1 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2 Sandra Loosemore <sandra@codesourcery.com>
4 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
5 * config/nios2/nios2.c (nios2_function_profiler):
6 Add -fPIC (flag_pic == 2) support.
7 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
8 (nios2_large_offset_p): New function.
9 (nios2_unspec_reloc_p): Move up position, update to use
11 (nios2_unspec_address): Remove function.
12 (nios2_unspec_offset): New function.
13 (nios2_large_got_address): New function.
14 (nios2_got_address): Add large offset support.
15 (nios2_legitimize_tls_address): Update usage of removed and new
17 (nios2_symbol_binds_local_p): New function.
18 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
19 (nios2_legitimize_address): Update to use nios2_large_offset_p.
20 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
21 (nios2_print_operand): Merge H/L processing, add hiadj/lo
22 processing for (const (unspec ...)).
23 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
25 2014-02-20 Richard Biener <rguenther@suse.de>
27 * tree-cfg.c (replace_uses_by): Mark altered BBs before
28 doing the substitution.
29 (verify_gimple_assign_single): Also verify bare MEM_REFs
32 2014-02-20 Martin Jambor <mjambor@suse.cz>
35 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
36 info when checking whether lattices are bottom.
38 2014-02-20 Richard Biener <rguenther@suse.de>
41 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
44 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
47 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale parameter
48 specifying the scaling.
49 (inline_call): Update.
50 (want_inline_recursively): Guard division by zero.
51 (recursive_inlining): Update.
52 * ipa-inline.h (clone_inlined_nodes): Update.
54 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
57 * config/i386/i386.c (classify_argument): Pass structures of size
58 64 bytes or less in register.
60 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
61 Kirill Yukhin <kirill.yukhin@intel.com>
63 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
64 (_mm_rcp28_round_ss): Ditto.
65 (_mm_rsqrt28_round_sd): Ditto.
66 (_mm_rsqrt28_round_ss): Ditto.
67 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
68 (_mm_rcp14_round_ss): Ditto.
69 (_mm_rsqrt14_round_sd): Ditto.
70 (_mm_rsqrt14_round_ss): Ditto.
71 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
72 the first input operand, get rid of match_dup.
73 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
75 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
77 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
78 operand as the first input operand, set type attribute.
79 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
81 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
82 operand as the first input operand, set type attribute.
84 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
86 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
89 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
92 * config/i386/i386.c (construct_container): Remove TFmode check
93 for X86_64_INTEGER_CLASS.
95 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
98 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
99 only when -Wpsabi is enabled.
101 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
104 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
105 passing arrays in registers are the same as for structs, so remove the
106 special case for them.
108 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
110 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
111 destination type, extract only the valid bits if the source type is not
112 integral and has a different mode.
114 2014-02-19 Richard Biener <rguenther@suse.de>
117 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
120 2014-02-19 Richard Biener <rguenther@suse.de>
123 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
124 (ipa_modify_call_arguments): Emit an argument load explicitely and
125 preserve virtual SSA form there and for the replacement call.
126 Do not update SSA form nor free dominance info.
128 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
130 * ipa.c (function_and_variable_visibility): Also clear WEAK
131 flag when disolving COMDAT_GROUP.
133 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
135 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
136 * ipa-prop.c (ipa_set_jf_known_type): Return early when
138 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
139 do more sanity checks.
140 (detect_type_change): Return true when giving up early.
141 (compute_complex_assign_jump_func): Fix type parameter of
143 (compute_complex_ancestor_jump_func): Likewise.
144 (update_jump_functions_after_inlining): Fix updating of
146 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
148 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
150 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
151 inline clones when edge disappears.
153 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
156 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
157 Split 64-bit moves into 2 patterns. Do not allow the use of
158 direct move for TDmode in little endian, since the decimal value
159 has little endian bytes within a word, but the 64-bit pieces are
160 ordered in a big endian fashion, and normal subreg's of TDmode are
162 (mov<mode>_64bit_dm): Likewise.
163 (movtd_64bit_nodm): Likewise.
165 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
167 PR tree-optimization/60174
168 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
169 statement of an SSA_NAME that occurs in an abnormal PHI node.
171 2014-02-18 Jakub Jelinek <jakub@redhat.com>
174 * final.c (SEEN_BB): Remove.
175 (SEEN_NOTE, SEEN_EMITTED): Renumber.
176 (final_scan_insn): Don't force_source_line on second
177 NOTE_INSN_BASIC_BLOCK.
179 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
182 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
183 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
184 (type_natural_mode): Warn ABI change when %zmm register is not
185 available for AVX512F vector value passing.
187 2014-02-18 Kai Tietz <ktietz@redhat.com>
190 * config/i386/i386.c (ix86_expand_prologue): Use value in
191 rax register as displacement when restoring %r10 or %rax.
192 Fix wrong offset when restoring both registers.
194 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
196 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
197 assertion with conditional return.
199 2014-02-18 Jakub Jelinek <jakub@redhat.com>
200 Uros Bizjak <ubizjak@gmail.com>
203 * config/i386/driver-i386.c (host_detect_local_cpu): If
204 YMM state is not saved by the OS, also clear has_f16c. Move
205 CPUID 0x80000001 handling before YMM state saving checking.
207 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
209 PR rtl-optimization/58960
210 * haifa-sched.c (alloc_global_sched_pressure_data): New,
211 factored out from ...
212 (sched_init): ... here.
213 (free_global_sched_pressure_data): New, factored out from ...
214 (sched_finish): ... here.
215 * sched-int.h (free_global_sched_pressure_data): Declare.
216 * sched-rgn.c (nr_regions_initial): New static global.
217 (haifa_find_rgns): Initialize it.
218 (schedule_region): Disable sched-pressure for the newly
221 2014-02-17 Richard Biener <rguenther@suse.de>
223 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
224 release SSA defs of pattern stmts.
226 2014-02-17 Richard Biener <rguenther@suse.de>
228 * tree-inline.c (expand_call_inline): Release the virtual
229 operand defined by the call we are about to inline.
231 2014-02-17 Richard Biener <rguenther@suse.de>
233 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
235 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
236 Ilya Tocar <ilya.tocar@intel.com>
238 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
239 arguments order in builtin.
240 (_mm512_permutexvar_epi64): Ditto.
241 (_mm512_mask_permutexvar_epi64): Ditto
242 (_mm512_maskz_permutexvar_epi32): Ditto
243 (_mm512_permutexvar_epi32): Ditto
244 (_mm512_mask_permutexvar_epi32): Ditto
246 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
248 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
249 (p8_vmrgow): Likewise.
251 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
253 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
256 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
259 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
260 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
261 into 64-bit and 32-bit moves. On 64-bit moves, add support for
262 using direct move instructions on ISA 2.07. Also adjust
263 instruction length for 64-bit.
264 (mov<mode>_64bit, TFmode/TDmode): Likewise.
265 (mov<mode>_32bit, TFmode/TDmode): Likewise.
267 2014-02-15 Alan Modra <amodra@gmail.com>
271 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
272 find_replacement on parts of insn rtl that might be reloaded.
274 2014-02-15 Richard Biener <rguenther@suse.de>
276 PR tree-optimization/60183
277 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
278 (tree_ssa_phiprop): Calculate and free post-dominators.
280 2014-02-14 Jeff Law <law@redhat.com>
282 PR rtl-optimization/60131
283 * ree.c (get_extended_src_reg): New function.
284 (combine_reaching_defs): Use it rather than assuming location of REG.
285 (find_and_remove_re): Verify first operand of extension is
286 a REG before adding the insns to the copy list.
288 2014-02-14 Roland McGrath <mcgrathr@google.com>
290 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
291 * configure: Regenerated.
292 * config.in: Regenerated.
293 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
294 instead of ASM_SHORT.
296 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
297 Richard Earnshaw <rearnsha@arm.com>
299 PR rtl-optimization/59535
300 * lra-constraints.c (process_alt_operands): Encourage alternative
301 when unassigned pseudo class is superset of the alternative class.
302 (inherit_reload_reg): Don't inherit when optimizing for code size.
303 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
304 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
305 modes not less than 4 for Thumb1.
307 2014-02-14 Kyle McMartin <kyle@redhat.com>
310 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
312 2014-02-14 Richard Biener <rguenther@suse.de>
314 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
315 (get_frame_arg): Drop the assert with langhook types_compatible_p.
316 Do not strip INDIRECT_REFs.
318 2014-02-14 Richard Biener <rguenther@suse.de>
321 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
322 DECL_FUNCTION_SPECIFIC_TARGET.
323 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
324 * tree-streamer-out.c (pack_ts_target_option): Remove.
325 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
326 (write_ts_function_decl_tree_pointers): Do not stream
327 DECL_FUNCTION_SPECIFIC_TARGET.
328 * tree-streamer-in.c (unpack_ts_target_option): Remove.
329 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
330 (lto_input_ts_function_decl_tree_pointers): Do not stream
331 DECL_FUNCTION_SPECIFIC_TARGET.
333 2014-02-14 Jakub Jelinek <jakub@redhat.com>
335 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
336 (get_initial_def_for_induction, vectorizable_induction): Ignore
337 debug stmts when looking for exit_phi.
338 (vectorizable_live_operation): Fix up condition.
340 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
342 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
343 nreverse() because it changes the content of original tree list.
345 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
347 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
348 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
350 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
352 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
353 GNU coding standards.
355 2014-02-13 Jakub Jelinek <jakub@redhat.com>
358 * dwarf2out.c (gen_subprogram_die): Don't call
359 add_calling_convention_attribute if subr_die is old_die.
361 2014-02-13 Sharad Singhai <singhai@google.com>
363 * doc/optinfo.texi: Fix order of nodes.
365 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
367 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
368 operands[2], not operands[3].
370 2014-02-13 Richard Biener <rguenther@suse.de>
373 * doc/install.texi (ISL): Update recommended version to 0.12.2,
374 mention the possibility of an in-tree build.
375 (CLooG): Update recommended version to 0.18.1, mention the
376 possibility of an in-tree build and clarify that the ISL
377 bundled with CLooG does not work.
379 2014-02-13 Jakub Jelinek <jakub@redhat.com>
382 * expr.c (compress_float_constant): If x is a hard register,
383 extend into a pseudo and then move to x.
385 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
387 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
388 caused by bad second argument to warning_at() with -mhotpatch and
389 nested functions (e.g. with gfortran).
391 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
393 * opts.c (option_name): Remove "enabled by default" rider.
395 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
397 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
399 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
400 Uros Bizjak <ubizjak@gmail.com>
403 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
404 * configure: Regenerated.
406 2014-02-12 Richard Biener <rguenther@suse.de>
408 * vec.c (vec_prefix::calculate_allocation): Move as
409 inline variant to vec.h.
410 (vec_prefix::calculate_allocation_1): New out-of-line version.
411 * vec.h (vec_prefix::calculate_allocation_1): Declare.
412 (vec_prefix::m_has_auto_buf): Rename to ...
413 (vec_prefix::m_using_auto_storage): ... this.
414 (vec_prefix::calculate_allocation): Inline the easy cases
415 and dispatch to calculate_allocation_1 which doesn't need the
417 (va_heap::reserve): Use gcc_checking_assert.
418 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
419 m_using_auto_storage.
420 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
422 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
423 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
424 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
426 2014-02-12 Richard Biener <rguenther@suse.de>
428 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
429 when we found a dependence.
431 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
433 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
435 (maybe_fold_stmt): ... into this new function.
436 * omp-low.c (lower_omp): Update comment.
438 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
441 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
444 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
446 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
447 identifiers in comments.
448 (cortexa53_extra_costs): Likewise.
449 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
450 (cortexa7_extra_costs): Likewise.
451 (cortexa12_extra_costs): Likewise.
452 (cortexa15_extra_costs): Likewise.
453 (v7m_extra_costs): Likewise.
455 2014-02-12 Richard Biener <rguenther@suse.de>
458 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
459 of posix_memalign being successful.
460 (lower_stmt): Restrict lowering of posix_memalign to when
461 -ftree-bit-ccp is enabled.
463 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
465 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
467 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
469 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
471 PR rtl-optimization/60116
472 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
473 other_insn once the combination has been validated.
475 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
478 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
480 * ipa-devirt.c: Include demangle.h
481 (odr_violation_reported): New static variable.
482 (add_type_duplicate): Update odr_violations.
483 (maybe_record_node): Add completep parameter; update it.
