| blob | 6748c0fc36a30b87995f83480352747060636bf3 |
1 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2 Kirill Yukhin <kirill.yukhin@intel.com>
4 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
5 (_mm_rcp28_round_ss): Ditto.
6 (_mm_rsqrt28_round_sd): Ditto.
7 (_mm_rsqrt28_round_ss): Ditto.
8 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
9 (_mm_rcp14_round_ss): Ditto.
10 (_mm_rsqrt14_round_sd): Ditto.
11 (_mm_rsqrt14_round_ss): Ditto.
12 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
13 the first input operand, get rid of match_dup.
14 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
15 attribute to sse.
16 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
17 Ditto.
18 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
19 operand as the first input operand, set type attribute.
20 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
21 Set type attribute.
22 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
23 operand as the first input operand, set type attribute.
25 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
28 bit of zero.
30 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
32 PR target/60207
33 * config/i386/i386.c (construct_container): Remove TFmode check
34 for X86_64_INTEGER_CLASS.
36 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
38 PR target/59794
39 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
40 only when -Wpsabi is enabled.
42 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
44 PR target/59799
45 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
46 passing arrays in registers are the same as for structs, so remove the
47 special case for them.
49 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
51 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
52 destination type, extract only the valid bits if the source type is not
53 integral and has a different mode.
55 2014-02-19 Richard Biener <rguenther@suse.de>
57 PR ipa/60243
58 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
59 for all calls.
61 2014-02-19 Richard Biener <rguenther@suse.de>
63 PR ipa/60243
64 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
65 (ipa_modify_call_arguments): Emit an argument load explicitely and
66 preserve virtual SSA form there and for the replacement call.
67 Do not update SSA form nor free dominance info.
69 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
71 * ipa.c (function_and_variable_visibility): Also clear WEAK
72 flag when disolving COMDAT_GROUP.
74 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
76 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
77 * ipa-prop.c (ipa_set_jf_known_type): Return early when
78 not devirtualizing.
79 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
80 do more sanity checks.
81 (detect_type_change): Return true when giving up early.
82 (compute_complex_assign_jump_func): Fix type parameter of
83 ipa_set_ancestor_jf.
84 (compute_complex_ancestor_jump_func): Likewise.
85 (update_jump_functions_after_inlining): Fix updating of
86 ancestor function.
87 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
89 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
91 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
92 inline clones when edge disappears.
94 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
96 PR target/60203
97 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
98 Split 64-bit moves into 2 patterns. Do not allow the use of
99 direct move for TDmode in little endian, since the decimal value
100 has little endian bytes within a word, but the 64-bit pieces are
101 ordered in a big endian fashion, and normal subreg's of TDmode are
102 not allowed.
103 (mov<mode>_64bit_dm): Likewise.
104 (movtd_64bit_nodm): Likewise.
106 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
108 PR tree-optimization/60174
109 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
110 statement of an SSA_NAME that occurs in an abnormal PHI node.
112 2014-02-18 Jakub Jelinek <jakub@redhat.com>
114 PR sanitizer/60142
115 * final.c (SEEN_BB): Remove.
116 (SEEN_NOTE, SEEN_EMITTED): Renumber.
117 (final_scan_insn): Don't force_source_line on second
118 NOTE_INSN_BASIC_BLOCK.
120 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
122 PR target/60205
123 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
124 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
125 (type_natural_mode): Warn ABI change when %zmm register is not
126 available for AVX512F vector value passing.
128 2014-02-18 Kai Tietz <ktietz@redhat.com>
130 PR target/60193
131 * config/i386/i386.c (ix86_expand_prologue): Use value in
132 rax register as displacement when restoring %r10 or %rax.
133 Fix wrong offset when restoring both registers.
135 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
137 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
138 assertion with conditional return.
140 2014-02-18 Jakub Jelinek <jakub@redhat.com>
141 Uros Bizjak <ubizjak@gmail.com>
143 PR driver/60233
144 * config/i386/driver-i386.c (host_detect_local_cpu): If
145 YMM state is not saved by the OS, also clear has_f16c. Move
146 CPUID 0x80000001 handling before YMM state saving checking.
148 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
150 PR rtl-optimization/58960
151 * haifa-sched.c (alloc_global_sched_pressure_data): New,
152 factored out from ...
153 (sched_init): ... here.
154 (free_global_sched_pressure_data): New, factored out from ...
155 (sched_finish): ... here.
156 * sched-int.h (free_global_sched_pressure_data): Declare.
157 * sched-rgn.c (nr_regions_initial): New static global.
158 (haifa_find_rgns): Initialize it.
159 (schedule_region): Disable sched-pressure for the newly
160 generated regions.
162 2014-02-17 Richard Biener <rguenther@suse.de>
164 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
165 release SSA defs of pattern stmts.
167 2014-02-17 Richard Biener <rguenther@suse.de>
169 * tree-inline.c (expand_call_inline): Release the virtual
170 operand defined by the call we are about to inline.
172 2014-02-17 Richard Biener <rguenther@suse.de>
174 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
176 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
177 Ilya Tocar <ilya.tocar@intel.com>
179 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
180 arguments order in builtin.
181 (_mm512_permutexvar_epi64): Ditto.
182 (_mm512_mask_permutexvar_epi64): Ditto
183 (_mm512_maskz_permutexvar_epi32): Ditto
184 (_mm512_permutexvar_epi32): Ditto
185 (_mm512_mask_permutexvar_epi32): Ditto
187 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
189 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
190 (p8_vmrgow): Likewise.
192 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
194 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
195 endian targets.
197 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
199 PR target/60203
200 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
201 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
202 into 64-bit and 32-bit moves. On 64-bit moves, add support for
203 using direct move instructions on ISA 2.07. Also adjust
204 instruction length for 64-bit.
205 (mov<mode>_64bit, TFmode/TDmode): Likewise.
206 (mov<mode>_32bit, TFmode/TDmode): Likewise.
208 2014-02-15 Alan Modra <amodra@gmail.com>
210 PR target/58675
211 PR target/57935
212 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
213 find_replacement on parts of insn rtl that might be reloaded.
215 2014-02-15 Richard Biener <rguenther@suse.de>
217 PR tree-optimization/60183
218 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
219 (tree_ssa_phiprop): Calculate and free post-dominators.
221 2014-02-14 Jeff Law <law@redhat.com>
223 PR rtl-optimization/60131
224 * ree.c (get_extended_src_reg): New function.
225 (combine_reaching_defs): Use it rather than assuming location of REG.
226 (find_and_remove_re): Verify first operand of extension is
227 a REG before adding the insns to the copy list.
229 2014-02-14 Roland McGrath <mcgrathr@google.com>
231 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
232 * configure: Regenerated.
233 * config.in: Regenerated.
234 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
235 instead of ASM_SHORT.
237 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
238 Richard Earnshaw <rearnsha@arm.com>
240 PR rtl-optimization/59535
241 * lra-constraints.c (process_alt_operands): Encourage alternative
242 when unassigned pseudo class is superset of the alternative class.
243 (inherit_reload_reg): Don't inherit when optimizing for code size.
244 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
245 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
246 modes not less than 4 for Thumb1.
248 2014-02-14 Kyle McMartin <kyle@redhat.com>
250 PR pch/60010
251 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
253 2014-02-14 Richard Biener <rguenther@suse.de>
255 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
256 (get_frame_arg): Drop the assert with langhook types_compatible_p.
257 Do not strip INDIRECT_REFs.
259 2014-02-14 Richard Biener <rguenther@suse.de>
261 PR lto/60179
262 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
263 DECL_FUNCTION_SPECIFIC_TARGET.
264 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
265 * tree-streamer-out.c (pack_ts_target_option): Remove.
266 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
267 (write_ts_function_decl_tree_pointers): Do not stream
268 DECL_FUNCTION_SPECIFIC_TARGET.
269 * tree-streamer-in.c (unpack_ts_target_option): Remove.
270 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
271 (lto_input_ts_function_decl_tree_pointers): Do not stream
272 DECL_FUNCTION_SPECIFIC_TARGET.
274 2014-02-14 Jakub Jelinek <jakub@redhat.com>
276 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
277 (get_initial_def_for_induction, vectorizable_induction): Ignore
278 debug stmts when looking for exit_phi.
279 (vectorizable_live_operation): Fix up condition.
281 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
283 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
284 nreverse() because it changes the content of original tree list.
286 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
288 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
289 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
291 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
293 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
294 GNU coding standards.
296 2014-02-13 Jakub Jelinek <jakub@redhat.com>
298 PR debug/60152
299 * dwarf2out.c (gen_subprogram_die): Don't call
300 add_calling_convention_attribute if subr_die is old_die.
302 2014-02-13 Sharad Singhai <singhai@google.com>
304 * doc/optinfo.texi: Fix order of nodes.
306 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
308 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
309 operands[2], not operands[3].
311 2014-02-13 Richard Biener <rguenther@suse.de>
313 PR bootstrap/59878
314 * doc/install.texi (ISL): Update recommended version to 0.12.2,
315 mention the possibility of an in-tree build.
316 (CLooG): Update recommended version to 0.18.1, mention the
317 possibility of an in-tree build and clarify that the ISL
318 bundled with CLooG does not work.
320 2014-02-13 Jakub Jelinek <jakub@redhat.com>
322 PR target/43546
323 * expr.c (compress_float_constant): If x is a hard register,
324 extend into a pseudo and then move to x.
326 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
328 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
329 caused by bad second argument to warning_at() with -mhotpatch and
330 nested functions (e.g. with gfortran).
332 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
334 * opts.c (option_name): Remove "enabled by default" rider.
336 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
338 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
340 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
341 Uros Bizjak <ubizjak@gmail.com>
343 PR target/60151
344 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
345 * configure: Regenerated.
347 2014-02-12 Richard Biener <rguenther@suse.de>
349 * vec.c (vec_prefix::calculate_allocation): Move as
350 inline variant to vec.h.
351 (vec_prefix::calculate_allocation_1): New out-of-line version.
352 * vec.h (vec_prefix::calculate_allocation_1): Declare.
353 (vec_prefix::m_has_auto_buf): Rename to ...
354 (vec_prefix::m_using_auto_storage): ... this.
355 (vec_prefix::calculate_allocation): Inline the easy cases
356 and dispatch to calculate_allocation_1 which doesn't need the
357 prefix address.
358 (va_heap::reserve): Use gcc_checking_assert.
359 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
360 m_using_auto_storage.
361 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
362 member and adjust.
363 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
364 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
365 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
367 2014-02-12 Richard Biener <rguenther@suse.de>
369 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
370 when we found a dependence.
372 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
374 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
375 common code...
376 (maybe_fold_stmt): ... into this new function.
377 * omp-low.c (lower_omp): Update comment.
379 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
380 last use.
382 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
383 dereference.
385 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
387 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
388 identifiers in comments.
389 (cortexa53_extra_costs): Likewise.
390 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
391 (cortexa7_extra_costs): Likewise.
392 (cortexa12_extra_costs): Likewise.
393 (cortexa15_extra_costs): Likewise.
394 (v7m_extra_costs): Likewise.
396 2014-02-12 Richard Biener <rguenther@suse.de>
398 PR middle-end/60092
399 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
400 of posix_memalign being successful.
401 (lower_stmt): Restrict lowering of posix_memalign to when
402 -ftree-bit-ccp is enabled.
404 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
406 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
407 arg_loc.
408 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
410 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
412 PR rtl-optimization/60116
413 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
414 other_insn once the combination has been validated.
416 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
418 PR lto/59468
419 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
420 and wrapper.
421 * ipa-devirt.c: Include demangle.h
422 (odr_violation_reported): New static variable.
423 (add_type_duplicate): Update odr_violations.
424 (maybe_record_node): Add completep parameter; update it.
425 (record_target_from_binfo): Add COMPLETEP parameter;
426 update it as needed.
427 (possible_polymorphic_call_targets_1): Likewise.
428 (struct polymorphic_call_target_d): Add nonconstruction_targets;
429 rename FINAL to COMPLETE.
430 (record_targets_from_bases): Sanity check we found the binfo;
431 fix COMPLETEP updating.
432 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
433 parameter, fix computing of COMPLETEP.
434 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
435 at LTO time do demangling.
436 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
437 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
438 parameter.
439 (gimple_get_virt_method_for_binfo): Likewise.
440 * gimple-fold.h (gimple_get_virt_method_for_binfo,
441 gimple_get_virt_method_for_vtable): Update prototypes.
443 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
445 PR target/49008
446 * genautomata.c (add_presence_absence): Fix typo with
447 {final_}presence_list.
449 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
451 PR target/60137
452 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
453 for VSX/Altivec vectors that land in GPR registers.
455 2014-02-11 Richard Henderson <rth@redhat.com>
456 Jakub Jelinek <jakub@redhat.com>
458 PR debug/59776
459 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
460 around drhs if type conversion to lacc->type is not useless.
462 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
464 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
465 tuning struct.
466 (cortex-a57.cortex-a53): Likewise.
467 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
469 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
471 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
472 arm_restrict_it.
474 2014-02-11 Renlin Li <Renlin.Li@arm.com>
476 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
477 add_options_for_arm_vfp3.
