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
726 generation for -maltivec=be.
727 (altivec_vsumsws): Simplify redundant test.
728
729 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
730
731 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
732 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
733 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
734 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
735 gen_altivec_vpkuwum.
736 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
737 BYTES_BIG_ENDIAN.
738 (altivec_vpks<VI_char>ss): Likewise.
739 (altivec_vpks<VI_char>us): Likewise.
740 (altivec_vpku<VI_char>us): Likewise.
741 (altivec_vpku<VI_char>um): Likewise.
742 (altivec_vpku<VI_char>um_direct): New (copy of
743 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
744 internal use).
745 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
746 target is little endian and -maltivec=be is not specified.
747 (*altivec_vupkhs<VU_char>_direct): New (copy of
748 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
749 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
750 target is little endian and -maltivec=be is not specified.
751 (*altivec_vupkls<VU_char>_direct): New (copy of
752 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
753 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
754 little endian and -maltivec=be is not specified.
755 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
756 little endian and -maltivec=be is not specified.
757
758 2014-02-05 Richard Henderson <rth@redhat.com>
759
760 PR debug/52727
761 * combine-stack-adj.c: Revert r206943.
762 * sched-int.h (struct deps_desc): Add last_args_size.
763 * sched-deps.c (init_deps): Initialize it.
764 (sched_analyze_insn): Add OUTPUT dependencies between insns that
765 contain REG_ARGS_SIZE notes.
766
767 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
768
769 * lto-cgraph.c (asm_nodes_output): Make global.
770 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
771 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
772 (driver_handle_option): Handle OPT_fwpa.
773
774 2014-02-05 Jakub Jelinek <jakub@redhat.com>
775
776 PR ipa/59947
777 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
778 a comment typo and formatting issue. If odr_hash hasn't been
779 created, return vNULL and set *completep to false.
780
781 PR middle-end/57499
782 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
783 bb with no successors.
784
785 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
786
787 PR target/59718
788 * doc/invoke.texi (-march): Clarify documentation for ARM.
789 (-mtune): Likewise.
790 (-mcpu): Likewise.
791
792 2014-02-05 Richard Biener <rguenther@suse.de>
793
794 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
795 when not vectorizing because of too many alias checks.
796 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
797 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
798
799 2014-02-05 Nick Clifton <nickc@redhat.com>
800
801 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
802 accept extended registers in any mode when compiling for the MN10300.
803
804 2014-02-05 Yury Gribov <y.gribov@samsung.com>
805
806 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
807 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
808 sanitization attributes.
809 (can_inline_edge_p): Likewise.
810 (sanitize_attrs_match_for_inline_p): New function.
811
812 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
813
814 * ipa-prop.c (detect_type_change): Shor circuit testing of
815 type changes on THIS pointer.
816
817 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
818
819 PR target/59777
820 * config/pa/pa.c (legitimize_tls_address): Return original address
821 if not passed a SYMBOL_REF rtx.
822 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
823 addresses.
824 (pa_emit_move_sequence): Simplify TLS source operands.
825 (pa_legitimate_constant_p): Reject all TLS constants.
826 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
827 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
828
829 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
830
831 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
832 groups when we know they are controlled by LTO.
833 * varasm.c (default_binds_local_p_1): If object is in other partition,
834 it will be resolved locally.
835
836 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
837
838 * config/host-linux.c (linux_gt_pch_use_address): Don't
839 use SSIZE_MAX because it is not always defined.
840
841 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
842
843 PR bootstrap/59913
844 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
845 threshold for pseudo splitting.
846 (update_ebb_live_info): Process call argument hard registers and
847 hard registers from insn definition too.
848 (max_small_class_regs_num): New constant.
849 (inherit_in_ebb): Update live hard regs through EBBs. Update
850 reloads_num only for small register classes. Don't split for
851 outputs of jumps.
852
853 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
854
855 PR ipa/60058
856 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
857 is non-null.
858
859 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
860
861 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
862 visibility is safe.
863
864 2014-02-04 Marek Polacek <polacek@redhat.com>
865
866 * gdbinit.in (pel): Define.
867
868 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
869
870 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
871 behavior.
872
873 2014-02-04 Richard Biener <rguenther@suse.de>
874
875 PR lto/59723
876 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
877 in function context local.
878 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
879 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
880 similar to LTO_imported_decl_ref.
881
882 2014-02-04 Jakub Jelinek <jakub@redhat.com>
883
884 PR tree-optimization/60002
885 * cgraphclones.c (build_function_decl_skip_args): Clear
886 DECL_LANG_SPECIFIC.
887
888 PR tree-optimization/60023
889 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
890 false to gsi_replace.
891 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
892 has been in some EH region and vec_stmt could throw, add
893 vec_stmt into the same EH region.
894 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
895 has no lhs, ignore it.
896 * internal-fn.c (expand_MASK_LOAD): Likewise.
897
898 PR ipa/60026
899 * tree-inline.c (copy_forbidden): Fail for
900 __attribute__((optimize (0))) functions.
901
902 PR other/58712
903 * omp-low.c (simd_clone_struct_copy): If from->inbranch
904 is set, copy one less argument.
905 (expand_simd_clones): Don't subtract clone_info->inbranch
906 from simd_clone_struct_alloc argument.
907
908 PR rtl-optimization/57915
909 * recog.c (simplify_while_replacing): If all unary/binary/relational
910 operation arguments are constant, attempt to simplify those.
911
912 PR middle-end/59261
913 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
914 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
915
916 2014-02-04 Richard Biener <rguenther@suse.de>
917
918 PR tree-optimization/60012
919 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
920 TBAA disambiguation to all DDRs.
921
922 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
923
924 PR target/59788
925 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
926 (LINK_SPEC): Use it for -shared, -shared-libgcc.
927
928 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
929
930 PR ipa/59882
931 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
932
933 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
934
935 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
936 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
937
938 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
939
940 PR ipa/59831
941 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
942 to figure out targets of polymorphic calls with known decl.
943 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
944 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
945 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
946 (get_polymorphic_call_info): ... here.
947 (get_polymorphic_call_info_from_invariant): New function.
948
949 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
950
951 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
952 lookup via vtable pointer; check for type consistency
953 and turn inconsitent facts into UNREACHABLE.
954 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
955 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
956 type inconsistent querries; return UNREACHABLE instead.
957
958 2014-02-03 Richard Henderson <rth@twiddle.net>
959
960 PR tree-opt/59924
961 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
962 already processed this node.
963 (normalize_one_pred_1): Pass along mark_set.
964 (normalize_one_pred): Create and destroy a pointer_set_t.
965 (normalize_one_pred_chain): Likewise.
966
967 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
968
969 PR gcov-profile/58602
970 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
971
972 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
973
974 PR ipa/59831
975 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
976 -fno-devirtualize; try to devirtualize by the knowledge of
977 virtual table pointer given by aggregate propagation.
978 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
979 (ipa_print_node_jump_functions): Dump also offset that
980 is relevant for polymorphic calls.
981 (determine_known_aggregate_parts): Add arg_type parameter; use it
982 instead of determining the type from pointer type.
983 (ipa_compute_jump_functions_for_edge): Update call of
984 determine_known_aggregate_parts.
985 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
986 (gimple_get_virt_method_for_binfo): ... here; simplify using
987 vtable_pointer_value_to_vtable.
988 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
989 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
990 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
991 (vtable_pointer_value_to_vtable): Break out from ...; handle also
992 POINTER_PLUS_EXPR.
993 (vtable_pointer_value_to_binfo): ... here.
994 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
995
996 2014-02-03 Teresa Johnson <tejohnson@google.com>
997
998 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
999 redef of outer loop index variable.
1000
1001 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1002
1003 PR c++/53017
1004 PR c++/59211
1005 * doc/extend.texi (Function Attributes): Typo.
1006
1007 2014-02-03 Cong Hou <congh@google.com>
1008
1009 PR tree-optimization/60000
1010 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1011 if the vectorized statement is a store. A store statement can only
1012 appear at the end of pattern statements.
1013
1014 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1015
1016 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1017 (ix86_option_override_internal): Default long double to 64-bit for
1018 32-bit Bionic and to 128-bit for 64-bit Bionic.
1019
1020 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1021 TARGET_LONG_DOUBLE_128 is true.
1022 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1023
1024 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1025 (mlong-double-64): Negate -mlong-double-128.
1026 (mlong-double-128): New option.
1027
1028 * config/i386/i386-c.c (ix86_target_macros): Define
1029 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1030
1031 * doc/invoke.texi: Document -mlong-double-128.
1032
1033 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1034
1035 PR rtl-optimization/60024
1036 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1037
1038 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1039
1040 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1041
1042 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1043
1044 PR rtl-optimization/57662
1045 * sel-sched.c (code_motion_path_driver): Do not mark already not
1046 existing blocks in the visiting bitmap.
1047
1048 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1049
1050 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1051 on the insn being emitted.
1052
1053 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1054 Will Deacon <will.deacon@arm.com>
1055
1056 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1057
1058 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1059
1060 * config/arm/arm-tables.opt: Regenerate.
1061
1062 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1063
1064 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1065 for vector types other than V16QImode.
1066 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1067 define_expand, and call altivec_expand_vec_perm_le when producing
1068 code with little endian element order.
1069 (*altivec_vperm_<mode>_internal): New insn having previous
1070 behavior of altivec_vperm_<mode>.
1071 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1072 altivec_expand_vec_perm_le when producing code with little endian
1073 element order.
1074 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1075 behavior of altivec_vperm_<mode>_uns.
1076
1077 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1078
1079 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1080 (altivec_vsumsws): Add handling for -maltivec=be with a little
1081 endian target.
1082 (altivec_vsumsws_direct): New.
1083 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1084 gen_altivec_vsumsws.
1085
1086 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1087
1088 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1089 vtable_pointer_value_to_binfo): New functions.
1090 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1091 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1092
1093 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1094
1095 * config/nios2/nios2.md (load_got_register): Initialize GOT
1096 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1097 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1098
1099 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1100
1101 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1102 preserverd by passthrough, do not propagate the type.
1103
1104 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1105
1106 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1107 (mips_atomic_assign_expand_fenv): New function.
1108 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1109
1110 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1111
1112 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1113 (__builtin_mips_set_fcsr): Likewise.
1114 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1115 MIPS_USI_FTYPE_VOID.
1116 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1117 (mips16_expand_set_fcsr): Likewise.
1118 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1119 (mips16_set_fcsr_stub): Likewise.
1120 (mips16_get_fcsr_one_only_stub): New class.
1121 (mips16_set_fcsr_one_only_stub): Likewise.