484 (record_target_from_binfo): Add COMPLETEP parameter;
486 (possible_polymorphic_call_targets_1): Likewise.
487 (struct polymorphic_call_target_d): Add nonconstruction_targets;
488 rename FINAL to COMPLETE.
489 (record_targets_from_bases): Sanity check we found the binfo;
490 fix COMPLETEP updating.
491 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
492 parameter, fix computing of COMPLETEP.
493 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
494 at LTO time do demangling.
495 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
496 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
498 (gimple_get_virt_method_for_binfo): Likewise.
499 * gimple-fold.h (gimple_get_virt_method_for_binfo,
500 gimple_get_virt_method_for_vtable): Update prototypes.
502 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
505 * genautomata.c (add_presence_absence): Fix typo with
506 {final_}presence_list.
508 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
511 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
512 for VSX/Altivec vectors that land in GPR registers.
514 2014-02-11 Richard Henderson <rth@redhat.com>
515 Jakub Jelinek <jakub@redhat.com>
518 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
519 around drhs if type conversion to lacc->type is not useless.
521 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
523 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
525 (cortex-a57.cortex-a53): Likewise.
526 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
528 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
530 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
533 2014-02-11 Renlin Li <Renlin.Li@arm.com>
535 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
536 add_options_for_arm_vfp3.
538 2014-02-11 Jeff Law <law@redhat.com>
541 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
542 object with an undesirable mode.
544 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
547 * config/i386/sol2-9.h: New file.
548 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
549 *-*-solaris2.9*): Use it.
551 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
553 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
554 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
556 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
558 * config/microblaze/microblaze.c: Extend mcpu version format
560 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
562 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
564 2014-02-10 Richard Henderson <rth@redhat.com>
567 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
568 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
569 ms-abi vs -mno-accumulate-outgoing-args.
570 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
571 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
574 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
577 * cfgexpand.c (expand_asm_operands): Attach source location to
578 ASM_INPUT rtx objects.
579 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
581 2014-02-10 Nick Clifton <nickc@redhat.com>
583 * config/mn10300/mn10300.c (popcount): New function.
584 (mn10300_expand_prologue): Include saved registers in stack usage
587 2014-02-10 Jeff Law <law@redhat.com>
590 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
591 when changing the SET_DEST of a prior insn to avoid an input reload.
593 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
595 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
596 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
597 -mcall-openbsd, or -mcall-linux.
598 (CC1_ENDIAN_BIG_SPEC): Remove.
599 (CC1_ENDIAN_LITTLE_SPEC): Remove.
600 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
601 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
602 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
603 and %cc1_endian_default.
604 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
606 2014-02-10 Richard Biener <rguenther@suse.de>
608 PR tree-optimization/60115
609 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
610 MEM_REF handling. Properly verify that the accesses are not
611 out of the objects bound.
613 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
615 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
618 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
620 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
621 proper constants and fix formatting.
622 (possible_polymorphic_call_targets): Fix formatting.
624 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
625 Ilya Tocar <ilya.tocar@intel.com>
627 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
628 (_mm512_loadu_epi32): Renamed into...
629 (_mm512_loadu_si512): This.
630 (_mm512_storeu_epi32): Renamed into...
631 (_mm512_storeu_si512): This.
632 (_mm512_maskz_ceil_ps): Removed.
633 (_mm512_maskz_ceil_pd): Ditto.
634 (_mm512_maskz_floor_ps): Ditto.
635 (_mm512_maskz_floor_pd): Ditto.
636 (_mm512_floor_round_ps): Ditto.
637 (_mm512_floor_round_pd): Ditto.
638 (_mm512_ceil_round_ps): Ditto.
639 (_mm512_ceil_round_pd): Ditto.
640 (_mm512_mask_floor_round_ps): Ditto.
641 (_mm512_mask_floor_round_pd): Ditto.
642 (_mm512_mask_ceil_round_ps): Ditto.
643 (_mm512_mask_ceil_round_pd): Ditto.
644 (_mm512_maskz_floor_round_ps): Ditto.
645 (_mm512_maskz_floor_round_pd): Ditto.
646 (_mm512_maskz_ceil_round_ps): Ditto.
647 (_mm512_maskz_ceil_round_pd): Ditto.
648 (_mm512_expand_pd): Ditto.
649 (_mm512_expand_ps): Ditto.
650 * config/i386/i386.c (ix86_builtins): Remove
651 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
653 * config/i386/predicates.md (const1256_operand): New.
654 (const_1_to_2_operand): Ditto.
655 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
656 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
657 (*avx512pf_gatherpf<mode>sf): Ditto.
658 (avx512pf_gatherpf<mode>df): Ditto.
659 (*avx512pf_gatherpf<mode>df_mask): Ditto.
660 (*avx512pf_gatherpf<mode>df): Ditto.
661 (avx512pf_scatterpf<mode>sf): Ditto.
662 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
663 (*avx512pf_scatterpf<mode>sf): Ditto.
664 (avx512pf_scatterpf<mode>df): Ditto.
665 (*avx512pf_scatterpf<mode>df_mask): Ditto.
666 (*avx512pf_scatterpf<mode>df): Ditto.
667 (avx512f_expand<mode>): Removed.
668 (<shift_insn><mode>3<mask_name>): Change predicate type.
670 2014-02-08 Jakub Jelinek <jakub@redhat.com>
672 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
673 not at the end of datarefs vector use ordered_remove to avoid
674 reordering datarefs vector.
677 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
678 mark local addressable non-static vars as GOVD_PRIVATE
679 instead of GOVD_LOCAL.
680 * omp-low.c (lower_omp_for): Move gimple_bind_vars
681 and BLOCK_VARS of gimple_bind_block to new_stmt rather
685 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
686 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
687 assume_aligned or alloc_align attributes.
688 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
689 arguments. Handle also assume_aligned and alloc_align attributes.
690 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
691 calls to functions with assume_aligned or alloc_align attributes.
692 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
694 2014-02-08 Terry Guo <terry.guo@arm.com>
696 * doc/invoke.texi: Document ARM -march=armv7e-m.
698 2014-02-08 Jakub Jelinek <jakub@redhat.com>
700 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
701 flag on __cilkrts_rethrow builtin.
704 * ipa-cp.c (determine_versionability): Fail at -O0
705 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
706 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
709 2014-02-04 Jakub Jelinek <jakub@redhat.com>
712 * tree-inline.c (copy_forbidden): Fail for
713 __attribute__((optimize (0))) functions.
715 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
717 * varpool.c: Include pointer-set.h.
718 (varpool_remove_unreferenced_decls): Variables in other partitions
719 will not be output; be however careful to not lose information
722 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
724 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
725 lookup in the vtable constructor.
727 2014-02-07 Jeff Law <law@redhat.com>
730 * config/m68k/m68k.md (ashldi_extsi): Turn into a
731 define_insn_and_split.
733 * ipa-inline.c (inline_small_functions): Fix typos.
735 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
737 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
738 (s390_can_use_return_insn): Declare.
739 * config/s390/s390.h (EPILOGUE_USES): Define.
740 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
742 (s390_chunkify_start): Handle return JUMP_LABELs.
743 (s390_early_mach): Emit a main_pool instruction on the entry edge.
744 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
745 (s390_can_use_return_insn): New functions.
746 (s390_fix_long_loop_prediction): Handle conditional returns.
747 (TARGET_SET_UP_BY_PROLOGUE): Define.
748 * config/s390/s390.md (ANY_RETURN): New code iterator.
749 (*creturn, *csimple_return, return, simple_return): New patterns.
751 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
753 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
754 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
755 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
756 REG_CFA_RESTORE list when deciding not to restore a register.
758 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
760 * config/s390/s390.c: Include tree-pass.h and context.h.
761 (s390_early_mach): New function, split out from...
762 (s390_emit_prologue): ...here.
763 (pass_data_s390_early_mach): New pass structure.
764 (pass_s390_early_mach): New class.
765 (s390_option_override): Create and register early_mach pass.
768 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
770 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
771 to match for the exit block.
773 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
775 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
776 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
777 Reject misaligned operands.
779 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
781 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
783 2014-02-07 Richard Biener <rguenther@suse.de>
786 * gimple-low.c (lower_builtin_posix_memalign): New function.
787 (lower_stmt): Call it to lower posix_memalign in a way
788 to make alignment info accessible.
790 2014-02-07 Jakub Jelinek <jakub@redhat.com>
793 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
794 __builtin_setjmp_receiver.
796 2014-02-07 Richard Biener <rguenther@suse.de>
799 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
800 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
801 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
802 Handle BUILT_IN_POSIX_MEMALIGN.
803 (find_func_clobbers): Likewise.
804 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
805 (call_may_clobber_ref_p_1): Likewise.
807 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
810 * ipa-devirt.c (record_target_from_binfo): Remove overactive
813 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
816 * lto-cgraph.c (lto_output_node): Use
817 symtab_get_symbol_partitioning_class.
818 (lto_output_varpool_node): likewise.
819 (symtab_get_symbol_partitioning_class): Move here from
821 * cgraph.h (symbol_partitioning_class): Likewise.
822 (symtab_get_symbol_partitioning_class): Declare.
824 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
826 * ggc.h (ggc_internal_cleared_alloc): New macro.
827 * vec.h (vec_safe_copy): Handle memory stats.
828 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
829 * target-globals.c (save_target_globals): Likewise.
831 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
834 * expr.c (emit_move_resolve_push): Export; be bit more selective
835 on when to clear alias set.
836 * expr.h (emit_move_resolve_push): Declare.
837 * function.h (struct function): Add tail_call_marked.
838 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
839 * config/i386/i386-protos.h (ix86_expand_push): Remove.
840 * config/i386/i386.md (TImode move expander): De not call
842 (FP push expanders): Preserve memory attributes.
843 * config/i386/sse.md (push<mode>1): Remove.
844 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
845 (ix86_expand_push): Remove.
846 * config/i386/mmx.md (push<mode>1): Remove.
848 2014-02-06 Jakub Jelinek <jakub@redhat.com>
850 PR rtl-optimization/60030
851 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
852 lopart with paradoxical subreg before shifting it up by hprec.
854 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
856 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
857 Remove extra newline at end of file.
858 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
859 (arm_issue_rate): Handle cortexa57.
860 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
861 (cortex-a57.cortex-a53): Likewise.
863 2014-02-06 Jakub Jelinek <jakub@redhat.com>
866 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
867 don't record in REG_FRAME_RELATED_EXPR registers not set in that
869 (arm_expand_prologue): Adjust all callers.
870 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
871 info, registers also at the lowest numbered registers side. Use
872 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
876 * var-tracking.c (adjust_mems): Before adding a SET to
877 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
879 2014-02-06 Alan Modra <amodra@gmail.com>
882 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
883 change SDmode to DDmode when lra_in_progress.
885 2014-02-06 Jakub Jelinek <jakub@redhat.com>
888 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
889 free_data_ref on the dr first, and before goto again also set dr
890 to the next dr. For simd_lane_access, free old datarefs[i] before
891 overwriting it. For get_vectype_for_scalar_type failure, don't
892 free_data_ref if simd_lane_access.
894 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
897 * tree.h (opts_for_fn): New inline function.
898 (opt_for_fn): Define.
899 * config/i386/i386.c (ix86_function_regparm): Use
900 opt_for_fn (decl, optimize) instead of optimize.
902 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
904 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
905 for SYMBOL_REF in large memory model.
907 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
909 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
911 (cortex-a57): Likewise.
912 (cortex-a57.cortex-a53): Likewise.
914 2014-02-06 Yury Gribov <y.gribov@samsung.com>
915 Kugan Vivekanandarajah <kuganv@linaro.org>
917 * config/arm/arm.c (arm_vector_alignment_reachable): Check
919 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
921 2014-02-06 Richard Biener <rguenther@suse.de>
923 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
924 set_loop_copy and initialize_original_copy_tables.
926 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
928 * config/aarch64/aarch64-simd.md
929 (aarch64_ashr_simddi): Change QI to SI.
931 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
932 Jakub Jelinek <jakub@redhat.com>
935 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
938 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
940 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
941 CODE_FOR_altivec_vpku[hw]um to
942 CODE_FOR_altivec_vpku[hw]um_direct.
943 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
944 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
945 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
946 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
948 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
950 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
951 generation for -maltivec=be.
952 (altivec_vsumsws): Simplify redundant test.