479 2014-02-11 Jeff Law <law@redhat.com>
481 PR middle-end/54041
482 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
483 object with an undesirable mode.
485 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
487 PR libgomp/60107
488 * config/i386/sol2-9.h: New file.
489 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
490 *-*-solaris2.9*): Use it.
492 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
494 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
495 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
497 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
499 * config/microblaze/microblaze.c: Extend mcpu version format
501 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
503 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
505 2014-02-10 Richard Henderson <rth@redhat.com>
507 PR target/59927
508 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
509 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
510 ms-abi vs -mno-accumulate-outgoing-args.
511 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
512 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
513 respect to ms-abi.
515 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
517 PR middle-end/60080
518 * cfgexpand.c (expand_asm_operands): Attach source location to
519 ASM_INPUT rtx objects.
520 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
522 2014-02-10 Nick Clifton <nickc@redhat.com>
524 * config/mn10300/mn10300.c (popcount): New function.
525 (mn10300_expand_prologue): Include saved registers in stack usage
526 count.
528 2014-02-10 Jeff Law <law@redhat.com>
530 PR middle-end/52306
531 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
532 when changing the SET_DEST of a prior insn to avoid an input reload.
534 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
536 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
537 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
538 -mcall-openbsd, or -mcall-linux.
539 (CC1_ENDIAN_BIG_SPEC): Remove.
540 (CC1_ENDIAN_LITTLE_SPEC): Remove.
541 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
542 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
543 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
544 and %cc1_endian_default.
545 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
547 2014-02-10 Richard Biener <rguenther@suse.de>
549 PR tree-optimization/60115
550 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
551 MEM_REF handling. Properly verify that the accesses are not
552 out of the objects bound.
554 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
556 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
557 coretex to cortex.
559 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
561 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
562 proper constants and fix formatting.
563 (possible_polymorphic_call_targets): Fix formatting.
565 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
566 Ilya Tocar <ilya.tocar@intel.com>
568 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
569 (_mm512_loadu_epi32): Renamed into...
570 (_mm512_loadu_si512): This.
571 (_mm512_storeu_epi32): Renamed into...
572 (_mm512_storeu_si512): This.
573 (_mm512_maskz_ceil_ps): Removed.
574 (_mm512_maskz_ceil_pd): Ditto.
575 (_mm512_maskz_floor_ps): Ditto.
576 (_mm512_maskz_floor_pd): Ditto.
577 (_mm512_floor_round_ps): Ditto.
578 (_mm512_floor_round_pd): Ditto.
579 (_mm512_ceil_round_ps): Ditto.
580 (_mm512_ceil_round_pd): Ditto.
581 (_mm512_mask_floor_round_ps): Ditto.
582 (_mm512_mask_floor_round_pd): Ditto.
583 (_mm512_mask_ceil_round_ps): Ditto.
584 (_mm512_mask_ceil_round_pd): Ditto.
585 (_mm512_maskz_floor_round_ps): Ditto.
586 (_mm512_maskz_floor_round_pd): Ditto.
587 (_mm512_maskz_ceil_round_ps): Ditto.
588 (_mm512_maskz_ceil_round_pd): Ditto.
589 (_mm512_expand_pd): Ditto.
590 (_mm512_expand_ps): Ditto.
591 * config/i386/i386.c (ix86_builtins): Remove
592 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
593 (bdesc_args): Ditto.
594 * config/i386/predicates.md (const1256_operand): New.
595 (const_1_to_2_operand): Ditto.
596 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
597 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
598 (*avx512pf_gatherpf<mode>sf): Ditto.
599 (avx512pf_gatherpf<mode>df): Ditto.
600 (*avx512pf_gatherpf<mode>df_mask): Ditto.
601 (*avx512pf_gatherpf<mode>df): Ditto.
602 (avx512pf_scatterpf<mode>sf): Ditto.
603 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
604 (*avx512pf_scatterpf<mode>sf): Ditto.
605 (avx512pf_scatterpf<mode>df): Ditto.
606 (*avx512pf_scatterpf<mode>df_mask): Ditto.
607 (*avx512pf_scatterpf<mode>df): Ditto.
608 (avx512f_expand<mode>): Removed.
609 (<shift_insn><mode>3<mask_name>): Change predicate type.
611 2014-02-08 Jakub Jelinek <jakub@redhat.com>
613 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
614 not at the end of datarefs vector use ordered_remove to avoid
615 reordering datarefs vector.
617 PR c/59984
618 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
619 mark local addressable non-static vars as GOVD_PRIVATE
620 instead of GOVD_LOCAL.
621 * omp-low.c (lower_omp_for): Move gimple_bind_vars
622 and BLOCK_VARS of gimple_bind_block to new_stmt rather
623 than copying them.
625 PR middle-end/60092
626 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
627 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
628 assume_aligned or alloc_align attributes.
629 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
630 arguments. Handle also assume_aligned and alloc_align attributes.
631 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
632 calls to functions with assume_aligned or alloc_align attributes.
633 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
635 2014-02-08 Terry Guo <terry.guo@arm.com>
637 * doc/invoke.texi: Document ARM -march=armv7e-m.
639 2014-02-08 Jakub Jelinek <jakub@redhat.com>
641 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
642 flag on __cilkrts_rethrow builtin.
644 PR ipa/60026
645 * ipa-cp.c (determine_versionability): Fail at -O0
646 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
647 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
649 Revert:
650 2014-02-04 Jakub Jelinek <jakub@redhat.com>
652 PR ipa/60026
653 * tree-inline.c (copy_forbidden): Fail for
654 __attribute__((optimize (0))) functions.
656 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
658 * varpool.c: Include pointer-set.h.
659 (varpool_remove_unreferenced_decls): Variables in other partitions
660 will not be output; be however careful to not lose information
661 about partitioning.
663 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
665 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
666 lookup in the vtable constructor.
668 2014-02-07 Jeff Law <law@redhat.com>
670 PR target/40977
671 * config/m68k/m68k.md (ashldi_extsi): Turn into a
672 define_insn_and_split.
674 * ipa-inline.c (inline_small_functions): Fix typos.
676 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
678 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
679 (s390_can_use_return_insn): Declare.
680 * config/s390/s390.h (EPILOGUE_USES): Define.
681 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
682 instructions.
683 (s390_chunkify_start): Handle return JUMP_LABELs.
684 (s390_early_mach): Emit a main_pool instruction on the entry edge.
685 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
686 (s390_can_use_return_insn): New functions.
687 (s390_fix_long_loop_prediction): Handle conditional returns.
688 (TARGET_SET_UP_BY_PROLOGUE): Define.
689 * config/s390/s390.md (ANY_RETURN): New code iterator.
690 (*creturn, *csimple_return, return, simple_return): New patterns.
692 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
694 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
695 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
696 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
697 REG_CFA_RESTORE list when deciding not to restore a register.
699 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
701 * config/s390/s390.c: Include tree-pass.h and context.h.
702 (s390_early_mach): New function, split out from...
703 (s390_emit_prologue): ...here.
704 (pass_data_s390_early_mach): New pass structure.
705 (pass_s390_early_mach): New class.
706 (s390_option_override): Create and register early_mach pass.
707 Move to end of file.
709 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
711 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
712 to match for the exit block.
714 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
716 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
717 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
718 Reject misaligned operands.
720 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
722 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
724 2014-02-07 Richard Biener <rguenther@suse.de>
726 PR middle-end/60092
727 * gimple-low.c (lower_builtin_posix_memalign): New function.
728 (lower_stmt): Call it to lower posix_memalign in a way
729 to make alignment info accessible.
731 2014-02-07 Jakub Jelinek <jakub@redhat.com>
733 PR c++/60082
734 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
735 __builtin_setjmp_receiver.
737 2014-02-07 Richard Biener <rguenther@suse.de>
739 PR middle-end/60092
740 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
741 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
742 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
743 Handle BUILT_IN_POSIX_MEMALIGN.
744 (find_func_clobbers): Likewise.
745 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
746 (call_may_clobber_ref_p_1): Likewise.
748 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
750 PR ipa/59918
751 * ipa-devirt.c (record_target_from_binfo): Remove overactive
752 sanity check.
754 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
756 PR ipa/59469
757 * lto-cgraph.c (lto_output_node): Use
758 symtab_get_symbol_partitioning_class.
759 (lto_output_varpool_node): likewise.
760 (symtab_get_symbol_partitioning_class): Move here from
761 lto/lto-partition.c
762 * cgraph.h (symbol_partitioning_class): Likewise.
763 (symtab_get_symbol_partitioning_class): Declare.
765 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
767 * ggc.h (ggc_internal_cleared_alloc): New macro.
768 * vec.h (vec_safe_copy): Handle memory stats.
769 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
770 * target-globals.c (save_target_globals): Likewise.
772 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
774 PR target/60077
775 * expr.c (emit_move_resolve_push): Export; be bit more selective
776 on when to clear alias set.
777 * expr.h (emit_move_resolve_push): Declare.
778 * function.h (struct function): Add tail_call_marked.
779 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
780 * config/i386/i386-protos.h (ix86_expand_push): Remove.
781 * config/i386/i386.md (TImode move expander): De not call
782 ix86_expand_push.
783 (FP push expanders): Preserve memory attributes.
784 * config/i386/sse.md (push<mode>1): Remove.
785 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
786 (ix86_expand_push): Remove.
787 * config/i386/mmx.md (push<mode>1): Remove.
789 2014-02-06 Jakub Jelinek <jakub@redhat.com>
791 PR rtl-optimization/60030
792 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
793 lopart with paradoxical subreg before shifting it up by hprec.
795 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
797 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
798 Remove extra newline at end of file.
799 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
800 (arm_issue_rate): Handle cortexa57.
801 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
802 (cortex-a57.cortex-a53): Likewise.
804 2014-02-06 Jakub Jelinek <jakub@redhat.com>
806 PR target/59575
807 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
808 don't record in REG_FRAME_RELATED_EXPR registers not set in that
809 bitmask.
810 (arm_expand_prologue): Adjust all callers.
811 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
812 info, registers also at the lowest numbered registers side. Use
813 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
814 XEXP.
816 PR debug/59992
817 * var-tracking.c (adjust_mems): Before adding a SET to
818 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
820 2014-02-06 Alan Modra <amodra@gmail.com>
822 PR target/60032
823 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
824 change SDmode to DDmode when lra_in_progress.
826 2014-02-06 Jakub Jelinek <jakub@redhat.com>
828 PR middle-end/59150
829 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
830 free_data_ref on the dr first, and before goto again also set dr
831 to the next dr. For simd_lane_access, free old datarefs[i] before
832 overwriting it. For get_vectype_for_scalar_type failure, don't
833 free_data_ref if simd_lane_access.
835 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
837 PR target/60062
838 * tree.h (opts_for_fn): New inline function.
839 (opt_for_fn): Define.
840 * config/i386/i386.c (ix86_function_regparm): Use
841 opt_for_fn (decl, optimize) instead of optimize.
843 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
845 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
846 for SYMBOL_REF in large memory model.
848 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
850 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
851 and crypto support.
852 (cortex-a57): Likewise.
853 (cortex-a57.cortex-a53): Likewise.
855 2014-02-06 Yury Gribov <y.gribov@samsung.com>
856 Kugan Vivekanandarajah <kuganv@linaro.org>
858 * config/arm/arm.c (arm_vector_alignment_reachable): Check
859 unaligned_access.
860 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
862 2014-02-06 Richard Biener <rguenther@suse.de>
864 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
865 set_loop_copy and initialize_original_copy_tables.
867 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
869 * config/aarch64/aarch64-simd.md
870 (aarch64_ashr_simddi): Change QI to SI.
872 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
873 Jakub Jelinek <jakub@redhat.com>
875 PR middle-end/60013
876 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
877 of the dataflow.
879 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
881 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
882 CODE_FOR_altivec_vpku[hw]um to
883 CODE_FOR_altivec_vpku[hw]um_direct.
884 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
885 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
886 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
887 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
889 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
891 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
892 generation for -maltivec=be.
893 (altivec_vsumsws): Simplify redundant test.
895 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
897 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
898 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
899 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
900 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
901 gen_altivec_vpkuwum.
902 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
903 BYTES_BIG_ENDIAN.
904 (altivec_vpks<VI_char>ss): Likewise.
905 (altivec_vpks<VI_char>us): Likewise.
906 (altivec_vpku<VI_char>us): Likewise.
907 (altivec_vpku<VI_char>um): Likewise.
908 (altivec_vpku<VI_char>um_direct): New (copy of
909 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
910 internal use).
911 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
912 target is little endian and -maltivec=be is not specified.
913 (*altivec_vupkhs<VU_char>_direct): New (copy of
914 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
915 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
916 target is little endian and -maltivec=be is not specified.
917 (*altivec_vupkls<VU_char>_direct): New (copy of
918 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
919 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
920 little endian and -maltivec=be is not specified.
921 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
922 little endian and -maltivec=be is not specified.
924 2014-02-05 Richard Henderson <rth@redhat.com>
926 PR debug/52727
927 * combine-stack-adj.c: Revert r206943.