1122 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1123 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1124 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1125 (hard_float): New availability predicate.
1126 (mips_builtins): Add get_fcsr and set_fcsr.
1127 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1128 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1129 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1130 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1131 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1132 patterns.
1133
1134 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1135
1136 * config/mips/mips.c (mips_one_only_stub): New class.
1137 (mips_need_mips16_rdhwr_p): Replace with...
1138 (mips16_rdhwr_stub): ...this new variable.
1139 (mips16_stub_call_address): New function.
1140 (mips16_rdhwr_one_only_stub): New class.
1141 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1142 (mips_output_mips16_rdhwr): Delete.
1143 (mips_finish_stub): New function.
1144 (mips_code_end): Use it to handle rdhwr stubs.
1145
1146 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1147
1148 PR target/60017
1149 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1150 when calculating size of integer atomic types.
1151
1152 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1153
1154 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1155
1156 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1157
1158 PR tree-optimization/60003
1159 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1160 * profile.c (branch_prob): Use gimple_call_builtin_p
1161 to check for BUILT_IN_SETJMP_RECEIVER.
1162 * tree-inline.c (copy_bb): Call notice_special_calls.
1163
1164 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1165
1166 PR bootstrap/59985
1167 * lra-constraints.c (process_alt_operands): Update reload_sum only
1168 on the first pass.
1169
1170 2014-01-31 Richard Henderson <rth@redhat.com>
1171
1172 PR middle-end/60004
1173 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1174 until after else_eh is processed.
1175
1176 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1177
1178 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1179 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1180 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1181 in smmintrin.h, remove them.
1182 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1183 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1184 * config/i386/i386.md (ROUND_SAE): Fix value.
1185 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1186 (const48_operand): New.
1187 * config/i386/subst.md (round), (round_expand): Use
1188 const_4_or_8_to_11_operand.
1189 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1190
1191 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1192
1193 * config/i386/constraints.md (Yk): Swap meaning with k.
1194 * config/i386/i386.md (movhi_internal): Change Yk to k.
1195 (movqi_internal): Ditto.
1196 (*k<logic><mode>): Ditto.
1197 (*andhi_1): Ditto.
1198 (*andqi_1): Ditto.
1199 (kandn<mode>): Ditto.
1200 (*<code>hi_1): Ditto.
1201 (*<code>qi_1): Ditto.
1202 (kxnor<mode>): Ditto.
1203 (kortestzhi): Ditto.
1204 (kortestchi): Ditto.
1205 (kunpckhi): Ditto.
1206 (*one_cmplhi2_1): Ditto.
1207 (*one_cmplqi2_1): Ditto.
1208 * config/i386/sse.md (): Change k to Yk.
1209 (avx512f_load<mode>_mask): Ditto.
1210 (avx512f_blendm<mode>): Ditto.
1211 (avx512f_store<mode>_mask): Ditto.
1212 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1213 (avx512f_storedqu<mode>_mask): Ditto.
1214 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1215 Ditto.
1216 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1217 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1218 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1219 (avx512f_maskcmp<mode>3): Ditto.
1220 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1221 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1222 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1223 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1224 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1225 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1226 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1227 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1228 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1229 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1230 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1231 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1232 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1233 (vec_extract_lo_<mode>_maskm): Ditto.
1234 (vec_extract_hi_<mode>_maskm): Ditto.
1235 (avx512f_vternlog<mode>_mask): Ditto.
1236 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1237 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1238 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1239 (avx512f_<code>v8div16qi2_mask): Ditto.
1240 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1241 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1242 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1243 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1244 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1245 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1246 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1247 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1248 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1249 (avx512cd_maskb_vec_dupv8di): Ditto.
1250 (avx512cd_maskw_vec_dupv16si): Ditto.
1251 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1252 (avx512f_vpermi2var<mode>3_mask): Ditto.
1253 (avx512f_vpermi2var<mode>3_mask): Ditto.
1254 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1255 (*avx512f_gathersi<mode>): Ditto.
1256 (*avx512f_gathersi<mode>_2): Ditto.
1257 (*avx512f_gatherdi<mode>): Ditto.
1258 (*avx512f_gatherdi<mode>_2): Ditto.
1259 (*avx512f_scattersi<mode>): Ditto.
1260 (*avx512f_scatterdi<mode>): Ditto.
1261 (avx512f_compress<mode>_mask): Ditto.
1262 (avx512f_compressstore<mode>_mask): Ditto.
1263 (avx512f_expand<mode>_mask): Ditto.
1264 * config/i386/subst.md (mask): Change k to Yk.
1265 (mask_scalar_merge): Ditto.
1266 (sd): Ditto.
1267
1268 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1269
1270 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1271
1272 2014-01-31 Richard Biener <rguenther@suse.de>
1273
1274 PR middle-end/59990
1275 * builtins.c (fold_builtin_memory_op): Make sure to not
1276 use a floating-point mode or a boolean or enumeral type for
1277 the copy operation.
1278
1279 2014-01-30 DJ Delorie <dj@redhat.com>
1280
1281 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1282 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1283 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1284 whenever main() has an epilogue.
1285
1286 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1287
1288 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1289 unused variable "field".
1290 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1291 (vsx_mergeh_<mode>): Likewise.
1292 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1293 (altivec_vmrghh): Likewise.
1294 (altivec_vmrghw): Likewise.
1295 (altivec_vmrglb): Likewise.
1296 (altivec_vmrglh): Likewise.
1297 (altivec_vmrglw): Likewise.
1298 (altivec_vspltb): Add missing uses.
1299 (altivec_vsplth): Likewise.
1300 (altivec_vspltw): Likewise.
1301 (altivec_vspltsf): Likewise.
1302
1303 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1304
1305 PR target/59923
1306 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1307 frame related instructions.
1308
1309 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1310
1311 PR rtl-optimization/59959
1312 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1313 any reload of register whose subreg is invalid.
1314
1315 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1316
1317 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1318 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1319 Add missing return type - void.
1320
1321 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1322
1323 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1324 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1325 remove element index adjustment for endian (now handled in vsx.md
1326 and altivec.md).
1327 (altivec_expand_vec_perm_const): Use
1328 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1329 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1330 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1331 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1332 define_expand and a new define_insn *altivec_vspltb_internal;
1333 adjust for -maltivec=be on a little endian target.
1334 (altivec_vspltb_direct): New.
1335 (altivec_vsplth): Divide into a define_expand and a new
1336 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1337 little endian target.
1338 (altivec_vsplth_direct): New.
1339 (altivec_vspltw): Divide into a define_expand and a new
1340 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1341 little endian target.
1342 (altivec_vspltw_direct): New.
1343 (altivec_vspltsf): Divide into a define_expand and a new
1344 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1345 a little endian target.
1346
1347 2014-01-30 Richard Biener <rguenther@suse.de>
1348
1349 PR tree-optimization/59993
1350 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1351 can propagate form the earlier stmt and avoid the transform
1352 when the intermediate result is needed.
1353
1354 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1355
1356 * README.Portability: Fix typo.
1357
1358 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1359
1360 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1361 comparison_operator with ordered_comparison_operator.
1362
1363 2014-01-30 Nick Clifton <nickc@redhat.com>
1364
1365 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1366 Rename to mn10300_store_multiple_regs.
1367 * config/mn10300/mn10300.c: Likewise.
1368 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1369 store_multiple_regs.
1370 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1371 Call mn10300_store_multiple_regs.
1372
1373 2014-01-30 Nick Clifton <nickc@redhat.com>
1374 DJ Delorie <dj@redhat.com>
1375
1376 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1377 %fp 2 to keep registers after it properly word-aligned.
1378 (rl78_alloc_physical_registers_umul): Handle the case where both
1379 input operands are the same.
1380
1381 2014-01-30 Richard Biener <rguenther@suse.de>
1382
1383 PR tree-optimization/59903
1384 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1385 check properly.
1386
1387 2014-01-30 Jason Merrill <jason@redhat.com>
1388
1389 PR c++/59633
1390 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1391
1392 PR c++/59645
1393 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1394
1395 2014-01-30 Richard Biener <rguenther@suse.de>
1396
1397 PR tree-optimization/59951
1398 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1399
1400 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1401
1402 PR target/59784
1403 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1404 SFmode to DFmode case.
1405
1406 2014-01-29 DJ Delorie <dj@redhat.com>
1407
1408 * config/msp430/msp430.opt (-minrt): New.
1409 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1410 if -minrt given.
1411 (ENDFILE_SPEC): Likewise.
1412
1413 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1414
1415 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1416 (estimate_function_body_sizes): Use it.
1417
1418 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1419
1420 PR c++/58561
1421 * dwarf2out.c (is_cxx_auto): New.
1422 (is_base_type): Use it.
1423 (gen_type_die_with_usage): Likewise.
1424
1425 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1426
1427 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1428 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1429 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1430 -maltivec=be with LE targets.
1431 (vsx_mergeh_<mode>): Likewise.
1432 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1433 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1434 (altivec_vmrghb): Replace with define_expand and new
1435 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1436 (altivec_vmrghb_direct): New define_insn.
1437 (altivec_vmrghh): Replace with define_expand and new
1438 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1439 (altivec_vmrghh_direct): New define_insn.
1440 (altivec_vmrghw): Replace with define_expand and new
1441 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1442 (altivec_vmrghw_direct): New define_insn.
1443 (*altivec_vmrghsf): Adjust for endianness.
1444 (altivec_vmrglb): Replace with define_expand and new
1445 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1446 (altivec_vmrglb_direct): New define_insn.
1447 (altivec_vmrglh): Replace with define_expand and new
1448 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1449 (altivec_vmrglh_direct): New define_insn.
1450 (altivec_vmrglw): Replace with define_expand and new
1451 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1452 (altivec_vmrglw_direct): New define_insn.
1453 (*altivec_vmrglsf): Adjust for endianness.
1454 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1455 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1456 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1457 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1458 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1459 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1460 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1461 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1462
1463 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1464
1465 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1466 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1467 whitespace.
1468
1469 2014-01-29 Richard Biener <rguenther@suse.de>
1470
1471 PR tree-optimization/58742
1472 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1473 associate_pointerplus_align.
1474 (associate_pointerplus_diff): New function.
1475 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1476 and associate_pointerplus_diff.
1477
1478 2014-01-29 Richard Biener <rguenther@suse.de>
1479
1480 * lto-streamer.h (LTO_major_version): Bump to 3.
1481 (LTO_minor_version): Reset to 0.
1482
1483 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1484
1485 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1486 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1487 (arm_file_start): Generate correct asm header for armv7ve.