954 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
956 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
957 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
958 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
959 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
961 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
963 (altivec_vpks<VI_char>ss): Likewise.
964 (altivec_vpks<VI_char>us): Likewise.
965 (altivec_vpku<VI_char>us): Likewise.
966 (altivec_vpku<VI_char>um): Likewise.
967 (altivec_vpku<VI_char>um_direct): New (copy of
968 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
970 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
971 target is little endian and -maltivec=be is not specified.
972 (*altivec_vupkhs<VU_char>_direct): New (copy of
973 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
974 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
975 target is little endian and -maltivec=be is not specified.
976 (*altivec_vupkls<VU_char>_direct): New (copy of
977 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
978 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
979 little endian and -maltivec=be is not specified.
980 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
981 little endian and -maltivec=be is not specified.
983 2014-02-05 Richard Henderson <rth@redhat.com>
986 * combine-stack-adj.c: Revert r206943.
987 * sched-int.h (struct deps_desc): Add last_args_size.
988 * sched-deps.c (init_deps): Initialize it.
989 (sched_analyze_insn): Add OUTPUT dependencies between insns that
990 contain REG_ARGS_SIZE notes.
992 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
994 * lto-cgraph.c (asm_nodes_output): Make global.
995 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
996 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
997 (driver_handle_option): Handle OPT_fwpa.
999 2014-02-05 Jakub Jelinek <jakub@redhat.com>
1002 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
1003 a comment typo and formatting issue. If odr_hash hasn't been
1004 created, return vNULL and set *completep to false.
1007 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
1008 bb with no successors.
1010 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
1013 * doc/invoke.texi (-march): Clarify documentation for ARM.
1017 2014-02-05 Richard Biener <rguenther@suse.de>
1019 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
1020 when not vectorizing because of too many alias checks.
1021 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
1022 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
1024 2014-02-05 Nick Clifton <nickc@redhat.com>
1026 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
1027 accept extended registers in any mode when compiling for the MN10300.
1029 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1031 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1032 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1033 sanitization attributes.
1034 (can_inline_edge_p): Likewise.
1035 (sanitize_attrs_match_for_inline_p): New function.
1037 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1039 * ipa-prop.c (detect_type_change): Shor circuit testing of
1040 type changes on THIS pointer.
1042 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1045 * config/pa/pa.c (legitimize_tls_address): Return original address
1046 if not passed a SYMBOL_REF rtx.
1047 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1049 (pa_emit_move_sequence): Simplify TLS source operands.
1050 (pa_legitimate_constant_p): Reject all TLS constants.
1051 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1052 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1054 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1056 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1057 groups when we know they are controlled by LTO.
1058 * varasm.c (default_binds_local_p_1): If object is in other partition,
1059 it will be resolved locally.
1061 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1063 * config/host-linux.c (linux_gt_pch_use_address): Don't
1064 use SSIZE_MAX because it is not always defined.
1066 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1069 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1070 threshold for pseudo splitting.
1071 (update_ebb_live_info): Process call argument hard registers and
1072 hard registers from insn definition too.
1073 (max_small_class_regs_num): New constant.
1074 (inherit_in_ebb): Update live hard regs through EBBs. Update
1075 reloads_num only for small register classes. Don't split for
1078 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1081 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1084 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1086 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1089 2014-02-04 Marek Polacek <polacek@redhat.com>
1091 * gdbinit.in (pel): Define.
1093 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1095 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1098 2014-02-04 Richard Biener <rguenther@suse.de>
1101 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1102 in function context local.
1103 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1104 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1105 similar to LTO_imported_decl_ref.
1107 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1109 PR tree-optimization/60002
1110 * cgraphclones.c (build_function_decl_skip_args): Clear
1113 PR tree-optimization/60023
1114 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1115 false to gsi_replace.
1116 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1117 has been in some EH region and vec_stmt could throw, add
1118 vec_stmt into the same EH region.
1119 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1120 has no lhs, ignore it.
1121 * internal-fn.c (expand_MASK_LOAD): Likewise.
1124 * tree-inline.c (copy_forbidden): Fail for
1125 __attribute__((optimize (0))) functions.
1128 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1129 is set, copy one less argument.
1130 (expand_simd_clones): Don't subtract clone_info->inbranch
1131 from simd_clone_struct_alloc argument.
1133 PR rtl-optimization/57915
1134 * recog.c (simplify_while_replacing): If all unary/binary/relational
1135 operation arguments are constant, attempt to simplify those.
1138 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1139 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1141 2014-02-04 Richard Biener <rguenther@suse.de>
1143 PR tree-optimization/60012
1144 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1145 TBAA disambiguation to all DDRs.
1147 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1150 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1151 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1153 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1156 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1158 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1160 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1161 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1163 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1166 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1167 to figure out targets of polymorphic calls with known decl.
1168 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1169 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1170 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1171 (get_polymorphic_call_info): ... here.
1172 (get_polymorphic_call_info_from_invariant): New function.
1174 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1176 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1177 lookup via vtable pointer; check for type consistency
1178 and turn inconsitent facts into UNREACHABLE.
1179 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1180 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1181 type inconsistent querries; return UNREACHABLE instead.
1183 2014-02-03 Richard Henderson <rth@twiddle.net>
1186 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1187 already processed this node.
1188 (normalize_one_pred_1): Pass along mark_set.
1189 (normalize_one_pred): Create and destroy a pointer_set_t.
1190 (normalize_one_pred_chain): Likewise.
1192 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1194 PR gcov-profile/58602
1195 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1197 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1200 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1201 -fno-devirtualize; try to devirtualize by the knowledge of
1202 virtual table pointer given by aggregate propagation.
1203 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1204 (ipa_print_node_jump_functions): Dump also offset that
1205 is relevant for polymorphic calls.
1206 (determine_known_aggregate_parts): Add arg_type parameter; use it
1207 instead of determining the type from pointer type.
1208 (ipa_compute_jump_functions_for_edge): Update call of
1209 determine_known_aggregate_parts.
1210 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1211 (gimple_get_virt_method_for_binfo): ... here; simplify using
1212 vtable_pointer_value_to_vtable.
1213 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1214 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1215 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1216 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1218 (vtable_pointer_value_to_binfo): ... here.
1219 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1221 2014-02-03 Teresa Johnson <tejohnson@google.com>
1223 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1224 redef of outer loop index variable.
1226 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1230 * doc/extend.texi (Function Attributes): Typo.
1232 2014-02-03 Cong Hou <congh@google.com>
1234 PR tree-optimization/60000
1235 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1236 if the vectorized statement is a store. A store statement can only
1237 appear at the end of pattern statements.
1239 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1241 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1242 (ix86_option_override_internal): Default long double to 64-bit for
1243 32-bit Bionic and to 128-bit for 64-bit Bionic.
1245 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1246 TARGET_LONG_DOUBLE_128 is true.
1247 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1249 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1250 (mlong-double-64): Negate -mlong-double-128.
1251 (mlong-double-128): New option.
1253 * config/i386/i386-c.c (ix86_target_macros): Define
1254 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1256 * doc/invoke.texi: Document -mlong-double-128.
1258 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1260 PR rtl-optimization/60024
1261 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1263 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1265 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1267 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1269 PR rtl-optimization/57662
1270 * sel-sched.c (code_motion_path_driver): Do not mark already not
1271 existing blocks in the visiting bitmap.
1273 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1275 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1276 on the insn being emitted.
1278 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1279 Will Deacon <will.deacon@arm.com>
1281 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1283 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1285 * config/arm/arm-tables.opt: Regenerate.
1287 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1289 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1290 for vector types other than V16QImode.
1291 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1292 define_expand, and call altivec_expand_vec_perm_le when producing
1293 code with little endian element order.
1294 (*altivec_vperm_<mode>_internal): New insn having previous
1295 behavior of altivec_vperm_<mode>.
1296 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1297 altivec_expand_vec_perm_le when producing code with little endian
1299 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1300 behavior of altivec_vperm_<mode>_uns.
1302 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1304 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1305 (altivec_vsumsws): Add handling for -maltivec=be with a little
1307 (altivec_vsumsws_direct): New.
1308 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1309 gen_altivec_vsumsws.
1311 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1313 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1314 vtable_pointer_value_to_binfo): New functions.
1315 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1316 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1318 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1320 * config/nios2/nios2.md (load_got_register): Initialize GOT
1321 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1322 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1324 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1326 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1327 preserverd by passthrough, do not propagate the type.
1329 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1331 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1332 (mips_atomic_assign_expand_fenv): New function.
1333 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1335 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1337 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1338 (__builtin_mips_set_fcsr): Likewise.
1339 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1340 MIPS_USI_FTYPE_VOID.
1341 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1342 (mips16_expand_set_fcsr): Likewise.
1343 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1344 (mips16_set_fcsr_stub): Likewise.
1345 (mips16_get_fcsr_one_only_stub): New class.
1346 (mips16_set_fcsr_one_only_stub): Likewise.
1347 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1348 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1349 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1350 (hard_float): New availability predicate.
1351 (mips_builtins): Add get_fcsr and set_fcsr.
1352 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1353 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1354 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1355 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1356 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1359 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1361 * config/mips/mips.c (mips_one_only_stub): New class.
1362 (mips_need_mips16_rdhwr_p): Replace with...
1363 (mips16_rdhwr_stub): ...this new variable.
1364 (mips16_stub_call_address): New function.
1365 (mips16_rdhwr_one_only_stub): New class.
1366 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1367 (mips_output_mips16_rdhwr): Delete.
1368 (mips_finish_stub): New function.
1369 (mips_code_end): Use it to handle rdhwr stubs.
1371 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1374 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1375 when calculating size of integer atomic types.
1377 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1379 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1381 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1383 PR tree-optimization/60003
1384 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1385 * profile.c (branch_prob): Use gimple_call_builtin_p
1386 to check for BUILT_IN_SETJMP_RECEIVER.
1387 * tree-inline.c (copy_bb): Call notice_special_calls.
1389 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1392 * lra-constraints.c (process_alt_operands): Update reload_sum only
1395 2014-01-31 Richard Henderson <rth@redhat.com>
1398 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1399 until after else_eh is processed.
1401 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1403 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1404 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1405 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1406 in smmintrin.h, remove them.
1407 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1408 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1409 * config/i386/i386.md (ROUND_SAE): Fix value.
1410 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1411 (const48_operand): New.
1412 * config/i386/subst.md (round), (round_expand): Use
1413 const_4_or_8_to_11_operand.
1414 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1416 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1418 * config/i386/constraints.md (Yk): Swap meaning with k.
1419 * config/i386/i386.md (movhi_internal): Change Yk to k.
1420 (movqi_internal): Ditto.
1421 (*k<logic><mode>): Ditto.
1424 (kandn<mode>): Ditto.
1425 (*<code>hi_1): Ditto.
1426 (*<code>qi_1): Ditto.
1427 (kxnor<mode>): Ditto.
1428 (kortestzhi): Ditto.
1429 (kortestchi): Ditto.
1431 (*one_cmplhi2_1): Ditto.
1432 (*one_cmplqi2_1): Ditto.
1433 * config/i386/sse.md (): Change k to Yk.
1434 (avx512f_load<mode>_mask): Ditto.
1435 (avx512f_blendm<mode>): Ditto.
1436 (avx512f_store<mode>_mask): Ditto.
1437 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1438 (avx512f_storedqu<mode>_mask): Ditto.
1439 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1441 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1442 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1443 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1444 (avx512f_maskcmp<mode>3): Ditto.
1445 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1446 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1447 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1448 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1449 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1450 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1451 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1452 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1453 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1454 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1455 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1456 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1457 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1458 (vec_extract_lo_<mode>_maskm): Ditto.
1459 (vec_extract_hi_<mode>_maskm): Ditto.
1460 (avx512f_vternlog<mode>_mask): Ditto.
1461 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1462 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1463 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1464 (avx512f_<code>v8div16qi2_mask): Ditto.
1465 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1466 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1467 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1468 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1469 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1470 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1471 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1472 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1473 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1474 (avx512cd_maskb_vec_dupv8di): Ditto.
1475 (avx512cd_maskw_vec_dupv16si): Ditto.
1476 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1477 (avx512f_vpermi2var<mode>3_mask): Ditto.
1478 (avx512f_vpermi2var<mode>3_mask): Ditto.
1479 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1480 (*avx512f_gathersi<mode>): Ditto.
1481 (*avx512f_gathersi<mode>_2): Ditto.
