1 2013-08-28 Jan Hubicka <jh@suse.cz>
3 * builtins.def (free): Declare leaf.
5 2013-08-27 David Malcolm <dmalcolm@redhat.com>
8 * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
9 * configure: Regenerate.
11 2013-08-27 Martin Jambor <mjambor@suse.cz>
13 * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
14 (ipa_ancestor_jf_data): Likewise.
15 (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
16 (ipa_get_jf_pass_through_type_preserved): New function.
17 (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
18 (ipa_get_jf_ancestor_type_preserved): New function.
19 * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
20 (ipa_get_jf_ancestor_result): Likewise.
21 (propagate_vals_accross_pass_through): Use
22 ipa_get_jf_pass_through_result to do all the value mappings.
23 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
25 (ipa_set_jf_cst_copy): New function.
26 (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
27 (ipa_set_jf_arith_pass_through): Likewise.
28 (ipa_set_ancestor_jf): Likewise.
29 (compute_complex_assign_jump_func): Set type_preserved instead of
31 (ipa_compute_jump_functions_for_edge): Likewise.
32 (combine_known_type_and_ancestor_jfs): Honor type_preserved.
33 (update_jump_functions_after_inlining): Update type_preserved.
34 Explicitely create jump functions when combining one with pass_through.
35 (ipa_write_jump_function): Stream the type_preserved flags.
36 (ipa_read_jump_function): Likewise.
38 2013-08-27 Jakub Jelinek <jakub@redhat.com>
39 Aldy Hernandez <aldyh@redhat.com>
41 * Makefile.in (omp-low.o): Depend on $(TARGET_H).
42 * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
43 * function.h (struct function): Add has_force_vect_loops and
45 * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
46 * gimple.c (gimple_build_omp_critical): Add KIND argument and
48 * gimple.def: Update CLAUSES comments.
49 * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
50 (gimple_build_omp_for): Add argument to prototype.
51 (gimple_omp_for_kind): New.
52 (gimple_omp_for_set_kind): New.
53 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
54 GOVD_DATA_SHARE_CLASS.
55 (enum omp_region_type): Add ORT_SIMD.
56 (gimple_add_tmp_var): Handle ORT_SIMD.
57 (gimplify_var_or_parm_decl): Same.
58 (is_gimple_stmt): Same.
59 (omp_firstprivatize_variable): Same.
60 (omp_add_variable): Only use splay_tree_insert if lookup failed.
61 (omp_notice_variable): Handle ORT_SIMD.
62 (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
63 (omp_check_private): Handle ORT_SIMD.
64 (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
66 (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
67 Handle OMP_CLAUSE_LASTPRIVATE.
68 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
70 (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
71 (gimplify_expr): Handle OMP_SIMD.
72 * internal-fn.c (expand_GOMP_SIMD_LANE): New.
73 (expand_GOMP_SIMD_VF): New.
74 (expand_GOMP_SIMD_LAST_LANE): New.
75 * internal-fn.def (GOMP_SIMD_LANE): New.
77 (GOMP_SIMD_LAST_LANE): New.
78 * omp-low.c: Include target.h.
79 (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
81 (check_omp_nesting_restrictions): Same.
83 (lower_rec_simd_input_clauses): New.
84 (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
86 (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
87 GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
88 (expand_omp_build_assign): New.
89 (expand_omp_for_init_counts): New.
90 (expand_omp_for_init_vars): New.
91 (extract_omp_for_update_vars): New.
92 (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
93 and rewrite accordingly.
94 (expand_omp_simd): New.
95 (expand_omp_for): Use expand_omp_simd.
96 (lower_omp_for_lastprivate): Unshare vinit when appropriate.
97 (lower_omp_for): Do not lower the body.
98 * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
100 * tree-flow.h (find_omp_clause): Remove prototype.
101 * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
102 forcing vectorization of the loop, or if flag_tree_vectorize.
103 (gate_tree_if_conversion): Similarly.
104 * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
105 gimple_build_omp_for.
106 (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
107 * tree-parloops (create_parallel_loop): Pass kind argument to
108 gimple_build_omp_for.
109 * tree-pretty-print.c (dump_omp_clause): Add cases for
110 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
111 OMP_CLAUSE__SIMDUID_.
112 (dump_generic_node): Handle OMP_SIMD.
113 * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
114 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
115 unroll OMP_SIMD loops here.
116 * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
117 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
119 (vect_analyze_data_refs): Handle simd loops.
120 * tree-vect-loop.c (vectorizable_live_operation): Handle
122 * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
123 (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
124 (vectorizable_load): Same.
125 * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
126 (struct simduid_to_vf): New.
127 (simduid_to_vf::hash): New.
128 (simduid_to-vf::equal): New.
129 (struct simd_array_to_simduid): New.
130 (simd_array_to_simduid::hash): New.
131 (simd_array_to_simduid::equal): New.
132 (adjust_simduid_builtins): New.
133 (struct note_simd_array_uses_struct): New.
134 (note_simd_array_uses_cb): New.
135 (note_simd_array_uses): New.
136 (vectorize_loops): Handle simd hints and adjust simd builtins
138 * tree-vectorizer.h (struct _stmt_vec_info): Add
139 simd_lane_access_p field.
140 (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
141 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
142 OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
143 (omp_clause_code_name): Same.
144 (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
145 OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
146 * tree.def (OMP_SIMD): New entry.
147 * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
148 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
149 (OMP_CLAUSE_DECL): Adjust range for new clauses.
150 (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
151 (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
152 (OMP_CLAUSE_LINEAR_STEP): New.
153 (OMP_CLAUSE_SAFELEN_EXPR): New.
154 (OMP_CLAUSE__SIMDUID__DECL): New.
155 (find_omp_clause): New prototype.
157 2013-08-27 H.J. Lu <hongjiu.lu@intel.com>
159 * config/i386/driver-i386.c (host_detect_local_cpu): Update
160 Haswell processor detection.
162 2013-08-27 Christian Widmer <shadow@umbrox.de>
165 * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
166 of Ivy Bridge and Haswell processors. Assume core-avx2 for unknown
167 AVX2 capable processors.
169 2013-08-27 Tejas Belagod <tejas.belagod@arm.com>
171 * config/aarch64/arm_neon.h: Replace all inline asm implementations
172 of vget_low_* with implementations in terms of other intrinsics.
174 2013-08-27 Marc Glisse <marc.glisse@inria.fr>
177 * doc/extend.texi (__builtin_isinf_sign): Restrict the return
178 values to -1, 0 and 1.
180 2013-08-27 Vidya Praveen <vidyapraveen@arm.com>
182 * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
183 UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
184 (<optab><mode>3_insn): Remove.
185 (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
186 (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
187 (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
188 (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
189 (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
190 (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
191 (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
192 (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
193 (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
194 (ror<mode>3_insn): Likewise.
195 * config/aarch64/predicates.md (aarch64_simd_register): New.
197 2013-08-27 Richard Biener <rguenther@suse.de>
199 PR tree-optimization/57521
200 * tree-if-conv.c (if_convertible_bb_p): Verify that at least
201 one edge is non-critical.
202 (find_phi_replacement_condition): Make sure to use a non-critical
203 edge. Cleanup and remove old bug workarounds.
204 (bb_postdominates_preds): Remove.
205 (if_convertible_loop_p_1): Do not compute post-dominators.
206 (combine_blocks): Do not free post-dominators.
207 (main_tree_if_conversion): Likewise.
208 (pass_data_if_conversion): Add TODO_verify_ssa.
210 2013-08-27 DJ Delorie <dj@redhat.com>
212 * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
214 2013-08-27 Yufeng Zhang <yufeng.zhang@arm.com>
216 * function.c (assign_parm_find_data_types): Set passed_mode and
217 nominal_mode to the TYPE_MODE of nominal_type for the built
218 pointer type in case of the struct-pass-by-reference.
220 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
222 * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
223 (UINT16_TYPE): Change default to "unsigned int".
225 * config/avr/avr.opt (mfract-convert-truncate): New option.
226 * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
227 is set, round negative fractional integers according to n1169
228 when converting to integer types.
230 2013-08-26 Jan Hubicka <jh@suse.cz>
232 * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
233 methods can not be called indirectly when their address is not taken.
235 2013-08-26 Jan Hubicka <jh@suse.cz>
237 * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
240 2013-08-26 Jan Hubicka <jh@suse.cz>
242 * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
245 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
247 * reload.c (find_valid_class): Allow classes that do not include
248 FIRST_PSEUDO_REGISTER - 1.
250 2013-08-26 Jan Hubicka <jh@suse.cz>
252 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
253 fix edge count/frequency when speculation failed; fix type check
256 2013-08-26 Jan Hubicka <jh@suse.cz>
258 * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
260 2013-08-26 Jan Hubicka <jh@suse.cz>
262 * ipa-inline-transform.c (inline_transform): Be ready for basic block
263 to be changed by edge redirection.
265 2013-08-26 Jan Hubicka <jh@suse.cz>
267 * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
268 formating; add sanity check.
269 (cgraph_resolve_speculation): Add FIXME about scaling profiles.
270 (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
271 * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
272 (resolve_noninline_speculation): Update callee keys, too.
274 2013-08-26 Jan Hubicka <jh@suse.cz>
276 * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
277 (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
279 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
281 * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
284 2013-08-26 Uros Bizjak <ubizjak@gmail.com>
286 * config/i386/i386.c (ix86_debug_options): Remove prototype.
287 (x86_64_elf_select_section): Ditto.
288 (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
290 (ix86_pass_by_reference): Ditto.
291 (output_set_got): Ditto.
292 (ix86_unary_operator_ok): Ditto.
293 (ix86_expand_builtin): Ditto.
295 2013-08-23 Jan Hubicka <jh@suse.cz>
297 * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
299 2013-08-23 Jan Hubicka <jh@suse.cz>
301 * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
302 (tree_decl_with_vis): Add FINAL field.
304 2013-08-23 Jeff Law <law@redhat.com>
306 * tree-ssa-pre.c (do_regular_insertion): Include the expression in
307 the debugging dump when the expression is fully redundant.
309 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
311 * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
312 * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
313 * pretty-print.c (pp_formatted_text_data): Likewise.
314 (pp_write_text_to_stream): Likewise.
315 (pp_write_text_as_dot_label_to_stream): Likewise.
316 (pp_append_r): Likewise.
317 (pp_format): Likewise.
318 (pp_flush): Likewise.
319 (pp_clear_output_area): Likewise.
320 (pp_append_text): Likewise.
321 (pp_formatted_text): Likewise.
322 (pp_remaining_character_count_for_line): Likewise.
323 (pp_newline): Likewise.
324 (pp_character): Likewise.
325 (output_buffer::~output_buffer): Define.
326 (pretty_printer::~pretty_printer): Destruct output buffer.
327 * pretty-print.h (output_buffer::~output_buffer): Declare.
328 (pretty_printer::~pretty_printer): Declare virtual.
330 2013-08-24 Marc Glisse <marc.glisse@inria.fr>
333 * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
334 HOST_WIDE_INT_M1U): New macros.
335 * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
336 fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
337 unsigned -1 for lshift.
338 * cse.c (cse_insn): Likewise.
339 * double-int.c (rshift_double, lshift_double): Likewise.
340 * builtins.c (fold_builtin_bitop): Likewise.
341 * combine.c (force_to_mode): Likewise.
342 * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
343 * simplify-rtx.c (simplify_const_unary_operation,
344 simplify_const_binary_operation): Likewise.
345 * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
346 check_va_list_escapes): Likewise.
347 * rtlanal.c (nonzero_bits1): Likewise.
348 * expmed.c (expand_smod_pow2): Likewise.
349 * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
351 2013-08-23 Jan Hubicka <jh@suse.cz>
353 * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
354 as having address taken.
356 2013-08-23 Jan Hubicka <jh@suse.cz>
358 * ipa-utils.h (method_class_type): Declare.
359 * ipa-devirt.c (method_class_type): Export.
361 * cgraphunit.c (analyze_functions): Do basic devirtualization;
362 do not walk base classes of anonymous types.
364 2013-08-23 Kaz Kojima <kkojima@gcc.gnu.org>
366 PR rtl-optimization/58220
368 * final.c (reemit_insn_block_notes): Use NEXT_INSN to
369 handle SEQUENCE insns properly.
371 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
373 * pretty-print.h (pp_newline_and_flush): Declare. Remove macro
375 (pp_newline_and_indent): Likewise.
376 (pp_separate_with): Likewise.
377 * pretty-print.c (pp_newline_and_flush): Define.
378 (pp_newline_and_indent): Likewise.
379 (pp_separate_with): Likewise.
381 2013-08-23 Jakub Jelinek <jakub@redhat.com>
384 * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
385 * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
387 2013-08-23 Kirill Yukhin <kirill.yukhin@intel.com>
389 * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
390 * gcc/config/i386/i386.md (*movti_internal): Use
391 predicate to determine if EVEX is needed.
392 (*movsi_internal): Ditto.
393 (*movdf_internal): Ditto.
394 (*movsf_internal): Ditto.
395 * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
397 2013-08-23 Jakub Jelinek <jakub@redhat.com>
399 PR tree-optimization/58209
400 * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
401 (find_tail_calls): Give up for pointer result types if m is non-NULL.