1488 * config/arm/bpabi.h: Add multilib support for armv7ve.
1489 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1490 and cortex-a15 to armv7ve.
1491 * config/arm/t-aprofile: Add multilib support for armv7ve.
1492 * doc/invoke.texi: Document -march=armv7ve.
1493
1494 2014-01-29 Richard Biener <rguenther@suse.de>
1495
1496 PR tree-optimization/58742
1497 * tree-ssa-forwprop.c (associate_plusminus): Return true
1498 if we changed sth, defer EH cleanup to ...
1499 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1500 (simplify_mult): New function.
1501
1502 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1503
1504 PR middle-end/59917
1505 PR tree-optimization/59920
1506 * tree.c (build_common_builtin_nodes): Remove
1507 __builtin_setjmp_dispatcher initialization.
1508 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1509 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1510 instead of gsi_after_labels + manually skipping debug stmts.
1511 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1512 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1513 * tree-inline.c (copy_edges_for_bb): Remove
1514 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1515 argument. Ignore computed_goto_p stmts. Don't call
1516 make_abnormal_goto_edges. If a call might need abnormal edges
1517 for non-local gotos, see if it already has an edge to
1518 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1519 with true argument, don't do anything then, otherwise add
1520 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1521 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1522 caller.
1523 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1524 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1525 (lower_stmt): Don't set data->calls_builtin_setjmp.
1526 (lower_builtin_setjmp): Adjust comment.
1527 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1528 * tree-cfg.c (found_computed_goto): Remove.
1529 (factor_computed_gotos): Remove.
1530 (make_goto_expr_edges): Return bool, true for computed gotos.
1531 Don't call make_abnormal_goto_edges.
1532 (build_gimple_cfg): Don't set found_computed_goto, don't call
1533 factor_computed_gotos.
1534 (computed_goto_p): No longer static.
1535 (make_blocks): Don't set found_computed_goto.
1536 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1537 (make_edges): If make_goto_expr_edges returns true, push bb
1538 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1539 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1540 vector. Record mapping between bbs and OpenMP regions if there
1541 are any, adjust make_gimple_omp_edges caller. Call
1542 handle_abnormal_edges.
1543 (make_abnormal_goto_edges): Remove.
1544 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1545 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1546 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1547 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1548 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1549 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1550 filling *region also set *region_idx to (*region)->entry->index.
1551
1552 PR other/58712
1553 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1554 For REGs set ORIGINAL_REGNO.
1555
1556 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1557
1558 * doc/md.texi: Mention that a target shouldn't implement
1559 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1560 than hi/lo pair.
1561
1562 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1563
1564 PR tree-optimization/59594
1565 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1566 a copy of the datarefs vector rather than the vector itself.
1567
1568 2014-01-28 Jason Merrill <jason@redhat.com>
1569
1570 PR c++/53756
1571 * dwarf2out.c (auto_die): New static.
1572 (gen_type_die_with_usage): Handle C++1y 'auto'.
1573 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1574 on definition.
1575
1576 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1577
1578 PR target/59672
1579 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1580 (SPEC_X32): Likewise.
1581 (SPEC_64): Likewise.
1582 * config/i386/i386.c (ix86_option_override_internal): Turn off
1583 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1584 for TARGET_16BIT.
1585 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1586 * config/i386/i386.h (TARGET_16BIT): New macro.
1587 (TARGET_16BIT_P): Likewise.
1588 * config/i386/i386.opt: Add m16.
1589 * doc/invoke.texi: Document -m16.
1590
1591 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1592
1593 PR preprocessor/59935
1594 * input.c (location_get_source_line): Bail out on when line number
1595 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1596
1597 2014-01-28 Richard Biener <rguenther@suse.de>
1598
1599 PR tree-optimization/58742
1600 * tree-ssa-forwprop.c (associate_plusminus): Handle
1601 pointer subtraction of the form (T)(P + A) - (T)P.
1602
1603 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1604
1605 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1606 at const_int_cost.
1607
1608 2014-01-28 Richard Biener <rguenther@suse.de>
1609
1610 Revert
1611 2014-01-28 Richard Biener <rguenther@suse.de>
1612
1613 PR rtl-optimization/45364
1614 PR rtl-optimization/59890
1615 * var-tracking.c (local_get_addr_clear_given_value): Handle
1616 already cleared slot.
1617 (val_reset): Handle not allocated local_get_addr_cache.
1618 (vt_find_locations): Use post-order on the inverted CFG.
1619
1620 2014-01-28 Richard Biener <rguenther@suse.de>
1621
1622 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1623
1624 2014-01-28 Richard Biener <rguenther@suse.de>
1625
1626 PR rtl-optimization/45364
1627 PR rtl-optimization/59890
1628 * var-tracking.c (local_get_addr_clear_given_value): Handle
1629 already cleared slot.
1630 (val_reset): Handle not allocated local_get_addr_cache.
1631 (vt_find_locations): Use post-order on the inverted CFG.
1632
1633 2014-01-28 Alan Modra <amodra@gmail.com>
1634
1635 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1636 * configure.ac <recursive call for build != host>: Define
1637 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1638 and LD_FOR_BUILD too.
1639 * configure: Regenerate.
1640
1641 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1642
1643 * config/i386/i386.c (get_builtin_code_for_version): Separate
1644 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1645 Broadwell from Haswell.
1646
1647 2014-01-27 Steve Ellcey <sellcey@mips.com>
1648
1649 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1650 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1651 * config/mips/mips.c (mips_option_override): Change setting
1652 of TARGET_DSP.
1653 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1654 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1655 Change from Mask to Var.
1656
1657 2014-01-27 Jeff Law <law@redhat.com>
1658
1659 * ipa-inline.c (inline_small_functions): Fix typo.
1660
1661 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1662
1663 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1664 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1665 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1666 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1667 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1668 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1669 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1670 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1671 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1672 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1673 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1674 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1675 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1676 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1677 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1678 (_mm512_storeu_epi64): Ditto.
1679 (_mm512_cmpge_epi32_mask): Ditto.
1680 (_mm512_cmpge_epu32_mask): Ditto.
1681 (_mm512_cmpge_epi64_mask): Ditto.
1682 (_mm512_cmpge_epu64_mask): Ditto.
1683 (_mm512_cmple_epi32_mask): Ditto.
1684 (_mm512_cmple_epu32_mask): Ditto.
1685 (_mm512_cmple_epi64_mask): Ditto.
1686 (_mm512_cmple_epu64_mask): Ditto.
1687 (_mm512_cmplt_epi32_mask): Ditto.
1688 (_mm512_cmplt_epu32_mask): Ditto.
1689 (_mm512_cmplt_epi64_mask): Ditto.
1690 (_mm512_cmplt_epu64_mask): Ditto.
1691 (_mm512_cmpneq_epi32_mask): Ditto.
1692 (_mm512_cmpneq_epu32_mask): Ditto.
1693 (_mm512_cmpneq_epi64_mask): Ditto.
1694 (_mm512_cmpneq_epu64_mask): Ditto.
1695 (_mm512_expand_pd): Ditto.
1696 (_mm512_expand_ps): Ditto.
1697 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1698 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1699 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1700 * config/i386/i386.c (ix86_builtins): Add
1701 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1702 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1703 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1704 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1705 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1706 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1707 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1708 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1709 IX86_BUILTIN_PMOVUSQW512_MEM.
1710 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1711 __builtin_ia32_pmovsqd512mem_mask,
1712 __builtin_ia32_pmovqd512mem_mask,
1713 __builtin_ia32_pmovusqw512mem_mask,
1714 __builtin_ia32_pmovsqw512mem_mask,
1715 __builtin_ia32_pmovqw512mem_mask,
1716 __builtin_ia32_pmovusdw512mem_mask,
1717 __builtin_ia32_pmovsdw512mem_mask,
1718 __builtin_ia32_pmovdw512mem_mask,
1719 __builtin_ia32_pmovqb512mem_mask,
1720 __builtin_ia32_pmovusqb512mem_mask,
1721 __builtin_ia32_pmovsqb512mem_mask,
1722 __builtin_ia32_pmovusdb512mem_mask,
1723 __builtin_ia32_pmovsdb512mem_mask,
1724 __builtin_ia32_pmovdb512mem_mask.
1725 (bdesc_args): Add __builtin_ia32_expanddf512,
1726 __builtin_ia32_expandsf512.
1727 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1728 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1729 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1730 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1731 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1732 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1733 (avx512f_<code>v8div16qi2_mask_store): This.
1734 (avx512f_expand<mode>): New.
1735
1736 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1737
1738 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1739 New.
1740 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1741 (_mm512_prefetch_i32scatter_pd): Ditto.
1742 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1743 (_mm512_prefetch_i64scatter_pd): Ditto.
1744 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1745 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1746 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1747 (_mm512_prefetch_i32scatter_ps): Ditto.
1748 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1749 (_mm512_prefetch_i64scatter_ps): Ditto.
1750 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1751 * config/i386/i386-builtin-types.def: Define
1752 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1753 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1754 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1755 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1756 IX86_BUILTIN_SCATTERPFQPD.
1757 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1758 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1759 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1760 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1761 __builtin_ia32_scatterpfqps.
1762 (ix86_expand_builtin): Expand new built-ins.
1763 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1764 fix memory access data type.
1765 (*avx512pf_gatherpf<mode>_mask): Ditto.
1766 (*avx512pf_gatherpf<mode>): Ditto.
1767 (avx512pf_scatterpf<mode>): Ditto.
1768 (*avx512pf_scatterpf<mode>_mask): Ditto.
1769 (*avx512pf_scatterpf<mode>): Ditto.
1770 (GATHER_SCATTER_SF_MEM_MODE): New.
1771 (avx512pf_gatherpf<mode>df): Ditto.
1772 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1773 (*avx512pf_scatterpf<mode>df): Ditto.
1774
1775 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1776
1777 PR bootstrap/59934
1778 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1779 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1780 reached.
1781
1782 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1783
1784 * common/config/arm/arm-common.c
1785 (arm_rewrite_mcpu): Handle multiple names.
1786 * config/arm/arm.h
1787 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1788
1789 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1790
1791 * gimple-builder.h (create_gimple_tmp): Delete.
1792
1793 2014-01-27 Christian Bruel <christian.bruel@st.com>
1794
1795 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1796 words comparisons.
1797
1798 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1799
1800 * config/pa/pa.md (call): Generate indirect long calls to non-local
1801 functions when outputing 32-bit code.
1802 (call_value): Likewise except for special call to buggy powf function.
1803
1804 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1805 portable runtime and PIC indirect calls.