1482 (*avx512f_gatherdi<mode>): Ditto.
1483 (*avx512f_gatherdi<mode>_2): Ditto.
1484 (*avx512f_scattersi<mode>): Ditto.
1485 (*avx512f_scatterdi<mode>): Ditto.
1486 (avx512f_compress<mode>_mask): Ditto.
1487 (avx512f_compressstore<mode>_mask): Ditto.
1488 (avx512f_expand<mode>_mask): Ditto.
1489 * config/i386/subst.md (mask): Change k to Yk.
1490 (mask_scalar_merge): Ditto.
1493 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1495 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1497 2014-01-31 Richard Biener <rguenther@suse.de>
1500 * builtins.c (fold_builtin_memory_op): Make sure to not
1501 use a floating-point mode or a boolean or enumeral type for
1504 2014-01-30 DJ Delorie <dj@redhat.com>
1506 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1507 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1508 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1509 whenever main() has an epilogue.
1511 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1513 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1514 unused variable "field".
1515 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1516 (vsx_mergeh_<mode>): Likewise.
1517 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1518 (altivec_vmrghh): Likewise.
1519 (altivec_vmrghw): Likewise.
1520 (altivec_vmrglb): Likewise.
1521 (altivec_vmrglh): Likewise.
1522 (altivec_vmrglw): Likewise.
1523 (altivec_vspltb): Add missing uses.
1524 (altivec_vsplth): Likewise.
1525 (altivec_vspltw): Likewise.
1526 (altivec_vspltsf): Likewise.
1528 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1531 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1532 frame related instructions.
1534 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1536 PR rtl-optimization/59959
1537 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1538 any reload of register whose subreg is invalid.
1540 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1542 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1543 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1544 Add missing return type - void.
1546 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1548 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1549 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1550 remove element index adjustment for endian (now handled in vsx.md
1552 (altivec_expand_vec_perm_const): Use
1553 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1554 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1555 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1556 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1557 define_expand and a new define_insn *altivec_vspltb_internal;
1558 adjust for -maltivec=be on a little endian target.
1559 (altivec_vspltb_direct): New.
1560 (altivec_vsplth): Divide into a define_expand and a new
1561 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1562 little endian target.
1563 (altivec_vsplth_direct): New.
1564 (altivec_vspltw): Divide into a define_expand and a new
1565 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1566 little endian target.
1567 (altivec_vspltw_direct): New.
1568 (altivec_vspltsf): Divide into a define_expand and a new
1569 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1570 a little endian target.
1572 2014-01-30 Richard Biener <rguenther@suse.de>
1574 PR tree-optimization/59993
1575 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1576 can propagate form the earlier stmt and avoid the transform
1577 when the intermediate result is needed.
1579 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1581 * README.Portability: Fix typo.
1583 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1585 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1586 comparison_operator with ordered_comparison_operator.
1588 2014-01-30 Nick Clifton <nickc@redhat.com>
1590 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1591 Rename to mn10300_store_multiple_regs.
1592 * config/mn10300/mn10300.c: Likewise.
1593 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1594 store_multiple_regs.
1595 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1596 Call mn10300_store_multiple_regs.
1598 2014-01-30 Nick Clifton <nickc@redhat.com>
1599 DJ Delorie <dj@redhat.com>
1601 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1602 %fp 2 to keep registers after it properly word-aligned.
1603 (rl78_alloc_physical_registers_umul): Handle the case where both
1604 input operands are the same.
1606 2014-01-30 Richard Biener <rguenther@suse.de>
1608 PR tree-optimization/59903
1609 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1612 2014-01-30 Jason Merrill <jason@redhat.com>
1615 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1618 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1620 2014-01-30 Richard Biener <rguenther@suse.de>
1622 PR tree-optimization/59951
1623 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1625 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1628 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1629 SFmode to DFmode case.
1631 2014-01-29 DJ Delorie <dj@redhat.com>
1633 * config/msp430/msp430.opt (-minrt): New.
1634 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1636 (ENDFILE_SPEC): Likewise.
1638 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1640 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1641 (estimate_function_body_sizes): Use it.
1643 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1646 * dwarf2out.c (is_cxx_auto): New.
1647 (is_base_type): Use it.
1648 (gen_type_die_with_usage): Likewise.
1650 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1652 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1653 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1654 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1655 -maltivec=be with LE targets.
1656 (vsx_mergeh_<mode>): Likewise.
1657 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1658 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1659 (altivec_vmrghb): Replace with define_expand and new
1660 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1661 (altivec_vmrghb_direct): New define_insn.
1662 (altivec_vmrghh): Replace with define_expand and new
1663 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1664 (altivec_vmrghh_direct): New define_insn.
1665 (altivec_vmrghw): Replace with define_expand and new
1666 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1667 (altivec_vmrghw_direct): New define_insn.
1668 (*altivec_vmrghsf): Adjust for endianness.
1669 (altivec_vmrglb): Replace with define_expand and new
1670 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1671 (altivec_vmrglb_direct): New define_insn.
1672 (altivec_vmrglh): Replace with define_expand and new
1673 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1674 (altivec_vmrglh_direct): New define_insn.
1675 (altivec_vmrglw): Replace with define_expand and new
1676 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1677 (altivec_vmrglw_direct): New define_insn.
1678 (*altivec_vmrglsf): Adjust for endianness.
1679 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1680 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1681 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1682 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1683 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1684 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1685 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1686 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1688 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1690 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1691 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1694 2014-01-29 Richard Biener <rguenther@suse.de>
1696 PR tree-optimization/58742
1697 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1698 associate_pointerplus_align.
1699 (associate_pointerplus_diff): New function.
1700 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1701 and associate_pointerplus_diff.
1703 2014-01-29 Richard Biener <rguenther@suse.de>
1705 * lto-streamer.h (LTO_major_version): Bump to 3.
1706 (LTO_minor_version): Reset to 0.
1708 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1710 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1711 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1712 (arm_file_start): Generate correct asm header for armv7ve.
1713 * config/arm/bpabi.h: Add multilib support for armv7ve.
1714 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1715 and cortex-a15 to armv7ve.
1716 * config/arm/t-aprofile: Add multilib support for armv7ve.
1717 * doc/invoke.texi: Document -march=armv7ve.
1719 2014-01-29 Richard Biener <rguenther@suse.de>
1721 PR tree-optimization/58742
1722 * tree-ssa-forwprop.c (associate_plusminus): Return true
1723 if we changed sth, defer EH cleanup to ...
1724 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1725 (simplify_mult): New function.
1727 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1730 PR tree-optimization/59920
1731 * tree.c (build_common_builtin_nodes): Remove
1732 __builtin_setjmp_dispatcher initialization.
1733 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1734 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1735 instead of gsi_after_labels + manually skipping debug stmts.
1736 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1737 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1738 * tree-inline.c (copy_edges_for_bb): Remove
1739 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1740 argument. Ignore computed_goto_p stmts. Don't call
1741 make_abnormal_goto_edges. If a call might need abnormal edges
1742 for non-local gotos, see if it already has an edge to
1743 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1744 with true argument, don't do anything then, otherwise add
1745 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1746 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1748 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1749 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1750 (lower_stmt): Don't set data->calls_builtin_setjmp.
1751 (lower_builtin_setjmp): Adjust comment.
1752 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1753 * tree-cfg.c (found_computed_goto): Remove.
1754 (factor_computed_gotos): Remove.
1755 (make_goto_expr_edges): Return bool, true for computed gotos.
1756 Don't call make_abnormal_goto_edges.
1757 (build_gimple_cfg): Don't set found_computed_goto, don't call
1758 factor_computed_gotos.
1759 (computed_goto_p): No longer static.
1760 (make_blocks): Don't set found_computed_goto.
1761 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1762 (make_edges): If make_goto_expr_edges returns true, push bb
1763 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1764 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1765 vector. Record mapping between bbs and OpenMP regions if there
1766 are any, adjust make_gimple_omp_edges caller. Call
1767 handle_abnormal_edges.
1768 (make_abnormal_goto_edges): Remove.
1769 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1770 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1771 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1772 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1773 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1774 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1775 filling *region also set *region_idx to (*region)->entry->index.
1778 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1779 For REGs set ORIGINAL_REGNO.
1781 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1783 * doc/md.texi: Mention that a target shouldn't implement
1784 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1787 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1789 PR tree-optimization/59594
1790 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1791 a copy of the datarefs vector rather than the vector itself.
1793 2014-01-28 Jason Merrill <jason@redhat.com>
1796 * dwarf2out.c (auto_die): New static.
1797 (gen_type_die_with_usage): Handle C++1y 'auto'.
1798 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1801 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1804 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1805 (SPEC_X32): Likewise.
1806 (SPEC_64): Likewise.
1807 * config/i386/i386.c (ix86_option_override_internal): Turn off
1808 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1810 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1811 * config/i386/i386.h (TARGET_16BIT): New macro.
1812 (TARGET_16BIT_P): Likewise.
1813 * config/i386/i386.opt: Add m16.
1814 * doc/invoke.texi: Document -m16.
1816 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1818 PR preprocessor/59935
1819 * input.c (location_get_source_line): Bail out on when line number
1820 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1822 2014-01-28 Richard Biener <rguenther@suse.de>
1824 PR tree-optimization/58742
1825 * tree-ssa-forwprop.c (associate_plusminus): Handle
1826 pointer subtraction of the form (T)(P + A) - (T)P.
1828 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1830 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1833 2014-01-28 Richard Biener <rguenther@suse.de>
1836 2014-01-28 Richard Biener <rguenther@suse.de>
1838 PR rtl-optimization/45364
1839 PR rtl-optimization/59890
1840 * var-tracking.c (local_get_addr_clear_given_value): Handle
1841 already cleared slot.
1842 (val_reset): Handle not allocated local_get_addr_cache.
1843 (vt_find_locations): Use post-order on the inverted CFG.
1845 2014-01-28 Richard Biener <rguenther@suse.de>
1847 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1849 2014-01-28 Richard Biener <rguenther@suse.de>
1851 PR rtl-optimization/45364
1852 PR rtl-optimization/59890
1853 * var-tracking.c (local_get_addr_clear_given_value): Handle
1854 already cleared slot.
1855 (val_reset): Handle not allocated local_get_addr_cache.
1856 (vt_find_locations): Use post-order on the inverted CFG.
1858 2014-01-28 Alan Modra <amodra@gmail.com>
1860 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1861 * configure.ac <recursive call for build != host>: Define
1862 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1863 and LD_FOR_BUILD too.
1864 * configure: Regenerate.
1866 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1868 * config/i386/i386.c (get_builtin_code_for_version): Separate
1869 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1870 Broadwell from Haswell.
1872 2014-01-27 Steve Ellcey <sellcey@mips.com>
1874 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1875 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1876 * config/mips/mips.c (mips_option_override): Change setting
1878 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1879 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1880 Change from Mask to Var.
1882 2014-01-27 Jeff Law <law@redhat.com>
1884 * ipa-inline.c (inline_small_functions): Fix typo.
1886 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1888 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1889 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1890 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1891 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1892 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1893 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1894 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1895 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1896 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1897 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1898 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1899 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1900 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1901 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1902 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1903 (_mm512_storeu_epi64): Ditto.
1904 (_mm512_cmpge_epi32_mask): Ditto.
1905 (_mm512_cmpge_epu32_mask): Ditto.
1906 (_mm512_cmpge_epi64_mask): Ditto.
1907 (_mm512_cmpge_epu64_mask): Ditto.
1908 (_mm512_cmple_epi32_mask): Ditto.
1909 (_mm512_cmple_epu32_mask): Ditto.
1910 (_mm512_cmple_epi64_mask): Ditto.
1911 (_mm512_cmple_epu64_mask): Ditto.
1912 (_mm512_cmplt_epi32_mask): Ditto.
1913 (_mm512_cmplt_epu32_mask): Ditto.
1914 (_mm512_cmplt_epi64_mask): Ditto.
1915 (_mm512_cmplt_epu64_mask): Ditto.
1916 (_mm512_cmpneq_epi32_mask): Ditto.
1917 (_mm512_cmpneq_epu32_mask): Ditto.
1918 (_mm512_cmpneq_epi64_mask): Ditto.
1919 (_mm512_cmpneq_epu64_mask): Ditto.
1920 (_mm512_expand_pd): Ditto.
1921 (_mm512_expand_ps): Ditto.