402 (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
403 emit POINTER_PLUS_EXPR.
404 (create_tailcall_accumulator): For pointer result type accumulate in
407 2013-08-22 Paolo Carlini <paolo.carlini@oracle.com>
409 * configure.ac: Add backslashes missing from the last change.
410 * configure: Regenerate.
412 2013-08-22 Jan Hubicka <jh@susue.cz>
414 * ipa.c (function_and_variable_visibility): First remember function
415 was global and then make it local.
417 2013-08-22 Julian Brown <julian@codesourcery.com>
419 * configure.ac: Add aarch64 to list of arches which use "nop" in
421 * configure: Regenerate.
423 2013-08-22 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
425 * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
426 gnu_libc_has_function.
427 * config/s390/tpf.h: Likewise.
429 2013-08-22 Jan Hubicka <jh@susue.cz>
431 * timevar.c (validate_phases): Add cast.
433 2013-08-22 Jan Hubicka <jh@susue.cz>
435 * timevar.c (validate_phases): Use size_t for memory.
436 * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
438 2013-08-22 Gabriel Dos Reis <gdr@integrable-solutions.net>
440 * pretty-print.h (output_buffer::output_buffer): Declare.
441 (pretty_printer::pretty_printer): Likewise.
442 (pp_construct): Remove.
443 * pretty-print.c (output_buffer::output_buffer): Define.
444 (pretty_printer::pretty_printer): Rename from pp_construct. Simplify.
445 * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
446 (print_gimple_expr): Likewise.
447 (print_gimple_seq): Likewise.
448 (gimple_dump_bb): Likewise.
449 * sched-vis.c (dump_value_slim): Likewise.
450 (dump_insn_slim): Likewise.
451 (dump_rtl_slim): Likewise.
452 (str_pattern_slim): Likewise.
453 * tree-mudflap.c (mf_varname_tree): Likewise.
454 * graph.c (print_graph_cfg): Likewise.
455 (start_graph_dump): Likewise.
456 * tree-pretty-print.c (maybe_init_pretty_print): Likewise. Use
458 * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
459 pretty printer initialization.
460 * coretypes.h (diagnostic_context): Remove superflous type alias
462 (pretty_printer): Likewise. Declare directly as a class.
463 (pretty_print_info): Remove declaration as class.
464 * asan.c (asan_emit_stack_protection): Remove call to pp_construct
465 and pp_clear_output_area.
466 (asan_add_global): Likewise.
468 2013-08-22 Jan Hubicka <jh@suse.cz>
470 * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
471 * ipa-utils.h (update_type_inheritance_graph): Declare.
472 (possible_polymorphic_call_target_p): Declare.
473 (possible_polymorphic_call_target_p): New.
474 * ipa-devirt.c: Update toplevel comments.
475 (cached_polymorphic_call_targets): Move up.
476 (odr_type_d): Move ID down.
477 (polymorphic_type_binfo_p): Update comment.
478 (odr_hasher::remove): Likewise;
479 (get_odr_type): Set anonymous_namespace.
480 (dump_odr_type): Dump it.
481 (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
482 (maybe_record_node): Record node in cached_polymorphic_call_targets.
483 (record_binfo): Add comment.
484 (free_polymorphic_call_targets_hash): Do not ICE when cache is not
486 (devirt_node_removal_hook): Do not iCE when cache is freed.
487 (possible_polymorphic_call_target_p): New predicate.
488 (update_type_inheritance_graph): New function.
490 2013-08-22 Alexander Ivchenko <alexander.ivchenko@intel.com>
491 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
492 Sergey Lega <sergey.s.lega@intel.com>
493 Anna Tikhonova <anna.tikhonova@intel.com>
494 Ilya Tocar <ilya.tocar@intel.com>
495 Andrey Turetskiy <andrey.turetskiy@intel.com>
496 Ilya Verbin <ilya.verbin@intel.com>
497 Kirill Yukhin <kirill.yukhin@intel.com>
498 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
500 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
501 (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
502 (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
503 (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
504 (OPTION_MASK_ISA_AVX2_UNSET): Update.
505 (OPTION_MASK_ISA_AVX512F_UNSET): New.
506 (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
507 (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
508 (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
509 (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
510 OPT_mavx512pf, OPT_mavx512er cases.
511 * config/i386/constraints.md (v): New constraint.
512 (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
513 * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
515 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
516 AVX512F, AVX512ER, AVX512PF, AVX512CD features.
517 * config/i386/i386-c.c (ix86_target_macros_internal):
518 Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
520 * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
521 (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
522 * config/i386/i386.c (regclass_map, dbx_register_map)
523 (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
524 (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
525 (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
527 (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
528 PTA_AVX512PF, PTA_AVX512CD. Handle -mavx512f, -mavx512er, -mavx512cd,
529 -mavx512pf options. Fix formatting.
530 (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
531 targets. Squash EVEX_SSE_REGS if AVX512F is disabled.
532 (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
533 -mavx512cd, -mavx512pf options.
534 (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
535 (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
536 (ix86_preferred_output_reload_class): Replace SSE_REGS with
538 (ix86_hard_regno_mode_ok): Support 512-bit registers.
539 (ix86_set_reg_reg_cost): Ditto.
540 (x86_order_regs_for_local_alloc): Ditto.
541 (MAX_VECT_LEN): Extend to 64-byte.
542 (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
543 * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
544 (TARGET_AVX512ER, TARGET_AVX512CD): New.
545 (BIGGEST_ALIGNMENT): Extend to 512-bits.
546 (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
547 (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
548 (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
549 (SSE_REG_MODE_P): Support new modes.
550 (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
551 (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
552 (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
553 (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
554 (REG_CLASS_CONTENTS): Add new registers.
555 (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
556 (EXT_REX_SSE_REGNO_P): New.
557 (HI_REGISTER_NAMES): Add new registers.
558 * config/i386/i386.md: Define constants for new registers.
559 (mode): Add new 512-bit modes.
560 (prefix): Support evex prefix.
561 (isa): Support avx512f, noavx512f, fma_avx512f.
562 (ssemodesuffix): Add new 512-bit modes.
564 (*movxi_internal_avx512f): Ditto.
565 (*movdi_internal): Replace constraint "x" with the new constraint "v".
567 (*movsi_internal): Likewise.
568 (*movdf_internal): Likewise.
569 (*movsf_internal): Likewise.
570 (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
571 (<code><mode>3): Likewise.
572 * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
574 * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
575 with the new constraint "v".
576 * config/i386/sse.md (*mov<mode>_internal): Support new registers and
578 (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
579 with the new constraint "v".
580 (<sse2>_loaddqu<avxsizesuffix>): Likewise.
581 (<sse2>_storedqu<avxsizesuffix>): Likewise.
582 (*<plusminus_insn><mode>3): Likewise.
583 (<sse>_vm<plusminus_insn><mode>3): Likewise.
584 (*mul<mode>3): Likewise.
585 (<sse>_vmmul<mode>3): Likewise.
586 (<sse>_div<mode>3): Likewise.
587 (<sse>_vmdiv<mode>3): Likewise.
588 (<sse>_sqrt<mode>2): Likewise.
589 (<sse>_vmsqrt<mode>2): Likewise.
590 (*<code><mode>3_finite): Likewise.
591 (*<code><mode>3) <smaxmin>: Likewise.
592 (<sse>_vm<code><mode>3): Likewise.
593 (*<code><mode>3) <any_logic>: Likewise.
594 (*fma_fmadd_<mode>): Likewise.
595 (*fma_fmsub_<mode>): Likewise.
596 (*fma_fnmadd_<mode>): Likewise.
597 (*fma_fnmsub_<mode>): Likewise.
598 (*fma_fmaddsub_<mode>): Likewise.
599 (*fma_fmsubadd_<mode>): Likewise.
600 (*fmai_fmadd_<mode>): Likewise.
601 (*fmai_fmsub_<mode>): Likewise.
602 (*fmai_fnmadd_<mode>): Likewise.
603 (*fmai_fnmsub_<mode>): Likewise.
604 (sse_cvtsi2ss): Likewise.
605 (sse_cvtsi2ssq): Likewise.
606 (sse_cvtss2si): Likewise.
607 (sse_cvtss2si_2): Likewise.
608 (sse_cvtss2siq): Likewise.
609 (sse_cvtss2siq_2): Likewise.
610 (sse_cvttss2si): Likewise.
611 (sse_cvtss2siq_2): Likewise.
612 (float<sseintvecmodelower><mode>2): Likewise.
613 (sse2_cvtsd2si_2): Likewise.
614 (sse2_cvtsd2siq_2): Likewise.
615 (*<plusminus_insn><mode>3): Likewise.
616 (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
617 (*<sse4_1_avx2>_mul<mode>3): Likewise.
618 (ashr<mode>3): Likewise.
619 (<shift_insn><mode>3): Likewise.
620 (avx2_<code><mode>3): Likewise.
621 (*avx2_<code><mode>3): Likewise.
622 (*andnot<mode>3): Likewise.
623 (*<code><mode>3) <any_logic>: Likewise.
624 (abs<mode>2): Likewise.
625 (avx2_permvar<mode>): Likewise.
626 (avx2_perm<mode>_1): Likewise.
627 (*avx_vpermilp<mode>): Likewise.
628 (avx_vpermilvar<mode>3): Likewise.
629 (avx2_ashrv<mode>): Likewise.
630 (avx2_<shift_insn>v<mode>): Likewise.
631 * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
633 * doc/rtl.texi: Document XImode.
635 2013-08-21 Jeff Law <law@redhat.com>
637 * tree-flow.h (register_jump_thread): Pass vector of edges
638 instead of each important edge.
639 * tree-ssa-threadedge.c (thread_across_edge): Build the jump
640 thread path into a vector and pass that to register_jump_thread.
641 * tree-ssa-threadupdate.c (register_jump_thread): Conver the
642 passed in edge vector to the current 3-edge form.
645 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
647 * dce.c (fini_dce): Call df_analyze again just in case
648 delete_unmarked_insns removed anything.
650 2013-08-21 Joern Rennecke <joern.rennecke@embecosm.com>
652 * reload.h (struct reg_equivs): Rename to ..
653 (struct reg_equivs_s): .. this.
655 2013-08-20 Martin Liska <marxin.liska@gmail.com>
657 * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
659 2013-08-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
661 * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
663 2013-08-21 Jeff Law <law@redhat.com>
665 * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
666 simplify assignments too. If the RHS collapses to a singleton
667 range, then return the value for the range.
669 2013-08-21 Kirill Yukhin <kirill.yukhin@intel.com>
671 * config/i386/sse.md (V16): Rename to...
673 (mov<mode>): Update iterator name.
674 (*mov<mode>_internal): Ditto.
675 (push<mode>1): Ditto.
676 (movmisalign<mode>): Ditto.
678 2013-08-20 Jan Hubicka <jh@suse.cz>
681 * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
682 entry for direct edges.
683 (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
685 2013-08-20 David Malcolm <dmalcolm@redhat.com>
687 Revert my last two changes, r201865 and r201864:
690 2013-08-20 David Malcolm <dmalcolm@redhat.com>
692 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
693 instances can own GC refs.
695 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
696 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
697 (gcc::context::gt_pch_nx): Likewise.
698 (gcc::context::gt_pch_nx): Likewise.
699 * ggc.h (gt_ggc_mx <T>): New.
700 (gt_pch_nx_with_op <T>): New.
701 (gt_pch_nx <T>): New.
702 * passes.c (opt_pass::gt_ggc_mx): New.
703 (opt_pass::gt_pch_nx): New.
704 (opt_pass::gt_pch_nx_with_op): New.
705 (pass_manager::gt_ggc_mx): New.
706 (pass_manager::gt_pch_nx): New.
707 (pass_manager::gt_pch_nx_with_op): New.
708 (pass_manager::operator new): Use
709 ggc_internal_cleared_alloc_stat rather than xcalloc.
710 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
711 (pass_manager::gt_ggc_mx): New.
712 (pass_manager::gt_pch_nx): New.
713 (pass_manager::gt_pch_nx_with_op): New.
714 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
715 (opt_pass::operator new): New.
716 (opt_pass::gt_ggc_mx): New.
717 (opt_pass::gt_pch_nx): New.
718 (opt_pass::gt_pch_nx_with_op): New.
721 2013-08-20 David Malcolm <dmalcolm@redhat.com>
723 * Makefile.in (GTFILES): Add context.h.
724 * context.c (gcc::context::operator new): New.
725 (gcc::context::gt_ggc_mx): New.
726 (gcc::context::gt_pch_nx): New.
727 (gcc::context::gt_pch_nx): New.
728 * context.h (gcc::context): Add GTY((user)) marking.
729 (gcc::context::operator new): New.
730 (gcc::context::gt_ggc_mx): New.
731 (gcc::context::gt_pch_nx): New.
732 (gcc::context::gt_pch_nx): New.
733 (g): Add GTY marking.
734 (gt_ggc_mx (gcc::context *)): New.
735 (gt_pch_nx (gcc::context *)): New.
736 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
738 * gengtype.c (open_base_files) <ifiles>: Add context.h.