1806 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1807 and PIC call sequences. Use ldo instead of blr to set return register
1808 in PIC call sequence.
1809
1810 2014-01-25 Walter Lee <walt@tilera.com>
1811
1812 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1813 avoid clobbering a live register.
1814
1815 2014-01-25 Walter Lee <walt@tilera.com>
1816
1817 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1818 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1819 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1820 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1821
1822 2014-01-25 Walter Lee <walt@tilera.com>
1823
1824 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1825 arguments on even registers.
1826 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1827 STACK_BOUNDARY.
1828 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1829 (BIGGEST_ALIGNMENT): Ditto.
1830 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1831
1832 2014-01-25 Walter Lee <walt@tilera.com>
1833
1834 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1835 insns before bundling.
1836 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1837
1838 2014-01-25 Walter Lee <walt@tilera.com>
1839
1840 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1841 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1842 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1843
1844 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1845
1846 * config/mips/constraints.md (kl): Delete.
1847 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1848 define expands, using...
1849 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1850 instructions for MIPS16.
1851 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1852 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1853
1854 2014-01-25 Walter Lee <walt@tilera.com>
1855
1856 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1857 (clzdi2): Ditto.
1858 (ffsdi2): Ditto.
1859
1860 2014-01-25 Walter Lee <walt@tilera.com>
1861
1862 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1863 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1864
1865 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1866
1867 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1868 Handle XOR.
1869
1870 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1871
1872 * print-rtl.c (in_call_function_usage): New var.
1873 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1874 EXPR_LIST mode as mode and not as reg note name.
1875
1876 PR middle-end/59561
1877 * cfgloopmanip.c (copy_loop_info): If
1878 loop->warned_aggressive_loop_optimizations, make sure
1879 the flag is set in target loop too.
1880
1881 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1882
1883 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1884 flag_cilkplus.
1885 * builtins.def: Likewise.
1886 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1887 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1888 * ira.c (ira_setup_eliminable_regset): Likewise.
1889 * omp-low.c (gate_expand_omp): Likewise.
1890 (execute_lower_omp): Likewise.
1891 (diagnose_sb_0): Likewise.
1892 (gate_diagnose_omp_blocks): Likewise.
1893 (simd_clone_clauses_extract): Likewise.
1894 (gate): Likewise.
1895
1896 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1897
1898 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1899 correction for little endian...
1900 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1901 here.
1902
1903 2014-01-24 Jeff Law <law@redhat.com>
1904
1905 PR tree-optimization/59919
1906 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1907 for non-returning calls.
1908
1909 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1910
1911 * common/config/aarch64/aarch64-common.c
1912 (aarch64_rewrite_mcpu): Handle multiple names.
1913 * config/aarch64/aarch64.h
1914 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1915
1916 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1917
1918 * input.c (add_file_to_cache_tab): Handle the case where fopen
1919 returns NULL.
1920
1921 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1922
1923 PR target/59929
1924 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1925 from push operand if code of push isn't PRE_DEC.
1926
1927 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1928
1929 PR target/59909
1930 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1931 -mquad-memory-atomic. Update -mquad-memory documentation to say
1932 it is only used for non-atomic loads/stores.
1933
1934 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1935 -mquad-memory or -mquad-memory-atomic switches.
1936
1937 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1938 -mquad-memory-atomic to ISA 2.07 support.
1939
1940 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1941 to separate support of normal quad word memory operations (ldq, stq)
1942 from the atomic quad word memory operations.
1943
1944 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1945 support to separate non-atomic quad word operations from atomic
1946 quad word operations. Disable non-atomic quad word operations in
1947 little endian mode so that we don't have to swap words after the
1948 load and before the store.
1949 (quad_load_store_p): Add comment about atomic quad word support.
1950 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1951 options printed with -mdebug=reg.
1952
1953 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1954 -mquad-memory-atomic as the test for whether we have quad word
1955 atomic instructions.
1956 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1957 or -mp8-vector are used, allow byte/half-word atomic operations.
1958
1959 * config/rs6000/sync.md (load_lockedti): Insure that the address
1960 is a proper indexed or indirect address for the lqarx instruction.
1961 On little endian systems, swap the hi/lo registers after the lqarx
1962 instruction.
1963 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1964 insure the address is valid for the lqarx instruction.
1965 (store_conditionalti): Insure that the address is a proper indexed
1966 or indirect address for the stqcrx. instruction. On little endian
1967 systems, swap the hi/lo registers before doing the stqcrx.
1968 instruction.
1969 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1970 insure the address is valid for the stqcrx. instruction.
1971
1972 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1973 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1974 type of quad memory support is available.
1975
1976 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1977
1978 PR regression/59915
1979 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1980 there is a danger of looping.
1981
1982 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1983
1984 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1985 force flag_ira_loop_pressure if set via command line.
1986
1987 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1988
1989 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1990 (ashr_simd): New builtin handling DI mode.
1991 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1992 (aarch64_sshr_simddi): New match pattern.
1993 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1994 (vshrd_n_s64): Likewise.
1995 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1996
1997 2014-01-23 Nick Clifton <nickc@redhat.com>
1998
1999 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2000 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2001 favour of mcu specific scripts.
2002 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2003 430x multilibs.
2004
2005 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2006 Alex Velenko <Alex.Velenko@arm.com>
2007
2008 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2009 (vaddv_s16): Likewise.
2010 (vaddv_s32): Likewise.
2011 (vaddv_u8): Likewise.
2012 (vaddv_u16): Likewise.
2013 (vaddv_u32): Likewise.
2014 (vaddvq_s8): Likewise.
2015 (vaddvq_s16): Likewise.
2016 (vaddvq_s32): Likewise.
2017 (vaddvq_s64): Likewise.
2018 (vaddvq_u8): Likewise.
2019 (vaddvq_u16): Likewise.
2020 (vaddvq_u32): Likewise.
2021 (vaddvq_u64): Likewise.
2022 (vaddv_f32): Likewise.
2023 (vaddvq_f32): Likewise.
2024 (vaddvq_f64): Likewise.
2025 (vmaxv_f32): Likewise.
2026 (vmaxv_s8): Likewise.
2027 (vmaxv_s16): Likewise.
2028 (vmaxv_s32): Likewise.
2029 (vmaxv_u8): Likewise.
2030 (vmaxv_u16): Likewise.
2031 (vmaxv_u32): Likewise.
2032 (vmaxvq_f32): Likewise.
2033 (vmaxvq_f64): Likewise.
2034 (vmaxvq_s8): Likewise.
2035 (vmaxvq_s16): Likewise.
2036 (vmaxvq_s32): Likewise.
2037 (vmaxvq_u8): Likewise.
2038 (vmaxvq_u16): Likewise.
2039 (vmaxvq_u32): Likewise.
2040 (vmaxnmv_f32): Likewise.
2041 (vmaxnmvq_f32): Likewise.
2042 (vmaxnmvq_f64): Likewise.
2043 (vminv_f32): Likewise.
2044 (vminv_s8): Likewise.
2045 (vminv_s16): Likewise.
2046 (vminv_s32): Likewise.
2047 (vminv_u8): Likewise.
2048 (vminv_u16): Likewise.
2049 (vminv_u32): Likewise.
2050 (vminvq_f32): Likewise.
2051 (vminvq_f64): Likewise.
2052 (vminvq_s8): Likewise.
2053 (vminvq_s16): Likewise.
2054 (vminvq_s32): Likewise.
2055 (vminvq_u8): Likewise.
2056 (vminvq_u16): Likewise.
2057 (vminvq_u32): Likewise.
2058 (vminnmv_f32): Likewise.
2059 (vminnmvq_f32): Likewise.
2060 (vminnmvq_f64): Likewise.
2061
2062 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2063
2064 * config/aarch64/aarch64-simd.md
2065 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2066 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2067 (*aarch64_mul3_elt<mode>): Likewise.
2068 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2069 (*aarch64_mul3_elt_to_64v2df): Likewise.
2070 (*aarch64_mla_elt<mode>): Likewise.
2071 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2072 (*aarch64_mls_elt<mode>): Likewise.
2073 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2074 (*aarch64_fma4_elt<mode>): Likewise.
2075 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2076 (*aarch64_fma4_elt_to_64v2df): Likewise.
2077 (*aarch64_fnma4_elt<mode>): Likewise.
2078 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2079 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2080 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2081 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2082 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2083 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2084 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2085 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2086 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2087
2088 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2089
2090 * config/aarch64/aarch64-simd.md
2091 (aarch64_be_checked_get_lane<mode>): New define_expand.
2092 * config/aarch64/aarch64-simd-builtins.def
2093 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2094 New builtin definition.
2095 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2096 Use new safe be builtin.
2097
2098 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2099
2100 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2101 New define_insn.
2102 (aarch64_be_st1<mode>): Likewise.
2103 (aarch_ld1<VALL:mode>): Define_expand modified.
2104 (aarch_st1<VALL:mode>): Likewise.
2105 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2106 (UNSPEC_ST1): Likewise.
2107
2108 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2109
2110 * config/microblaze/microblaze.md: Add trap insn and attribute
2111
2112 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2113
2114 PR preprocessor/58580
2115 * input.h (location_get_source_line): Take an additional line_size
2116 parameter.
2117 (void diagnostics_file_cache_fini): Declare new function.
2118 * input.c (struct fcache): New type.
2119 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2120 New static constants.
2121 (diagnostic_file_cache_init, total_lines_num)
2122 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2123 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2124 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2125 (get_next_line, read_next_line, goto_next_line, read_line_num):
2126 New static function definitions.
2127 (diagnostic_file_cache_fini): New function.
2128 (location_get_source_line): Take an additional output line_len
2129 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2130 read_line_num.
2131 * diagnostic.c (diagnostic_finish): Call
2132 diagnostic_file_cache_fini.
2133 (adjust_line): Take an additional input parameter for the length
2134 of the line, rather than calculating it with strlen.
2135 (diagnostic_show_locus): Adjust the use of
2136 location_get_source_line and adjust_line with respect to their new
2137 signature. While displaying a line now, do not stop at the first
2138 null byte. Rather, display the zero byte as a space and keep
2139 going until we reach the size of the line.
2140 * Makefile.in: Add vec.o to OBJS-libcommon
2141
2142 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2143 Ilya Tocar <ilya.tocar@intel.com>
2144
2145 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2146 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2147 (__builtin_ia32_kmov16): Ditto.
2148 * config/i386/i386.md (UNSPEC_KMOV): New.
2149 (kmovw): Ditto.
2150
2151 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2152
2153 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2154 (_mm512_storeu_si512): Ditto.