1922 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1923 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1924 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1925 * config/i386/i386.c (ix86_builtins): Add
1926 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1927 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1928 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1929 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1930 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1931 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1932 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1933 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1934 IX86_BUILTIN_PMOVUSQW512_MEM.
1935 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1936 __builtin_ia32_pmovsqd512mem_mask,
1937 __builtin_ia32_pmovqd512mem_mask,
1938 __builtin_ia32_pmovusqw512mem_mask,
1939 __builtin_ia32_pmovsqw512mem_mask,
1940 __builtin_ia32_pmovqw512mem_mask,
1941 __builtin_ia32_pmovusdw512mem_mask,
1942 __builtin_ia32_pmovsdw512mem_mask,
1943 __builtin_ia32_pmovdw512mem_mask,
1944 __builtin_ia32_pmovqb512mem_mask,
1945 __builtin_ia32_pmovusqb512mem_mask,
1946 __builtin_ia32_pmovsqb512mem_mask,
1947 __builtin_ia32_pmovusdb512mem_mask,
1948 __builtin_ia32_pmovsdb512mem_mask,
1949 __builtin_ia32_pmovdb512mem_mask.
1950 (bdesc_args): Add __builtin_ia32_expanddf512,
1951 __builtin_ia32_expandsf512.
1952 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1953 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1954 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1955 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1956 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1957 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1958 (avx512f_<code>v8div16qi2_mask_store): This.
1959 (avx512f_expand<mode>): New.
1961 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1963 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1965 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1966 (_mm512_prefetch_i32scatter_pd): Ditto.
1967 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1968 (_mm512_prefetch_i64scatter_pd): Ditto.
1969 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1970 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1971 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1972 (_mm512_prefetch_i32scatter_ps): Ditto.
1973 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1974 (_mm512_prefetch_i64scatter_ps): Ditto.
1975 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1976 * config/i386/i386-builtin-types.def: Define
1977 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1978 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1979 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1980 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1981 IX86_BUILTIN_SCATTERPFQPD.
1982 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1983 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1984 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1985 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1986 __builtin_ia32_scatterpfqps.
1987 (ix86_expand_builtin): Expand new built-ins.
1988 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1989 fix memory access data type.
1990 (*avx512pf_gatherpf<mode>_mask): Ditto.
1991 (*avx512pf_gatherpf<mode>): Ditto.
1992 (avx512pf_scatterpf<mode>): Ditto.
1993 (*avx512pf_scatterpf<mode>_mask): Ditto.
1994 (*avx512pf_scatterpf<mode>): Ditto.
1995 (GATHER_SCATTER_SF_MEM_MODE): New.
1996 (avx512pf_gatherpf<mode>df): Ditto.
1997 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1998 (*avx512pf_scatterpf<mode>df): Ditto.
2000 2014-01-27 Jakub Jelinek <jakub@redhat.com>
2003 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
2004 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
2007 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2009 * common/config/arm/arm-common.c
2010 (arm_rewrite_mcpu): Handle multiple names.
2012 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2014 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
2016 * gimple-builder.h (create_gimple_tmp): Delete.
2018 2014-01-27 Christian Bruel <christian.bruel@st.com>
2020 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
2023 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
2025 * config/pa/pa.md (call): Generate indirect long calls to non-local
2026 functions when outputing 32-bit code.
2027 (call_value): Likewise except for special call to buggy powf function.
2029 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
2030 portable runtime and PIC indirect calls.
2031 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2032 and PIC call sequences. Use ldo instead of blr to set return register
2033 in PIC call sequence.
2035 2014-01-25 Walter Lee <walt@tilera.com>
2037 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2038 avoid clobbering a live register.
2040 2014-01-25 Walter Lee <walt@tilera.com>
2042 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2043 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2044 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2045 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2047 2014-01-25 Walter Lee <walt@tilera.com>
2049 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2050 arguments on even registers.
2051 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2053 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2054 (BIGGEST_ALIGNMENT): Ditto.
2055 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2057 2014-01-25 Walter Lee <walt@tilera.com>
2059 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2060 insns before bundling.
2061 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2063 2014-01-25 Walter Lee <walt@tilera.com>
2065 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2066 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2067 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2069 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2071 * config/mips/constraints.md (kl): Delete.
2072 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2073 define expands, using...
2074 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2075 instructions for MIPS16.
2076 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2077 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2079 2014-01-25 Walter Lee <walt@tilera.com>
2081 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2085 2014-01-25 Walter Lee <walt@tilera.com>
2087 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2088 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2090 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2092 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2095 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2097 * print-rtl.c (in_call_function_usage): New var.
2098 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2099 EXPR_LIST mode as mode and not as reg note name.
2102 * cfgloopmanip.c (copy_loop_info): If
2103 loop->warned_aggressive_loop_optimizations, make sure
2104 the flag is set in target loop too.
2106 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2108 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2110 * builtins.def: Likewise.
2111 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2112 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2113 * ira.c (ira_setup_eliminable_regset): Likewise.
2114 * omp-low.c (gate_expand_omp): Likewise.
2115 (execute_lower_omp): Likewise.
2116 (diagnose_sb_0): Likewise.
2117 (gate_diagnose_omp_blocks): Likewise.
2118 (simd_clone_clauses_extract): Likewise.
2121 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2123 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2124 correction for little endian...
2125 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2128 2014-01-24 Jeff Law <law@redhat.com>
2130 PR tree-optimization/59919
2131 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2132 for non-returning calls.
2134 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2136 * common/config/aarch64/aarch64-common.c
2137 (aarch64_rewrite_mcpu): Handle multiple names.
2138 * config/aarch64/aarch64.h
2139 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2141 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2143 * input.c (add_file_to_cache_tab): Handle the case where fopen
2146 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2149 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2150 from push operand if code of push isn't PRE_DEC.
2152 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2155 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2156 -mquad-memory-atomic. Update -mquad-memory documentation to say
2157 it is only used for non-atomic loads/stores.
2159 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2160 -mquad-memory or -mquad-memory-atomic switches.
2162 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2163 -mquad-memory-atomic to ISA 2.07 support.
2165 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2166 to separate support of normal quad word memory operations (ldq, stq)
2167 from the atomic quad word memory operations.
2169 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2170 support to separate non-atomic quad word operations from atomic
2171 quad word operations. Disable non-atomic quad word operations in
2172 little endian mode so that we don't have to swap words after the
2173 load and before the store.
2174 (quad_load_store_p): Add comment about atomic quad word support.
2175 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2176 options printed with -mdebug=reg.
2178 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2179 -mquad-memory-atomic as the test for whether we have quad word
2180 atomic instructions.
2181 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2182 or -mp8-vector are used, allow byte/half-word atomic operations.
2184 * config/rs6000/sync.md (load_lockedti): Insure that the address
2185 is a proper indexed or indirect address for the lqarx instruction.
2186 On little endian systems, swap the hi/lo registers after the lqarx
2188 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2189 insure the address is valid for the lqarx instruction.
2190 (store_conditionalti): Insure that the address is a proper indexed
2191 or indirect address for the stqcrx. instruction. On little endian
2192 systems, swap the hi/lo registers before doing the stqcrx.
2194 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2195 insure the address is valid for the stqcrx. instruction.
2197 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2198 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2199 type of quad memory support is available.
2201 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2204 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2205 there is a danger of looping.
2207 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2209 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2210 force flag_ira_loop_pressure if set via command line.
2212 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2214 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2215 (ashr_simd): New builtin handling DI mode.
2216 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2217 (aarch64_sshr_simddi): New match pattern.
2218 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2219 (vshrd_n_s64): Likewise.
2220 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2222 2014-01-23 Nick Clifton <nickc@redhat.com>
2224 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2225 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2226 favour of mcu specific scripts.
2227 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2230 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2231 Alex Velenko <Alex.Velenko@arm.com>
2233 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2234 (vaddv_s16): Likewise.
2235 (vaddv_s32): Likewise.
2236 (vaddv_u8): Likewise.
2237 (vaddv_u16): Likewise.
2238 (vaddv_u32): Likewise.
2239 (vaddvq_s8): Likewise.
2240 (vaddvq_s16): Likewise.
2241 (vaddvq_s32): Likewise.
2242 (vaddvq_s64): Likewise.
2243 (vaddvq_u8): Likewise.
2244 (vaddvq_u16): Likewise.
2245 (vaddvq_u32): Likewise.
2246 (vaddvq_u64): Likewise.
2247 (vaddv_f32): Likewise.
2248 (vaddvq_f32): Likewise.
2249 (vaddvq_f64): Likewise.
2250 (vmaxv_f32): Likewise.
2251 (vmaxv_s8): Likewise.
2252 (vmaxv_s16): Likewise.
2253 (vmaxv_s32): Likewise.
2254 (vmaxv_u8): Likewise.
2255 (vmaxv_u16): Likewise.
2256 (vmaxv_u32): Likewise.
2257 (vmaxvq_f32): Likewise.
2258 (vmaxvq_f64): Likewise.
2259 (vmaxvq_s8): Likewise.
2260 (vmaxvq_s16): Likewise.
2261 (vmaxvq_s32): Likewise.
2262 (vmaxvq_u8): Likewise.
2263 (vmaxvq_u16): Likewise.
2264 (vmaxvq_u32): Likewise.
2265 (vmaxnmv_f32): Likewise.
2266 (vmaxnmvq_f32): Likewise.
2267 (vmaxnmvq_f64): Likewise.
2268 (vminv_f32): Likewise.
2269 (vminv_s8): Likewise.
2270 (vminv_s16): Likewise.
2271 (vminv_s32): Likewise.
2272 (vminv_u8): Likewise.
2273 (vminv_u16): Likewise.
2274 (vminv_u32): Likewise.
2275 (vminvq_f32): Likewise.
2276 (vminvq_f64): Likewise.
2277 (vminvq_s8): Likewise.
2278 (vminvq_s16): Likewise.
2279 (vminvq_s32): Likewise.
2280 (vminvq_u8): Likewise.
2281 (vminvq_u16): Likewise.
2282 (vminvq_u32): Likewise.
2283 (vminnmv_f32): Likewise.
2284 (vminnmvq_f32): Likewise.
2285 (vminnmvq_f64): Likewise.
2287 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2289 * config/aarch64/aarch64-simd.md
2290 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2291 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2292 (*aarch64_mul3_elt<mode>): Likewise.
2293 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2294 (*aarch64_mul3_elt_to_64v2df): Likewise.
2295 (*aarch64_mla_elt<mode>): Likewise.
2296 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2297 (*aarch64_mls_elt<mode>): Likewise.
2298 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2299 (*aarch64_fma4_elt<mode>): Likewise.
2300 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2301 (*aarch64_fma4_elt_to_64v2df): Likewise.
2302 (*aarch64_fnma4_elt<mode>): Likewise.
2303 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2304 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2305 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2306 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2307 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2308 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2309 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2310 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2311 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2313 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2315 * config/aarch64/aarch64-simd.md
2316 (aarch64_be_checked_get_lane<mode>): New define_expand.
2317 * config/aarch64/aarch64-simd-builtins.def
2318 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2319 New builtin definition.
2320 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2321 Use new safe be builtin.
2323 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2325 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2327 (aarch64_be_st1<mode>): Likewise.
2328 (aarch_ld1<VALL:mode>): Define_expand modified.
2329 (aarch_st1<VALL:mode>): Likewise.
2330 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2331 (UNSPEC_ST1): Likewise.
2333 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2335 * config/microblaze/microblaze.md: Add trap insn and attribute
2337 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2339 PR preprocessor/58580
2340 * input.h (location_get_source_line): Take an additional line_size
2342 (void diagnostics_file_cache_fini): Declare new function.
2343 * input.c (struct fcache): New type.
2344 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2345 New static constants.
2346 (diagnostic_file_cache_init, total_lines_num)
2347 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2348 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2349 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2350 (get_next_line, read_next_line, goto_next_line, read_line_num):
2351 New static function definitions.
2352 (diagnostic_file_cache_fini): New function.
2353 (location_get_source_line): Take an additional output line_len
2354 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2356 * diagnostic.c (diagnostic_finish): Call
2357 diagnostic_file_cache_fini.
2358 (adjust_line): Take an additional input parameter for the length
2359 of the line, rather than calculating it with strlen.
2360 (diagnostic_show_locus): Adjust the use of
2361 location_get_source_line and adjust_line with respect to their new
2362 signature. While displaying a line now, do not stop at the first
2363 null byte. Rather, display the zero byte as a space and keep
2364 going until we reach the size of the line.