740 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
742 * dce.c (fini_dce): Call df_analyze again just in case
743 delete_unmarked_insns removed anything.
745 2013-08-20 Teresa Johnson <tejohnson@google.com>
747 PR rtl-optimizations/57451
748 * final.c (reemit_insn_block_notes): Prevent lexical blocks
749 from crossing split section boundaries.
751 2013-08-20 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
753 * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
754 * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
755 with MULTLIB_DEFAULTS.
757 2013-08-20 Nick Clifton <nickc@redhat.com>
759 * target.def (narrow_volatile_bitfield): Note that the default
760 value is false, not !TARGET_STRICT_ALIGN.
761 * doc/tm.texi: Regenerate.
763 2013-08-20 Pavel Chupin <pavel.v.chupin@intel.com>
765 Fix LIB_SPEC for systems without libpthread.
767 * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
768 * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
770 * config/i386/linux-common.h: Likewise.
771 * config/mips/linux-common.h: Likewise.
773 2013-08-20 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
775 * tree-ssa-ccp.c (get_default_value): Remove redundant condition
778 2013-08-20 David Malcolm <dmalcolm@redhat.com>
780 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
781 instances can own GC refs.
783 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
784 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
785 (gcc::context::gt_pch_nx): Likewise.
786 (gcc::context::gt_pch_nx): Likewise.
787 * ggc.h (gt_ggc_mx <T>): New.
788 (gt_pch_nx_with_op <T>): New.
789 (gt_pch_nx <T>): New.
790 * passes.c (opt_pass::gt_ggc_mx): New.
791 (opt_pass::gt_pch_nx): New.
792 (opt_pass::gt_pch_nx_with_op): New.
793 (pass_manager::gt_ggc_mx): New.
794 (pass_manager::gt_pch_nx): New.
795 (pass_manager::gt_pch_nx_with_op): New.
796 (pass_manager::operator new): Use
797 ggc_internal_cleared_alloc_stat rather than xcalloc.
798 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
799 (pass_manager::gt_ggc_mx): New.
800 (pass_manager::gt_pch_nx): New.
801 (pass_manager::gt_pch_nx_with_op): New.
802 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
803 (opt_pass::operator new): New.
804 (opt_pass::gt_ggc_mx): New.
805 (opt_pass::gt_pch_nx): New.
806 (opt_pass::gt_pch_nx_with_op): New.
808 2013-08-20 David Malcolm <dmalcolm@redhat.com>
810 * Makefile.in (GTFILES): Add context.h.
811 * context.c (gcc::context::operator new): New.
812 (gcc::context::gt_ggc_mx): New.
813 (gcc::context::gt_pch_nx): New.
814 (gcc::context::gt_pch_nx): New.
815 * context.h (gcc::context): Add GTY((user)) marking.
816 (gcc::context::operator new): New.
817 (gcc::context::gt_ggc_mx): New.
818 (gcc::context::gt_pch_nx): New.
819 (gcc::context::gt_pch_nx): New.
820 (g): Add GTY marking.
821 (gt_ggc_mx (gcc::context *)): New.
822 (gt_pch_nx (gcc::context *)): New.
823 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
825 * gengtype.c (open_base_files) <ifiles>: Add context.h.
827 2013-08-20 Alan Modra <amodra@gmail.com>
830 * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
831 (rs6000_emit_epilogue): Likewise.
833 2013-08-19 Dehao Chen <dehao@google.com>
835 * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
837 2013-08-19 Peter Bergner <bergner@vnet.ibm.com>
838 Jakub Jelinek <jakub@redhat.com>
840 * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
841 (BUILT_IN_FABSD64): Likewise.
842 (BUILT_IN_FABSD128): Likewise.
843 * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
844 (fold_builtin_1): Likewise.
845 * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
846 destination and source operands.
847 (*abstd2_fpr): Likewise.
848 (*nabstd2_fpr): Likewise.
850 2013-08-19 Richard Sandiford <rdsandiford@googlemail.com>
852 * config/mips/mips.c (mips_adjust_insn_length): Add checks for
855 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
857 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
860 2013-08-19 Alexander Ivchenko <alexander.ivchenko@intel.com>
862 * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
863 * builtins.c (default_libc_has_function): New.
864 (gnu_libc_has_function): Ditto.
865 (no_c99_libc_has_function): Ditto.
866 (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
867 instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
868 (fold_builtin_sincos): Likewise.
869 (fold_builtin_cexp): Likewise.
870 * builtins.def (DEF_C94_BUILTIN): Likewise.
871 (DEF_C99_BUILTIN): Likewise.
872 (DEF_C99_C90RES_BUILTIN): Likewise.
873 (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
874 definitions to using this define.
875 * config/darwin-protos.h (darwin_libc_has_function): New.
876 * config/darwin.c (darwin_libc_has_function): Ditto.
877 * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
878 TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
879 * config/darwin.h: Ditto.
880 * config/elfos.h: Ditto.
881 * config/freebsd.h: Ditto.
882 * config/i386/cygming.h: Ditto.
883 * config/i386/djgpp.h: Ditto.
884 * config/i386/i386-interix.h: Ditto.
885 * config/microblaze/microblaze.h: Ditto.
886 * config/mmix/mmix.h: Ditto.
887 * config/gnu-user.h: Ditto.
888 * config/ia64/hpux.h: Ditto.
889 * config/pa/pa-hpux.h: Ditto.
890 * config/pdp11/pdp11.h: Ditto.
891 * config/picochip/picochip.h: Ditto.
892 * config/linux.h: Ditto.
893 * config/netbsd.h: Ditto.
894 * config/openbsd.h: Ditto.
895 * config/rs6000/aix43.h: Ditto.
896 * config/rs6000/aix51.h: Ditto.
897 * config/rs6000/aix52.h: Ditto.
898 * config/rs6000/aix53.h: Ditto.
899 * config/rs6000/aix61.h: Ditto.
900 * config/rs6000/darwin.h: Ditto.
901 * config/rs6000/linux.h: Ditto.
902 * config/rs6000/linux64.h: Ditto.
903 * config/s390/tpf.h: Ditto.
904 * config/sol2-10.h: Ditto.
905 * config/sol2.h: Ditto.
906 * config/vms/vms.h: Ditto.
907 * config/vxworks.h: Ditto.
908 * config/linux-android.c (linux_android_libc_has_function):
909 New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
910 * config/linux-protos.h (linux_android_libc_has_function):
912 * config/i386/i386.c (ix86_libc_has_function): New.
913 * config/i386/i386-protos.h
914 (ix86_libc_has_function): New declaration.
915 * config/i386/i386.md
916 ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
917 ("isinf<mode>2): Likewise.
918 * convert.c (convert_to_integer): Using new target hook
919 TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
920 TARGET_C99_FUNCTIONS.
921 * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
922 * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
923 * coretypes.h (function_class): New enum for different
924 classes of functions.
925 * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
926 * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
927 (TARGET_HAS_SINCOS): Likewise.
928 (TARGET_LIBC_HAS_FUNCTION): New.
929 * doc/tm.texi: Regenerated.
930 * targhooks.h (default_libc_has_function): New declaration.
931 (no_c99_libc_has_function): Ditto.
932 (gnu_libc_has_function): Ditto.
933 * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
934 and TARGET_HAS_SINCOS.
936 2013-08-18 Jan Hubicka <jh@suse.cz>
938 * Makeifle-in (ipa-devirt.o): New.
939 (GTFILES): Add ipa-utils.h and ipa-devirt.c
940 * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
941 (analyze_functions): Look into possible targets of polymorphic call.
942 * dumpfile.c (dump_files): Add type-inheritance dump.
943 * dumpfile.h (TDI_inheritance): New.
944 * ipa-devirt.c: New file.
945 * ipa-utils.h (odr_type_d): Forward declare.
946 (odr_type): New type.
947 (build_type_inheritance_graph): Declare.
948 (possible_polymorphic_call_targets): Declare and introduce inline
949 variant when only edge is pased.
950 (dump_possible_polymorphic_call_targets): Likewise.
951 * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
952 * tree.c (type_in_anonymous_namespace_p): Break out from ...
953 (types_same_for_odr): ... here.
954 * tree.h (type_in_anonymous_namespace_p): Declare.
956 2013-08-18 Jakub Jelinek <jakub@redhat.com>
958 PR tree-optimization/58006
959 * tree-parloops.c (take_address_of): Don't ICE if get_name
961 (eliminate_local_variables_stmt): Remove clobber stmts.
963 2013-08-18 Eric Botcazou <ebotcazou@adacore.com>
965 * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
966 error message is issued for an alias to undefined symbol.
968 2013-08-18 Jan Hubicka <jh@suse.cz>
970 * cgraph.c (cgraph_create_indirect_edge): Discover
971 polymorphic calls and record basic info into indirect_info.
972 * gimple-fold.c (gimple_fold_call): When doing BINFO based
973 devirtualization, ignore objc function calls.
974 * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
975 call with no parm index info.
976 * ipa-prop.c (ipa_analyze_call_uses): Likewise.
977 * tree.c (virtual_method_call_p): New function.
978 * tree.h (virtual_method_call_p): Declare.
980 2013-08-16 Jan Hubicka <jh@suse.cz>
983 * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
985 2013-08-16 David Edelsohn <dje.gcc@gmail.com>
987 * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
990 2013-08-16 David Malcolm <dmalcolm@redhat.com>
992 * gengtype.c (type_for_name): Add special-case support for
993 locating types within the "gcc::" namespace.
994 (open_base_files): Emit a "using namespace gcc" directive.
996 2013-08-16 Michael Meissner <meissner@linux.vnet.ibm.com>
999 * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1000 memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1002 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1003 array instead of each individual operand as a separate argument.
1004 (emit_fusion_gpr_load): Likewise.
1005 (expand_fusion_gpr_load): Add new function declaration.
1007 * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1008 signature to have the operands passed as an array, instead of as
1009 separate arguments. Allow ZERO_EXTEND to be in the memory
1010 address, and also SIGN_EXTEND if -mpower8-fusion-sign. Do not
1011 depend on the register live/dead flags when peepholes are run.
1012 (expand_fusion_gpr_load): New function to be called from the
1013 peephole2 pass, to change the register that addis sets to be the
1015 (emit_fusion_gpr_load): Change the calling signature to have the
1016 operands passed as an array, instead of as separate arguments.
1017 Allow ZERO_EXTEND to be in the memory address, and also
1018 SIGN_EXTEND if -mpower8-fusion-sign.
1020 * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1022 (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1023 adjust the register addis loads up in the peephole2 pass. Do not
1024 depend on the register live/dead state when the peephole pass is done.
1026 2013-08-16 David Malcolm <dmalcolm@redhat.com>
1028 * gengtype.c (create_user_defined_type): Ensure that the kind
1029 is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1030 declaration is seen before the GTY((user)) marking.
1032 2013-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
1035 * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1037 2013-08-16 Jan Hubicka <jh@suse.cz>
1039 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1040 arugment expected_type.
1041 (gimple_fold_call): Use it.
1042 * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1043 * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1044 * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1045 (try_make_edge_direct_virtual_call): Likewise.
1046 * tree.c (obj_type_ref_class): New.
1047 * tree.h (obj_type_ref_class): Use it.
1049 2013-08-16 Gabriel Dos Reis <gdr@integrable-solutions.net>
1051 * sched-vis.c (rtl_slim_pp_initialized): Remove.
1052 (rtl_slim_pp): Likewise.
1053 (init_rtl_slim_pretty_print): Likewise.
1054 (dump_value_slim): Don't call it. Use local pretty printer.
1055 (dump_insn_slim): Likewise.
1056 (dump_rtl_slim): Likewise.
1057 (str_pattern_slim): Likewise.
1058 * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1061 2013-08-16 Jakub Jelinek <jakub@redhat.com>
1063 PR tree-optimization/58164
1064 * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1065 walk gimple_goto_dest of GIMPLE_GOTO.
1067 PR tree-optimization/58165
1068 * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1069 bi_call must be the last stmt in a bb, don't split_block, instead
1070 use fallthru edge from it and give up if there is none.
1071 Release conds vector when returning early.
1073 2013-08-14 Xinliang David Li <davidxl@google.com>
1075 * config/i386/i386.c (ix86_option_override_internal):
1076 Remove unused variable and field.
1078 2013-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1081 * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1082 * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1083 * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1084 and Linux, correct BLKmode alignment when 128-bit alignment is
1085 required and compatibility flag is not set.
1086 (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1087 for zero-size arguments when compatibility flag is not set.
1089 2013-08-14 Jakub Jelinek <jakub@redhat.com>
1091 PR tree-optimization/58145
1092 * tree-sra.c (build_ref_for_offset): If prev_base has
1093 TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1095 2013-08-14 Xinliang David Li <davidxl@google.com>
1097 * config/i386/i386.c (ix86_option_override_internal):
1098 Fix uninitialized variable error.
1100 2013-08-14 Xinliang David Li <davidxl@google.com>
1102 * config/i386/i386.opt: Define two new options.
1103 * config/i386/x86-tune.def: Add arch selector field in macros.
1104 * config/i386/i386.h: Adjust macro definition.
1105 * config/i386/i386.c (ix86_option_override_internal):
1107 (parse_mtune_ctrl_str): New function.