928 * sched-int.h (struct deps_desc): Add last_args_size.
929 * sched-deps.c (init_deps): Initialize it.
930 (sched_analyze_insn): Add OUTPUT dependencies between insns that
931 contain REG_ARGS_SIZE notes.
933 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
935 * lto-cgraph.c (asm_nodes_output): Make global.
936 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
937 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
938 (driver_handle_option): Handle OPT_fwpa.
940 2014-02-05 Jakub Jelinek <jakub@redhat.com>
942 PR ipa/59947
943 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
944 a comment typo and formatting issue. If odr_hash hasn't been
945 created, return vNULL and set *completep to false.
947 PR middle-end/57499
948 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
949 bb with no successors.
951 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
953 PR target/59718
954 * doc/invoke.texi (-march): Clarify documentation for ARM.
955 (-mtune): Likewise.
956 (-mcpu): Likewise.
958 2014-02-05 Richard Biener <rguenther@suse.de>
960 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
961 when not vectorizing because of too many alias checks.
962 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
963 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
965 2014-02-05 Nick Clifton <nickc@redhat.com>
967 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
968 accept extended registers in any mode when compiling for the MN10300.
970 2014-02-05 Yury Gribov <y.gribov@samsung.com>
972 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
973 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
974 sanitization attributes.
975 (can_inline_edge_p): Likewise.
976 (sanitize_attrs_match_for_inline_p): New function.
978 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
980 * ipa-prop.c (detect_type_change): Shor circuit testing of
981 type changes on THIS pointer.
983 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
985 PR target/59777
986 * config/pa/pa.c (legitimize_tls_address): Return original address
987 if not passed a SYMBOL_REF rtx.
988 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
989 addresses.
990 (pa_emit_move_sequence): Simplify TLS source operands.
991 (pa_legitimate_constant_p): Reject all TLS constants.
992 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
993 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
995 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
997 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
998 groups when we know they are controlled by LTO.
999 * varasm.c (default_binds_local_p_1): If object is in other partition,
1000 it will be resolved locally.
1002 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1004 * config/host-linux.c (linux_gt_pch_use_address): Don't
1005 use SSIZE_MAX because it is not always defined.
1007 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1009 PR bootstrap/59913
1010 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1011 threshold for pseudo splitting.
1012 (update_ebb_live_info): Process call argument hard registers and
1013 hard registers from insn definition too.
1014 (max_small_class_regs_num): New constant.
1015 (inherit_in_ebb): Update live hard regs through EBBs. Update
1016 reloads_num only for small register classes. Don't split for
1017 outputs of jumps.
1019 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1021 PR ipa/60058
1022 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1023 is non-null.
1025 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1027 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1028 visibility is safe.
1030 2014-02-04 Marek Polacek <polacek@redhat.com>
1032 * gdbinit.in (pel): Define.
1034 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1036 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1037 behavior.
1039 2014-02-04 Richard Biener <rguenther@suse.de>
1041 PR lto/59723
1042 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1043 in function context local.
1044 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1045 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1046 similar to LTO_imported_decl_ref.
1048 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1050 PR tree-optimization/60002
1051 * cgraphclones.c (build_function_decl_skip_args): Clear
1052 DECL_LANG_SPECIFIC.
1054 PR tree-optimization/60023
1055 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1056 false to gsi_replace.
1057 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1058 has been in some EH region and vec_stmt could throw, add
1059 vec_stmt into the same EH region.
1060 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1061 has no lhs, ignore it.
1062 * internal-fn.c (expand_MASK_LOAD): Likewise.
1064 PR ipa/60026
1065 * tree-inline.c (copy_forbidden): Fail for
1066 __attribute__((optimize (0))) functions.
1068 PR other/58712
1069 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1070 is set, copy one less argument.
1071 (expand_simd_clones): Don't subtract clone_info->inbranch
1072 from simd_clone_struct_alloc argument.
1074 PR rtl-optimization/57915
1075 * recog.c (simplify_while_replacing): If all unary/binary/relational
1076 operation arguments are constant, attempt to simplify those.
1078 PR middle-end/59261
1079 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1080 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1082 2014-02-04 Richard Biener <rguenther@suse.de>
1084 PR tree-optimization/60012
1085 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1086 TBAA disambiguation to all DDRs.
1088 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1090 PR target/59788
1091 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1092 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1094 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1096 PR ipa/59882
1097 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1099 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1101 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1102 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1104 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1106 PR ipa/59831
1107 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1108 to figure out targets of polymorphic calls with known decl.
1109 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1110 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1111 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1112 (get_polymorphic_call_info): ... here.
1113 (get_polymorphic_call_info_from_invariant): New function.
1115 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1117 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1118 lookup via vtable pointer; check for type consistency
1119 and turn inconsitent facts into UNREACHABLE.
1120 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1121 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1122 type inconsistent querries; return UNREACHABLE instead.
1124 2014-02-03 Richard Henderson <rth@twiddle.net>
1126 PR tree-opt/59924
1127 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1128 already processed this node.
1129 (normalize_one_pred_1): Pass along mark_set.
1130 (normalize_one_pred): Create and destroy a pointer_set_t.
1131 (normalize_one_pred_chain): Likewise.
1133 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1135 PR gcov-profile/58602
1136 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1138 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1140 PR ipa/59831
1141 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1142 -fno-devirtualize; try to devirtualize by the knowledge of
1143 virtual table pointer given by aggregate propagation.
1144 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1145 (ipa_print_node_jump_functions): Dump also offset that
1146 is relevant for polymorphic calls.
1147 (determine_known_aggregate_parts): Add arg_type parameter; use it
1148 instead of determining the type from pointer type.
1149 (ipa_compute_jump_functions_for_edge): Update call of
1150 determine_known_aggregate_parts.
1151 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1152 (gimple_get_virt_method_for_binfo): ... here; simplify using
1153 vtable_pointer_value_to_vtable.
1154 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1155 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1156 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1157 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1158 POINTER_PLUS_EXPR.
1159 (vtable_pointer_value_to_binfo): ... here.
1160 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1162 2014-02-03 Teresa Johnson <tejohnson@google.com>
1164 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1165 redef of outer loop index variable.
1167 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1169 PR c++/53017
1170 PR c++/59211
1171 * doc/extend.texi (Function Attributes): Typo.
1173 2014-02-03 Cong Hou <congh@google.com>
1175 PR tree-optimization/60000
1176 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1177 if the vectorized statement is a store. A store statement can only
1178 appear at the end of pattern statements.
1180 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1182 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1183 (ix86_option_override_internal): Default long double to 64-bit for
1184 32-bit Bionic and to 128-bit for 64-bit Bionic.
1186 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1187 TARGET_LONG_DOUBLE_128 is true.
1188 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1190 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1191 (mlong-double-64): Negate -mlong-double-128.
1192 (mlong-double-128): New option.
1194 * config/i386/i386-c.c (ix86_target_macros): Define
1195 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1197 * doc/invoke.texi: Document -mlong-double-128.
1199 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1201 PR rtl-optimization/60024
1202 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1204 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1206 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1208 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1210 PR rtl-optimization/57662
1211 * sel-sched.c (code_motion_path_driver): Do not mark already not
1212 existing blocks in the visiting bitmap.
1214 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1216 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1217 on the insn being emitted.
1219 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1220 Will Deacon <will.deacon@arm.com>
1222 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1224 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1226 * config/arm/arm-tables.opt: Regenerate.
1228 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1230 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1231 for vector types other than V16QImode.
1232 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1233 define_expand, and call altivec_expand_vec_perm_le when producing
1234 code with little endian element order.
1235 (*altivec_vperm_<mode>_internal): New insn having previous
1236 behavior of altivec_vperm_<mode>.
1237 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1238 altivec_expand_vec_perm_le when producing code with little endian
1239 element order.
1240 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1241 behavior of altivec_vperm_<mode>_uns.
1243 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1245 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1246 (altivec_vsumsws): Add handling for -maltivec=be with a little
1247 endian target.
1248 (altivec_vsumsws_direct): New.
1249 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1250 gen_altivec_vsumsws.
1252 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1254 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1255 vtable_pointer_value_to_binfo): New functions.
1256 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1257 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1259 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1261 * config/nios2/nios2.md (load_got_register): Initialize GOT
1262 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1263 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1265 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1267 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1268 preserverd by passthrough, do not propagate the type.
1270 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1272 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1273 (mips_atomic_assign_expand_fenv): New function.
1274 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1276 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1278 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1279 (__builtin_mips_set_fcsr): Likewise.
1280 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1281 MIPS_USI_FTYPE_VOID.
1282 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1283 (mips16_expand_set_fcsr): Likewise.
1284 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1285 (mips16_set_fcsr_stub): Likewise.
1286 (mips16_get_fcsr_one_only_stub): New class.
1287 (mips16_set_fcsr_one_only_stub): Likewise.
1288 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1289 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1290 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1291 (hard_float): New availability predicate.
1292 (mips_builtins): Add get_fcsr and set_fcsr.
1293 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1294 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1295 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1296 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1297 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1298 patterns.
1300 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1302 * config/mips/mips.c (mips_one_only_stub): New class.
1303 (mips_need_mips16_rdhwr_p): Replace with...
1304 (mips16_rdhwr_stub): ...this new variable.
1305 (mips16_stub_call_address): New function.
1306 (mips16_rdhwr_one_only_stub): New class.
1307 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1308 (mips_output_mips16_rdhwr): Delete.
1309 (mips_finish_stub): New function.
1310 (mips_code_end): Use it to handle rdhwr stubs.
1312 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1314 PR target/60017
1315 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1316 when calculating size of integer atomic types.
1318 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1320 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1322 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1324 PR tree-optimization/60003
1325 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1326 * profile.c (branch_prob): Use gimple_call_builtin_p
1327 to check for BUILT_IN_SETJMP_RECEIVER.
1328 * tree-inline.c (copy_bb): Call notice_special_calls.
1330 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1332 PR bootstrap/59985
1333 * lra-constraints.c (process_alt_operands): Update reload_sum only
1334 on the first pass.
1336 2014-01-31 Richard Henderson <rth@redhat.com>
1338 PR middle-end/60004
1339 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1340 until after else_eh is processed.
1342 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1344 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1345 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1346 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1347 in smmintrin.h, remove them.
1348 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1349 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1350 * config/i386/i386.md (ROUND_SAE): Fix value.
1351 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1352 (const48_operand): New.
1353 * config/i386/subst.md (round), (round_expand): Use
1354 const_4_or_8_to_11_operand.
1355 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1357 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1359 * config/i386/constraints.md (Yk): Swap meaning with k.
1360 * config/i386/i386.md (movhi_internal): Change Yk to k.
1361 (movqi_internal): Ditto.
1362 (*k<logic><mode>): Ditto.
1363 (*andhi_1): Ditto.
1364 (*andqi_1): Ditto.
1365 (kandn<mode>): Ditto.
1366 (*<code>hi_1): Ditto.
1367 (*<code>qi_1): Ditto.
1368 (kxnor<mode>): Ditto.
1369 (kortestzhi): Ditto.
1370 (kortestchi): Ditto.
1371 (kunpckhi): Ditto.
1372 (*one_cmplhi2_1): Ditto.
1373 (*one_cmplqi2_1): Ditto.
1374 * config/i386/sse.md (): Change k to Yk.
1375 (avx512f_load<mode>_mask): Ditto.
1376 (avx512f_blendm<mode>): Ditto.
1377 (avx512f_store<mode>_mask): Ditto.
1378 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1379 (avx512f_storedqu<mode>_mask): Ditto.
1380 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1381 Ditto.
1382 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1383 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1384 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1385 (avx512f_maskcmp<mode>3): Ditto.
1386 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1387 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1388 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1389 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1390 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1391 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1392 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1393 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1394 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1395 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1396 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1397 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1398 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1399 (vec_extract_lo_<mode>_maskm): Ditto.
1400 (vec_extract_hi_<mode>_maskm): Ditto.
1401 (avx512f_vternlog<mode>_mask): Ditto.
1402 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1403 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1404 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1405 (avx512f_<code>v8div16qi2_mask): Ditto.
1406 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1407 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1408 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1409 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1410 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1411 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1412 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1413 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1414 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1415 (avx512cd_maskb_vec_dupv8di): Ditto.
1416 (avx512cd_maskw_vec_dupv16si): Ditto.
1417 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1418 (avx512f_vpermi2var<mode>3_mask): Ditto.
1419 (avx512f_vpermi2var<mode>3_mask): Ditto.
1420 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1421 (*avx512f_gathersi<mode>): Ditto.
1422 (*avx512f_gathersi<mode>_2): Ditto.
1423 (*avx512f_gatherdi<mode>): Ditto.
1424 (*avx512f_gatherdi<mode>_2): Ditto.
1425 (*avx512f_scattersi<mode>): Ditto.
1426 (*avx512f_scatterdi<mode>): Ditto.
1427 (avx512f_compress<mode>_mask): Ditto.
1428 (avx512f_compressstore<mode>_mask): Ditto.
1429 (avx512f_expand<mode>_mask): Ditto.