2155
2156 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2157
2158 PR target/52125
2159 * rtl.h (get_referenced_operands): Declare.
2160 * recog.c (get_referenced_operands): New function.
2161 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2162 operands have been referenced when recording LO_SUM references.
2163
2164 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2165
2166 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2167
2168 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2169
2170 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2171 Enable for generic and recent AMD targets.
2172
2173 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2174
2175 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2176 ARG_SIZE note when adjustment was eliminated.
2177
2178 2014-01-22 Jeff Law <law@redhat.com>
2179
2180 PR tree-optimization/59597
2181 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2182 in file. Accept new argument REGISTERING and use it to modify
2183 dump output appropriately.
2184 (register_jump_thread): Corresponding changes.
2185 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2186 thread paths involving joiner blocks. Add code to dump cancelled
2187 jump threading paths.
2188
2189 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2190
2191 PR rtl-optimization/59477
2192 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2193 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2194
2195 2014-01-22 Tom Tromey <tromey@redhat.com>
2196
2197 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2198 PARAMS.
2199 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2200
2201 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2202
2203 PR rtl-optimization/59896
2204 * lra-constraints.c (process_alt_operands): Check unused note for
2205 matched operands of insn with no output reloads.
2206
2207 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2208
2209 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2210 (mips_move_from_gpr_cost): Likewise.
2211
2212 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2213
2214 PR rtl-optimization/59858
2215 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2216 ira_class_hard_regs_num.
2217 (process_alt_operands): Increase reject for dying matched operand.
2218
2219 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2220
2221 PR target/59003
2222 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2223 smaller than size, perform several stores or loads and stores
2224 at dst + count - size to store or copy all of size bytes, rather
2225 than just last modesize bytes.
2226
2227 2014-01-20 DJ Delorie <dj@redhat.com>
2228
2229 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2230 that CLOBBERs are REGs before propogating their values.
2231
2232 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2233
2234 PR middle-end/59789
2235 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2236 (cgraph_inline_failed_type): New function.
2237 * cgraph.h (DEFCIFCODE): Add type.
2238 (cgraph_inline_failed_type_t): New enum.
2239 (cgraph_inline_failed_type): New prototype.
2240 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2241 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2242 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2243 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2244 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2245 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2246 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2247 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2248 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2249 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2250 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2251 OPTIMIZATION_MISMATCH.
2252 * tree-inline.c (expand_call_inline): Emit errors during
2253 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2254
2255 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2256
2257 PR target/59685
2258 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2259 mode attribute in insn output.
2260
2261 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2262
2263 * output.h (output_constant): Delete.
2264 * varasm.c (output_constant): Make private.
2265
2266 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2267
2268 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2269
2270 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2271
2272 PR middle-end/59860
2273 * tree.h (fold_builtin_strcat): New prototype.
2274 * builtins.c (fold_builtin_strcat): No longer static. Add len
2275 argument, if non-NULL, don't call c_strlen. Optimize
2276 directly into __builtin_memcpy instead of __builtin_strcpy.
2277 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2278 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2279
2280 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2281
2282 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2283 for SImode_address_operand operands, having only a REG argument.
2284
2285 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2286
2287 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2288 loader name using mbig-endian.
2289 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2290
2291 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2292
2293 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2294 (-mtune): Likewise.
2295 (-mcpu): Likewise.
2296
2297 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2298
2299 * config/aarch64/aarch64-protos.h
2300 (aarch64_cannot_change_mode_class_ptr): Declare.
2301 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2302 aarch64_cannot_change_mode_class_ptr): New.
2303 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2304 backend hook aarch64_cannot_change_mode_class.
2305
2306 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2307
2308 * common/config/aarch64/aarch64-common.c
2309 (aarch64_handle_option): Don't handle any option order logic here.
2310 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2311 selected_cpu, warn on architecture version mismatch.
2312 (aarch64_override_options): Fix parsing order for option strings.
2313
2314 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2315 Iain Sandoe <iain@codesourcery.com>
2316
2317 PR bootstrap/59496
2318 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2319 warning. Amend comment to reflect current functionality.
2320
2321 2014-01-20 Richard Biener <rguenther@suse.de>
2322
2323 PR middle-end/59860
2324 * builtins.c (fold_builtin_strcat): Remove case better handled
2325 by tree-ssa-strlen.c.
2326
2327 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2328
2329 * config/aarch64/aarch64.opt
2330 (mcpu, march, mtune): Make case-insensitive.
2331
2332 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2333
2334 PR target/59880
2335 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2336 if operands[1] is a REG or ZERO_EXTEND of a REG.
2337
2338 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2339
2340 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2341
2342 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2343
2344 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2345 long non-pic millicode calls.
2346
2347 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2348
2349 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2350
2351 2014-01-19 Kito Cheng <kito@0xlab.org>
2352
2353 * builtins.c (expand_movstr): Check movstr expand done or fail.
2354
2355 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2356 H.J. Lu <hongjiu.lu@intel.com>
2357
2358 PR target/59379
2359 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2360 to DImode for zero-extended addresses.
2361
2362 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2363
2364 PR rtl-optimization/57763
2365 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2366 on the new indirect jump_insn and increment LABEL_NUSES (label).
2367
2368 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2369
2370 PR bootstrap/59580
2371 PR bootstrap/59583
2372 * config.gcc (x86_archs): New variable.
2373 (x86_64_archs): Likewise.
2374 (x86_cpus): Likewise.
2375 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2376 --with-arch/--with-cpu= options.
2377 Support --with-arch=/--with-cpu={nehalem,westmere,
2378 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2379
2380 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2381
2382 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2383 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2384
2385 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2386
2387 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2388
2389 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2390
2391 PR target/58944
2392 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2393 clear cpp_get_options (parse_in)->warn_unused_macros for
2394 ix86_target_macros_internal with cpp_define.
2395
2396 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2397
2398 * jump.c (delete_related_insns): Keep (use (insn))s.
2399 * reorg.c (redundant_insn): Check for barriers too.
2400
2401 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2402
2403 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2404
2405 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2406
2407 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2408 call to $$dyncall when TARGET_LONG_CALLS is true.
2409
2410 2014-01-17 Jeff Law <law@redhat.com>
2411
2412 * ree.c (combine_set_extension): Temporarily disable test for
2413 changing number of hard registers.
2414
2415 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2416
2417 PR middle-end/58125
2418 * ipa-inline-analysis.c (inline_free_summary):
2419 Do not free summary of aliases.
2420
2421 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2422
2423 PR middle-end/59706
2424 * gimplify.c (gimplify_expr): Use create_tmp_var
2425 instead of create_tmp_var_raw. If cond doesn't have
2426 integral type, don't add the IFN_ANNOTATE builtin at all.
2427
2428 2014-01-17 Martin Jambor <mjambor@suse.cz>
2429
2430 PR ipa/59736
2431 * ipa-cp.c (prev_edge_clone): New variable.
2432 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2433 Also resize prev_edge_clone vector.
2434 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2435 (ipcp_edge_removal_hook): New function.
2436 (ipcp_driver): Register ipcp_edge_removal_hook.
2437
2438 2014-01-17 Andrew Pinski <apinski@cavium.com>
2439 Steve Ellcey <sellcey@mips.com>
2440
2441 PR target/59462
2442 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2443 of operator mode.
2444
2445 2014-01-17 Jeff Law <law@redhat.com>
2446
2447 PR middle-end/57904
2448 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2449 so that pass_ccp runs first.
2450
2451 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2452
2453 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2454 (ix86_adjust_cost): Use !TARGET_XXX.
2455 (do_reorder_for_imul): Likewise.
2456 (swap_top_of_ready_list): Likewise.
2457 (ix86_sched_reorder): Likewise.
2458
2459 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2460
2461 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2462 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2463 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2464 (intel_memset): New. Duplicate slm_memset.
2465 (intel_cost): New. Duplicate slm_cost.
2466 (m_INTEL): New macro.
2467 (processor_target_table): Add "intel".
2468 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2469 with PROCESSOR_INTEL for "intel".
2470 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2471 PROCESSOR_SILVERMONT.
2472 (ix86_issue_rate): Likewise.
2473 (ix86_adjust_cost): Likewise.
2474 (ia32_multipass_dfa_lookahead): Likewise.
2475 (swap_top_of_ready_list): Likewise.
2476 (ix86_sched_reorder): Likewise.
2477 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2478 instead of TARGET_OPT_AGU.
2479 * config/i386/i386.h (TARGET_INTEL): New.
2480 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2481 (processor_type): Add PROCESSOR_INTEL.
2482 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2483 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2484
2485 2014-01-17 Marek Polacek <polacek@redhat.com>
2486
2487 PR c/58346
2488 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2489 size is zero.
2490
2491 2014-01-17 Richard Biener <rguenther@suse.de>
2492
2493 PR tree-optimization/46590
2494 * opts.c (default_options_table): Add entries for
2495 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2496 all enabled at -O1 but not for -Og.
2497 * common.opt (fbranch-count-reg): Remove Init(1).
2498 (fmove-loop-invariants): Likewise.
2499 (ftree-pta): Likewise.
2500
2501 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2502
2503 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2504 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2505 decls to at least the GCC 4.8 used alignments.
2506
2507 PR fortran/59440
2508 * tree-nested.c (convert_nonlocal_reference_stmt,
2509 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2510 of GIMPLE_BIND stmts, adjust associated decls.
2511
2512 2014-01-17 Richard Biener <rguenther@suse.de>
2513
2514 PR tree-optimization/46590
2515 * vec.h (vec<>::bseach): New member function implementing
2516 binary search according to C89 bsearch.
2517 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2518 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2519 bitmap pointer again. Make accesses_in_loop a flat array.
2520 (mem_ref_obstack): New global.
2521 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2522 (mark_ref_stored): Likewise.
2523 (ref_indep_loop_p_2): Likewise.
2524 (set_ref_stored_in_loop): New helper function.
2525 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2526 (memref_free): Adjust.
2527 (record_mem_ref_loc): Simplify.
2528 (gather_mem_refs_stmt): Adjust.
2529 (sort_locs_in_loop_postorder_cmp): New function.
2530 (analyze_memory_references): Sort accesses_in_loop after
2531 loop postorder number.
2532 (find_ref_loc_in_loop_cmp): New function.
2533 (for_all_locs_in_loop): Find relevant cluster of locs in
2534 accesses_in_loop and iterate without recursion.
2535 (execute_sm): Avoid uninit warning.
2536 (struct ref_always_accessed): Simplify.
2537 (ref_always_accessed::operator ()): Likewise.
2538 (ref_always_accessed_p): Likewise.