1108 (set_ix86_tune_features): New function.
1109 (ix86_function_specific_restore): Call the new helper function.
1111 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
1113 PR rtl-optimization/57662
1114 * sel-sched.c (code_motion_process_successors): When the current insn
1115 is removed after the recursive traversal, break from the loop.
1116 Add comments and debug printouts.
1118 2013-08-14 Jakub Jelinek <jakub@redhat.com>
1119 Alexandre Oliva <aoliva@redhat.com>
1122 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1123 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1124 there also UNSPEC_PLTOFF.
1126 2013-08-14 Marek Polacek <polacek@redhat.com>
1128 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1129 NUM_CONDITIONS bit positions.
1131 2013-08-13 Cary Coutant <ccoutant@google.com>
1133 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1134 (attr_checksum): Hash vector contents instead of pointer.
1135 (attr_checksum_ordered): Likewise.
1137 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
1139 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1140 when Pmode != word_mode. Add length_address attribute.
1141 (sse3_monitor_<mode>): Merge from sse3_monitor and
1142 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
1143 Pmode != word_mode. Update insn length attribute.
1144 * config/i386/i386.c (ix86_option_override_internal): Update
1145 ix86_gen_monitor selection for merged sse3_monitor insn.
1147 2013-08-13 Julian Brown <julian@codesourcery.com>
1149 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1150 perform invalid legitimization on greater-than-word-size modes for
1153 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
1155 * ira.c (setup_class_translate_array): Use aclass instead of cl
1156 for classes not fully covered by allocno classes.
1158 2013-08-13 Jakub Jelinek <jakub@redhat.com>
1160 PR tree-optimization/57661
1161 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1162 * tree-inline.c (tree_function_versioning): Initialize it.
1163 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1164 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1165 that is not being copied.
1168 * asan.c (instrument_strlen_call): Fix typo in comment.
1169 Use char * type even for the lhs of POINTER_PLUS_EXPR.
1171 2013-08-13 Steve Ellcey <sellcey@mips.com>
1173 * config/mips/mips.md (prefetch): Use lw instead of ld on
1174 loongson in 32bit mode.
1176 2013-08-13 Nick Clifton <nickc@redhat.com>
1178 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1180 2013-08-13 Jan Hubicka <jh@suse.cz>
1182 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1183 introduced edge; fix typo in sanity check.
1184 (cgraph_resolve_speculation): Export; improve diagnostic.
1185 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1186 speculation at type mismatch.
1187 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1188 (cgraph_resolve_speculation): Declare.
1189 (symtab_can_be_discarded): New function.
1190 * value-prof.c (gimple_ic_transform): Remove actual transform code.
1191 * ipa-inline-transform.c (speculation_removed): New global var.
1192 (clone_inlined_nodes): See if speculation can be removed.
1193 (inline_call): If speculations was removed, we growths may not match.
1194 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1195 (speculation_useful_p): New function.
1196 (resolve_noninline_speculation): New function.
1197 (inline_small_functions): Resolve useless speculations.
1198 * ipa-inline.h (speculation_useful_p): Declare
1199 * ipa.c (can_replace_by_local_alias): Simplify.
1200 (ipa_profile): Produce speculative calls in non-lto, too;
1201 add simple cost model; produce local aliases.
1203 2013-08-13 David Malcolm <dmalcolm@redhat.com>
1205 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1208 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
1210 * config/i386/i386.c (ix86_function_versions): Use error + inform.
1212 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
1214 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1215 iterator instead of X87MODEF.
1217 2013-08-12 Perez Read <netfirewall@gmail.com>
1220 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
1221 operand 0 for intel asm alternative.
1222 (*movabs<mode>_2): Ditto for operand 1.
1224 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
1226 * config/aarch64/arm_none.h
1227 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1229 2013-08-12 Nick Clifton <nickc@redhat.com>
1231 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1233 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
1235 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1236 expand for QI/HImode operand to produce more effictive code for
1237 unsigned char(short) --> float(double) conversion.
1239 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
1241 * doc/invoke.texi: Mention that -ftls-model does not force the final
1244 2013-08-12 Marek Polacek <polacek@redhat.com>
1245 Marc Glisse <marc.glisse@inria.fr>
1247 PR tree-optimization/57980
1248 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1249 when creating -1 constant.
1251 2013-08-10 Jan Hubicka <jh@suse.cz>
1253 Workaround binutils PR14342.
1254 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1255 (gimple_init_edge_profiler): Likewise.
1256 (gimple_gen_ic_func_profiler): Likewise.
1258 2013-08-09 Jan Hubicka <jh@suse.cz>
1260 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1262 2013-08-09 Xinliang David Li <davidxl@google.com>
1264 * config/i386/stringop.def: New file.
1265 * config/i386/stringop.opt: New file.
1266 * config/i386/i386-opts.h: Include stringopt.def.
1267 * config/i386/i386.opt: Include stringopt.opt.
1268 * config/i386/i386.c (ix86_option_override_internal):
1269 Override default size based stringop inline strategies with options.
1270 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1273 2013-08-09 Jan Hubicka <jh@suse.cz>
1275 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1277 2013-08-09 Jan Hubicka <jh@suse.cz>
1279 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1281 (dump_cgraph_node): Dump profile-id.
1282 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1283 and common_target_probability.
1284 * lto-cgraph.c (lto_output_edge): Stream common targets.
1285 (lto_output_node): Stream profile ids.
1286 (input_node): Stream profile ids.
1287 (input_edge): Stream common targets.
1288 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1289 * ipa.c: Include value-prof.h
1290 (ipa_profile_generate_summary): Turn indirect call statement histograms
1291 into common targets.
1292 (ipa_profile): Turn common targets into speculative edges.
1294 2013-08-09 Jan Hubicka <jh@suse.cz>
1296 * cgraph.h (cgraph_node): Add profile_id.
1297 * value-prof.c (cgraph_node_map): Turn into pointer_map.
1298 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1299 (del_node_map): Update.
1300 (find_func_by_funcdef_no): Replace by ...
1301 (find_func_by_profile_id): ... this one.
1302 (gimple_ic_transform): Do not remove useful histograms when
1303 speculation is not done; dump info when indirect call removal
1305 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1306 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1307 (__gcov_indirect_call_profiler_v2): .. this one.
1308 * profile.h (init_node_map): Update.
1309 * coverage.c (coverage_compute_profile_id): New function.
1310 * coverage.h (coverage_compute_profile_id): Declare.
1311 * tree-profile.c (init_ic_make_global_vars): Make
1312 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
1313 (gimple_init_edge_profiler): Update prototype of
1314 __gcov_indirect_call_profiler.
1315 (gimple_gen_ic_func_profiler): Simplify.
1316 (tree_profiling): Use init_node_map
1318 2013-08-09 Jan Hubicka <jh@suse.cz>
1320 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1321 non-speculative refs.
1322 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1323 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1324 (cgraph_set_call_stmt): Likewise.
1325 (cgraph_create_edge_1): Fix release checking compilatoin;
1327 (cgraph_free_edge): Free indirect info.
1328 (cgraph_turn_edge_to_speculative): New function.
1329 (cgraph_speculative_call_info): New function.
1330 (cgraph_make_edge_direct): Return direct edge; handle speculation.
1331 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1332 (dump_cgraph_node): Dump speculation.
1333 (verify_edge_count_and_frequency): Accept speculative edges.
1334 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1335 (verify_cgraph_node): Handle speculation.
1336 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1337 (cgraph_set_call_stmt): Update prototype.
1338 (cgraph_make_edge_direct): Update prototype.
1339 (cgraph_speculative_call_info): Declare.
1340 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1341 to change; update call of ipa_find_references.
1342 * ipa-ref.c (ipa_record_reference): Fix return value; clear
1343 lto_stmt_uid and speculative flags.
1344 (ipa_dump_references): Dump speculation.
1345 (ipa_clone_references): Clone speculative flag.
1346 (ipa_clone_referring): Likewise.
1347 (ipa_clone_ref): New function.
1348 (ipa_find_reference): Look into lto_stmt_uids
1349 (ipa_clear_stmts_in_references): Do not clear speculative calls.
1350 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1351 (ipa_find_reference): Update declaration.
1352 (ipa_clone_ref): Declare.
1353 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1354 stream speculative flag.
1355 (lto_output_ref): Stream statements uids and speculation.
1356 (input_ref): Likewise.
1357 (input_edge): Stream speuclation.
1358 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1359 (cgraph_set_call_stmt_including_clones): Handle speculation.
1360 * ipa-inline.c (heap_edge_removal_hook): New function.
1361 (inline_small_functions): Register it.
1362 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1363 also initialize refs.
1364 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1366 (try_make_edge_direct_simple_call): Likewise.
1367 (try_make_edge_direct_simple_call): Likewise.
1368 (update_indirect_edges_after_inlining): Likewise.
1369 (remove_described_reference): Look proper lto_stmt_uid.
1370 (propagate_controlled_uses): Likewise.
1371 (propagate_controlled_uses): Liekwise.
1372 * tree-inline.c (copy_bb): Copy speculative edges.
1373 (redirect_all_calls): New function.
1374 (copy_cfg_body): Do redirection after loop info is updated.
1375 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1377 2013-08-09 Jan Hubicka <jh@suse.cz>
1379 * lto-streamer-out.c (output_function): Renumber PHIs.
1380 * lto-streamer-in.c (input_function): Likewise.
1382 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
1384 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1385 (get_lane_unsigned): Likewise.
1386 (dup_lane_scalar): Likewise.
1387 (get_lane): enable for VALL.
1388 * config/aarch64/aarch64-simd.md
1389 (aarch64_dup_lane_scalar<mode>): Remove.
1390 (aarch64_get_lane_signed<mode>): Likewise.
1391 (aarch64_get_lane_unsigned<mode>): Likewise.
1392 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1393 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1394 (aarch64_get_lane<mode>): Enable for all vector modes.
1395 (aarch64_get_lanedi): Remove misleading constraints.
1396 * config/aarch64/arm_neon.h
1397 (__aarch64_vget_lane_any): Define.
1398 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1399 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1400 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1401 * config/aarch64/iterators.md (VDQQH): New.
1405 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
1407 * configure.ac: Add GAS check for LEON instructions on SPARC.
1408 * configure: Regenerate.
1409 * config.in: Likewise.
1410 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1412 * config/sparc/sparc.opt (LEON, LEON3): New masks.
1413 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1415 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1416 (AS_LEON_FLAG): New macro.
1417 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1418 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1419 Deal with LEON and LEON3 for the memory model.
1420 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1421 (atomic_compare_and_swap<mode>_1): Likewise.
1422 (*atomic_compare_and_swap<mode>_1): Likewise.
1424 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1426 * config/arm/neon.md (vcond): Fix floating-point vector
1427 comparisons against 0.
1429 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
1431 * lra-constraints.c (emit_spill_move): Remove assert.
1432 (process_alt_operands): Add more debugging
1433 output. Increase reject for spilling into memory. Decrease
1434 reject for reloading scratch.
1435 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1437 2013-08-08 Steve Ellcey <sellcey@mips.com>
1439 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1440 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1441 micromips incompatible. Add nan2008.
1442 (MULTILIB_DIRNAMES): Add nan2008.
1443 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1444 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1445 and micromips incompatible. Add nan2008.
1446 (MULTILIB_DIRNAMES): Add nan2008.
1447 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1449 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
1451 PR rtl-optimization/58079
1452 * combine.c (combine_simplify_rtx): Avoid using SUBST if
1453 simplify_comparison has widened a comparison with an integer.
1455 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1457 * config/arm/neon.md (movmisalign<mode>): Disable when we
1458 don't allow unaligned accesses.
1459 (*movmisalign<mode>_neon_store): Likewise.
1460 (*movmisalign<mode>_neon_load): Likewise.
1461 (*movmisalign<mode>_neon_store): Likewise.
1462 (*movmisalign<mode>_neon_load): Likewise.
1464 2013-08-08 Jan Hubicka <jh@suse.cz>
1466 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1467 (make_pass_rebuild_cgraph_edges): Also clear references.
1468 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1469 * ipa-inline-transform.c (inline_transform): Remove all references
1471 * cgraphunit.c (expand_function): Remove all references after
1473 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1474 (ipa_find_reference): Rewrite to iterator.
1475 (remove_stmt_references): Likewise.
1476 (ipa_clear_stmts_in_references): New function.
1477 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1478 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1480 * ipa-split.c (split_function): Remove references in split function.
1482 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
1485 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1486 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1488 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
1491 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1492 suggested mode for the assignment isn't compatible with the
1495 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1498 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1500 2013-08-07 Xinliang David Li <davidxl@google.com>
1502 * config/i386/i386.opt: New option -mtune-ctrl=.
1503 * config/i386/x86-tune.def: New file.
1504 * config/i386/i386.h: include x86-tune.def.
1505 * config/i386/i386.c (ix86_option_override_internal):
1506 Parsing -mtune-ctrl= option and set tune features.
1508 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
1511 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1512 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1515 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
1517 * rtl.h (update_alignments): Declare.
1518 * final.c (grow_label_align): New function extracted from...