1430 * config/i386/subst.md (mask): Change k to Yk.
1431 (mask_scalar_merge): Ditto.
1432 (sd): Ditto.
1434 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1436 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1438 2014-01-31 Richard Biener <rguenther@suse.de>
1440 PR middle-end/59990
1441 * builtins.c (fold_builtin_memory_op): Make sure to not
1442 use a floating-point mode or a boolean or enumeral type for
1443 the copy operation.
1445 2014-01-30 DJ Delorie <dj@redhat.com>
1447 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1448 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1449 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1450 whenever main() has an epilogue.
1452 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1454 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1455 unused variable "field".
1456 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1457 (vsx_mergeh_<mode>): Likewise.
1458 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1459 (altivec_vmrghh): Likewise.
1460 (altivec_vmrghw): Likewise.
1461 (altivec_vmrglb): Likewise.
1462 (altivec_vmrglh): Likewise.
1463 (altivec_vmrglw): Likewise.
1464 (altivec_vspltb): Add missing uses.
1465 (altivec_vsplth): Likewise.
1466 (altivec_vspltw): Likewise.
1467 (altivec_vspltsf): Likewise.
1469 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1471 PR target/59923
1472 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1473 frame related instructions.
1475 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1477 PR rtl-optimization/59959
1478 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1479 any reload of register whose subreg is invalid.
1481 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1483 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1484 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1485 Add missing return type - void.
1487 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1489 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1490 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1491 remove element index adjustment for endian (now handled in vsx.md
1492 and altivec.md).
1493 (altivec_expand_vec_perm_const): Use
1494 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1495 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1496 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1497 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1498 define_expand and a new define_insn *altivec_vspltb_internal;
1499 adjust for -maltivec=be on a little endian target.
1500 (altivec_vspltb_direct): New.
1501 (altivec_vsplth): Divide into a define_expand and a new
1502 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1503 little endian target.
1504 (altivec_vsplth_direct): New.
1505 (altivec_vspltw): Divide into a define_expand and a new
1506 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1507 little endian target.
1508 (altivec_vspltw_direct): New.
1509 (altivec_vspltsf): Divide into a define_expand and a new
1510 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1511 a little endian target.
1513 2014-01-30 Richard Biener <rguenther@suse.de>
1515 PR tree-optimization/59993
1516 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1517 can propagate form the earlier stmt and avoid the transform
1518 when the intermediate result is needed.
1520 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1522 * README.Portability: Fix typo.
1524 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1526 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1527 comparison_operator with ordered_comparison_operator.
1529 2014-01-30 Nick Clifton <nickc@redhat.com>
1531 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1532 Rename to mn10300_store_multiple_regs.
1533 * config/mn10300/mn10300.c: Likewise.
1534 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1535 store_multiple_regs.
1536 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1537 Call mn10300_store_multiple_regs.
1539 2014-01-30 Nick Clifton <nickc@redhat.com>
1540 DJ Delorie <dj@redhat.com>
1542 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1543 %fp 2 to keep registers after it properly word-aligned.
1544 (rl78_alloc_physical_registers_umul): Handle the case where both
1545 input operands are the same.
1547 2014-01-30 Richard Biener <rguenther@suse.de>
1549 PR tree-optimization/59903
1550 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1551 check properly.
1553 2014-01-30 Jason Merrill <jason@redhat.com>
1555 PR c++/59633
1556 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1558 PR c++/59645
1559 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1561 2014-01-30 Richard Biener <rguenther@suse.de>
1563 PR tree-optimization/59951
1564 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1566 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1568 PR target/59784
1569 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1570 SFmode to DFmode case.
1572 2014-01-29 DJ Delorie <dj@redhat.com>
1574 * config/msp430/msp430.opt (-minrt): New.
1575 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1576 if -minrt given.
1577 (ENDFILE_SPEC): Likewise.
1579 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1581 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1582 (estimate_function_body_sizes): Use it.
1584 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1586 PR c++/58561
1587 * dwarf2out.c (is_cxx_auto): New.
1588 (is_base_type): Use it.
1589 (gen_type_die_with_usage): Likewise.
1591 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1593 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1594 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1595 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1596 -maltivec=be with LE targets.
1597 (vsx_mergeh_<mode>): Likewise.
1598 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1599 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1600 (altivec_vmrghb): Replace with define_expand and new
1601 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1602 (altivec_vmrghb_direct): New define_insn.
1603 (altivec_vmrghh): Replace with define_expand and new
1604 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1605 (altivec_vmrghh_direct): New define_insn.
1606 (altivec_vmrghw): Replace with define_expand and new
1607 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1608 (altivec_vmrghw_direct): New define_insn.
1609 (*altivec_vmrghsf): Adjust for endianness.
1610 (altivec_vmrglb): Replace with define_expand and new
1611 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1612 (altivec_vmrglb_direct): New define_insn.
1613 (altivec_vmrglh): Replace with define_expand and new
1614 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1615 (altivec_vmrglh_direct): New define_insn.
1616 (altivec_vmrglw): Replace with define_expand and new
1617 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1618 (altivec_vmrglw_direct): New define_insn.
1619 (*altivec_vmrglsf): Adjust for endianness.
1620 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1621 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1622 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1623 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1624 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1625 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1626 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1627 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1629 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1631 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1632 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1633 whitespace.
1635 2014-01-29 Richard Biener <rguenther@suse.de>
1637 PR tree-optimization/58742
1638 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1639 associate_pointerplus_align.
1640 (associate_pointerplus_diff): New function.
1641 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1642 and associate_pointerplus_diff.
1644 2014-01-29 Richard Biener <rguenther@suse.de>
1646 * lto-streamer.h (LTO_major_version): Bump to 3.
1647 (LTO_minor_version): Reset to 0.
1649 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1651 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1652 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1653 (arm_file_start): Generate correct asm header for armv7ve.
1654 * config/arm/bpabi.h: Add multilib support for armv7ve.
1655 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1656 and cortex-a15 to armv7ve.
1657 * config/arm/t-aprofile: Add multilib support for armv7ve.
1658 * doc/invoke.texi: Document -march=armv7ve.
1660 2014-01-29 Richard Biener <rguenther@suse.de>
1662 PR tree-optimization/58742
1663 * tree-ssa-forwprop.c (associate_plusminus): Return true
1664 if we changed sth, defer EH cleanup to ...
1665 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1666 (simplify_mult): New function.
1668 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1670 PR middle-end/59917
1671 PR tree-optimization/59920
1672 * tree.c (build_common_builtin_nodes): Remove
1673 __builtin_setjmp_dispatcher initialization.
1674 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1675 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1676 instead of gsi_after_labels + manually skipping debug stmts.
1677 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1678 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1679 * tree-inline.c (copy_edges_for_bb): Remove
1680 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1681 argument. Ignore computed_goto_p stmts. Don't call
1682 make_abnormal_goto_edges. If a call might need abnormal edges
1683 for non-local gotos, see if it already has an edge to
1684 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1685 with true argument, don't do anything then, otherwise add
1686 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1687 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1688 caller.
1689 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1690 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1691 (lower_stmt): Don't set data->calls_builtin_setjmp.
1692 (lower_builtin_setjmp): Adjust comment.
1693 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1694 * tree-cfg.c (found_computed_goto): Remove.
1695 (factor_computed_gotos): Remove.
1696 (make_goto_expr_edges): Return bool, true for computed gotos.
1697 Don't call make_abnormal_goto_edges.
1698 (build_gimple_cfg): Don't set found_computed_goto, don't call
1699 factor_computed_gotos.
1700 (computed_goto_p): No longer static.
1701 (make_blocks): Don't set found_computed_goto.
1702 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1703 (make_edges): If make_goto_expr_edges returns true, push bb
1704 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1705 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1706 vector. Record mapping between bbs and OpenMP regions if there
1707 are any, adjust make_gimple_omp_edges caller. Call
1708 handle_abnormal_edges.
1709 (make_abnormal_goto_edges): Remove.
1710 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1711 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1712 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1713 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1714 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1715 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1716 filling *region also set *region_idx to (*region)->entry->index.
1718 PR other/58712
1719 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1720 For REGs set ORIGINAL_REGNO.
1722 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1724 * doc/md.texi: Mention that a target shouldn't implement
1725 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1726 than hi/lo pair.
1728 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1730 PR tree-optimization/59594
1731 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1732 a copy of the datarefs vector rather than the vector itself.
1734 2014-01-28 Jason Merrill <jason@redhat.com>
1736 PR c++/53756
1737 * dwarf2out.c (auto_die): New static.
1738 (gen_type_die_with_usage): Handle C++1y 'auto'.
1739 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1740 on definition.
1742 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1744 PR target/59672
1745 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1746 (SPEC_X32): Likewise.
1747 (SPEC_64): Likewise.
1748 * config/i386/i386.c (ix86_option_override_internal): Turn off
1749 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1750 for TARGET_16BIT.
1751 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1752 * config/i386/i386.h (TARGET_16BIT): New macro.
1753 (TARGET_16BIT_P): Likewise.
1754 * config/i386/i386.opt: Add m16.
1755 * doc/invoke.texi: Document -m16.
1757 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1759 PR preprocessor/59935
1760 * input.c (location_get_source_line): Bail out on when line number
1761 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1763 2014-01-28 Richard Biener <rguenther@suse.de>
1765 PR tree-optimization/58742
1766 * tree-ssa-forwprop.c (associate_plusminus): Handle
1767 pointer subtraction of the form (T)(P + A) - (T)P.
1769 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1771 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1772 at const_int_cost.
1774 2014-01-28 Richard Biener <rguenther@suse.de>
1776 Revert
1777 2014-01-28 Richard Biener <rguenther@suse.de>
1779 PR rtl-optimization/45364
1780 PR rtl-optimization/59890
1781 * var-tracking.c (local_get_addr_clear_given_value): Handle
1782 already cleared slot.
1783 (val_reset): Handle not allocated local_get_addr_cache.
1784 (vt_find_locations): Use post-order on the inverted CFG.
1786 2014-01-28 Richard Biener <rguenther@suse.de>
1788 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1790 2014-01-28 Richard Biener <rguenther@suse.de>
1792 PR rtl-optimization/45364
1793 PR rtl-optimization/59890
1794 * var-tracking.c (local_get_addr_clear_given_value): Handle
1795 already cleared slot.
1796 (val_reset): Handle not allocated local_get_addr_cache.
1797 (vt_find_locations): Use post-order on the inverted CFG.
1799 2014-01-28 Alan Modra <amodra@gmail.com>
1801 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1802 * configure.ac <recursive call for build != host>: Define
1803 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1804 and LD_FOR_BUILD too.
1805 * configure: Regenerate.
1807 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1809 * config/i386/i386.c (get_builtin_code_for_version): Separate
1810 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1811 Broadwell from Haswell.
1813 2014-01-27 Steve Ellcey <sellcey@mips.com>
1815 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1816 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1817 * config/mips/mips.c (mips_option_override): Change setting
1818 of TARGET_DSP.
1819 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1820 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1821 Change from Mask to Var.
1823 2014-01-27 Jeff Law <law@redhat.com>
1825 * ipa-inline.c (inline_small_functions): Fix typo.
1827 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1829 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1830 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1831 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1832 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1833 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1834 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1835 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1836 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1837 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1838 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1839 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1840 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1841 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1842 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1843 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1844 (_mm512_storeu_epi64): Ditto.
1845 (_mm512_cmpge_epi32_mask): Ditto.
1846 (_mm512_cmpge_epu32_mask): Ditto.
1847 (_mm512_cmpge_epi64_mask): Ditto.
1848 (_mm512_cmpge_epu64_mask): Ditto.
1849 (_mm512_cmple_epi32_mask): Ditto.
1850 (_mm512_cmple_epu32_mask): Ditto.
1851 (_mm512_cmple_epi64_mask): Ditto.
1852 (_mm512_cmple_epu64_mask): Ditto.
1853 (_mm512_cmplt_epi32_mask): Ditto.
1854 (_mm512_cmplt_epu32_mask): Ditto.
1855 (_mm512_cmplt_epi64_mask): Ditto.
1856 (_mm512_cmplt_epu64_mask): Ditto.
1857 (_mm512_cmpneq_epi32_mask): Ditto.
1858 (_mm512_cmpneq_epu32_mask): Ditto.
1859 (_mm512_cmpneq_epi64_mask): Ditto.
1860 (_mm512_cmpneq_epu64_mask): Ditto.
1861 (_mm512_expand_pd): Ditto.
1862 (_mm512_expand_ps): Ditto.
1863 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1864 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1865 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1866 * config/i386/i386.c (ix86_builtins): Add
1867 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1868 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1869 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1870 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1871 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1872 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1873 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1874 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1875 IX86_BUILTIN_PMOVUSQW512_MEM.