2365 * Makefile.in: Add vec.o to OBJS-libcommon
2367 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2368 Ilya Tocar <ilya.tocar@intel.com>
2370 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2371 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2372 (__builtin_ia32_kmov16): Ditto.
2373 * config/i386/i386.md (UNSPEC_KMOV): New.
2376 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2378 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2379 (_mm512_storeu_si512): Ditto.
2381 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2384 * rtl.h (get_referenced_operands): Declare.
2385 * recog.c (get_referenced_operands): New function.
2386 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2387 operands have been referenced when recording LO_SUM references.
2389 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2391 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2393 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2395 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2396 Enable for generic and recent AMD targets.
2398 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2400 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2401 ARG_SIZE note when adjustment was eliminated.
2403 2014-01-22 Jeff Law <law@redhat.com>
2405 PR tree-optimization/59597
2406 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2407 in file. Accept new argument REGISTERING and use it to modify
2408 dump output appropriately.
2409 (register_jump_thread): Corresponding changes.
2410 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2411 thread paths involving joiner blocks. Add code to dump cancelled
2412 jump threading paths.
2414 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2416 PR rtl-optimization/59477
2417 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2418 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2420 2014-01-22 Tom Tromey <tromey@redhat.com>
2422 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2424 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2426 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2428 PR rtl-optimization/59896
2429 * lra-constraints.c (process_alt_operands): Check unused note for
2430 matched operands of insn with no output reloads.
2432 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2434 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2435 (mips_move_from_gpr_cost): Likewise.
2437 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2439 PR rtl-optimization/59858
2440 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2441 ira_class_hard_regs_num.
2442 (process_alt_operands): Increase reject for dying matched operand.
2444 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2447 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2448 smaller than size, perform several stores or loads and stores
2449 at dst + count - size to store or copy all of size bytes, rather
2450 than just last modesize bytes.
2452 2014-01-20 DJ Delorie <dj@redhat.com>
2454 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2455 that CLOBBERs are REGs before propogating their values.
2457 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2460 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2461 (cgraph_inline_failed_type): New function.
2462 * cgraph.h (DEFCIFCODE): Add type.
2463 (cgraph_inline_failed_type_t): New enum.
2464 (cgraph_inline_failed_type): New prototype.
2465 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2466 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2467 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2468 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2469 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2470 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2471 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2472 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2473 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2474 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2475 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2476 OPTIMIZATION_MISMATCH.
2477 * tree-inline.c (expand_call_inline): Emit errors during
2478 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2480 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2483 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2484 mode attribute in insn output.
2486 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2488 * output.h (output_constant): Delete.
2489 * varasm.c (output_constant): Make private.
2491 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2493 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2495 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2498 * tree.h (fold_builtin_strcat): New prototype.
2499 * builtins.c (fold_builtin_strcat): No longer static. Add len
2500 argument, if non-NULL, don't call c_strlen. Optimize
2501 directly into __builtin_memcpy instead of __builtin_strcpy.
2502 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2503 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2505 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2507 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2508 for SImode_address_operand operands, having only a REG argument.
2510 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2512 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2513 loader name using mbig-endian.
2514 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2516 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2518 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2522 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2524 * config/aarch64/aarch64-protos.h
2525 (aarch64_cannot_change_mode_class_ptr): Declare.
2526 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2527 aarch64_cannot_change_mode_class_ptr): New.
2528 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2529 backend hook aarch64_cannot_change_mode_class.
2531 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2533 * common/config/aarch64/aarch64-common.c
2534 (aarch64_handle_option): Don't handle any option order logic here.
2535 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2536 selected_cpu, warn on architecture version mismatch.
2537 (aarch64_override_options): Fix parsing order for option strings.
2539 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2540 Iain Sandoe <iain@codesourcery.com>
2543 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2544 warning. Amend comment to reflect current functionality.
2546 2014-01-20 Richard Biener <rguenther@suse.de>
2549 * builtins.c (fold_builtin_strcat): Remove case better handled
2550 by tree-ssa-strlen.c.
2552 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2554 * config/aarch64/aarch64.opt
2555 (mcpu, march, mtune): Make case-insensitive.
2557 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2560 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2561 if operands[1] is a REG or ZERO_EXTEND of a REG.
2563 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2565 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2567 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2569 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2570 long non-pic millicode calls.
2572 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2574 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2576 2014-01-19 Kito Cheng <kito@0xlab.org>
2578 * builtins.c (expand_movstr): Check movstr expand done or fail.
2580 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2581 H.J. Lu <hongjiu.lu@intel.com>
2584 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2585 to DImode for zero-extended addresses.
2587 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2589 PR rtl-optimization/57763
2590 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2591 on the new indirect jump_insn and increment LABEL_NUSES (label).
2593 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2597 * config.gcc (x86_archs): New variable.
2598 (x86_64_archs): Likewise.
2599 (x86_cpus): Likewise.
2600 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2601 --with-arch/--with-cpu= options.
2602 Support --with-arch=/--with-cpu={nehalem,westmere,
2603 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2605 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2607 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2608 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2610 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2612 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2614 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2617 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2618 clear cpp_get_options (parse_in)->warn_unused_macros for
2619 ix86_target_macros_internal with cpp_define.
2621 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2623 * jump.c (delete_related_insns): Keep (use (insn))s.
2624 * reorg.c (redundant_insn): Check for barriers too.
2626 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2628 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2630 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2632 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2633 call to $$dyncall when TARGET_LONG_CALLS is true.
2635 2014-01-17 Jeff Law <law@redhat.com>
2637 * ree.c (combine_set_extension): Temporarily disable test for
2638 changing number of hard registers.
2640 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2643 * ipa-inline-analysis.c (inline_free_summary):
2644 Do not free summary of aliases.
2646 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2649 * gimplify.c (gimplify_expr): Use create_tmp_var
2650 instead of create_tmp_var_raw. If cond doesn't have
2651 integral type, don't add the IFN_ANNOTATE builtin at all.
2653 2014-01-17 Martin Jambor <mjambor@suse.cz>
2656 * ipa-cp.c (prev_edge_clone): New variable.
2657 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2658 Also resize prev_edge_clone vector.
2659 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2660 (ipcp_edge_removal_hook): New function.
2661 (ipcp_driver): Register ipcp_edge_removal_hook.
2663 2014-01-17 Andrew Pinski <apinski@cavium.com>
2664 Steve Ellcey <sellcey@mips.com>
2667 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2670 2014-01-17 Jeff Law <law@redhat.com>
2673 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2674 so that pass_ccp runs first.
2676 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2678 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2679 (ix86_adjust_cost): Use !TARGET_XXX.
2680 (do_reorder_for_imul): Likewise.
2681 (swap_top_of_ready_list): Likewise.
2682 (ix86_sched_reorder): Likewise.
2684 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2686 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2687 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2688 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2689 (intel_memset): New. Duplicate slm_memset.
2690 (intel_cost): New. Duplicate slm_cost.
2691 (m_INTEL): New macro.
2692 (processor_target_table): Add "intel".
2693 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2694 with PROCESSOR_INTEL for "intel".
2695 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2696 PROCESSOR_SILVERMONT.
2697 (ix86_issue_rate): Likewise.
2698 (ix86_adjust_cost): Likewise.
2699 (ia32_multipass_dfa_lookahead): Likewise.
2700 (swap_top_of_ready_list): Likewise.
2701 (ix86_sched_reorder): Likewise.
2702 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2703 instead of TARGET_OPT_AGU.
2704 * config/i386/i386.h (TARGET_INTEL): New.
2705 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2706 (processor_type): Add PROCESSOR_INTEL.
2707 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2708 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2710 2014-01-17 Marek Polacek <polacek@redhat.com>
2713 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2716 2014-01-17 Richard Biener <rguenther@suse.de>
2718 PR tree-optimization/46590
2719 * opts.c (default_options_table): Add entries for
2720 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2721 all enabled at -O1 but not for -Og.
2722 * common.opt (fbranch-count-reg): Remove Init(1).
2723 (fmove-loop-invariants): Likewise.
2724 (ftree-pta): Likewise.
2726 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2728 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2729 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2730 decls to at least the GCC 4.8 used alignments.
2733 * tree-nested.c (convert_nonlocal_reference_stmt,
2734 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2735 of GIMPLE_BIND stmts, adjust associated decls.
2737 2014-01-17 Richard Biener <rguenther@suse.de>
2739 PR tree-optimization/46590
2740 * vec.h (vec<>::bseach): New member function implementing
2741 binary search according to C89 bsearch.
2742 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2743 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2744 bitmap pointer again. Make accesses_in_loop a flat array.
2745 (mem_ref_obstack): New global.
2746 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2747 (mark_ref_stored): Likewise.
2748 (ref_indep_loop_p_2): Likewise.
2749 (set_ref_stored_in_loop): New helper function.
2750 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2751 (memref_free): Adjust.
2752 (record_mem_ref_loc): Simplify.
2753 (gather_mem_refs_stmt): Adjust.
2754 (sort_locs_in_loop_postorder_cmp): New function.
2755 (analyze_memory_references): Sort accesses_in_loop after
2756 loop postorder number.
2757 (find_ref_loc_in_loop_cmp): New function.
2758 (for_all_locs_in_loop): Find relevant cluster of locs in
2759 accesses_in_loop and iterate without recursion.
2760 (execute_sm): Avoid uninit warning.
2761 (struct ref_always_accessed): Simplify.
2762 (ref_always_accessed::operator ()): Likewise.
2763 (ref_always_accessed_p): Likewise.
2764 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2765 loop postorder numbers here.
2766 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2769 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2772 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2773 on decls for which assemble_alias has been called.
2775 2014-01-17 Nick Clifton <nickc@redhat.com>
2777 * config/msp430/msp430.opt: (mcpu): New option.
2778 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2779 (msp430_option_override): Parse target_cpu. If the MCU name
2780 matches a generic string, clear target_mcu.
2781 (msp430_attr): Allow numeric interrupt values up to 63.
2782 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2783 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2785 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2787 * config/msp430/msp430.md (popm): Use %J rather than %I.
2788 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2789 (addhi_cy_i): Use immediate_operand for operand 2.
2790 * doc/invoke.texi: Document -mcpu option.
2792 2014-01-17 Richard Biener <rguenther@suse.de>
2794 PR rtl-optimization/38518
2795 * df.h (df_analyze_loop): Declare.
2796 * df-core.c: Include cfgloop.h.
2797 (df_analyze_1): Split out main part of df_analyze.
2798 (df_analyze): Adjust.
2799 (loop_inverted_post_order_compute): New function.
2800 (loop_post_order_compute): Likewise.
2801 (df_analyze_loop): New function avoiding whole-function
2803 * loop-invariant.c (find_defs): Use df_analyze_loop.
2804 (find_invariants): Adjust.
2805 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2807 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2809 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2810 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2812 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2814 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2815 traversal when removing references.
2817 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2820 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2822 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2825 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2826 pushing a reload for an autoinc when we had previously reloaded an
2827 inner part of the address.
2829 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2831 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2833 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2834 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2835 when not giving up or versioning for alias only because of
2837 (vect_analyze_data_ref_dependences): Set to true.
2838 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2840 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2841 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2845 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2848 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2849 operand 0 predicate for gathers, use a new pseudo as subtarget.
2851 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2854 * lra-constraints.c (process_alt_operands): Add printing debug info.
2855 Check absence of input/output reloads for matched operands too.
2857 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2859 PR rtl-optimization/59835
2860 * ira.c (ira_init_register_move_cost): Increase cost for
2863 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2865 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2867 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2870 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2871 non-register objects. Use gen_(high/low)part more consistently.
2874 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2877 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2878 endian support, remove tests for WORDS_BIG_ENDIAN.
2879 (p8_mfvsrd_3_<mode>): Likewise.
2880 (reload_gpr_from_vsx<mode>): Likewise.
2881 (reload_gpr_from_vsxsf): Likewise.
2882 (p8_mfvsrd_4_disf): Likewise.
2884 2014-01-16 Richard Biener <rguenther@suse.de>
2886 PR rtl-optimization/46590
2887 * lcm.c (compute_antinout_edge): Use postorder iteration.
2888 (compute_laterin): Use inverted postorder iteration.