1519 (shorten_branches): ...here. Call it.
1520 (update_alignments): New function.
1521 * reorg.c (sibling_labels): New variable.
1522 (get_label_before): Add SIBLING parameter. If it is non-zero, push
1523 the new label along with it onto the sibling_labels vector.
1524 (fill_simple_delay_slots): Adjust call to get_label_before.
1525 (fill_slots_from_thread): Likewise.
1526 (relax_delay_slots): Likewise.
1527 (make_return_insns): Likewise.
1528 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1530 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
1532 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1533 document its semantics.
1534 (diagnostic_report_diagnostic): Adjust accordingly.
1536 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1538 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1539 (sparc_option_override): ...and port to new C++ pass API.
1540 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1542 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
1544 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1546 2013-08-06 Caroline Tice <cmtice@google.com>
1548 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1549 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1550 * tree-pass.h: Add pass_vtable_verify.
1551 * varasm.c (assemble_variable): Add code to properly set the comdat
1552 section and name for the .vtable_map_vars section.
1553 (assemble_vtyv_preinit_initializer): New function.
1554 (default_sectin_type_flags): Make sure .vtable_map_vars section has
1556 * output.h: Add function decl for assemble_vtv_preinit_initializer.
1557 * vtable-verify.c: New file.
1558 * vtable-verify.h: New file.
1559 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1560 initialiation levels.
1561 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1562 * passes.def: Insert pass_vtable_verify.
1563 * aclocal.m4: Reorder includes.
1564 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
1565 -fvtv-counts options.
1566 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1567 as appropriate, if -fvtable-verify=... is used.
1568 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1569 -fvtable-verify=... is used.
1570 * Makefile.in (OBJS): Add vtable-verify.o to list.
1571 (vtable-verify.o): Add new build rule.
1572 (GTFILES): Add vtable-verify.c to list.
1573 * common.opt (fvtable-verify=): New flag.
1574 (vtv_priority): Values for fvtable-verify= flag.
1575 (fvtv-counts): New flag.
1576 (fvtv-debug): New flag.
1577 * tree.h (save_vtable_map_decl): New extern function decl.
1579 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1581 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1582 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1583 (pass_data_rl78_devirt): ...new pass_data instance and...
1584 (make_pass_rl78_devirt): ...new function.
1585 (rl78_asm_file_start): Port pass registration to new C++ API.
1587 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1589 * coretypes.h (rtl_opt_pass): Add.
1590 (gcc::context): Add.
1591 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1592 (epiphany_init): Port to new C++ pass API.
1593 (epiphany_optimize_mode_switching): Likewise.
1594 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
1595 (pass_manager::get_pass_mode_switching): New.
1596 (pass_manager::get_pass_peephole2): New.
1597 * mode-switching.c (pass_mode_switching): Add clone method.
1598 * recog.c (pass_peephole2): Add clone method.
1599 (pass_split_all_insns): Add clone method.
1601 2013-08-06 David Malcolm <dmalcolm@redhat.com>
1603 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
1604 (mips_option_override): ...here, porting to new C++ API for passes.
1606 2013-08-06 Jan Hubicka <jh@suse.cz>
1608 * cgraph.c (cgraph_get_body): New function based on lto.c
1610 * cgraph.h (cgraph_get_body): Declare.
1611 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
1613 * cgraphunit.c (expand_function): Get body prior expanding.
1614 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
1615 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
1617 * passes.c (do_per_function_toporder): Get body.
1618 * tree-inline.c (expand_call_inline): Get body prior inlining it.
1619 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
1621 2013-08-06 Martin Jambor <mjambor@suse.cz>
1624 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
1625 is not re-finalized. Rename second parameter to no_collect.
1627 2013-08-06 Martin Jambor <mjambor@suse.cz>
1630 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
1631 MEM_REF has proper alignment information.
1633 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
1636 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
1638 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
1639 rtx (*) (rtx, ...) with insn_gen_fn.
1640 * genoutput.c (output_insn_data): Cast gen_? function pointers to
1641 insn_gen_fn::stored_funcptr. Add initializer braces.
1643 2013-08-05 David Malcolm <dmalcolm@redhat.com>
1645 Rewrite how instances of passes are cloned to remove assumptions
1646 about their sizes (thus allowing pass subclasses to have
1647 additional data fields, albeit non-GC-managed ones at this point).
1649 * passes.c (make_pass_instance): Now that passes have clone
1650 methods, rewrite this function to eliminate XNEW and memcpy
1651 calls that used hardcoded sizes. Since this function no longer
1652 creates pass instances, rename it to...
1653 (add_pass_instance): ...this. Document the old way that passes were
1654 numbered and flagged, and rework this function to continue using it.
1655 (next_pass_1): Add an initial_pass argument for use by
1657 (position_pass): When adding multiple instances of a pass, use
1658 the pass's clone method, rather than relying on the XNEW/memcpy
1659 within the former make_pass_instance (now add_pass_instance).
1660 (pass_manager::pass_manager): When invoking next_pass_1, also supply
1661 the initial instance of the current pass within the pass manager.
1663 2013-08-05 David Malcolm <dmalcolm@redhat.com>
1665 This is the automated part of the conversion of passes from C
1666 structs to C++ classes.
1668 Patch autogenerated by refactor_passes.py from
1669 https://github.com/davidmalcolm/gcc-refactoring-scripts
1670 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
1672 * asan.c (pass_asan): Convert from a global struct to a subclass of
1673 gimple_opt_pass along with...
1674 (pass_data_asan): ...new pass_data instance and...
1675 (make_pass_asan): ...new function.
1676 (pass_asan_O0): Convert from a global struct to a subclass of
1677 gimple_opt_pass along with...
1678 (pass_data_asan_O0): ...new pass_data instance and...
1679 (make_pass_asan_O0): ...new function.
1680 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
1681 subclass of rtl_opt_pass along with...
1682 (pass_data_inc_dec): ...new pass_data instance and...
1683 (make_pass_inc_dec): ...new function.
1684 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
1685 a subclass of rtl_opt_pass along with...
1686 (pass_data_reorder_blocks): ...new pass_data instance and...
1687 (make_pass_reorder_blocks): ...new function.
1688 (pass_duplicate_computed_gotos): Convert from a global struct to a
1689 subclass of rtl_opt_pass along with...
1690 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
1691 (make_pass_duplicate_computed_gotos): ...new function.
1692 (pass_partition_blocks): Convert from a global struct to a subclass of
1693 rtl_opt_pass along with...
1694 (pass_data_partition_blocks): ...new pass_data instance and...
1695 (make_pass_partition_blocks): ...new function.
1696 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
1697 struct to a subclass of rtl_opt_pass along with...
1698 (pass_data_branch_target_load_optimize1): ...new pass_data instance
1700 (make_pass_branch_target_load_optimize1): ...new function.
1701 (pass_branch_target_load_optimize2): Convert from a global struct to a
1702 subclass of rtl_opt_pass along with...
1703 (pass_data_branch_target_load_optimize2): ...new pass_data instance
1705 (make_pass_branch_target_load_optimize2): ...new function.
1706 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
1707 of rtl_opt_pass along with...
1708 (pass_data_jump): ...new pass_data instance and...
1709 (make_pass_jump): ...new function.
1710 (pass_jump2): Convert from a global struct to a subclass of
1711 rtl_opt_pass along with...
1712 (pass_data_jump2): ...new pass_data instance and...
1713 (make_pass_jump2): ...new function.
1714 * cfgexpand.c (pass_expand): Convert from a global struct to a
1715 subclass of rtl_opt_pass along with...
1716 (pass_data_expand): ...new pass_data instance and...
1717 (make_pass_expand): ...new function.
1718 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
1719 of rtl_opt_pass along with...
1720 (pass_data_free_cfg): ...new pass_data instance and...
1721 (make_pass_free_cfg): ...new function.
1722 (pass_into_cfg_layout_mode): Convert from a global struct to a
1723 subclass of rtl_opt_pass along with...
1724 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
1725 (make_pass_into_cfg_layout_mode): ...new function.
1726 (pass_outof_cfg_layout_mode): Convert from a global struct to a
1727 subclass of rtl_opt_pass along with...
1728 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
1729 (make_pass_outof_cfg_layout_mode): ...new function.
1730 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
1731 struct to a subclass of gimple_opt_pass along with...
1732 (pass_data_build_cgraph_edges): ...new pass_data instance and...
1733 (make_pass_build_cgraph_edges): ...new function.
1734 (pass_rebuild_cgraph_edges): Convert from a global struct to a
1735 subclass of gimple_opt_pass along with...
1736 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
1737 (make_pass_rebuild_cgraph_edges): ...new function.
1738 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
1739 subclass of gimple_opt_pass along with...
1740 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
1742 (make_pass_remove_cgraph_callee_edges): ...new function.
1743 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
1744 struct to a subclass of rtl_opt_pass along with...
1745 (pass_data_stack_adjustments): ...new pass_data instance and...
1746 (make_pass_stack_adjustments): ...new function.
1747 * combine.c (pass_combine): Convert from a global struct to a subclass
1748 of rtl_opt_pass along with...
1749 (pass_data_combine): ...new pass_data instance and...
1750 (make_pass_combine): ...new function.
1751 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
1752 global struct to a subclass of rtl_opt_pass along with...
1753 (pass_data_compare_elim_after_reload): ...new pass_data instance
1755 (make_pass_compare_elim_after_reload): ...new function.
1756 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
1757 of rtl_opt_pass along with...
1758 (pass_data_rtl_cprop): ...new pass_data instance and...
1759 (make_pass_rtl_cprop): ...new function.
1760 * cse.c (pass_cse): Convert from a global struct to a subclass of
1761 rtl_opt_pass along with...
1762 (pass_data_cse): ...new pass_data instance and...
1763 (make_pass_cse): ...new function.
1764 (pass_cse2): Convert from a global struct to a subclass of
1765 rtl_opt_pass along with...
1766 (pass_data_cse2): ...new pass_data instance and...
1767 (make_pass_cse2): ...new function.
1768 (pass_cse_after_global_opts): Convert from a global struct to a
1769 subclass of rtl_opt_pass along with...
1770 (pass_data_cse_after_global_opts): ...new pass_data instance and...
1771 (make_pass_cse_after_global_opts): ...new function.
1772 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
1773 of rtl_opt_pass along with...
1774 (pass_data_ud_rtl_dce): ...new pass_data instance and...
1775 (make_pass_ud_rtl_dce): ...new function.
1776 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
1777 rtl_opt_pass along with...
1778 (pass_data_fast_rtl_dce): ...new pass_data instance and...
1779 (make_pass_fast_rtl_dce): ...new function.
1780 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
1781 a subclass of rtl_opt_pass along with...
1782 (pass_data_df_initialize_opt): ...new pass_data instance and...
1783 (make_pass_df_initialize_opt): ...new function.
1784 (pass_df_initialize_no_opt): Convert from a global struct to a
1785 subclass of rtl_opt_pass along with...
1786 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
1787 (make_pass_df_initialize_no_opt): ...new function.
1788 (pass_df_finish): Convert from a global struct to a subclass of
1789 rtl_opt_pass along with...
1790 (pass_data_df_finish): ...new pass_data instance and...
1791 (make_pass_df_finish): ...new function.
1792 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
1793 rtl_opt_pass along with...
1794 (pass_data_rtl_dse1): ...new pass_data instance and...
1795 (make_pass_rtl_dse1): ...new function.
1796 (pass_rtl_dse2): Convert from a global struct to a subclass of
1797 rtl_opt_pass along with...
1798 (pass_data_rtl_dse2): ...new pass_data instance and...
1799 (make_pass_rtl_dse2): ...new function.
1800 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
1801 subclass of rtl_opt_pass along with...
1802 (pass_data_dwarf2_frame): ...new pass_data instance and...
1803 (make_pass_dwarf2_frame): ...new function.
1804 * except.c (pass_set_nothrow_function_flags): Convert from a global
1805 struct to a subclass of rtl_opt_pass along with...
1806 (pass_data_set_nothrow_function_flags): ...new pass_data instance
1808 (make_pass_set_nothrow_function_flags): ...new function.
1809 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
1810 subclass of rtl_opt_pass along with...
1811 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
1813 (make_pass_convert_to_eh_region_ranges): ...new function.
1814 * final.c (pass_compute_alignments): Convert from a global struct to a
1815 subclass of rtl_opt_pass along with...
1816 (pass_data_compute_alignments): ...new pass_data instance and...
1817 (make_pass_compute_alignments): ...new function.
1818 (pass_final): Convert from a global struct to a subclass of
1819 rtl_opt_pass along with...
1820 (pass_data_final): ...new pass_data instance and...
1821 (make_pass_final): ...new function.
1822 (pass_shorten_branches): Convert from a global struct to a subclass of
1823 rtl_opt_pass along with...
1824 (pass_data_shorten_branches): ...new pass_data instance and...
1825 (make_pass_shorten_branches): ...new function.
1826 (pass_clean_state): Convert from a global struct to a subclass of
1827 rtl_opt_pass along with...
1828 (pass_data_clean_state): ...new pass_data instance and...
1829 (make_pass_clean_state): ...new function.