1876 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1877 __builtin_ia32_pmovsqd512mem_mask,
1878 __builtin_ia32_pmovqd512mem_mask,
1879 __builtin_ia32_pmovusqw512mem_mask,
1880 __builtin_ia32_pmovsqw512mem_mask,
1881 __builtin_ia32_pmovqw512mem_mask,
1882 __builtin_ia32_pmovusdw512mem_mask,
1883 __builtin_ia32_pmovsdw512mem_mask,
1884 __builtin_ia32_pmovdw512mem_mask,
1885 __builtin_ia32_pmovqb512mem_mask,
1886 __builtin_ia32_pmovusqb512mem_mask,
1887 __builtin_ia32_pmovsqb512mem_mask,
1888 __builtin_ia32_pmovusdb512mem_mask,
1889 __builtin_ia32_pmovsdb512mem_mask,
1890 __builtin_ia32_pmovdb512mem_mask.
1891 (bdesc_args): Add __builtin_ia32_expanddf512,
1892 __builtin_ia32_expandsf512.
1893 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1894 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1895 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1896 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1897 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1898 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1899 (avx512f_<code>v8div16qi2_mask_store): This.
1900 (avx512f_expand<mode>): New.
1902 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1904 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1905 New.
1906 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1907 (_mm512_prefetch_i32scatter_pd): Ditto.
1908 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1909 (_mm512_prefetch_i64scatter_pd): Ditto.
1910 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1911 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1912 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1913 (_mm512_prefetch_i32scatter_ps): Ditto.
1914 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1915 (_mm512_prefetch_i64scatter_ps): Ditto.
1916 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1917 * config/i386/i386-builtin-types.def: Define
1918 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1919 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1920 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1921 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1922 IX86_BUILTIN_SCATTERPFQPD.
1923 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1924 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1925 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1926 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1927 __builtin_ia32_scatterpfqps.
1928 (ix86_expand_builtin): Expand new built-ins.
1929 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1930 fix memory access data type.
1931 (*avx512pf_gatherpf<mode>_mask): Ditto.
1932 (*avx512pf_gatherpf<mode>): Ditto.
1933 (avx512pf_scatterpf<mode>): Ditto.
1934 (*avx512pf_scatterpf<mode>_mask): Ditto.
1935 (*avx512pf_scatterpf<mode>): Ditto.
1936 (GATHER_SCATTER_SF_MEM_MODE): New.
1937 (avx512pf_gatherpf<mode>df): Ditto.
1938 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1939 (*avx512pf_scatterpf<mode>df): Ditto.
1941 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1943 PR bootstrap/59934
1944 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1945 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1946 reached.
1948 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1950 * common/config/arm/arm-common.c
1951 (arm_rewrite_mcpu): Handle multiple names.
1952 * config/arm/arm.h
1953 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1955 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1957 * gimple-builder.h (create_gimple_tmp): Delete.
1959 2014-01-27 Christian Bruel <christian.bruel@st.com>
1961 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1962 words comparisons.
1964 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1966 * config/pa/pa.md (call): Generate indirect long calls to non-local
1967 functions when outputing 32-bit code.
1968 (call_value): Likewise except for special call to buggy powf function.
1970 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1971 portable runtime and PIC indirect calls.
1972 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1973 and PIC call sequences. Use ldo instead of blr to set return register
1974 in PIC call sequence.
1976 2014-01-25 Walter Lee <walt@tilera.com>
1978 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1979 avoid clobbering a live register.
1981 2014-01-25 Walter Lee <walt@tilera.com>
1983 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1984 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1985 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1986 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1988 2014-01-25 Walter Lee <walt@tilera.com>
1990 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1991 arguments on even registers.
1992 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1993 STACK_BOUNDARY.
1994 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1995 (BIGGEST_ALIGNMENT): Ditto.
1996 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1998 2014-01-25 Walter Lee <walt@tilera.com>
2000 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2001 insns before bundling.
2002 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2004 2014-01-25 Walter Lee <walt@tilera.com>
2006 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2007 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2008 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2010 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2012 * config/mips/constraints.md (kl): Delete.
2013 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2014 define expands, using...
2015 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2016 instructions for MIPS16.
2017 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2018 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2020 2014-01-25 Walter Lee <walt@tilera.com>
2022 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2023 (clzdi2): Ditto.
2024 (ffsdi2): Ditto.
2026 2014-01-25 Walter Lee <walt@tilera.com>
2028 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2029 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2031 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2033 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2034 Handle XOR.
2036 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2038 * print-rtl.c (in_call_function_usage): New var.
2039 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2040 EXPR_LIST mode as mode and not as reg note name.
2042 PR middle-end/59561
2043 * cfgloopmanip.c (copy_loop_info): If
2044 loop->warned_aggressive_loop_optimizations, make sure
2045 the flag is set in target loop too.
2047 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2049 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2050 flag_cilkplus.
2051 * builtins.def: Likewise.
2052 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2053 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2054 * ira.c (ira_setup_eliminable_regset): Likewise.
2055 * omp-low.c (gate_expand_omp): Likewise.
2056 (execute_lower_omp): Likewise.
2057 (diagnose_sb_0): Likewise.
2058 (gate_diagnose_omp_blocks): Likewise.
2059 (simd_clone_clauses_extract): Likewise.
2060 (gate): Likewise.
2062 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2064 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2065 correction for little endian...
2066 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2067 here.
2069 2014-01-24 Jeff Law <law@redhat.com>
2071 PR tree-optimization/59919
2072 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2073 for non-returning calls.
2075 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2077 * common/config/aarch64/aarch64-common.c
2078 (aarch64_rewrite_mcpu): Handle multiple names.
2079 * config/aarch64/aarch64.h
2080 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2082 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2084 * input.c (add_file_to_cache_tab): Handle the case where fopen
2085 returns NULL.
2087 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2089 PR target/59929
2090 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2091 from push operand if code of push isn't PRE_DEC.
2093 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2095 PR target/59909
2096 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2097 -mquad-memory-atomic. Update -mquad-memory documentation to say
2098 it is only used for non-atomic loads/stores.
2100 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2101 -mquad-memory or -mquad-memory-atomic switches.
2103 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2104 -mquad-memory-atomic to ISA 2.07 support.
2106 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2107 to separate support of normal quad word memory operations (ldq, stq)
2108 from the atomic quad word memory operations.
2110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2111 support to separate non-atomic quad word operations from atomic
2112 quad word operations. Disable non-atomic quad word operations in
2113 little endian mode so that we don't have to swap words after the
2114 load and before the store.
2115 (quad_load_store_p): Add comment about atomic quad word support.
2116 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2117 options printed with -mdebug=reg.
2119 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2120 -mquad-memory-atomic as the test for whether we have quad word
2121 atomic instructions.
2122 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2123 or -mp8-vector are used, allow byte/half-word atomic operations.
2125 * config/rs6000/sync.md (load_lockedti): Insure that the address
2126 is a proper indexed or indirect address for the lqarx instruction.
2127 On little endian systems, swap the hi/lo registers after the lqarx
2128 instruction.
2129 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2130 insure the address is valid for the lqarx instruction.
2131 (store_conditionalti): Insure that the address is a proper indexed
2132 or indirect address for the stqcrx. instruction. On little endian
2133 systems, swap the hi/lo registers before doing the stqcrx.
2134 instruction.
2135 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2136 insure the address is valid for the stqcrx. instruction.
2138 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2139 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2140 type of quad memory support is available.
2142 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2144 PR regression/59915
2145 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2146 there is a danger of looping.
2148 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2150 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2151 force flag_ira_loop_pressure if set via command line.
2153 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2155 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2156 (ashr_simd): New builtin handling DI mode.
2157 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2158 (aarch64_sshr_simddi): New match pattern.
2159 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2160 (vshrd_n_s64): Likewise.
2161 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2163 2014-01-23 Nick Clifton <nickc@redhat.com>
2165 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2166 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2167 favour of mcu specific scripts.
2168 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2169 430x multilibs.
2171 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2172 Alex Velenko <Alex.Velenko@arm.com>
2174 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2175 (vaddv_s16): Likewise.
2176 (vaddv_s32): Likewise.
2177 (vaddv_u8): Likewise.
2178 (vaddv_u16): Likewise.
2179 (vaddv_u32): Likewise.
2180 (vaddvq_s8): Likewise.
2181 (vaddvq_s16): Likewise.
2182 (vaddvq_s32): Likewise.
2183 (vaddvq_s64): Likewise.
2184 (vaddvq_u8): Likewise.
2185 (vaddvq_u16): Likewise.
2186 (vaddvq_u32): Likewise.
2187 (vaddvq_u64): Likewise.
2188 (vaddv_f32): Likewise.
2189 (vaddvq_f32): Likewise.
2190 (vaddvq_f64): Likewise.
2191 (vmaxv_f32): Likewise.
2192 (vmaxv_s8): Likewise.
2193 (vmaxv_s16): Likewise.
2194 (vmaxv_s32): Likewise.
2195 (vmaxv_u8): Likewise.
2196 (vmaxv_u16): Likewise.
2197 (vmaxv_u32): Likewise.
2198 (vmaxvq_f32): Likewise.
2199 (vmaxvq_f64): Likewise.
2200 (vmaxvq_s8): Likewise.
2201 (vmaxvq_s16): Likewise.
2202 (vmaxvq_s32): Likewise.
2203 (vmaxvq_u8): Likewise.
2204 (vmaxvq_u16): Likewise.
2205 (vmaxvq_u32): Likewise.
2206 (vmaxnmv_f32): Likewise.
2207 (vmaxnmvq_f32): Likewise.
2208 (vmaxnmvq_f64): Likewise.
2209 (vminv_f32): Likewise.
2210 (vminv_s8): Likewise.
2211 (vminv_s16): Likewise.
2212 (vminv_s32): Likewise.
2213 (vminv_u8): Likewise.
2214 (vminv_u16): Likewise.
2215 (vminv_u32): Likewise.
2216 (vminvq_f32): Likewise.
2217 (vminvq_f64): Likewise.
2218 (vminvq_s8): Likewise.
2219 (vminvq_s16): Likewise.
2220 (vminvq_s32): Likewise.
2221 (vminvq_u8): Likewise.
2222 (vminvq_u16): Likewise.
2223 (vminvq_u32): Likewise.
2224 (vminnmv_f32): Likewise.
2225 (vminnmvq_f32): Likewise.
2226 (vminnmvq_f64): Likewise.
2228 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2230 * config/aarch64/aarch64-simd.md
2231 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2232 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2233 (*aarch64_mul3_elt<mode>): Likewise.
2234 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2235 (*aarch64_mul3_elt_to_64v2df): Likewise.
2236 (*aarch64_mla_elt<mode>): Likewise.
2237 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2238 (*aarch64_mls_elt<mode>): Likewise.
2239 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2240 (*aarch64_fma4_elt<mode>): Likewise.
2241 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2242 (*aarch64_fma4_elt_to_64v2df): Likewise.
2243 (*aarch64_fnma4_elt<mode>): Likewise.
2244 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2245 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2246 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2247 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2248 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2249 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2250 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2251 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2252 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2254 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2256 * config/aarch64/aarch64-simd.md
2257 (aarch64_be_checked_get_lane<mode>): New define_expand.
2258 * config/aarch64/aarch64-simd-builtins.def
2259 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2260 New builtin definition.
2261 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2262 Use new safe be builtin.
2264 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2266 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2267 New define_insn.
2268 (aarch64_be_st1<mode>): Likewise.
2269 (aarch_ld1<VALL:mode>): Define_expand modified.
2270 (aarch_st1<VALL:mode>): Likewise.
2271 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2272 (UNSPEC_ST1): Likewise.
2274 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2276 * config/microblaze/microblaze.md: Add trap insn and attribute
2278 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2280 PR preprocessor/58580
2281 * input.h (location_get_source_line): Take an additional line_size
2282 parameter.
2283 (void diagnostics_file_cache_fini): Declare new function.
2284 * input.c (struct fcache): New type.
2285 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2286 New static constants.
2287 (diagnostic_file_cache_init, total_lines_num)
2288 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2289 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2290 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2291 (get_next_line, read_next_line, goto_next_line, read_line_num):
2292 New static function definitions.
2293 (diagnostic_file_cache_fini): New function.
2294 (location_get_source_line): Take an additional output line_len
2295 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2296 read_line_num.
2297 * diagnostic.c (diagnostic_finish): Call
2298 diagnostic_file_cache_fini.
2299 (adjust_line): Take an additional input parameter for the length
2300 of the line, rather than calculating it with strlen.
2301 (diagnostic_show_locus): Adjust the use of
2302 location_get_source_line and adjust_line with respect to their new
2303 signature. While displaying a line now, do not stop at the first
2304 null byte. Rather, display the zero byte as a space and keep
2305 going until we reach the size of the line.
2306 * Makefile.in: Add vec.o to OBJS-libcommon
2308 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2309 Ilya Tocar <ilya.tocar@intel.com>
2311 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2312 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2313 (__builtin_ia32_kmov16): Ditto.
2314 * config/i386/i386.md (UNSPEC_KMOV): New.
2315 (kmovw): Ditto.
2317 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2319 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2320 (_mm512_storeu_si512): Ditto.
2322 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2324 PR target/52125
2325 * rtl.h (get_referenced_operands): Declare.
2326 * recog.c (get_referenced_operands): New function.
2327 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2328 operands have been referenced when recording LO_SUM references.