2890 2014-01-16 Nick Clifton <nickc@redhat.com>
2893 * varasm.c (output_constant): Return the number of bytes actually
2895 (output_constructor_array_range): Update the field size with the
2896 number of bytes emitted by output_constant.
2897 (output_constructor_regular_field): Likewise. Also do not
2898 complain if the total number of bytes emitted is now greater
2899 than the expected fieldpos.
2900 * output.h (output_constant): Update prototype and descriptive comment.
2902 2014-01-16 Marek Polacek <polacek@redhat.com>
2905 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2906 it is error_mark_node.
2908 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2910 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2911 VALID_AVX256_REG_OR_OI_MODE.
2913 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2915 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2916 current procedure should be profiled.
2918 2014-01-15 Andrew Pinski <apinski@cavium.com>
2920 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2921 of moving from/to the STACK_REG register class.
2923 2014-01-15 Richard Henderson <rth@redhat.com>
2926 * reginfo.c (global_regs_decl): Globalize.
2927 * rtl.h (global_regs_decl): Declare.
2928 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2929 reserved via global_regs.
2931 2014-01-15 Teresa Johnson <tejohnson@google.com>
2933 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2935 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2937 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2938 and vmulosh rather than call gen_vec_widen_smult_*.
2939 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2940 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2941 (vec_widen_smult_even_v16qi): Likewise.
2942 (vec_widen_umult_even_v8hi): Likewise.
2943 (vec_widen_smult_even_v8hi): Likewise.
2944 (vec_widen_umult_odd_v16qi): Likewise.
2945 (vec_widen_smult_odd_v16qi): Likewise.
2946 (vec_widen_umult_odd_v8hi): Likewise.
2947 (vec_widen_smult_odd_v8hi): Likewise.
2948 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2949 vmuloub rather than call gen_vec_widen_umult_*.
2950 (vec_widen_umult_lo_v16qi): Likewise.
2951 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2952 vmulosb rather than call gen_vec_widen_smult_*.
2953 (vec_widen_smult_lo_v16qi): Likewise.
2954 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2955 rather than call gen_vec_widen_umult_*.
2956 (vec_widen_umult_lo_v8hi): Likewise.
2957 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2958 rather than call gen_vec_widen_smult_*.
2959 (vec_widen_smult_lo_v8hi): Likewise.
2961 2014-01-15 Jeff Law <law@redhat.com>
2963 PR tree-optimization/59747
2964 * ree.c (find_and_remove_re): Properly handle case where a second
2965 eliminated extension requires widening a copy created for elimination
2966 of a prior extension.
2967 (combine_set_extension): Ensure that the number of hard regs needed
2968 for a destination register does not change when we widen it.
2970 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2972 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2973 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2974 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2975 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2976 (avr-*-rtems*): Likewise.
2977 (bfin*-rtems*): Likewise.
2978 (moxie-*-rtems*): Likewise.
2979 (h8300-*-rtems*): Likewise.
2980 (i[34567]86-*-rtems*): Likewise.
2981 (lm32-*-rtems*): Likewise.
2982 (m32r-*-rtems*): Likewise.
2983 (m68k-*-rtems*): Likewise.
2984 (microblaze*-*-rtems*): Likewise.
2985 (mips*-*-rtems*): Likewise.
2986 (powerpc-*-rtems*): Likewise.
2987 (sh-*-rtems*): Likewise.
2988 (sparc-*-rtems*): Likewise.
2989 (sparc64-*-rtems*): Likewise.
2990 (v850-*-rtems*): Likewise.
2991 (m32c-*-rtems*): Likewise.
2993 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2995 PR rtl-optimization/59511
2996 * ira.c (ira_init_register_move_cost): Use memory costs for some
2997 cases of register move cost calculations.
2998 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2999 instead of BB frequency.
3000 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
3001 * lra-assigns.c (find_hard_regno_for): Ditto.
3003 2014-01-15 Richard Biener <rguenther@suse.de>
3005 PR tree-optimization/59822
3006 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
3007 (vectorizable_load): Use it to hoist defs of uses of invariant
3008 loads out of the loop.
3010 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3011 Kugan Vivekanandarajah <kuganv@linaro.org>
3014 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
3017 2014-01-15 Richard Biener <rguenther@suse.de>
3019 PR rtl-optimization/59802
3020 * lcm.c (compute_available): Use inverted postorder to seed
3021 the initial worklist.
3023 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3026 * config/s390/s390.c (s390_preferred_reload_class): Don't return
3027 ADDR_REGS for invalid symrefs in non-PIC code.
3029 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3032 * builtins.c (determine_block_size): Initialize *probable_max_size
3033 even if len_rtx is CONST_INT.
3035 2014-01-14 Andrew Pinski <apinski@cavium.com>
3037 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3038 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3039 (cortexa53_tunings): Likewise.
3040 (aarch64_sched_issue_rate): New function.
3041 (TARGET_SCHED_ISSUE_RATE): Define.
3043 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3045 * ira-costs.c (find_costs_and_classes): Add missed
3046 ira_init_register_move_cost_if_necessary.
3048 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3051 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3053 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3056 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3057 to indicate if type is used for function return value. Warn ABI
3058 change if the vector mode isn't available for function return value.
3059 (ix86_function_arg_advance): Pass false to type_natural_mode.
3060 (ix86_function_arg): Likewise.
3061 (ix86_gimplify_va_arg): Likewise.
3062 (function_arg_32): Don't warn ABI change.
3063 (ix86_function_value): Pass true to type_natural_mode.
3064 (ix86_return_in_memory): Likewise.
3065 (ix86_struct_value_rtx): Removed.
3066 (TARGET_STRUCT_VALUE_RTX): Likewise.
3068 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3070 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3071 converting a conditional jump into a conditional return.
3073 2014-01-14 Richard Biener <rguenther@suse.de>
3075 PR tree-optimization/58921
3076 PR tree-optimization/59006
3077 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3078 hoisting invariant stmts.
3079 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3080 invariant loads on the preheader edge if possible.
3082 2014-01-14 Joey Ye <joey.ye@arm.com>
3084 * doc/plugin.texi (Building GCC plugins): Update to C++.
3086 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3088 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3089 (_mm_rcp28_round_ss): Ditto.
3090 (_mm_rsqrt28_round_sd): Ditto.
3091 (_mm_rsqrt28_round_ss): Ditto.
3092 (_mm_rcp28_sd): Ditto.
3093 (_mm_rcp28_ss): Ditto.
3094 (_mm_rsqrt28_sd): Ditto.
3095 (_mm_rsqrt28_ss): Ditto.
3096 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3097 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3098 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3099 (IX86_BUILTIN_RCP28SD): Ditto.
3100 (IX86_BUILTIN_RCP28SS): Ditto.
3101 (IX86_BUILTIN_RSQRT28SD): Ditto.
3102 (IX86_BUILTIN_RSQRT28SS): Ditto.
3103 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3104 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3105 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3106 (ix86_expand_special_args_builtin): Expand new FTYPE.
3107 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3108 (srcp14<mode>): Make insn unary.
3109 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3110 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3111 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3112 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3113 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3114 Fix rounding: make it SAE only.
3115 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3117 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3119 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3120 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3121 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3122 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3123 (round_saeonly_mask_scalar_operand4): Ditto.
3124 (round_saeonly_mask_scalar_op3): Ditto.
3125 (round_saeonly_mask_scalar_op4): Ditto.
3127 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3129 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3130 Implement -maltivec=be for vec_insert and vec_extract.
3132 2014-01-10 DJ Delorie <dj@redhat.com>
3134 * config/msp430/msp430.md (call_internal): Don't allow memory
3135 references with SP as the base register.
3136 (call_value_internal): Likewise.
3137 * config/msp430/constraints.md (Yc): New. For memory references
3138 that don't use SP as a base register.
3140 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3141 "an integer without a # prefix"
3142 * config/msp430/msp430.md (epilogue_helper): Use it.
3144 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3147 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3148 AVX512F gather builtins.
3149 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3150 on gather decls with INTEGER_TYPE masktype.
3151 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3152 directly into the builtin rather than hoisting it before loop.
3154 PR tree-optimization/59387
3155 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3156 (scev_const_prop): If folded_casts and type has undefined overflow,
3157 use force_gimple_operand instead of force_gimple_operand_gsi and
3158 for each added stmt if it is assign with
3159 arith_code_with_undefined_signed_overflow, call
3160 rewrite_to_defined_overflow.
3161 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3162 gimple-fold.h instead.
3163 (arith_code_with_undefined_signed_overflow,
3164 rewrite_to_defined_overflow): Moved to ...
3165 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3166 rewrite_to_defined_overflow): ... here. No longer static.
3167 Include gimplify-me.h.
3168 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3169 rewrite_to_defined_overflow): New prototypes.
3171 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3173 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3175 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3177 * builtins.c (get_object_alignment_2): Minor tweak.
3178 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3180 2014-01-13 Christian Bruel <christian.bruel@st.com>
3182 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3183 optimized non constant lengths.
3185 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3188 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3189 load as __atomic_load_N if possible.
3191 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3193 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3195 (rs6000_expand_builtin): Adjust call.
3197 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3200 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3201 * config/rs6000/rs6000.c: Include target-globals.h.
3202 (rs6000_set_current_function): Instead of doing target_reinit
3203 unconditionally, use save_target_globals_default_opts and
3204 restore_target_globals.
3206 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3208 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3209 (rs6000_expand_builtin): Handle mffs and mtfsf.
3210 (rs6000_init_builtins): Define mffs and mtfsf.
3211 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3212 (rs6000_mffs): New pattern.
3213 (rs6000_mtfsf): New pattern.
3215 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3217 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3218 Start narrowing with START. Apply candidate-use pair
3219 and check overall cost in narrowing.
3220 (iv_ca_prune): Pass new argument.
3222 2014-01-10 Jeff Law <law@redhat.com>
3225 * ree.c (combine_reaching_defs): Ensure the defining statement
3226 occurs before the extension when optimizing extensions with
3227 different source and destination hard registers.
3229 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3232 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3233 vtables into the type inheritance graph.
3235 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3237 PR rtl-optimization/59754
3238 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3239 modes in the REGNO != REGNO case.
3241 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3243 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3245 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3247 PR tree-optimization/59745
3248 * tree-predcom.c (tree_predictive_commoning_loop): Call
3249 free_affine_expand_cache if giving up because components is NULL.
3251 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3252 GC in payload of target_globals struct instead of allocating them on
3253 the heap and the larger structs separately using GC.
3254 * target-globals.h (struct target_globals): Make regs, hard_regs,
3255 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3256 of GTY((skip)) and change type to void *.
3257 (reset_target_globals): Cast loads from those fields to corresponding
3260 2014-01-10 Steve Ellcey <sellcey@mips.com>
3263 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3264 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3265 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3267 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3270 * aarch64-modes.def (CC_Zmode): New flags mode.
3271 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3272 represents an equality.
3273 (aarch64_get_condition_code): Handle CC_Zmode.
3274 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3276 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3278 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3279 extraction in good case.
3281 2014-01-10 Richard Biener <rguenther@suse.de>
3283 PR tree-optimization/59374
3284 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3285 checking after SLP discovery. Mark stmts not participating
3286 in any SLP instance properly.
3288 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3290 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3291 when handling a SET rtx.
3293 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3295 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3296 (cortex-a57): Likewise.
3297 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3299 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3301 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3302 non-iwmmxt builtins.
3304 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3308 * ipa-devirt.c record_target_from_binfo): Take as argument
3309 stack of binfos and lookup matching one for virtual inheritance.
3310 (possible_polymorphic_call_targets_1): Update.
3312 2014-01-10 Huacai Chen <chenhc@lemote.com>
3314 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3315 kernel strings for Loongson-2E/2F/3A.
3317 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3320 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3321 is_gimple_call before calling gimple_call_internal_p.
3323 2014-01-09 Steve Ellcey <sellcey@mips.com>
3325 * Makefile.in (TREE_FLOW_H): Remove.
3326 (TREE_SSA_H): Add file names from tree-flow.h.
3327 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3328 * tree.h: Remove tree-flow.h reference.
3329 * hash-table.h: Remove tree-flow.h reference.
3330 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3331 reference with tree-ssa-loop.h.
3333 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3335 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3336 default element-order behavior for -maltivec.
3337 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3338 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3339 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3340 when targeting big endian, at least for now.
3341 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3343 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3346 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3347 var satisfies use_register_for_decl, just take into account type
3348 alignment, rather than decl alignment.