2539 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2540 loop postorder numbers here.
2541 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2542 numbers.
2543
2544 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2545
2546 PR c++/57945
2547 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2548 on decls for which assemble_alias has been called.
2549
2550 2014-01-17 Nick Clifton <nickc@redhat.com>
2551
2552 * config/msp430/msp430.opt: (mcpu): New option.
2553 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2554 (msp430_option_override): Parse target_cpu. If the MCU name
2555 matches a generic string, clear target_mcu.
2556 (msp430_attr): Allow numeric interrupt values up to 63.
2557 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2558 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2559 option.
2560 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2561 Add mcpu matches.
2562 * config/msp430/msp430.md (popm): Use %J rather than %I.
2563 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2564 (addhi_cy_i): Use immediate_operand for operand 2.
2565 * doc/invoke.texi: Document -mcpu option.
2566
2567 2014-01-17 Richard Biener <rguenther@suse.de>
2568
2569 PR rtl-optimization/38518
2570 * df.h (df_analyze_loop): Declare.
2571 * df-core.c: Include cfgloop.h.
2572 (df_analyze_1): Split out main part of df_analyze.
2573 (df_analyze): Adjust.
2574 (loop_inverted_post_order_compute): New function.
2575 (loop_post_order_compute): Likewise.
2576 (df_analyze_loop): New function avoiding whole-function
2577 postorder computes.
2578 * loop-invariant.c (find_defs): Use df_analyze_loop.
2579 (find_invariants): Adjust.
2580 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2581
2582 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2583
2584 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2585 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2586
2587 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2588
2589 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2590 traversal when removing references.
2591
2592 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2593
2594 PR ipa/59775
2595 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2596
2597 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2598
2599 PR middle-end/56791
2600 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2601 pushing a reload for an autoinc when we had previously reloaded an
2602 inner part of the address.
2603
2604 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2605
2606 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2607 field.
2608 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2609 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2610 when not giving up or versioning for alias only because of
2611 loop->safelen.
2612 (vect_analyze_data_ref_dependences): Set to true.
2613 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2614 is a GIMPLE_PHI.
2615 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2616 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2617 to the condition.
2618
2619 PR middle-end/58344
2620 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2621
2622 PR target/59839
2623 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2624 operand 0 predicate for gathers, use a new pseudo as subtarget.
2625
2626 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2627
2628 PR middle-end/59609
2629 * lra-constraints.c (process_alt_operands): Add printing debug info.
2630 Check absence of input/output reloads for matched operands too.
2631
2632 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2633
2634 PR rtl-optimization/59835
2635 * ira.c (ira_init_register_move_cost): Increase cost for
2636 impossible modes.
2637
2638 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2639
2640 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2641
2642 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2643
2644 PR target/59780
2645 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2646 non-register objects. Use gen_(high/low)part more consistently.
2647 Fix assertions.
2648
2649 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2650
2651 PR target/59844
2652 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2653 endian support, remove tests for WORDS_BIG_ENDIAN.
2654 (p8_mfvsrd_3_<mode>): Likewise.
2655 (reload_gpr_from_vsx<mode>): Likewise.
2656 (reload_gpr_from_vsxsf): Likewise.
2657 (p8_mfvsrd_4_disf): Likewise.
2658
2659 2014-01-16 Richard Biener <rguenther@suse.de>
2660
2661 PR rtl-optimization/46590
2662 * lcm.c (compute_antinout_edge): Use postorder iteration.
2663 (compute_laterin): Use inverted postorder iteration.
2664
2665 2014-01-16 Nick Clifton <nickc@redhat.com>
2666
2667 PR middle-end/28865
2668 * varasm.c (output_constant): Return the number of bytes actually
2669 emitted.
2670 (output_constructor_array_range): Update the field size with the
2671 number of bytes emitted by output_constant.
2672 (output_constructor_regular_field): Likewise. Also do not
2673 complain if the total number of bytes emitted is now greater
2674 than the expected fieldpos.
2675 * output.h (output_constant): Update prototype and descriptive comment.
2676
2677 2014-01-16 Marek Polacek <polacek@redhat.com>
2678
2679 PR middle-end/59827
2680 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2681 it is error_mark_node.
2682
2683 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2684
2685 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2686 VALID_AVX256_REG_OR_OI_MODE.
2687
2688 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2689
2690 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2691 current procedure should be profiled.
2692
2693 2014-01-15 Andrew Pinski <apinski@cavium.com>
2694
2695 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2696 of moving from/to the STACK_REG register class.
2697
2698 2014-01-15 Richard Henderson <rth@redhat.com>
2699
2700 PR debug/54694
2701 * reginfo.c (global_regs_decl): Globalize.
2702 * rtl.h (global_regs_decl): Declare.
2703 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2704 reserved via global_regs.
2705
2706 2014-01-15 Teresa Johnson <tejohnson@google.com>
2707
2708 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2709
2710 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2711
2712 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2713 and vmulosh rather than call gen_vec_widen_smult_*.
2714 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2715 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2716 (vec_widen_smult_even_v16qi): Likewise.
2717 (vec_widen_umult_even_v8hi): Likewise.
2718 (vec_widen_smult_even_v8hi): Likewise.
2719 (vec_widen_umult_odd_v16qi): Likewise.
2720 (vec_widen_smult_odd_v16qi): Likewise.
2721 (vec_widen_umult_odd_v8hi): Likewise.
2722 (vec_widen_smult_odd_v8hi): Likewise.
2723 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2724 vmuloub rather than call gen_vec_widen_umult_*.
2725 (vec_widen_umult_lo_v16qi): Likewise.
2726 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2727 vmulosb rather than call gen_vec_widen_smult_*.
2728 (vec_widen_smult_lo_v16qi): Likewise.
2729 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2730 rather than call gen_vec_widen_umult_*.
2731 (vec_widen_umult_lo_v8hi): Likewise.
2732 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2733 rather than call gen_vec_widen_smult_*.
2734 (vec_widen_smult_lo_v8hi): Likewise.
2735
2736 2014-01-15 Jeff Law <law@redhat.com>
2737
2738 PR tree-optimization/59747
2739 * ree.c (find_and_remove_re): Properly handle case where a second
2740 eliminated extension requires widening a copy created for elimination
2741 of a prior extension.
2742 (combine_set_extension): Ensure that the number of hard regs needed
2743 for a destination register does not change when we widen it.
2744
2745 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2746
2747 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2748 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2749 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2750 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2751 (avr-*-rtems*): Likewise.
2752 (bfin*-rtems*): Likewise.
2753 (moxie-*-rtems*): Likewise.
2754 (h8300-*-rtems*): Likewise.
2755 (i[34567]86-*-rtems*): Likewise.
2756 (lm32-*-rtems*): Likewise.
2757 (m32r-*-rtems*): Likewise.
2758 (m68k-*-rtems*): Likewise.
2759 (microblaze*-*-rtems*): Likewise.
2760 (mips*-*-rtems*): Likewise.
2761 (powerpc-*-rtems*): Likewise.
2762 (sh-*-rtems*): Likewise.
2763 (sparc-*-rtems*): Likewise.
2764 (sparc64-*-rtems*): Likewise.
2765 (v850-*-rtems*): Likewise.
2766 (m32c-*-rtems*): Likewise.
2767
2768 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2769
2770 PR rtl-optimization/59511
2771 * ira.c (ira_init_register_move_cost): Use memory costs for some
2772 cases of register move cost calculations.
2773 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2774 instead of BB frequency.
2775 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2776 * lra-assigns.c (find_hard_regno_for): Ditto.
2777
2778 2014-01-15 Richard Biener <rguenther@suse.de>
2779
2780 PR tree-optimization/59822
2781 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2782 (vectorizable_load): Use it to hoist defs of uses of invariant
2783 loads out of the loop.
2784
2785 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2786 Kugan Vivekanandarajah <kuganv@linaro.org>
2787
2788 PR target/59695
2789 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2790 truncation.
2791
2792 2014-01-15 Richard Biener <rguenther@suse.de>
2793
2794 PR rtl-optimization/59802
2795 * lcm.c (compute_available): Use inverted postorder to seed
2796 the initial worklist.
2797
2798 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2799
2800 PR target/59803
2801 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2802 ADDR_REGS for invalid symrefs in non-PIC code.
2803
2804 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2805
2806 PR other/58712
2807 * builtins.c (determine_block_size): Initialize *probable_max_size
2808 even if len_rtx is CONST_INT.
2809
2810 2014-01-14 Andrew Pinski <apinski@cavium.com>
2811
2812 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2813 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2814 (cortexa53_tunings): Likewise.
2815 (aarch64_sched_issue_rate): New function.
2816 (TARGET_SCHED_ISSUE_RATE): Define.
2817
2818 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2819
2820 * ira-costs.c (find_costs_and_classes): Add missed
2821 ira_init_register_move_cost_if_necessary.
2822
2823 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2824
2825 PR target/59787
2826 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2827
2828 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2829
2830 PR target/59794
2831 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2832 to indicate if type is used for function return value. Warn ABI
2833 change if the vector mode isn't available for function return value.
2834 (ix86_function_arg_advance): Pass false to type_natural_mode.
2835 (ix86_function_arg): Likewise.
2836 (ix86_gimplify_va_arg): Likewise.
2837 (function_arg_32): Don't warn ABI change.
2838 (ix86_function_value): Pass true to type_natural_mode.
2839 (ix86_return_in_memory): Likewise.
2840 (ix86_struct_value_rtx): Removed.
2841 (TARGET_STRUCT_VALUE_RTX): Likewise.
2842
2843 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2844
2845 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2846 converting a conditional jump into a conditional return.
2847
2848 2014-01-14 Richard Biener <rguenther@suse.de>
2849
2850 PR tree-optimization/58921
2851 PR tree-optimization/59006
2852 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2853 hoisting invariant stmts.
2854 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2855 invariant loads on the preheader edge if possible.
2856
2857 2014-01-14 Joey Ye <joey.ye@arm.com>
2858
2859 * doc/plugin.texi (Building GCC plugins): Update to C++.
2860
2861 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2862
2863 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2864 (_mm_rcp28_round_ss): Ditto.
2865 (_mm_rsqrt28_round_sd): Ditto.
2866 (_mm_rsqrt28_round_ss): Ditto.
2867 (_mm_rcp28_sd): Ditto.
2868 (_mm_rcp28_ss): Ditto.
2869 (_mm_rsqrt28_sd): Ditto.
2870 (_mm_rsqrt28_ss): Ditto.
2871 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2872 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2873 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2874 (IX86_BUILTIN_RCP28SD): Ditto.