2330 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2332 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2334 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2336 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2337 Enable for generic and recent AMD targets.
2339 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2341 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2342 ARG_SIZE note when adjustment was eliminated.
2344 2014-01-22 Jeff Law <law@redhat.com>
2346 PR tree-optimization/59597
2347 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2348 in file. Accept new argument REGISTERING and use it to modify
2349 dump output appropriately.
2350 (register_jump_thread): Corresponding changes.
2351 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2352 thread paths involving joiner blocks. Add code to dump cancelled
2353 jump threading paths.
2355 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2357 PR rtl-optimization/59477
2358 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2359 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2361 2014-01-22 Tom Tromey <tromey@redhat.com>
2363 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2364 PARAMS.
2365 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2367 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2369 PR rtl-optimization/59896
2370 * lra-constraints.c (process_alt_operands): Check unused note for
2371 matched operands of insn with no output reloads.
2373 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2375 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2376 (mips_move_from_gpr_cost): Likewise.
2378 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2380 PR rtl-optimization/59858
2381 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2382 ira_class_hard_regs_num.
2383 (process_alt_operands): Increase reject for dying matched operand.
2385 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2387 PR target/59003
2388 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2389 smaller than size, perform several stores or loads and stores
2390 at dst + count - size to store or copy all of size bytes, rather
2391 than just last modesize bytes.
2393 2014-01-20 DJ Delorie <dj@redhat.com>
2395 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2396 that CLOBBERs are REGs before propogating their values.
2398 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2400 PR middle-end/59789
2401 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2402 (cgraph_inline_failed_type): New function.
2403 * cgraph.h (DEFCIFCODE): Add type.
2404 (cgraph_inline_failed_type_t): New enum.
2405 (cgraph_inline_failed_type): New prototype.
2406 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2407 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2408 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2409 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2410 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2411 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2412 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2413 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2414 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2415 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2416 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2417 OPTIMIZATION_MISMATCH.
2418 * tree-inline.c (expand_call_inline): Emit errors during
2419 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2421 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2423 PR target/59685
2424 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2425 mode attribute in insn output.
2427 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2429 * output.h (output_constant): Delete.
2430 * varasm.c (output_constant): Make private.
2432 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2434 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2436 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2438 PR middle-end/59860
2439 * tree.h (fold_builtin_strcat): New prototype.
2440 * builtins.c (fold_builtin_strcat): No longer static. Add len
2441 argument, if non-NULL, don't call c_strlen. Optimize
2442 directly into __builtin_memcpy instead of __builtin_strcpy.
2443 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2444 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2446 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2448 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2449 for SImode_address_operand operands, having only a REG argument.
2451 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2453 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2454 loader name using mbig-endian.
2455 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2457 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2459 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2460 (-mtune): Likewise.
2461 (-mcpu): Likewise.
2463 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2465 * config/aarch64/aarch64-protos.h
2466 (aarch64_cannot_change_mode_class_ptr): Declare.
2467 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2468 aarch64_cannot_change_mode_class_ptr): New.
2469 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2470 backend hook aarch64_cannot_change_mode_class.
2472 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2474 * common/config/aarch64/aarch64-common.c
2475 (aarch64_handle_option): Don't handle any option order logic here.
2476 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2477 selected_cpu, warn on architecture version mismatch.
2478 (aarch64_override_options): Fix parsing order for option strings.
2480 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2481 Iain Sandoe <iain@codesourcery.com>
2483 PR bootstrap/59496
2484 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2485 warning. Amend comment to reflect current functionality.
2487 2014-01-20 Richard Biener <rguenther@suse.de>
2489 PR middle-end/59860
2490 * builtins.c (fold_builtin_strcat): Remove case better handled
2491 by tree-ssa-strlen.c.
2493 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2495 * config/aarch64/aarch64.opt
2496 (mcpu, march, mtune): Make case-insensitive.
2498 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2500 PR target/59880
2501 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2502 if operands[1] is a REG or ZERO_EXTEND of a REG.
2504 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2506 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2508 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2510 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2511 long non-pic millicode calls.
2513 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2515 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2517 2014-01-19 Kito Cheng <kito@0xlab.org>
2519 * builtins.c (expand_movstr): Check movstr expand done or fail.
2521 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2522 H.J. Lu <hongjiu.lu@intel.com>
2524 PR target/59379
2525 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2526 to DImode for zero-extended addresses.
2528 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2530 PR rtl-optimization/57763
2531 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2532 on the new indirect jump_insn and increment LABEL_NUSES (label).
2534 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2536 PR bootstrap/59580
2537 PR bootstrap/59583
2538 * config.gcc (x86_archs): New variable.
2539 (x86_64_archs): Likewise.
2540 (x86_cpus): Likewise.
2541 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2542 --with-arch/--with-cpu= options.
2543 Support --with-arch=/--with-cpu={nehalem,westmere,
2544 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2546 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2548 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2549 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2551 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2553 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2555 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2557 PR target/58944
2558 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2559 clear cpp_get_options (parse_in)->warn_unused_macros for
2560 ix86_target_macros_internal with cpp_define.
2562 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2564 * jump.c (delete_related_insns): Keep (use (insn))s.
2565 * reorg.c (redundant_insn): Check for barriers too.
2567 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2569 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2571 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2573 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2574 call to $$dyncall when TARGET_LONG_CALLS is true.
2576 2014-01-17 Jeff Law <law@redhat.com>
2578 * ree.c (combine_set_extension): Temporarily disable test for
2579 changing number of hard registers.
2581 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2583 PR middle-end/58125
2584 * ipa-inline-analysis.c (inline_free_summary):
2585 Do not free summary of aliases.
2587 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2589 PR middle-end/59706
2590 * gimplify.c (gimplify_expr): Use create_tmp_var
2591 instead of create_tmp_var_raw. If cond doesn't have
2592 integral type, don't add the IFN_ANNOTATE builtin at all.
2594 2014-01-17 Martin Jambor <mjambor@suse.cz>
2596 PR ipa/59736
2597 * ipa-cp.c (prev_edge_clone): New variable.
2598 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2599 Also resize prev_edge_clone vector.
2600 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2601 (ipcp_edge_removal_hook): New function.
2602 (ipcp_driver): Register ipcp_edge_removal_hook.
2604 2014-01-17 Andrew Pinski <apinski@cavium.com>
2605 Steve Ellcey <sellcey@mips.com>
2607 PR target/59462
2608 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2609 of operator mode.
2611 2014-01-17 Jeff Law <law@redhat.com>
2613 PR middle-end/57904
2614 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2615 so that pass_ccp runs first.
2617 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2619 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2620 (ix86_adjust_cost): Use !TARGET_XXX.
2621 (do_reorder_for_imul): Likewise.
2622 (swap_top_of_ready_list): Likewise.
2623 (ix86_sched_reorder): Likewise.
2625 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2627 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2628 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2629 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2630 (intel_memset): New. Duplicate slm_memset.
2631 (intel_cost): New. Duplicate slm_cost.
2632 (m_INTEL): New macro.
2633 (processor_target_table): Add "intel".
2634 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2635 with PROCESSOR_INTEL for "intel".
2636 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2637 PROCESSOR_SILVERMONT.
2638 (ix86_issue_rate): Likewise.
2639 (ix86_adjust_cost): Likewise.
2640 (ia32_multipass_dfa_lookahead): Likewise.
2641 (swap_top_of_ready_list): Likewise.
2642 (ix86_sched_reorder): Likewise.
2643 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2644 instead of TARGET_OPT_AGU.
2645 * config/i386/i386.h (TARGET_INTEL): New.
2646 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2647 (processor_type): Add PROCESSOR_INTEL.
2648 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2649 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2651 2014-01-17 Marek Polacek <polacek@redhat.com>
2653 PR c/58346
2654 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2655 size is zero.
2657 2014-01-17 Richard Biener <rguenther@suse.de>
2659 PR tree-optimization/46590
2660 * opts.c (default_options_table): Add entries for
2661 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2662 all enabled at -O1 but not for -Og.
2663 * common.opt (fbranch-count-reg): Remove Init(1).
2664 (fmove-loop-invariants): Likewise.
2665 (ftree-pta): Likewise.
2667 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2669 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2670 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2671 decls to at least the GCC 4.8 used alignments.
2673 PR fortran/59440
2674 * tree-nested.c (convert_nonlocal_reference_stmt,
2675 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2676 of GIMPLE_BIND stmts, adjust associated decls.
2678 2014-01-17 Richard Biener <rguenther@suse.de>
2680 PR tree-optimization/46590
2681 * vec.h (vec<>::bseach): New member function implementing
2682 binary search according to C89 bsearch.
2683 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2684 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2685 bitmap pointer again. Make accesses_in_loop a flat array.
2686 (mem_ref_obstack): New global.
2687 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2688 (mark_ref_stored): Likewise.
2689 (ref_indep_loop_p_2): Likewise.
2690 (set_ref_stored_in_loop): New helper function.
2691 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2692 (memref_free): Adjust.
2693 (record_mem_ref_loc): Simplify.
2694 (gather_mem_refs_stmt): Adjust.
2695 (sort_locs_in_loop_postorder_cmp): New function.
2696 (analyze_memory_references): Sort accesses_in_loop after
2697 loop postorder number.
2698 (find_ref_loc_in_loop_cmp): New function.
2699 (for_all_locs_in_loop): Find relevant cluster of locs in
2700 accesses_in_loop and iterate without recursion.
2701 (execute_sm): Avoid uninit warning.
2702 (struct ref_always_accessed): Simplify.
2703 (ref_always_accessed::operator ()): Likewise.
2704 (ref_always_accessed_p): Likewise.
2705 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2706 loop postorder numbers here.
2707 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2708 numbers.
2710 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2712 PR c++/57945
2713 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2714 on decls for which assemble_alias has been called.
2716 2014-01-17 Nick Clifton <nickc@redhat.com>
2718 * config/msp430/msp430.opt: (mcpu): New option.
2719 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2720 (msp430_option_override): Parse target_cpu. If the MCU name
2721 matches a generic string, clear target_mcu.
2722 (msp430_attr): Allow numeric interrupt values up to 63.
2723 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2724 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2725 option.
2726 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2727 Add mcpu matches.
2728 * config/msp430/msp430.md (popm): Use %J rather than %I.
2729 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2730 (addhi_cy_i): Use immediate_operand for operand 2.
2731 * doc/invoke.texi: Document -mcpu option.
2733 2014-01-17 Richard Biener <rguenther@suse.de>
2735 PR rtl-optimization/38518
2736 * df.h (df_analyze_loop): Declare.
2737 * df-core.c: Include cfgloop.h.
2738 (df_analyze_1): Split out main part of df_analyze.
2739 (df_analyze): Adjust.
2740 (loop_inverted_post_order_compute): New function.
2741 (loop_post_order_compute): Likewise.
2742 (df_analyze_loop): New function avoiding whole-function
2743 postorder computes.
2744 * loop-invariant.c (find_defs): Use df_analyze_loop.
2745 (find_invariants): Adjust.
2746 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2748 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2750 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2751 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2753 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2755 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2756 traversal when removing references.
2758 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2760 PR ipa/59775
2761 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2763 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2765 PR middle-end/56791
2766 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2767 pushing a reload for an autoinc when we had previously reloaded an
2768 inner part of the address.
2770 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2772 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2773 field.
2774 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2775 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2776 when not giving up or versioning for alias only because of
2777 loop->safelen.
2778 (vect_analyze_data_ref_dependences): Set to true.
2779 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2780 is a GIMPLE_PHI.
2781 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2782 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2783 to the condition.
2785 PR middle-end/58344
2786 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2788 PR target/59839
2789 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2790 operand 0 predicate for gathers, use a new pseudo as subtarget.
2792 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2794 PR middle-end/59609
2795 * lra-constraints.c (process_alt_operands): Add printing debug info.
2796 Check absence of input/output reloads for matched operands too.
2798 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2800 PR rtl-optimization/59835
2801 * ira.c (ira_init_register_move_cost): Increase cost for
2802 impossible modes.
2804 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2806 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2808 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2810 PR target/59780
2811 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2812 non-register objects. Use gen_(high/low)part more consistently.
2813 Fix assertions.
2815 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2817 PR target/59844
2818 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2819 endian support, remove tests for WORDS_BIG_ENDIAN.
2820 (p8_mfvsrd_3_<mode>): Likewise.
2821 (reload_gpr_from_vsx<mode>): Likewise.
2822 (reload_gpr_from_vsxsf): Likewise.
2823 (p8_mfvsrd_4_disf): Likewise.
2825 2014-01-16 Richard Biener <rguenther@suse.de>
2827 PR rtl-optimization/46590
2828 * lcm.c (compute_antinout_edge): Use postorder iteration.
2829 (compute_laterin): Use inverted postorder iteration.
2831 2014-01-16 Nick Clifton <nickc@redhat.com>
2833 PR middle-end/28865
2834 * varasm.c (output_constant): Return the number of bytes actually
2835 emitted.
2836 (output_constructor_array_range): Update the field size with the
2837 number of bytes emitted by output_constant.