3350 PR tree-optimization/59622
3351 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3352 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3353 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3354 Don't devirtualize for inplace at all. For targets.length () == 1,
3355 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3357 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3359 * config/i386/i386.md (cpu): Remove the unused btver1.
3361 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3363 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3365 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3368 * tree-core.h (struct target_globals): New forward declaration.
3369 (struct tree_target_option): Add globals field.
3370 * tree.h (TREE_TARGET_GLOBALS): Define.
3371 (prepare_target_option_nodes_for_pch): New prototype.
3372 * target-globals.h (struct target_globals): Define even if
3374 * tree.c (prepare_target_option_node_for_pch,
3375 prepare_target_option_nodes_for_pch): New functions.
3376 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3377 * config/i386/i386.c: Include target-globals.h.
3378 (ix86_set_current_function): Instead of doing target_reinit
3379 unconditionally, use save_target_globals_default_opts and
3380 restore_target_globals.
3382 2014-01-09 Richard Biener <rguenther@suse.de>
3384 PR tree-optimization/59715
3385 * tree-cfg.h (split_critical_edges): Declare.
3386 * tree-cfg.c (split_critical_edges): Export.
3387 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3389 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3391 * cfgexpand.c (expand_stack_vars): Optionally disable
3392 asan stack protection.
3393 (expand_used_vars): Likewise.
3394 (partition_stack_vars): Likewise.
3395 * asan.c (asan_emit_stack_protection): Optionally disable
3396 after return stack usage.
3397 (instrument_derefs): Optionally disable memory access instrumentation.
3398 (instrument_builtin_call): Likewise.
3399 (instrument_strlen_call): Likewise.
3400 (asan_protect_global): Optionally disable global variables protection.
3401 * doc/invoke.texi: Added doc for new options.
3402 * params.def: Added new options.
3403 * params.h: Likewise.
3405 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3407 PR rtl-optimization/59724
3408 * ifcvt.c (cond_exec_process_if_block): Don't call
3409 flow_find_head_matching_sequence with 0 longest_match.
3410 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3411 non-active insns if !stop_after.
3412 (try_head_merge_bb): Revert 2014-01-07 changes.
3414 2014-01-08 Jeff Law <law@redhat.com>
3416 * ree.c (get_sub_rtx): New function, extracted from...
3417 (merge_def_and_ext): Here.
3418 (combine_reaching_defs): Use get_sub_rtx.
3420 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3422 * cgraph.h (varpool_variable_node): Do not choke on null node.
3424 2014-01-08 Catherine Moore <clm@codesourcery.com>
3426 * config/mips/mips.md (simple_return): Attempt to use JRC
3428 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3430 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3432 PR rtl-optimization/59137
3433 * reorg.c (steal_delay_list_from_target): Call update_block for
3435 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3437 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3439 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3440 two duplicate entries.
3442 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3445 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3447 * config/mips/mips.c (mips_truncated_op_cost): New function.
3448 (mips_rtx_costs): Adjust test for BADDU.
3449 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3451 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3453 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3454 (*baddu_si): ...this new pattern.
3456 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3459 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3461 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3464 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3466 (expand_expr, expand_normal): Adjust.
3467 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3468 inner_reference_p. Use inner_reference_p to expand inner references.
3469 (store_expr): Adjust.
3470 * cfgexpand.c (expand_call_stmt): Adjust.
3472 2014-01-08 Rong Xu <xur@google.com>
3474 * gcov-io.c (gcov_var): Move from gcov-io.h.
3475 (gcov_position): Ditto.
3476 (gcov_is_error): Ditto.
3477 (gcov_rewrite): Ditto.
3478 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3479 only part to libgcc/libgcov.h.
3481 2014-01-08 Marek Polacek <polacek@redhat.com>
3484 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3486 2014-01-08 Marek Polacek <polacek@redhat.com>
3489 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3491 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3493 PR rtl-optimization/59649
3494 * stor-layout.c (get_mode_bounds): For BImode return
3495 0 and STORE_FLAG_VALUE.
3497 2014-01-08 Richard Biener <rguenther@suse.de>
3500 * gimple.h (is_gimple_builtin_call): Remove.
3501 (gimple_builtin_call_types_compatible_p): New.
3502 (gimple_call_builtin_p): New overload.
3503 * gimple.c (is_gimple_builtin_call): Remove.
3504 (validate_call): Rename to ...
3505 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3507 (validate_type): New static function.
3508 (gimple_call_builtin_p): New overload and adjust.
3509 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3510 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3511 (gimple_fold_stmt_to_constant_1): Likewise.
3512 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3514 2014-01-08 Richard Biener <rguenther@suse.de>
3517 * gimplify.c (gimplify_expr): Gimplify register-register type
3518 VIEW_CONVERT_EXPRs to separate stmts.
3520 2014-01-07 Jeff Law <law@redhat.com>
3523 * ree.c (combine_set_extension): Handle case where source
3524 and destination registers in an extension insn are different.
3525 (combine_reaching_defs): Allow source and destination registers
3526 in extension to be different under limited circumstances.
3527 (add_removable_extension): Remove restriction that the
3528 source and destination registers in the extension are the same.
3529 (find_and_remove_re): Emit a copy from the extension's
3530 destination to its source after the defining insn if
3531 the source and destination registers are different.
3534 * ifcvt.c (merge_if_block): If we are merging a block with more than
3535 one successor with a block with no successors, remove any BARRIER
3536 after the second block.
3538 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3540 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3542 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3545 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3546 for 14-bit register offsets when INT14_OK_STRICT is false.
3548 2014-01-07 Roland Stigge <stigge@antcom.de>
3549 Michael Meissner <meissner@linux.vnet.ibm.com>
3552 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3553 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3555 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3557 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3560 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3562 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3563 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3564 rtx is const0_rtx or not.
3566 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3569 * target-globals.c (save_target_globals): Remove this_fn_optab
3571 * toplev.c: Include optabs.h.
3572 (target_reinit): Temporarily restore the global options if another
3573 set of options are in force.
3575 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3577 PR rtl-optimization/58668
3578 * cfgcleanup.c (flow_find_cross_jump): Don't count
3579 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3580 to determine what is counted.
3581 (flow_find_head_matching_sequence): Use active_insn_p to determine
3583 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3585 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3586 determine what is counted.
3588 PR tree-optimization/59643
3589 * tree-predcom.c (split_data_refs_to_components): If one dr is
3590 read and one write, determine_offset fails and the write isn't
3591 in the bad component, just put the read into the bad component.
3593 2014-01-07 Mike Stump <mikestump@comcast.net>
3594 Jakub Jelinek <jakub@redhat.com>
3597 * tree-core.h (struct tree_optimization_option): Change optabs
3598 type from unsigned char * to void *.
3599 * optabs.c (init_tree_optimization_optabs): Adjust
3600 TREE_OPTIMIZATION_OPTABS initialization.
3602 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3605 * config/i386/i386.h (struct machine_function): Add
3606 no_drap_save_restore field.
3607 * config/i386/i386.c (ix86_save_reg): Use
3608 !cfun->machine->no_drap_save_restore instead of
3609 crtl->stack_realign_needed.
3610 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3611 this function clears frame_pointer_needed. Set
3612 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3613 and DRAP reg is needed.
3615 2014-01-06 Marek Polacek <polacek@redhat.com>
3618 * doc/implement-c.texi: Mention that other integer types are
3619 permitted as bit-field types in strictly conforming mode.
3621 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3623 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3626 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3628 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3630 2014-01-06 Martin Jambor <mjambor@suse.cz>
3633 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3635 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3637 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3641 * var-tracking.c (add_stores): Preserve the value of the source even if
3642 we don't record the store.
3644 2014-01-06 Terry Guo <terry.guo@arm.com>
3646 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3648 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3651 * config/darwin.c (darwin_function_section): Adjust return values to
3652 correspond to optimisation changes made in r206070.
3654 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3656 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3657 from prefetch_block tune setting.
3658 (nocona_cost): Correct size of prefetch block to 64.
3660 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3662 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3663 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3664 used to save the static chain register in the computation of the offset
3665 from which the FP registers need to be restored.
3667 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3669 PR tree-optimization/59519
3670 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3671 ICE if get_current_def (current_new_name) is already non-NULL, as long
3672 as it is a phi result of some other phi in *new_exit_bb that has
3675 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3676 or vmovdqu* for misaligned_operand.
3677 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3678 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3679 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3680 aligned_mem for AVX512F masked aligned load and store builtins and for
3683 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3685 PR tree-optimization/59651
3686 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3687 Address range for negative step should be added by TYPE_SIZE_UNIT.
3689 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3691 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3692 overlapping registers also for registers other than the stack pointer.
3694 2014-01-03 Marek Polacek <polacek@redhat.com>
3697 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3700 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3703 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3706 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3707 (push splitter): Use <P:MODE_SIZE> instead of
3708 GET_MODE_SIZE (<P:MODE>mode).
3709 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3710 (mov -1, reg peephole2): Likewise.
3711 * config/i386/sse.md (*mov<mode>_internal,
3712 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3713 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3714 *<code><mode>3, *andnot<mode>3<mask_name>,
3715 <mask_codefor><code><mode>3<mask_name>): Likewise.
3716 * config/i386/subst.md (mask_mode512bit_condition,
3717 sd_mask_mode512bit_condition): Likewise.
3719 2014-01-02 Xinliang David Li <davidxl@google.com>
3721 PR tree-optimization/59303
3722 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3723 (dump_predicates): Better output format.
3724 (pred_equal_p): New function.
3725 (is_neq_relop_p): Ditto.
3726 (is_neq_zero_form_p): Ditto.
3727 (pred_expr_equal_p): Ditto.
3728 (pred_neg_p): Ditto.
3729 (simplify_pred): Ditto.
3730 (simplify_preds_2): Ditto.
3731 (simplify_preds_3): Ditto.
3732 (simplify_preds_4): Ditto.
3733 (simplify_preds): Ditto.
3735 (push_to_worklist): Ditto.
3736 (get_pred_info_from_cmp): Ditto.
3737 (is_degenerated_phi): Ditto.
3738 (normalize_one_pred_1): Ditto.
3739 (normalize_one_pred): Ditto.
3740 (normalize_one_pred_chain): Ditto.
3741 (normalize_preds): Ditto.
3742 (normalize_cond_1): Remove function.
3743 (normalize_cond): Ditto.
3744 (is_gcond_subset_of): Ditto.
3745 (is_subset_of_any): Ditto.
3746 (is_or_set_subset_of): Ditto.
3747 (is_and_set_subset_of): Ditto.
3748 (is_norm_cond_subset_of): Ditto.
3749 (pred_chain_length_cmp): Ditto.
3750 (convert_control_dep_chain_into_preds): Type change.
3751 (find_predicates): Ditto.
3752 (find_def_preds): Ditto.
3753 (destroy_predicates_vecs): Ditto.
3754 (find_matching_predicates_in_rest_chains): Ditto.
3755 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3756 (is_pred_expr_subset): Ditto.
3757 (is_pred_chain_subset_of): Ditto.
3758 (is_included_in): Ditto.
3759 (is_superset_of): Ditto.
3761 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3763 Update copyright years.
3765 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3767 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3768 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3769 config/arc/arc.md, config/arc/arc.opt,
3770 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3771 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3772 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3773 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3774 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3775 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3776 vtable-verify.c, vtable-verify.h: Use the standard form for the
3779 2014-01-02 Tobias Burnus <burnus@net-b.de>
3781 * gcc.c (process_command): Update copyright notice dates.
3782 * gcov-dump.c: Ditto.
3784 * doc/cpp.texi: Bump @copying's copyright year.
3785 * doc/cppinternals.texi: Ditto.
3786 * doc/gcc.texi: Ditto.
3787 * doc/gccint.texi: Ditto.
3788 * doc/gcov.texi: Ditto.
3789 * doc/install.texi: Ditto.
3790 * doc/invoke.texi: Ditto.
3792 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3794 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3796 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3798 * config/i386/sse.md (*mov<mode>_internal): Guard
3799 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3801 PR rtl-optimization/59647
3802 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3803 new_rtx into UNSIGNED_FLOAT rtxes.
3805 Copyright (C) 2014 Free Software Foundation, Inc.
3807 Copying and distribution of this file, with or without modification,
3808 are permitted in any medium without royalty provided the copyright
3809 notice and this notice are preserved.