1830 * function.c (pass_instantiate_virtual_regs): Convert from a global
1831 struct to a subclass of rtl_opt_pass along with...
1832 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
1833 (make_pass_instantiate_virtual_regs): ...new function.
1834 (pass_leaf_regs): Convert from a global struct to a subclass of
1835 rtl_opt_pass along with...
1836 (pass_data_leaf_regs): ...new pass_data instance and...
1837 (make_pass_leaf_regs): ...new function.
1838 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1839 subclass of rtl_opt_pass along with...
1840 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1842 (make_pass_thread_prologue_and_epilogue): ...new function.
1843 (pass_match_asm_constraints): Convert from a global struct to a
1844 subclass of rtl_opt_pass along with...
1845 (pass_data_match_asm_constraints): ...new pass_data instance and...
1846 (make_pass_match_asm_constraints): ...new function.
1847 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1848 subclass of rtl_opt_pass along with...
1849 (pass_data_rtl_fwprop): ...new pass_data instance and...
1850 (make_pass_rtl_fwprop): ...new function.
1851 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1852 rtl_opt_pass along with...
1853 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
1854 (make_pass_rtl_fwprop_addr): ...new function.
1855 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
1856 rtl_opt_pass along with...
1857 (pass_data_rtl_pre): ...new pass_data instance and...
1858 (make_pass_rtl_pre): ...new function.
1859 (pass_rtl_hoist): Convert from a global struct to a subclass of
1860 rtl_opt_pass along with...
1861 (pass_data_rtl_hoist): ...new pass_data instance and...
1862 (make_pass_rtl_hoist): ...new function.
1863 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
1864 subclass of gimple_opt_pass along with...
1865 (pass_data_lower_cf): ...new pass_data instance and...
1866 (make_pass_lower_cf): ...new function.
1867 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
1868 from a global struct to a subclass of gimple_opt_pass along with...
1869 (pass_data_strength_reduction): ...new pass_data instance and...
1870 (make_pass_strength_reduction): ...new function.
1871 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
1872 of rtl_opt_pass along with...
1873 (pass_data_rtl_ifcvt): ...new pass_data instance and...
1874 (make_pass_rtl_ifcvt): ...new function.
1875 (pass_if_after_combine): Convert from a global struct to a subclass of
1876 rtl_opt_pass along with...
1877 (pass_data_if_after_combine): ...new pass_data instance and...
1878 (make_pass_if_after_combine): ...new function.
1879 (pass_if_after_reload): Convert from a global struct to a subclass of
1880 rtl_opt_pass along with...
1881 (pass_data_if_after_reload): ...new pass_data instance and...
1882 (make_pass_if_after_reload): ...new function.
1883 * init-regs.c (pass_initialize_regs): Convert from a global struct to
1884 a subclass of rtl_opt_pass along with...
1885 (pass_data_initialize_regs): ...new pass_data instance and...
1886 (make_pass_initialize_regs): ...new function.
1887 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
1888 of ipa_opt_pass_d along with...
1889 (pass_data_ipa_cp): ...new pass_data instance and...
1890 (make_pass_ipa_cp): ...new function.
1891 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
1892 global struct to a subclass of gimple_opt_pass along with...
1893 (pass_data_inline_parameters): ...new pass_data instance and...
1894 (make_pass_inline_parameters): ...new function.
1895 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
1896 subclass of gimple_opt_pass along with...
1897 (pass_data_early_inline): ...new pass_data instance and...
1898 (make_pass_early_inline): ...new function.
1899 (pass_ipa_inline): Convert from a global struct to a subclass of
1900 ipa_opt_pass_d along with...
1901 (pass_data_ipa_inline): ...new pass_data instance and...
1902 (make_pass_ipa_inline): ...new function.
1903 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
1904 struct to a subclass of gimple_opt_pass along with...
1905 (pass_data_local_pure_const): ...new pass_data instance and...
1906 (make_pass_local_pure_const): ...new function.
1907 (pass_ipa_pure_const): Convert from a global struct to a subclass of
1908 ipa_opt_pass_d along with...
1909 (pass_data_ipa_pure_const): ...new pass_data instance and...
1910 (make_pass_ipa_pure_const): ...new function.
1911 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
1912 to a subclass of ipa_opt_pass_d along with...
1913 (pass_data_ipa_reference): ...new pass_data instance and...
1914 (make_pass_ipa_reference): ...new function.
1915 * ipa-split.c (pass_split_functions): Convert from a global struct to
1916 a subclass of gimple_opt_pass along with...
1917 (pass_data_split_functions): ...new pass_data instance and...
1918 (make_pass_split_functions): ...new function.
1919 (pass_feedback_split_functions): Convert from a global struct to a
1920 subclass of gimple_opt_pass along with...
1921 (pass_data_feedback_split_functions): ...new pass_data instance and...
1922 (make_pass_feedback_split_functions): ...new function.
1923 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
1924 global struct to a subclass of simple_ipa_opt_pass along with...
1925 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
1927 (make_pass_ipa_function_and_variable_visibility): ...new function.
1928 (pass_ipa_free_inline_summary): Convert from a global struct to a
1929 subclass of simple_ipa_opt_pass along with...
1930 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
1931 (make_pass_ipa_free_inline_summary): ...new function.
1932 (pass_ipa_whole_program_visibility): Convert from a global struct to a
1933 subclass of ipa_opt_pass_d along with...
1934 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
1936 (make_pass_ipa_whole_program_visibility): ...new function.
1937 (pass_ipa_profile): Convert from a global struct to a subclass of
1938 ipa_opt_pass_d along with...
1939 (pass_data_ipa_profile): ...new pass_data instance and...
1940 (make_pass_ipa_profile): ...new function.
1941 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
1942 ipa_opt_pass_d along with...
1943 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
1944 (make_pass_ipa_cdtor_merge): ...new function.
1945 * ira.c (pass_ira): Convert from a global struct to a subclass of
1946 rtl_opt_pass along with...
1947 (pass_data_ira): ...new pass_data instance and...
1948 (make_pass_ira): ...new function.
1949 (pass_reload): Convert from a global struct to a subclass of
1950 rtl_opt_pass along with...
1951 (pass_data_reload): ...new pass_data instance and...
1952 (make_pass_reload): ...new function.
1953 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
1954 subclass of rtl_opt_pass along with...
1955 (pass_data_cleanup_barriers): ...new pass_data instance and...
1956 (make_pass_cleanup_barriers): ...new function.
1957 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
1958 of rtl_opt_pass along with...
1959 (pass_data_loop2): ...new pass_data instance and...
1960 (make_pass_loop2): ...new function.
1961 (pass_rtl_loop_init): Convert from a global struct to a subclass of
1962 rtl_opt_pass along with...
1963 (pass_data_rtl_loop_init): ...new pass_data instance and...
1964 (make_pass_rtl_loop_init): ...new function.
1965 (pass_rtl_loop_done): Convert from a global struct to a subclass of
1966 rtl_opt_pass along with...
1967 (pass_data_rtl_loop_done): ...new pass_data instance and...
1968 (make_pass_rtl_loop_done): ...new function.
1969 (pass_rtl_move_loop_invariants): Convert from a global struct to a
1970 subclass of rtl_opt_pass along with...
1971 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
1972 (make_pass_rtl_move_loop_invariants): ...new function.
1973 (pass_rtl_unswitch): Convert from a global struct to a subclass of
1974 rtl_opt_pass along with...
1975 (pass_data_rtl_unswitch): ...new pass_data instance and...
1976 (make_pass_rtl_unswitch): ...new function.
1977 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
1978 subclass of rtl_opt_pass along with...
1979 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
1981 (make_pass_rtl_unroll_and_peel_loops): ...new function.
1982 (pass_rtl_doloop): Convert from a global struct to a subclass of
1983 rtl_opt_pass along with...
1984 (pass_data_rtl_doloop): ...new pass_data instance and...
1985 (make_pass_rtl_doloop): ...new function.
1986 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
1987 a subclass of rtl_opt_pass along with...
1988 (pass_data_lower_subreg): ...new pass_data instance and...
1989 (make_pass_lower_subreg): ...new function.
1990 (pass_lower_subreg2): Convert from a global struct to a subclass of
1991 rtl_opt_pass along with...
1992 (pass_data_lower_subreg2): ...new pass_data instance and...
1993 (make_pass_lower_subreg2): ...new function.
1994 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
1995 struct to a subclass of ipa_opt_pass_d along with...
1996 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
1997 (make_pass_ipa_lto_gimple_out): ...new function.
1998 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
1999 of ipa_opt_pass_d along with...
2000 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2001 (make_pass_ipa_lto_finish_out): ...new function.
2002 * mode-switching.c (pass_mode_switching): Convert from a global struct
2003 to a subclass of rtl_opt_pass along with...
2004 (pass_data_mode_switching): ...new pass_data instance and...
2005 (make_pass_mode_switching): ...new function.
2006 * modulo-sched.c (pass_sms): Convert from a global struct to a
2007 subclass of rtl_opt_pass along with...
2008 (pass_data_sms): ...new pass_data instance and...
2009 (make_pass_sms): ...new function.
2010 * omp-low.c (pass_expand_omp): Convert from a global struct to a
2011 subclass of gimple_opt_pass along with...
2012 (pass_data_expand_omp): ...new pass_data instance and...
2013 (make_pass_expand_omp): ...new function.
2014 (pass_lower_omp): Convert from a global struct to a subclass of
2015 gimple_opt_pass along with...
2016 (pass_data_lower_omp): ...new pass_data instance and...
2017 (make_pass_lower_omp): ...new function.
2018 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2019 of gimple_opt_pass along with...
2020 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2021 (make_pass_diagnose_omp_blocks): ...new function.
2022 * passes.c (pass_early_local_passes): Convert from a global struct to
2023 a subclass of simple_ipa_opt_pass along with...
2024 (pass_data_early_local_passes): ...new pass_data instance and...
2025 (make_pass_early_local_passes): ...new function.
2026 (pass_all_early_optimizations): Convert from a global struct to a
2027 subclass of gimple_opt_pass along with...
2028 (pass_data_all_early_optimizations): ...new pass_data instance and...
2029 (make_pass_all_early_optimizations): ...new function.
2030 (pass_all_optimizations): Convert from a global struct to a subclass
2031 of gimple_opt_pass along with...
2032 (pass_data_all_optimizations): ...new pass_data instance and...
2033 (make_pass_all_optimizations): ...new function.
2034 (pass_all_optimizations_g): Convert from a global struct to a subclass
2035 of gimple_opt_pass along with...
2036 (pass_data_all_optimizations_g): ...new pass_data instance and...
2037 (make_pass_all_optimizations_g): ...new function.
2038 (pass_rest_of_compilation): Convert from a global struct to a subclass
2039 of rtl_opt_pass along with...
2040 (pass_data_rest_of_compilation): ...new pass_data instance and...
2041 (make_pass_rest_of_compilation): ...new function.
2042 (pass_postreload): Convert from a global struct to a subclass of
2043 rtl_opt_pass along with...
2044 (pass_data_postreload): ...new pass_data instance and...
2045 (make_pass_postreload): ...new function.
2046 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2047 subclass of rtl_opt_pass along with...
2048 (pass_data_gcse2): ...new pass_data instance and...
2049 (make_pass_gcse2): ...new function.
2050 * postreload.c (pass_postreload_cse): Convert from a global struct to
2051 a subclass of rtl_opt_pass along with...
2052 (pass_data_postreload_cse): ...new pass_data instance and...
2053 (make_pass_postreload_cse): ...new function.
2054 * predict.c (pass_profile): Convert from a global struct to a subclass
2055 of gimple_opt_pass along with...
2056 (pass_data_profile): ...new pass_data instance and...
2057 (make_pass_profile): ...new function.
2058 (pass_strip_predict_hints): Convert from a global struct to a subclass
2059 of gimple_opt_pass along with...
2060 (pass_data_strip_predict_hints): ...new pass_data instance and...
2061 (make_pass_strip_predict_hints): ...new function.
2062 * recog.c (pass_peephole2): Convert from a global struct to a subclass
2063 of rtl_opt_pass along with...
2064 (pass_data_peephole2): ...new pass_data instance and...
2065 (make_pass_peephole2): ...new function.
2066 (pass_split_all_insns): Convert from a global struct to a subclass of
2067 rtl_opt_pass along with...
2068 (pass_data_split_all_insns): ...new pass_data instance and...
2069 (make_pass_split_all_insns): ...new function.
2070 (pass_split_after_reload): Convert from a global struct to a subclass
2071 of rtl_opt_pass along with...
2072 (pass_data_split_after_reload): ...new pass_data instance and...
2073 (make_pass_split_after_reload): ...new function.
2074 (pass_split_before_regstack): Convert from a global struct to a
2075 subclass of rtl_opt_pass along with...
2076 (pass_data_split_before_regstack): ...new pass_data instance and...
2077 (make_pass_split_before_regstack): ...new function.
2078 (pass_split_before_sched2): Convert from a global struct to a subclass
2079 of rtl_opt_pass along with...
2080 (pass_data_split_before_sched2): ...new pass_data instance and...
2081 (make_pass_split_before_sched2): ...new function.
2082 (pass_split_for_shorten_branches): Convert from a global struct to a
2083 subclass of rtl_opt_pass along with...