2875 (IX86_BUILTIN_RCP28SS): Ditto.
2876 (IX86_BUILTIN_RSQRT28SD): Ditto.
2877 (IX86_BUILTIN_RSQRT28SS): Ditto.
2878 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2879 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2880 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2881 (ix86_expand_special_args_builtin): Expand new FTYPE.
2882 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2883 (srcp14<mode>): Make insn unary.
2884 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2885 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2886 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2887 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2888 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2889 Fix rounding: make it SAE only.
2890 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2891 Ditto.
2892 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2893 Ditto.
2894 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2895 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2896 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2897 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2898 (round_saeonly_mask_scalar_operand4): Ditto.
2899 (round_saeonly_mask_scalar_op3): Ditto.
2900 (round_saeonly_mask_scalar_op4): Ditto.
2901
2902 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2903
2904 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2905 Implement -maltivec=be for vec_insert and vec_extract.
2906
2907 2014-01-10 DJ Delorie <dj@redhat.com>
2908
2909 * config/msp430/msp430.md (call_internal): Don't allow memory
2910 references with SP as the base register.
2911 (call_value_internal): Likewise.
2912 * config/msp430/constraints.md (Yc): New. For memory references
2913 that don't use SP as a base register.
2914
2915 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2916 "an integer without a # prefix"
2917 * config/msp430/msp430.md (epilogue_helper): Use it.
2918
2919 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2920
2921 PR target/59617
2922 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2923 AVX512F gather builtins.
2924 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2925 on gather decls with INTEGER_TYPE masktype.
2926 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2927 directly into the builtin rather than hoisting it before loop.
2928
2929 PR tree-optimization/59387
2930 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2931 (scev_const_prop): If folded_casts and type has undefined overflow,
2932 use force_gimple_operand instead of force_gimple_operand_gsi and
2933 for each added stmt if it is assign with
2934 arith_code_with_undefined_signed_overflow, call
2935 rewrite_to_defined_overflow.
2936 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2937 gimple-fold.h instead.
2938 (arith_code_with_undefined_signed_overflow,
2939 rewrite_to_defined_overflow): Moved to ...
2940 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2941 rewrite_to_defined_overflow): ... here. No longer static.
2942 Include gimplify-me.h.
2943 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2944 rewrite_to_defined_overflow): New prototypes.
2945
2946 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2947
2948 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2949
2950 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2951
2952 * builtins.c (get_object_alignment_2): Minor tweak.
2953 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2954
2955 2014-01-13 Christian Bruel <christian.bruel@st.com>
2956
2957 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2958 optimized non constant lengths.
2959
2960 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2961
2962 PR libgomp/59194
2963 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2964 load as __atomic_load_N if possible.
2965
2966 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2967
2968 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2969 target parameter.
2970 (rs6000_expand_builtin): Adjust call.
2971
2972 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2973
2974 PR target/58115
2975 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2976 * config/rs6000/rs6000.c: Include target-globals.h.
2977 (rs6000_set_current_function): Instead of doing target_reinit
2978 unconditionally, use save_target_globals_default_opts and
2979 restore_target_globals.
2980
2981 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2982 FPSCR.
2983 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2984 (rs6000_expand_builtin): Handle mffs and mtfsf.
2985 (rs6000_init_builtins): Define mffs and mtfsf.
2986 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2987 (rs6000_mffs): New pattern.
2988 (rs6000_mtfsf): New pattern.
2989
2990 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2991
2992 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2993 Start narrowing with START. Apply candidate-use pair
2994 and check overall cost in narrowing.
2995 (iv_ca_prune): Pass new argument.
2996
2997 2014-01-10 Jeff Law <law@redhat.com>
2998
2999 PR middle-end/59743
3000 * ree.c (combine_reaching_defs): Ensure the defining statement
3001 occurs before the extension when optimizing extensions with
3002 different source and destination hard registers.
3003
3004 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3005
3006 PR ipa/58585
3007 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3008 vtables into the type inheritance graph.
3009
3010 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3011
3012 PR rtl-optimization/59754
3013 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3014 modes in the REGNO != REGNO case.
3015
3016 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3017
3018 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3019
3020 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3021
3022 PR tree-optimization/59745
3023 * tree-predcom.c (tree_predictive_commoning_loop): Call
3024 free_affine_expand_cache if giving up because components is NULL.
3025
3026 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3027 GC in payload of target_globals struct instead of allocating them on
3028 the heap and the larger structs separately using GC.
3029 * target-globals.h (struct target_globals): Make regs, hard_regs,
3030 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3031 of GTY((skip)) and change type to void *.
3032 (reset_target_globals): Cast loads from those fields to corresponding
3033 types.
3034
3035 2014-01-10 Steve Ellcey <sellcey@mips.com>
3036
3037 PR plugins/59335
3038 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3039 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3040 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3041
3042 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3043
3044 PR target/59744
3045 * aarch64-modes.def (CC_Zmode): New flags mode.
3046 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3047 represents an equality.
3048 (aarch64_get_condition_code): Handle CC_Zmode.
3049 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3050
3051 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3052
3053 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3054 extraction in good case.
3055
3056 2014-01-10 Richard Biener <rguenther@suse.de>
3057
3058 PR tree-optimization/59374
3059 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3060 checking after SLP discovery. Mark stmts not participating
3061 in any SLP instance properly.
3062
3063 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3064
3065 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3066 when handling a SET rtx.
3067
3068 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3069
3070 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3071 (cortex-a57): Likewise.
3072 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3073
3074 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3075
3076 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3077 non-iwmmxt builtins.
3078
3079 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3080
3081 PR ipa/58252
3082 PR ipa/59226
3083 * ipa-devirt.c record_target_from_binfo): Take as argument
3084 stack of binfos and lookup matching one for virtual inheritance.
3085 (possible_polymorphic_call_targets_1): Update.
3086
3087 2014-01-10 Huacai Chen <chenhc@lemote.com>
3088
3089 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3090 kernel strings for Loongson-2E/2F/3A.
3091
3092 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3093
3094 PR middle-end/59670
3095 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3096 is_gimple_call before calling gimple_call_internal_p.
3097
3098 2014-01-09 Steve Ellcey <sellcey@mips.com>
3099
3100 * Makefile.in (TREE_FLOW_H): Remove.
3101 (TREE_SSA_H): Add file names from tree-flow.h.
3102 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3103 * tree.h: Remove tree-flow.h reference.
3104 * hash-table.h: Remove tree-flow.h reference.
3105 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3106 reference with tree-ssa-loop.h.
3107
3108 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3109
3110 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3111 default element-order behavior for -maltivec.
3112 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3113 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3114 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3115 when targeting big endian, at least for now.
3116 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3117
3118 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3119
3120 PR middle-end/47735
3121 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3122 var satisfies use_register_for_decl, just take into account type
3123 alignment, rather than decl alignment.
3124
3125 PR tree-optimization/59622
3126 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3127 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3128 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3129 Don't devirtualize for inplace at all. For targets.length () == 1,
3130 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3131
3132 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3133
3134 * config/i386/i386.md (cpu): Remove the unused btver1.
3135
3136 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3137
3138 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3139
3140 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3141
3142 PR target/58115
3143 * tree-core.h (struct target_globals): New forward declaration.
3144 (struct tree_target_option): Add globals field.
3145 * tree.h (TREE_TARGET_GLOBALS): Define.
3146 (prepare_target_option_nodes_for_pch): New prototype.
3147 * target-globals.h (struct target_globals): Define even if
3148 !SWITCHABLE_TARGET.
3149 * tree.c (prepare_target_option_node_for_pch,
3150 prepare_target_option_nodes_for_pch): New functions.
3151 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3152 * config/i386/i386.c: Include target-globals.h.
3153 (ix86_set_current_function): Instead of doing target_reinit
3154 unconditionally, use save_target_globals_default_opts and
3155 restore_target_globals.
3156
3157 2014-01-09 Richard Biener <rguenther@suse.de>
3158
3159 PR tree-optimization/59715
3160 * tree-cfg.h (split_critical_edges): Declare.
3161 * tree-cfg.c (split_critical_edges): Export.
3162 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3163
3164 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3165
3166 * cfgexpand.c (expand_stack_vars): Optionally disable
3167 asan stack protection.
3168 (expand_used_vars): Likewise.
3169 (partition_stack_vars): Likewise.
3170 * asan.c (asan_emit_stack_protection): Optionally disable
3171 after return stack usage.
3172 (instrument_derefs): Optionally disable memory access instrumentation.
3173 (instrument_builtin_call): Likewise.
3174 (instrument_strlen_call): Likewise.
3175 (asan_protect_global): Optionally disable global variables protection.
3176 * doc/invoke.texi: Added doc for new options.
3177 * params.def: Added new options.
3178 * params.h: Likewise.
3179
3180 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3181
3182 PR rtl-optimization/59724
3183 * ifcvt.c (cond_exec_process_if_block): Don't call
3184 flow_find_head_matching_sequence with 0 longest_match.
3185 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3186 non-active insns if !stop_after.
3187 (try_head_merge_bb): Revert 2014-01-07 changes.
3188
3189 2014-01-08 Jeff Law <law@redhat.com>
3190
3191 * ree.c (get_sub_rtx): New function, extracted from...
3192 (merge_def_and_ext): Here.
3193 (combine_reaching_defs): Use get_sub_rtx.
3194
3195 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3196
3197 * cgraph.h (varpool_variable_node): Do not choke on null node.
3198
3199 2014-01-08 Catherine Moore <clm@codesourcery.com>
3200
3201 * config/mips/mips.md (simple_return): Attempt to use JRC
3202 for microMIPS.
3203 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3204
3205 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3206
3207 PR rtl-optimization/59137
3208 * reorg.c (steal_delay_list_from_target): Call update_block for
3209 elided insns.
3210 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3211
3212 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3213
3214 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3215 two duplicate entries.
3216
3217 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3218
3219 Revert:
3220 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3221
3222 * config/mips/mips.c (mips_truncated_op_cost): New function.
3223 (mips_rtx_costs): Adjust test for BADDU.
3224 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3225
3226 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3227
3228 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3229 (*baddu_si): ...this new pattern.
3230
3231 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3232
3233 PR ipa/59722
3234 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3235
3236 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3237
3238 PR middle-end/57748
3239 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3240 inner_reference_p.
3241 (expand_expr, expand_normal): Adjust.
3242 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3243 inner_reference_p. Use inner_reference_p to expand inner references.
3244 (store_expr): Adjust.
3245 * cfgexpand.c (expand_call_stmt): Adjust.