2838 (output_constructor_regular_field): Likewise. Also do not
2839 complain if the total number of bytes emitted is now greater
2840 than the expected fieldpos.
2841 * output.h (output_constant): Update prototype and descriptive comment.
2843 2014-01-16 Marek Polacek <polacek@redhat.com>
2845 PR middle-end/59827
2846 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2847 it is error_mark_node.
2849 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2851 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2852 VALID_AVX256_REG_OR_OI_MODE.
2854 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2856 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2857 current procedure should be profiled.
2859 2014-01-15 Andrew Pinski <apinski@cavium.com>
2861 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2862 of moving from/to the STACK_REG register class.
2864 2014-01-15 Richard Henderson <rth@redhat.com>
2866 PR debug/54694
2867 * reginfo.c (global_regs_decl): Globalize.
2868 * rtl.h (global_regs_decl): Declare.
2869 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2870 reserved via global_regs.
2872 2014-01-15 Teresa Johnson <tejohnson@google.com>
2874 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2876 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2878 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2879 and vmulosh rather than call gen_vec_widen_smult_*.
2880 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2881 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2882 (vec_widen_smult_even_v16qi): Likewise.
2883 (vec_widen_umult_even_v8hi): Likewise.
2884 (vec_widen_smult_even_v8hi): Likewise.
2885 (vec_widen_umult_odd_v16qi): Likewise.
2886 (vec_widen_smult_odd_v16qi): Likewise.
2887 (vec_widen_umult_odd_v8hi): Likewise.
2888 (vec_widen_smult_odd_v8hi): Likewise.
2889 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2890 vmuloub rather than call gen_vec_widen_umult_*.
2891 (vec_widen_umult_lo_v16qi): Likewise.
2892 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2893 vmulosb rather than call gen_vec_widen_smult_*.
2894 (vec_widen_smult_lo_v16qi): Likewise.
2895 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2896 rather than call gen_vec_widen_umult_*.
2897 (vec_widen_umult_lo_v8hi): Likewise.
2898 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2899 rather than call gen_vec_widen_smult_*.
2900 (vec_widen_smult_lo_v8hi): Likewise.
2902 2014-01-15 Jeff Law <law@redhat.com>
2904 PR tree-optimization/59747
2905 * ree.c (find_and_remove_re): Properly handle case where a second
2906 eliminated extension requires widening a copy created for elimination
2907 of a prior extension.
2908 (combine_set_extension): Ensure that the number of hard regs needed
2909 for a destination register does not change when we widen it.
2911 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2913 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2914 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2915 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2916 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2917 (avr-*-rtems*): Likewise.
2918 (bfin*-rtems*): Likewise.
2919 (moxie-*-rtems*): Likewise.
2920 (h8300-*-rtems*): Likewise.
2921 (i[34567]86-*-rtems*): Likewise.
2922 (lm32-*-rtems*): Likewise.
2923 (m32r-*-rtems*): Likewise.
2924 (m68k-*-rtems*): Likewise.
2925 (microblaze*-*-rtems*): Likewise.
2926 (mips*-*-rtems*): Likewise.
2927 (powerpc-*-rtems*): Likewise.
2928 (sh-*-rtems*): Likewise.
2929 (sparc-*-rtems*): Likewise.
2930 (sparc64-*-rtems*): Likewise.
2931 (v850-*-rtems*): Likewise.
2932 (m32c-*-rtems*): Likewise.
2934 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2936 PR rtl-optimization/59511
2937 * ira.c (ira_init_register_move_cost): Use memory costs for some
2938 cases of register move cost calculations.
2939 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2940 instead of BB frequency.
2941 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2942 * lra-assigns.c (find_hard_regno_for): Ditto.
2944 2014-01-15 Richard Biener <rguenther@suse.de>
2946 PR tree-optimization/59822
2947 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2948 (vectorizable_load): Use it to hoist defs of uses of invariant
2949 loads out of the loop.
2951 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2952 Kugan Vivekanandarajah <kuganv@linaro.org>
2954 PR target/59695
2955 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2956 truncation.
2958 2014-01-15 Richard Biener <rguenther@suse.de>
2960 PR rtl-optimization/59802
2961 * lcm.c (compute_available): Use inverted postorder to seed
2962 the initial worklist.
2964 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2966 PR target/59803
2967 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2968 ADDR_REGS for invalid symrefs in non-PIC code.
2970 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2972 PR other/58712
2973 * builtins.c (determine_block_size): Initialize *probable_max_size
2974 even if len_rtx is CONST_INT.
2976 2014-01-14 Andrew Pinski <apinski@cavium.com>
2978 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2979 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2980 (cortexa53_tunings): Likewise.
2981 (aarch64_sched_issue_rate): New function.
2982 (TARGET_SCHED_ISSUE_RATE): Define.
2984 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2986 * ira-costs.c (find_costs_and_classes): Add missed
2987 ira_init_register_move_cost_if_necessary.
2989 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2991 PR target/59787
2992 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2994 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2996 PR target/59794
2997 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2998 to indicate if type is used for function return value. Warn ABI
2999 change if the vector mode isn't available for function return value.
3000 (ix86_function_arg_advance): Pass false to type_natural_mode.
3001 (ix86_function_arg): Likewise.
3002 (ix86_gimplify_va_arg): Likewise.
3003 (function_arg_32): Don't warn ABI change.
3004 (ix86_function_value): Pass true to type_natural_mode.
3005 (ix86_return_in_memory): Likewise.
3006 (ix86_struct_value_rtx): Removed.
3007 (TARGET_STRUCT_VALUE_RTX): Likewise.
3009 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3011 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3012 converting a conditional jump into a conditional return.
3014 2014-01-14 Richard Biener <rguenther@suse.de>
3016 PR tree-optimization/58921
3017 PR tree-optimization/59006
3018 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3019 hoisting invariant stmts.
3020 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3021 invariant loads on the preheader edge if possible.
3023 2014-01-14 Joey Ye <joey.ye@arm.com>
3025 * doc/plugin.texi (Building GCC plugins): Update to C++.
3027 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3029 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3030 (_mm_rcp28_round_ss): Ditto.
3031 (_mm_rsqrt28_round_sd): Ditto.
3032 (_mm_rsqrt28_round_ss): Ditto.
3033 (_mm_rcp28_sd): Ditto.
3034 (_mm_rcp28_ss): Ditto.
3035 (_mm_rsqrt28_sd): Ditto.
3036 (_mm_rsqrt28_ss): Ditto.
3037 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3038 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3039 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3040 (IX86_BUILTIN_RCP28SD): Ditto.
3041 (IX86_BUILTIN_RCP28SS): Ditto.
3042 (IX86_BUILTIN_RSQRT28SD): Ditto.
3043 (IX86_BUILTIN_RSQRT28SS): Ditto.
3044 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3045 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3046 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3047 (ix86_expand_special_args_builtin): Expand new FTYPE.
3048 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3049 (srcp14<mode>): Make insn unary.
3050 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3051 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3052 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3053 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3054 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3055 Fix rounding: make it SAE only.
3056 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3057 Ditto.
3058 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3059 Ditto.
3060 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3061 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3062 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3063 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3064 (round_saeonly_mask_scalar_operand4): Ditto.
3065 (round_saeonly_mask_scalar_op3): Ditto.
3066 (round_saeonly_mask_scalar_op4): Ditto.
3068 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3070 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3071 Implement -maltivec=be for vec_insert and vec_extract.
3073 2014-01-10 DJ Delorie <dj@redhat.com>
3075 * config/msp430/msp430.md (call_internal): Don't allow memory
3076 references with SP as the base register.
3077 (call_value_internal): Likewise.
3078 * config/msp430/constraints.md (Yc): New. For memory references
3079 that don't use SP as a base register.
3081 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3082 "an integer without a # prefix"
3083 * config/msp430/msp430.md (epilogue_helper): Use it.
3085 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3087 PR target/59617
3088 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3089 AVX512F gather builtins.
3090 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3091 on gather decls with INTEGER_TYPE masktype.
3092 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3093 directly into the builtin rather than hoisting it before loop.
3095 PR tree-optimization/59387
3096 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3097 (scev_const_prop): If folded_casts and type has undefined overflow,
3098 use force_gimple_operand instead of force_gimple_operand_gsi and
3099 for each added stmt if it is assign with
3100 arith_code_with_undefined_signed_overflow, call
3101 rewrite_to_defined_overflow.
3102 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3103 gimple-fold.h instead.
3104 (arith_code_with_undefined_signed_overflow,
3105 rewrite_to_defined_overflow): Moved to ...
3106 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3107 rewrite_to_defined_overflow): ... here. No longer static.
3108 Include gimplify-me.h.
3109 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3110 rewrite_to_defined_overflow): New prototypes.
3112 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3114 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3116 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3118 * builtins.c (get_object_alignment_2): Minor tweak.
3119 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3121 2014-01-13 Christian Bruel <christian.bruel@st.com>
3123 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3124 optimized non constant lengths.
3126 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3128 PR libgomp/59194
3129 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3130 load as __atomic_load_N if possible.
3132 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3134 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3135 target parameter.
3136 (rs6000_expand_builtin): Adjust call.
3138 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3140 PR target/58115
3141 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3142 * config/rs6000/rs6000.c: Include target-globals.h.
3143 (rs6000_set_current_function): Instead of doing target_reinit
3144 unconditionally, use save_target_globals_default_opts and
3145 restore_target_globals.
3147 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3148 FPSCR.
3149 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3150 (rs6000_expand_builtin): Handle mffs and mtfsf.
3151 (rs6000_init_builtins): Define mffs and mtfsf.
3152 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3153 (rs6000_mffs): New pattern.
3154 (rs6000_mtfsf): New pattern.
3156 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3158 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3159 Start narrowing with START. Apply candidate-use pair
3160 and check overall cost in narrowing.
3161 (iv_ca_prune): Pass new argument.
3163 2014-01-10 Jeff Law <law@redhat.com>
3165 PR middle-end/59743
3166 * ree.c (combine_reaching_defs): Ensure the defining statement
3167 occurs before the extension when optimizing extensions with
3168 different source and destination hard registers.
3170 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3172 PR ipa/58585
3173 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3174 vtables into the type inheritance graph.
3176 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3178 PR rtl-optimization/59754
3179 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3180 modes in the REGNO != REGNO case.
3182 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3184 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3186 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3188 PR tree-optimization/59745
3189 * tree-predcom.c (tree_predictive_commoning_loop): Call
3190 free_affine_expand_cache if giving up because components is NULL.
3192 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3193 GC in payload of target_globals struct instead of allocating them on
3194 the heap and the larger structs separately using GC.
3195 * target-globals.h (struct target_globals): Make regs, hard_regs,
3196 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3197 of GTY((skip)) and change type to void *.
3198 (reset_target_globals): Cast loads from those fields to corresponding
3199 types.
3201 2014-01-10 Steve Ellcey <sellcey@mips.com>
3203 PR plugins/59335
3204 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3205 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3206 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3208 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3210 PR target/59744
3211 * aarch64-modes.def (CC_Zmode): New flags mode.
3212 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3213 represents an equality.
3214 (aarch64_get_condition_code): Handle CC_Zmode.
3215 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3217 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3219 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3220 extraction in good case.
3222 2014-01-10 Richard Biener <rguenther@suse.de>
3224 PR tree-optimization/59374
3225 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3226 checking after SLP discovery. Mark stmts not participating
3227 in any SLP instance properly.
3229 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3231 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3232 when handling a SET rtx.
3234 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3236 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3237 (cortex-a57): Likewise.
3238 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3240 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3242 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3243 non-iwmmxt builtins.
3245 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3247 PR ipa/58252
3248 PR ipa/59226
3249 * ipa-devirt.c record_target_from_binfo): Take as argument
3250 stack of binfos and lookup matching one for virtual inheritance.
3251 (possible_polymorphic_call_targets_1): Update.
3253 2014-01-10 Huacai Chen <chenhc@lemote.com>
3255 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3256 kernel strings for Loongson-2E/2F/3A.
3258 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3260 PR middle-end/59670
3261 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3262 is_gimple_call before calling gimple_call_internal_p.
3264 2014-01-09 Steve Ellcey <sellcey@mips.com>
3266 * Makefile.in (TREE_FLOW_H): Remove.
3267 (TREE_SSA_H): Add file names from tree-flow.h.
3268 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3269 * tree.h: Remove tree-flow.h reference.
3270 * hash-table.h: Remove tree-flow.h reference.
3271 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3272 reference with tree-ssa-loop.h.
3274 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3276 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3277 default element-order behavior for -maltivec.
3278 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3279 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3280 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3281 when targeting big endian, at least for now.
3282 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3284 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3286 PR middle-end/47735
3287 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3288 var satisfies use_register_for_decl, just take into account type
3289 alignment, rather than decl alignment.
3291 PR tree-optimization/59622
3292 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3293 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3294 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3295 Don't devirtualize for inplace at all. For targets.length () == 1,
3296 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3298 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3300 * config/i386/i386.md (cpu): Remove the unused btver1.
3302 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3304 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3306 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3308 PR target/58115
3309 * tree-core.h (struct target_globals): New forward declaration.