2084 (pass_data_split_for_shorten_branches): ...new pass_data instance
2086 (make_pass_split_for_shorten_branches): ...new function.
2087 * ree.c (pass_ree): Convert from a global struct to a subclass of
2088 rtl_opt_pass along with...
2089 (pass_data_ree): ...new pass_data instance and...
2090 (make_pass_ree): ...new function.
2091 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2092 subclass of rtl_opt_pass along with...
2093 (pass_data_stack_regs): ...new pass_data instance and...
2094 (make_pass_stack_regs): ...new function.
2095 (pass_stack_regs_run): Convert from a global struct to a subclass of
2096 rtl_opt_pass along with...
2097 (pass_data_stack_regs_run): ...new pass_data instance and...
2098 (make_pass_stack_regs_run): ...new function.
2099 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2100 subclass of rtl_opt_pass along with...
2101 (pass_data_cprop_hardreg): ...new pass_data instance and...
2102 (make_pass_cprop_hardreg): ...new function.
2103 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2104 subclass of rtl_opt_pass along with...
2105 (pass_data_reginfo_init): ...new pass_data instance and...
2106 (make_pass_reginfo_init): ...new function.
2107 * regmove.c (pass_regmove): Convert from a global struct to a subclass
2108 of rtl_opt_pass along with...
2109 (pass_data_regmove): ...new pass_data instance and...
2110 (make_pass_regmove): ...new function.
2111 * regrename.c (pass_regrename): Convert from a global struct to a
2112 subclass of rtl_opt_pass along with...
2113 (pass_data_regrename): ...new pass_data instance and...
2114 (make_pass_regrename): ...new function.
2115 * reorg.c (pass_delay_slots): Convert from a global struct to a
2116 subclass of rtl_opt_pass along with...
2117 (pass_data_delay_slots): ...new pass_data instance and...
2118 (make_pass_delay_slots): ...new function.
2119 (pass_machine_reorg): Convert from a global struct to a subclass of
2120 rtl_opt_pass along with...
2121 (pass_data_machine_reorg): ...new pass_data instance and...
2122 (make_pass_machine_reorg): ...new function.
2123 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2124 of rtl_opt_pass along with...
2125 (pass_data_sched): ...new pass_data instance and...
2126 (make_pass_sched): ...new function.
2127 (pass_sched2): Convert from a global struct to a subclass of
2128 rtl_opt_pass along with...
2129 (pass_data_sched2): ...new pass_data instance and...
2130 (make_pass_sched2): ...new function.
2131 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2132 to a subclass of rtl_opt_pass along with...
2133 (pass_data_stack_ptr_mod): ...new pass_data instance and...
2134 (make_pass_stack_ptr_mod): ...new function.
2135 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2136 to a subclass of rtl_opt_pass along with...
2137 (pass_data_rtl_store_motion): ...new pass_data instance and...
2138 (make_pass_rtl_store_motion): ...new function.
2139 * tracer.c (pass_tracer): Convert from a global struct to a subclass
2140 of gimple_opt_pass along with...
2141 (pass_data_tracer): ...new pass_data instance and...
2142 (make_pass_tracer): ...new function.
2143 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2144 to a subclass of gimple_opt_pass along with...
2145 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2146 (make_pass_diagnose_tm_blocks): ...new function.
2147 (pass_lower_tm): Convert from a global struct to a subclass of
2148 gimple_opt_pass along with...
2149 (pass_data_lower_tm): ...new pass_data instance and...
2150 (make_pass_lower_tm): ...new function.
2151 (pass_tm_init): Convert from a global struct to a subclass of
2152 gimple_opt_pass along with...
2153 (pass_data_tm_init): ...new pass_data instance and...
2154 (make_pass_tm_init): ...new function.
2155 (pass_tm_mark): Convert from a global struct to a subclass of
2156 gimple_opt_pass along with...
2157 (pass_data_tm_mark): ...new pass_data instance and...
2158 (make_pass_tm_mark): ...new function.
2159 (pass_tm_edges): Convert from a global struct to a subclass of
2160 gimple_opt_pass along with...
2161 (pass_data_tm_edges): ...new pass_data instance and...
2162 (make_pass_tm_edges): ...new function.
2163 (pass_tm_memopt): Convert from a global struct to a subclass of
2164 gimple_opt_pass along with...
2165 (pass_data_tm_memopt): ...new pass_data instance and...
2166 (make_pass_tm_memopt): ...new function.
2167 (pass_ipa_tm): Convert from a global struct to a subclass of
2168 simple_ipa_opt_pass along with...
2169 (pass_data_ipa_tm): ...new pass_data instance and...
2170 (make_pass_ipa_tm): ...new function.
2171 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2172 subclass of gimple_opt_pass along with...
2173 (pass_data_call_cdce): ...new pass_data instance and...
2174 (make_pass_call_cdce): ...new function.
2175 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2176 subclass of gimple_opt_pass along with...
2177 (pass_data_build_cfg): ...new pass_data instance and...
2178 (make_pass_build_cfg): ...new function.
2179 (pass_split_crit_edges): Convert from a global struct to a subclass of
2180 gimple_opt_pass along with...
2181 (pass_data_split_crit_edges): ...new pass_data instance and...
2182 (make_pass_split_crit_edges): ...new function.
2183 (pass_warn_function_return): Convert from a global struct to a
2184 subclass of gimple_opt_pass along with...
2185 (pass_data_warn_function_return): ...new pass_data instance and...
2186 (make_pass_warn_function_return): ...new function.
2187 (pass_warn_function_noreturn): Convert from a global struct to a
2188 subclass of gimple_opt_pass along with...
2189 (pass_data_warn_function_noreturn): ...new pass_data instance and...
2190 (make_pass_warn_function_noreturn): ...new function.
2191 (pass_warn_unused_result): Convert from a global struct to a subclass
2192 of gimple_opt_pass along with...
2193 (pass_data_warn_unused_result): ...new pass_data instance and...
2194 (make_pass_warn_unused_result): ...new function.
2195 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2196 a subclass of gimple_opt_pass along with...
2197 (pass_data_merge_phi): ...new pass_data instance and...
2198 (make_pass_merge_phi): ...new function.
2199 * tree-complex.c (pass_lower_complex): Convert from a global struct to
2200 a subclass of gimple_opt_pass along with...
2201 (pass_data_lower_complex): ...new pass_data instance and...
2202 (make_pass_lower_complex): ...new function.
2203 (pass_lower_complex_O0): Convert from a global struct to a subclass of
2204 gimple_opt_pass along with...
2205 (pass_data_lower_complex_O0): ...new pass_data instance and...
2206 (make_pass_lower_complex_O0): ...new function.
2207 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2208 subclass of gimple_opt_pass along with...
2209 (pass_data_lower_eh): ...new pass_data instance and...
2210 (make_pass_lower_eh): ...new function.
2211 (pass_refactor_eh): Convert from a global struct to a subclass of
2212 gimple_opt_pass along with...
2213 (pass_data_refactor_eh): ...new pass_data instance and...
2214 (make_pass_refactor_eh): ...new function.
2215 (pass_lower_resx): Convert from a global struct to a subclass of
2216 gimple_opt_pass along with...
2217 (pass_data_lower_resx): ...new pass_data instance and...
2218 (make_pass_lower_resx): ...new function.
2219 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2220 of gimple_opt_pass along with...
2221 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2222 (make_pass_lower_eh_dispatch): ...new function.
2223 (pass_cleanup_eh): Convert from a global struct to a subclass of
2224 gimple_opt_pass along with...
2225 (pass_data_cleanup_eh): ...new pass_data instance and...
2226 (make_pass_cleanup_eh): ...new function.
2227 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2228 to a subclass of simple_ipa_opt_pass along with...
2229 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2230 (make_pass_ipa_lower_emutls): ...new function.
2231 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2232 a subclass of gimple_opt_pass along with...
2233 (pass_data_if_conversion): ...new pass_data instance and...
2234 (make_pass_if_conversion): ...new function.
2235 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2236 subclass of gimple_opt_pass along with...
2237 (pass_data_build_ssa): ...new pass_data instance and...
2238 (make_pass_build_ssa): ...new function.
2239 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2240 global struct to a subclass of gimple_opt_pass along with...
2241 (pass_data_loop_distribution): ...new pass_data instance and...
2242 (make_pass_loop_distribution): ...new function.
2243 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2244 subclass of gimple_opt_pass along with...
2245 (pass_data_mudflap_1): ...new pass_data instance and...
2246 (make_pass_mudflap_1): ...new function.
2247 (pass_mudflap_2): Convert from a global struct to a subclass of
2248 gimple_opt_pass along with...
2249 (pass_data_mudflap_2): ...new pass_data instance and...
2250 (make_pass_mudflap_2): ...new function.
2251 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2252 subclass of gimple_opt_pass along with...
2253 (pass_data_mudflap_1): ...new pass_data instance and...
2254 (make_pass_mudflap_1): ...new function.
2255 (pass_mudflap_2): Convert from a global struct to a subclass of
2256 gimple_opt_pass along with...
2257 (pass_data_mudflap_2): ...new pass_data instance and...
2258 (make_pass_mudflap_2): ...new function.
2259 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2260 gimple_opt_pass along with...
2261 (pass_data_nrv): ...new pass_data instance and...
2262 (make_pass_nrv): ...new function.
2263 (pass_return_slot): Convert from a global struct to a subclass of
2264 gimple_opt_pass along with...
2265 (pass_data_return_slot): ...new pass_data instance and...
2266 (make_pass_return_slot): ...new function.
2267 * tree-object-size.c (pass_object_sizes): Convert from a global struct
2268 to a subclass of gimple_opt_pass along with...
2269 (pass_data_object_sizes): ...new pass_data instance and...
2270 (make_pass_object_sizes): ...new function.
2271 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2272 global struct to a subclass of gimple_opt_pass along with...
2273 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2275 (make_pass_cleanup_cfg_post_optimizing): ...new function.
2276 (pass_fixup_cfg): Convert from a global struct to a subclass of
2277 gimple_opt_pass along with...
2278 (pass_data_fixup_cfg): ...new pass_data instance and...
2279 (make_pass_fixup_cfg): ...new function.
2280 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2281 (make_pass_mudflap_1): ...new function.
2282 (pass_mudflap_2): Replace declaration with that of...
2283 (make_pass_mudflap_2): ...new function.
2284 (pass_asan): Replace declaration with that of...
2285 (make_pass_asan): ...new function.
2286 (pass_asan_O0): Replace declaration with that of...
2287 (make_pass_asan_O0): ...new function.
2288 (pass_tsan): Replace declaration with that of...
2289 (make_pass_tsan): ...new function.
2290 (pass_tsan_O0): Replace declaration with that of...
2291 (make_pass_tsan_O0): ...new function.
2292 (pass_lower_cf): Replace declaration with that of...
2293 (make_pass_lower_cf): ...new function.
2294 (pass_refactor_eh): Replace declaration with that of...
2295 (make_pass_refactor_eh): ...new function.
2296 (pass_lower_eh): Replace declaration with that of...
2297 (make_pass_lower_eh): ...new function.
2298 (pass_lower_eh_dispatch): Replace declaration with that of...
2299 (make_pass_lower_eh_dispatch): ...new function.
2300 (pass_lower_resx): Replace declaration with that of...
2301 (make_pass_lower_resx): ...new function.
2302 (pass_build_cfg): Replace declaration with that of...
2303 (make_pass_build_cfg): ...new function.
2304 (pass_early_tree_profile): Replace declaration with that of...
2305 (make_pass_early_tree_profile): ...new function.
2306 (pass_cleanup_eh): Replace declaration with that of...
2307 (make_pass_cleanup_eh): ...new function.
2308 (pass_sra): Replace declaration with that of...
2309 (make_pass_sra): ...new function.
2310 (pass_sra_early): Replace declaration with that of...
2311 (make_pass_sra_early): ...new function.
2312 (pass_early_ipa_sra): Replace declaration with that of...
2313 (make_pass_early_ipa_sra): ...new function.
2314 (pass_tail_recursion): Replace declaration with that of...
2315 (make_pass_tail_recursion): ...new function.
2316 (pass_tail_calls): Replace declaration with that of...
2317 (make_pass_tail_calls): ...new function.
2318 (pass_tree_loop): Replace declaration with that of...
2319 (make_pass_tree_loop): ...new function.
2320 (pass_tree_loop_init): Replace declaration with that of...
2321 (make_pass_tree_loop_init): ...new function.
2322 (pass_lim): Replace declaration with that of...
2323 (make_pass_lim): ...new function.
2324 (pass_tree_unswitch): Replace declaration with that of...
2325 (make_pass_tree_unswitch): ...new function.
2326 (pass_predcom): Replace declaration with that of...
2327 (make_pass_predcom): ...new function.
2328 (pass_iv_canon): Replace declaration with that of...
2329 (make_pass_iv_canon): ...new function.
2330 (pass_scev_cprop): Replace declaration with that of...
2331 (make_pass_scev_cprop): ...new function.
2332 (pass_empty_loop): Replace declaration with that of...
2333 (make_pass_empty_loop): ...new function.
2334 (pass_record_bounds): Replace declaration with that of...