3246
3247 2014-01-08 Rong Xu <xur@google.com>
3248
3249 * gcov-io.c (gcov_var): Move from gcov-io.h.
3250 (gcov_position): Ditto.
3251 (gcov_is_error): Ditto.
3252 (gcov_rewrite): Ditto.
3253 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3254 only part to libgcc/libgcov.h.
3255
3256 2014-01-08 Marek Polacek <polacek@redhat.com>
3257
3258 PR middle-end/59669
3259 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3260
3261 2014-01-08 Marek Polacek <polacek@redhat.com>
3262
3263 PR sanitizer/59667
3264 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3265
3266 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3267
3268 PR rtl-optimization/59649
3269 * stor-layout.c (get_mode_bounds): For BImode return
3270 0 and STORE_FLAG_VALUE.
3271
3272 2014-01-08 Richard Biener <rguenther@suse.de>
3273
3274 PR middle-end/59630
3275 * gimple.h (is_gimple_builtin_call): Remove.
3276 (gimple_builtin_call_types_compatible_p): New.
3277 (gimple_call_builtin_p): New overload.
3278 * gimple.c (is_gimple_builtin_call): Remove.
3279 (validate_call): Rename to ...
3280 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3281 check return types.
3282 (validate_type): New static function.
3283 (gimple_call_builtin_p): New overload and adjust.
3284 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3285 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3286 (gimple_fold_stmt_to_constant_1): Likewise.
3287 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3288
3289 2014-01-08 Richard Biener <rguenther@suse.de>
3290
3291 PR middle-end/59471
3292 * gimplify.c (gimplify_expr): Gimplify register-register type
3293 VIEW_CONVERT_EXPRs to separate stmts.
3294
3295 2014-01-07 Jeff Law <law@redhat.com>
3296
3297 PR middle-end/53623
3298 * ree.c (combine_set_extension): Handle case where source
3299 and destination registers in an extension insn are different.
3300 (combine_reaching_defs): Allow source and destination registers
3301 in extension to be different under limited circumstances.
3302 (add_removable_extension): Remove restriction that the
3303 source and destination registers in the extension are the same.
3304 (find_and_remove_re): Emit a copy from the extension's
3305 destination to its source after the defining insn if
3306 the source and destination registers are different.
3307
3308 PR middle-end/59285
3309 * ifcvt.c (merge_if_block): If we are merging a block with more than
3310 one successor with a block with no successors, remove any BARRIER
3311 after the second block.
3312
3313 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3314
3315 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3316
3317 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3318
3319 PR target/59652
3320 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3321 for 14-bit register offsets when INT14_OK_STRICT is false.
3322
3323 2014-01-07 Roland Stigge <stigge@antcom.de>
3324 Michael Meissner <meissner@linux.vnet.ibm.com>
3325
3326 PR 57386/target
3327 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3328 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3329
3330 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3331
3332 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3333 -mcpu.
3334
3335 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3336
3337 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3338 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3339 rtx is const0_rtx or not.
3340
3341 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3342
3343 PR target/58115
3344 * target-globals.c (save_target_globals): Remove this_fn_optab
3345 handling.
3346 * toplev.c: Include optabs.h.
3347 (target_reinit): Temporarily restore the global options if another
3348 set of options are in force.
3349
3350 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3351
3352 PR rtl-optimization/58668
3353 * cfgcleanup.c (flow_find_cross_jump): Don't count
3354 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3355 to determine what is counted.
3356 (flow_find_head_matching_sequence): Use active_insn_p to determine
3357 what is counted.
3358 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3359 counting change.
3360 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3361 determine what is counted.
3362
3363 PR tree-optimization/59643
3364 * tree-predcom.c (split_data_refs_to_components): If one dr is
3365 read and one write, determine_offset fails and the write isn't
3366 in the bad component, just put the read into the bad component.
3367
3368 2014-01-07 Mike Stump <mikestump@comcast.net>
3369 Jakub Jelinek <jakub@redhat.com>
3370
3371 PR pch/59436
3372 * tree-core.h (struct tree_optimization_option): Change optabs
3373 type from unsigned char * to void *.
3374 * optabs.c (init_tree_optimization_optabs): Adjust
3375 TREE_OPTIMIZATION_OPTABS initialization.
3376
3377 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3378
3379 PR target/59644
3380 * config/i386/i386.h (struct machine_function): Add
3381 no_drap_save_restore field.
3382 * config/i386/i386.c (ix86_save_reg): Use
3383 !cfun->machine->no_drap_save_restore instead of
3384 crtl->stack_realign_needed.
3385 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3386 this function clears frame_pointer_needed. Set
3387 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3388 and DRAP reg is needed.
3389
3390 2014-01-06 Marek Polacek <polacek@redhat.com>
3391
3392 PR c/57773
3393 * doc/implement-c.texi: Mention that other integer types are
3394 permitted as bit-field types in strictly conforming mode.
3395
3396 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3397
3398 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3399 is newly allocated.
3400
3401 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3402
3403 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3404
3405 2014-01-06 Martin Jambor <mjambor@suse.cz>
3406
3407 PR ipa/59008
3408 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3409 to int.
3410 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3411
3412 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3413
3414 PR debug/59350
3415 PR debug/59510
3416 * var-tracking.c (add_stores): Preserve the value of the source even if
3417 we don't record the store.
3418
3419 2014-01-06 Terry Guo <terry.guo@arm.com>
3420
3421 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3422
3423 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3424
3425 PR bootstrap/59541
3426 * config/darwin.c (darwin_function_section): Adjust return values to
3427 correspond to optimisation changes made in r206070.
3428
3429 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3430
3431 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3432 from prefetch_block tune setting.
3433 (nocona_cost): Correct size of prefetch block to 64.
3434
3435 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3436
3437 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3438 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3439 used to save the static chain register in the computation of the offset
3440 from which the FP registers need to be restored.
3441
3442 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3443
3444 PR tree-optimization/59519
3445 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3446 ICE if get_current_def (current_new_name) is already non-NULL, as long
3447 as it is a phi result of some other phi in *new_exit_bb that has
3448 the same argument.
3449
3450 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3451 or vmovdqu* for misaligned_operand.
3452 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3453 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3454 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3455 aligned_mem for AVX512F masked aligned load and store builtins and for
3456 non-temporal moves.
3457
3458 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3459
3460 PR tree-optimization/59651
3461 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3462 Address range for negative step should be added by TYPE_SIZE_UNIT.
3463
3464 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3465
3466 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3467 overlapping registers also for registers other than the stack pointer.
3468
3469 2014-01-03 Marek Polacek <polacek@redhat.com>
3470
3471 PR other/59661
3472 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3473 __builtin_FILE.
3474
3475 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3476
3477 PR target/59625
3478 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3479 asm goto as jump.
3480
3481 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3482 (push splitter): Use <P:MODE_SIZE> instead of
3483 GET_MODE_SIZE (<P:MODE>mode).
3484 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3485 (mov -1, reg peephole2): Likewise.
3486 * config/i386/sse.md (*mov<mode>_internal,
3487 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3488 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3489 *<code><mode>3, *andnot<mode>3<mask_name>,
3490 <mask_codefor><code><mode>3<mask_name>): Likewise.
3491 * config/i386/subst.md (mask_mode512bit_condition,
3492 sd_mask_mode512bit_condition): Likewise.
3493
3494 2014-01-02 Xinliang David Li <davidxl@google.com>
3495
3496 PR tree-optimization/59303
3497 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3498 (dump_predicates): Better output format.
3499 (pred_equal_p): New function.
3500 (is_neq_relop_p): Ditto.
3501 (is_neq_zero_form_p): Ditto.
3502 (pred_expr_equal_p): Ditto.
3503 (pred_neg_p): Ditto.
3504 (simplify_pred): Ditto.
3505 (simplify_preds_2): Ditto.
3506 (simplify_preds_3): Ditto.
3507 (simplify_preds_4): Ditto.
3508 (simplify_preds): Ditto.
3509 (push_pred): Ditto.
3510 (push_to_worklist): Ditto.
3511 (get_pred_info_from_cmp): Ditto.
3512 (is_degenerated_phi): Ditto.
3513 (normalize_one_pred_1): Ditto.
3514 (normalize_one_pred): Ditto.
3515 (normalize_one_pred_chain): Ditto.
3516 (normalize_preds): Ditto.
3517 (normalize_cond_1): Remove function.
3518 (normalize_cond): Ditto.
3519 (is_gcond_subset_of): Ditto.
3520 (is_subset_of_any): Ditto.
3521 (is_or_set_subset_of): Ditto.
3522 (is_and_set_subset_of): Ditto.
3523 (is_norm_cond_subset_of): Ditto.
3524 (pred_chain_length_cmp): Ditto.
3525 (convert_control_dep_chain_into_preds): Type change.
3526 (find_predicates): Ditto.
3527 (find_def_preds): Ditto.
3528 (destroy_predicates_vecs): Ditto.
3529 (find_matching_predicates_in_rest_chains): Ditto.
3530 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3531 (is_pred_expr_subset): Ditto.
3532 (is_pred_chain_subset_of): Ditto.
3533 (is_included_in): Ditto.
3534 (is_superset_of): Ditto.
3535
3536 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3537
3538 Update copyright years.
3539
3540 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3541
3542 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3543 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3544 config/arc/arc.md, config/arc/arc.opt,
3545 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3546 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3547 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3548 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3549 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3550 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3551 vtable-verify.c, vtable-verify.h: Use the standard form for the
3552 copyright notice.
3553
3554 2014-01-02 Tobias Burnus <burnus@net-b.de>
3555
3556 * gcc.c (process_command): Update copyright notice dates.
3557 * gcov-dump.c: Ditto.
3558 * gcov.c: Ditto.
3559 * doc/cpp.texi: Bump @copying's copyright year.
3560 * doc/cppinternals.texi: Ditto.
3561 * doc/gcc.texi: Ditto.
3562 * doc/gccint.texi: Ditto.
3563 * doc/gcov.texi: Ditto.
3564 * doc/install.texi: Ditto.
3565 * doc/invoke.texi: Ditto.
3566
3567 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3568
3569 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3570
3571 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3572
3573 * config/i386/sse.md (*mov<mode>_internal): Guard
3574 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3575
3576 PR rtl-optimization/59647
3577 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3578 new_rtx into UNSIGNED_FLOAT rtxes.
3579 \f
3580 Copyright (C) 2014 Free Software Foundation, Inc.
3581
3582 Copying and distribution of this file, with or without modification,
3583 are permitted in any medium without royalty provided the copyright
3584 notice and this notice are preserved.
This page took 0.190719 seconds and 5 git commands to generate.