3310 (struct tree_target_option): Add globals field.
3311 * tree.h (TREE_TARGET_GLOBALS): Define.
3312 (prepare_target_option_nodes_for_pch): New prototype.
3313 * target-globals.h (struct target_globals): Define even if
3314 !SWITCHABLE_TARGET.
3315 * tree.c (prepare_target_option_node_for_pch,
3316 prepare_target_option_nodes_for_pch): New functions.
3317 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3318 * config/i386/i386.c: Include target-globals.h.
3319 (ix86_set_current_function): Instead of doing target_reinit
3320 unconditionally, use save_target_globals_default_opts and
3321 restore_target_globals.
3323 2014-01-09 Richard Biener <rguenther@suse.de>
3325 PR tree-optimization/59715
3326 * tree-cfg.h (split_critical_edges): Declare.
3327 * tree-cfg.c (split_critical_edges): Export.
3328 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3330 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3332 * cfgexpand.c (expand_stack_vars): Optionally disable
3333 asan stack protection.
3334 (expand_used_vars): Likewise.
3335 (partition_stack_vars): Likewise.
3336 * asan.c (asan_emit_stack_protection): Optionally disable
3337 after return stack usage.
3338 (instrument_derefs): Optionally disable memory access instrumentation.
3339 (instrument_builtin_call): Likewise.
3340 (instrument_strlen_call): Likewise.
3341 (asan_protect_global): Optionally disable global variables protection.
3342 * doc/invoke.texi: Added doc for new options.
3343 * params.def: Added new options.
3344 * params.h: Likewise.
3346 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3348 PR rtl-optimization/59724
3349 * ifcvt.c (cond_exec_process_if_block): Don't call
3350 flow_find_head_matching_sequence with 0 longest_match.
3351 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3352 non-active insns if !stop_after.
3353 (try_head_merge_bb): Revert 2014-01-07 changes.
3355 2014-01-08 Jeff Law <law@redhat.com>
3357 * ree.c (get_sub_rtx): New function, extracted from...
3358 (merge_def_and_ext): Here.
3359 (combine_reaching_defs): Use get_sub_rtx.
3361 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3363 * cgraph.h (varpool_variable_node): Do not choke on null node.
3365 2014-01-08 Catherine Moore <clm@codesourcery.com>
3367 * config/mips/mips.md (simple_return): Attempt to use JRC
3368 for microMIPS.
3369 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3371 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3373 PR rtl-optimization/59137
3374 * reorg.c (steal_delay_list_from_target): Call update_block for
3375 elided insns.
3376 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3378 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3380 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3381 two duplicate entries.
3383 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3385 Revert:
3386 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3388 * config/mips/mips.c (mips_truncated_op_cost): New function.
3389 (mips_rtx_costs): Adjust test for BADDU.
3390 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3392 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3394 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3395 (*baddu_si): ...this new pattern.
3397 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3399 PR ipa/59722
3400 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3402 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3404 PR middle-end/57748
3405 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3406 inner_reference_p.
3407 (expand_expr, expand_normal): Adjust.
3408 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3409 inner_reference_p. Use inner_reference_p to expand inner references.
3410 (store_expr): Adjust.
3411 * cfgexpand.c (expand_call_stmt): Adjust.
3413 2014-01-08 Rong Xu <xur@google.com>
3415 * gcov-io.c (gcov_var): Move from gcov-io.h.
3416 (gcov_position): Ditto.
3417 (gcov_is_error): Ditto.
3418 (gcov_rewrite): Ditto.
3419 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3420 only part to libgcc/libgcov.h.
3422 2014-01-08 Marek Polacek <polacek@redhat.com>
3424 PR middle-end/59669
3425 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3427 2014-01-08 Marek Polacek <polacek@redhat.com>
3429 PR sanitizer/59667
3430 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3432 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3434 PR rtl-optimization/59649
3435 * stor-layout.c (get_mode_bounds): For BImode return
3436 0 and STORE_FLAG_VALUE.
3438 2014-01-08 Richard Biener <rguenther@suse.de>
3440 PR middle-end/59630
3441 * gimple.h (is_gimple_builtin_call): Remove.
3442 (gimple_builtin_call_types_compatible_p): New.
3443 (gimple_call_builtin_p): New overload.
3444 * gimple.c (is_gimple_builtin_call): Remove.
3445 (validate_call): Rename to ...
3446 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3447 check return types.
3448 (validate_type): New static function.
3449 (gimple_call_builtin_p): New overload and adjust.
3450 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3451 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3452 (gimple_fold_stmt_to_constant_1): Likewise.
3453 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3455 2014-01-08 Richard Biener <rguenther@suse.de>
3457 PR middle-end/59471
3458 * gimplify.c (gimplify_expr): Gimplify register-register type
3459 VIEW_CONVERT_EXPRs to separate stmts.
3461 2014-01-07 Jeff Law <law@redhat.com>
3463 PR middle-end/53623
3464 * ree.c (combine_set_extension): Handle case where source
3465 and destination registers in an extension insn are different.
3466 (combine_reaching_defs): Allow source and destination registers
3467 in extension to be different under limited circumstances.
3468 (add_removable_extension): Remove restriction that the
3469 source and destination registers in the extension are the same.
3470 (find_and_remove_re): Emit a copy from the extension's
3471 destination to its source after the defining insn if
3472 the source and destination registers are different.
3474 PR middle-end/59285
3475 * ifcvt.c (merge_if_block): If we are merging a block with more than
3476 one successor with a block with no successors, remove any BARRIER
3477 after the second block.
3479 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3481 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3483 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3485 PR target/59652
3486 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3487 for 14-bit register offsets when INT14_OK_STRICT is false.
3489 2014-01-07 Roland Stigge <stigge@antcom.de>
3490 Michael Meissner <meissner@linux.vnet.ibm.com>
3492 PR 57386/target
3493 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3494 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3496 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3498 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3499 -mcpu.
3501 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3503 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3504 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3505 rtx is const0_rtx or not.
3507 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3509 PR target/58115
3510 * target-globals.c (save_target_globals): Remove this_fn_optab
3511 handling.
3512 * toplev.c: Include optabs.h.
3513 (target_reinit): Temporarily restore the global options if another
3514 set of options are in force.
3516 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3518 PR rtl-optimization/58668
3519 * cfgcleanup.c (flow_find_cross_jump): Don't count
3520 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3521 to determine what is counted.
3522 (flow_find_head_matching_sequence): Use active_insn_p to determine
3523 what is counted.
3524 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3525 counting change.
3526 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3527 determine what is counted.
3529 PR tree-optimization/59643
3530 * tree-predcom.c (split_data_refs_to_components): If one dr is
3531 read and one write, determine_offset fails and the write isn't
3532 in the bad component, just put the read into the bad component.
3534 2014-01-07 Mike Stump <mikestump@comcast.net>
3535 Jakub Jelinek <jakub@redhat.com>
3537 PR pch/59436
3538 * tree-core.h (struct tree_optimization_option): Change optabs
3539 type from unsigned char * to void *.
3540 * optabs.c (init_tree_optimization_optabs): Adjust
3541 TREE_OPTIMIZATION_OPTABS initialization.
3543 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3545 PR target/59644
3546 * config/i386/i386.h (struct machine_function): Add
3547 no_drap_save_restore field.
3548 * config/i386/i386.c (ix86_save_reg): Use
3549 !cfun->machine->no_drap_save_restore instead of
3550 crtl->stack_realign_needed.
3551 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3552 this function clears frame_pointer_needed. Set
3553 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3554 and DRAP reg is needed.
3556 2014-01-06 Marek Polacek <polacek@redhat.com>
3558 PR c/57773
3559 * doc/implement-c.texi: Mention that other integer types are
3560 permitted as bit-field types in strictly conforming mode.
3562 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3564 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3565 is newly allocated.
3567 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3569 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3571 2014-01-06 Martin Jambor <mjambor@suse.cz>
3573 PR ipa/59008
3574 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3575 to int.
3576 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3578 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3580 PR debug/59350
3581 PR debug/59510
3582 * var-tracking.c (add_stores): Preserve the value of the source even if
3583 we don't record the store.
3585 2014-01-06 Terry Guo <terry.guo@arm.com>
3587 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3589 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3591 PR bootstrap/59541
3592 * config/darwin.c (darwin_function_section): Adjust return values to
3593 correspond to optimisation changes made in r206070.
3595 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3597 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3598 from prefetch_block tune setting.
3599 (nocona_cost): Correct size of prefetch block to 64.
3601 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3603 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3604 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3605 used to save the static chain register in the computation of the offset
3606 from which the FP registers need to be restored.
3608 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3610 PR tree-optimization/59519
3611 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3612 ICE if get_current_def (current_new_name) is already non-NULL, as long
3613 as it is a phi result of some other phi in *new_exit_bb that has
3614 the same argument.
3616 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3617 or vmovdqu* for misaligned_operand.
3618 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3619 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3620 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3621 aligned_mem for AVX512F masked aligned load and store builtins and for
3622 non-temporal moves.
3624 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3626 PR tree-optimization/59651
3627 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3628 Address range for negative step should be added by TYPE_SIZE_UNIT.
3630 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3632 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3633 overlapping registers also for registers other than the stack pointer.
3635 2014-01-03 Marek Polacek <polacek@redhat.com>
3637 PR other/59661
3638 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3639 __builtin_FILE.
3641 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3643 PR target/59625
3644 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3645 asm goto as jump.
3647 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3648 (push splitter): Use <P:MODE_SIZE> instead of
3649 GET_MODE_SIZE (<P:MODE>mode).
3650 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3651 (mov -1, reg peephole2): Likewise.
3652 * config/i386/sse.md (*mov<mode>_internal,
3653 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3654 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3655 *<code><mode>3, *andnot<mode>3<mask_name>,
3656 <mask_codefor><code><mode>3<mask_name>): Likewise.
3657 * config/i386/subst.md (mask_mode512bit_condition,
3658 sd_mask_mode512bit_condition): Likewise.
3660 2014-01-02 Xinliang David Li <davidxl@google.com>
3662 PR tree-optimization/59303
3663 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3664 (dump_predicates): Better output format.
3665 (pred_equal_p): New function.
3666 (is_neq_relop_p): Ditto.
3667 (is_neq_zero_form_p): Ditto.
3668 (pred_expr_equal_p): Ditto.
3669 (pred_neg_p): Ditto.
3670 (simplify_pred): Ditto.
3671 (simplify_preds_2): Ditto.
3672 (simplify_preds_3): Ditto.
3673 (simplify_preds_4): Ditto.
3674 (simplify_preds): Ditto.
3675 (push_pred): Ditto.
3676 (push_to_worklist): Ditto.
3677 (get_pred_info_from_cmp): Ditto.
3678 (is_degenerated_phi): Ditto.
3679 (normalize_one_pred_1): Ditto.
3680 (normalize_one_pred): Ditto.
3681 (normalize_one_pred_chain): Ditto.
3682 (normalize_preds): Ditto.
3683 (normalize_cond_1): Remove function.
3684 (normalize_cond): Ditto.
3685 (is_gcond_subset_of): Ditto.
3686 (is_subset_of_any): Ditto.
3687 (is_or_set_subset_of): Ditto.
3688 (is_and_set_subset_of): Ditto.
3689 (is_norm_cond_subset_of): Ditto.
3690 (pred_chain_length_cmp): Ditto.
3691 (convert_control_dep_chain_into_preds): Type change.
3692 (find_predicates): Ditto.
3693 (find_def_preds): Ditto.
3694 (destroy_predicates_vecs): Ditto.
3695 (find_matching_predicates_in_rest_chains): Ditto.
3696 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3697 (is_pred_expr_subset): Ditto.
3698 (is_pred_chain_subset_of): Ditto.
3699 (is_included_in): Ditto.
3700 (is_superset_of): Ditto.
3702 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3704 Update copyright years.
3706 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3708 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3709 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3710 config/arc/arc.md, config/arc/arc.opt,
3711 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3712 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3713 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3714 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3715 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3716 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3717 vtable-verify.c, vtable-verify.h: Use the standard form for the
3718 copyright notice.
3720 2014-01-02 Tobias Burnus <burnus@net-b.de>
3722 * gcc.c (process_command): Update copyright notice dates.
3723 * gcov-dump.c: Ditto.
3724 * gcov.c: Ditto.
3725 * doc/cpp.texi: Bump @copying's copyright year.
3726 * doc/cppinternals.texi: Ditto.
3727 * doc/gcc.texi: Ditto.
3728 * doc/gccint.texi: Ditto.
3729 * doc/gcov.texi: Ditto.
3730 * doc/install.texi: Ditto.
3731 * doc/invoke.texi: Ditto.
3733 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3735 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3737 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3739 * config/i386/sse.md (*mov<mode>_internal): Guard
3740 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3742 PR rtl-optimization/59647
3743 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3744 new_rtx into UNSIGNED_FLOAT rtxes.
3745 \f
3746 Copyright (C) 2014 Free Software Foundation, Inc.
3748 Copying and distribution of this file, with or without modification,
3749 are permitted in any medium without royalty provided the copyright
3750 notice and this notice are preserved.