2335 (make_pass_record_bounds): ...new function.
2336 (pass_graphite): Replace declaration with that of...
2337 (make_pass_graphite): ...new function.
2338 (pass_graphite_transforms): Replace declaration with that of...
2339 (make_pass_graphite_transforms): ...new function.
2340 (pass_if_conversion): Replace declaration with that of...
2341 (make_pass_if_conversion): ...new function.
2342 (pass_loop_distribution): Replace declaration with that of...
2343 (make_pass_loop_distribution): ...new function.
2344 (pass_vectorize): Replace declaration with that of...
2345 (make_pass_vectorize): ...new function.
2346 (pass_slp_vectorize): Replace declaration with that of...
2347 (make_pass_slp_vectorize): ...new function.
2348 (pass_complete_unroll): Replace declaration with that of...
2349 (make_pass_complete_unroll): ...new function.
2350 (pass_complete_unrolli): Replace declaration with that of...
2351 (make_pass_complete_unrolli): ...new function.
2352 (pass_parallelize_loops): Replace declaration with that of...
2353 (make_pass_parallelize_loops): ...new function.
2354 (pass_loop_prefetch): Replace declaration with that of...
2355 (make_pass_loop_prefetch): ...new function.
2356 (pass_iv_optimize): Replace declaration with that of...
2357 (make_pass_iv_optimize): ...new function.
2358 (pass_tree_loop_done): Replace declaration with that of...
2359 (make_pass_tree_loop_done): ...new function.
2360 (pass_ch): Replace declaration with that of...
2361 (make_pass_ch): ...new function.
2362 (pass_ccp): Replace declaration with that of...
2363 (make_pass_ccp): ...new function.
2364 (pass_phi_only_cprop): Replace declaration with that of...
2365 (make_pass_phi_only_cprop): ...new function.
2366 (pass_build_ssa): Replace declaration with that of...
2367 (make_pass_build_ssa): ...new function.
2368 (pass_build_alias): Replace declaration with that of...
2369 (make_pass_build_alias): ...new function.
2370 (pass_build_ealias): Replace declaration with that of...
2371 (make_pass_build_ealias): ...new function.
2372 (pass_dominator): Replace declaration with that of...
2373 (make_pass_dominator): ...new function.
2374 (pass_dce): Replace declaration with that of...
2375 (make_pass_dce): ...new function.
2376 (pass_dce_loop): Replace declaration with that of...
2377 (make_pass_dce_loop): ...new function.
2378 (pass_cd_dce): Replace declaration with that of...
2379 (make_pass_cd_dce): ...new function.
2380 (pass_call_cdce): Replace declaration with that of...
2381 (make_pass_call_cdce): ...new function.
2382 (pass_merge_phi): Replace declaration with that of...
2383 (make_pass_merge_phi): ...new function.
2384 (pass_split_crit_edges): Replace declaration with that of...
2385 (make_pass_split_crit_edges): ...new function.
2386 (pass_pre): Replace declaration with that of...
2387 (make_pass_pre): ...new function.
2388 (pass_profile): Replace declaration with that of...
2389 (make_pass_profile): ...new function.
2390 (pass_strip_predict_hints): Replace declaration with that of...
2391 (make_pass_strip_predict_hints): ...new function.
2392 (pass_lower_complex_O0): Replace declaration with that of...
2393 (make_pass_lower_complex_O0): ...new function.
2394 (pass_lower_complex): Replace declaration with that of...
2395 (make_pass_lower_complex): ...new function.
2396 (pass_lower_vector): Replace declaration with that of...
2397 (make_pass_lower_vector): ...new function.
2398 (pass_lower_vector_ssa): Replace declaration with that of...
2399 (make_pass_lower_vector_ssa): ...new function.
2400 (pass_lower_omp): Replace declaration with that of...
2401 (make_pass_lower_omp): ...new function.
2402 (pass_diagnose_omp_blocks): Replace declaration with that of...
2403 (make_pass_diagnose_omp_blocks): ...new function.
2404 (pass_expand_omp): Replace declaration with that of...
2405 (make_pass_expand_omp): ...new function.
2406 (pass_expand_omp_ssa): Replace declaration with that of...
2407 (make_pass_expand_omp_ssa): ...new function.
2408 (pass_object_sizes): Replace declaration with that of...
2409 (make_pass_object_sizes): ...new function.
2410 (pass_strlen): Replace declaration with that of...
2411 (make_pass_strlen): ...new function.
2412 (pass_fold_builtins): Replace declaration with that of...
2413 (make_pass_fold_builtins): ...new function.
2414 (pass_stdarg): Replace declaration with that of...
2415 (make_pass_stdarg): ...new function.
2416 (pass_early_warn_uninitialized): Replace declaration with that of...
2417 (make_pass_early_warn_uninitialized): ...new function.
2418 (pass_late_warn_uninitialized): Replace declaration with that of...
2419 (make_pass_late_warn_uninitialized): ...new function.
2420 (pass_cse_reciprocals): Replace declaration with that of...
2421 (make_pass_cse_reciprocals): ...new function.
2422 (pass_cse_sincos): Replace declaration with that of...
2423 (make_pass_cse_sincos): ...new function.
2424 (pass_optimize_bswap): Replace declaration with that of...
2425 (make_pass_optimize_bswap): ...new function.
2426 (pass_optimize_widening_mul): Replace declaration with that of...
2427 (make_pass_optimize_widening_mul): ...new function.
2428 (pass_warn_function_return): Replace declaration with that of...
2429 (make_pass_warn_function_return): ...new function.
2430 (pass_warn_function_noreturn): Replace declaration with that of...
2431 (make_pass_warn_function_noreturn): ...new function.
2432 (pass_cselim): Replace declaration with that of...
2433 (make_pass_cselim): ...new function.
2434 (pass_phiopt): Replace declaration with that of...
2435 (make_pass_phiopt): ...new function.
2436 (pass_forwprop): Replace declaration with that of...
2437 (make_pass_forwprop): ...new function.
2438 (pass_phiprop): Replace declaration with that of...
2439 (make_pass_phiprop): ...new function.
2440 (pass_tree_ifcombine): Replace declaration with that of...
2441 (make_pass_tree_ifcombine): ...new function.
2442 (pass_dse): Replace declaration with that of...
2443 (make_pass_dse): ...new function.
2444 (pass_nrv): Replace declaration with that of...
2445 (make_pass_nrv): ...new function.
2446 (pass_rename_ssa_copies): Replace declaration with that of...
2447 (make_pass_rename_ssa_copies): ...new function.
2448 (pass_sink_code): Replace declaration with that of...
2449 (make_pass_sink_code): ...new function.
2450 (pass_fre): Replace declaration with that of...
2451 (make_pass_fre): ...new function.
2452 (pass_check_data_deps): Replace declaration with that of...
2453 (make_pass_check_data_deps): ...new function.
2454 (pass_copy_prop): Replace declaration with that of...
2455 (make_pass_copy_prop): ...new function.
2456 (pass_vrp): Replace declaration with that of...
2457 (make_pass_vrp): ...new function.
2458 (pass_uncprop): Replace declaration with that of...
2459 (make_pass_uncprop): ...new function.
2460 (pass_return_slot): Replace declaration with that of...
2461 (make_pass_return_slot): ...new function.
2462 (pass_reassoc): Replace declaration with that of...
2463 (make_pass_reassoc): ...new function.
2464 (pass_rebuild_cgraph_edges): Replace declaration with that of...
2465 (make_pass_rebuild_cgraph_edges): ...new function.
2466 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2467 (make_pass_remove_cgraph_callee_edges): ...new function.
2468 (pass_build_cgraph_edges): Replace declaration with that of...
2469 (make_pass_build_cgraph_edges): ...new function.
2470 (pass_local_pure_const): Replace declaration with that of...
2471 (make_pass_local_pure_const): ...new function.
2472 (pass_tracer): Replace declaration with that of...
2473 (make_pass_tracer): ...new function.
2474 (pass_warn_unused_result): Replace declaration with that of...
2475 (make_pass_warn_unused_result): ...new function.
2476 (pass_diagnose_tm_blocks): Replace declaration with that of...
2477 (make_pass_diagnose_tm_blocks): ...new function.
2478 (pass_lower_tm): Replace declaration with that of...
2479 (make_pass_lower_tm): ...new function.
2480 (pass_tm_init): Replace declaration with that of...
2481 (make_pass_tm_init): ...new function.
2482 (pass_tm_mark): Replace declaration with that of...
2483 (make_pass_tm_mark): ...new function.
2484 (pass_tm_memopt): Replace declaration with that of...
2485 (make_pass_tm_memopt): ...new function.
2486 (pass_tm_edges): Replace declaration with that of...
2487 (make_pass_tm_edges): ...new function.
2488 (pass_split_functions): Replace declaration with that of...
2489 (make_pass_split_functions): ...new function.
2490 (pass_feedback_split_functions): Replace declaration with that of...
2491 (make_pass_feedback_split_functions): ...new function.
2492 (pass_strength_reduction): Replace declaration with that of...
2493 (make_pass_strength_reduction): ...new function.
2494 (pass_ipa_lower_emutls): Replace declaration with that of...
2495 (make_pass_ipa_lower_emutls): ...new function.
2496 (pass_ipa_function_and_variable_visibility): Replace declaration with
2498 (make_pass_ipa_function_and_variable_visibility): ...new function.
2499 (pass_ipa_tree_profile): Replace declaration with that of...
2500 (make_pass_ipa_tree_profile): ...new function.
2501 (pass_early_local_passes): Replace declaration with that of...
2502 (make_pass_early_local_passes): ...new function.
2503 (pass_ipa_whole_program_visibility): Replace declaration with that
2505 (make_pass_ipa_whole_program_visibility): ...new function.
2506 (pass_ipa_lto_gimple_out): Replace declaration with that of...
2507 (make_pass_ipa_lto_gimple_out): ...new function.
2508 (pass_ipa_increase_alignment): Replace declaration with that of...
2509 (make_pass_ipa_increase_alignment): ...new function.
2510 (pass_ipa_inline): Replace declaration with that of...
2511 (make_pass_ipa_inline): ...new function.
2512 (pass_ipa_free_lang_data): Replace declaration with that of...
2513 (make_pass_ipa_free_lang_data): ...new function.
2514 (pass_ipa_free_inline_summary): Replace declaration with that of...
2515 (make_pass_ipa_free_inline_summary): ...new function.
2516 (pass_ipa_cp): Replace declaration with that of...
2517 (make_pass_ipa_cp): ...new function.
2518 (pass_ipa_reference): Replace declaration with that of...
2519 (make_pass_ipa_reference): ...new function.
2520 (pass_ipa_pure_const): Replace declaration with that of...
2521 (make_pass_ipa_pure_const): ...new function.
2522 (pass_ipa_pta): Replace declaration with that of...
2523 (make_pass_ipa_pta): ...new function.
2524 (pass_ipa_lto_finish_out): Replace declaration with that of...
2525 (make_pass_ipa_lto_finish_out): ...new function.
2526 (pass_ipa_tm): Replace declaration with that of...
2527 (make_pass_ipa_tm): ...new function.
2528 (pass_ipa_profile): Replace declaration with that of...
2529 (make_pass_ipa_profile): ...new function.
2530 (pass_ipa_cdtor_merge): Replace declaration with that of...
2531 (make_pass_ipa_cdtor_merge): ...new function.
2532 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2534 (make_pass_cleanup_cfg_post_optimizing): ...new function.
2535 (pass_init_datastructures): Replace declaration with that of...
2536 (make_pass_init_datastructures): ...new function.
2537 (pass_fixup_cfg): Replace declaration with that of...
2538 (make_pass_fixup_cfg): ...new function.
2539 (pass_expand): Replace declaration with that of...
2540 (make_pass_expand): ...new function.
2541 (pass_instantiate_virtual_regs): Replace declaration with that of...
2542 (make_pass_instantiate_virtual_regs): ...new function.
2543 (pass_rtl_fwprop): Replace declaration with that of...
2544 (make_pass_rtl_fwprop): ...new function.
2545 (pass_rtl_fwprop_addr): Replace declaration with that of...
2546 (make_pass_rtl_fwprop_addr): ...new function.
2547 (pass_jump): Replace declaration with that of...
2548 (make_pass_jump): ...new function.
2549 (pass_jump2): Replace declaration with that of...
2550 (make_pass_jump2): ...new function.
2551 (pass_lower_subreg): Replace declaration with that of...
2552 (make_pass_lower_subreg): ...new function.
2553 (pass_cse): Replace declaration with that of...
2554 (make_pass_cse): ...new function.
2555 (pass_fast_rtl_dce): Replace declaration with that of...
2556 (make_pass_fast_rtl_dce): ...new function.
2557 (pass_ud_rtl_dce): Replace declaration with that of...
2558 (make_pass_ud_rtl_dce): ...new function.
2559 (pass_rtl_dce): Replace declaration with that of...
2560 (make_pass_rtl_dce): ...new function.
2561 (pass_rtl_dse1): Replace declaration with that of...
2562 (make_pass_rtl_dse1): ...new function.
2563 (pass_rtl_dse2): Replace declaration with that of...
2564 (make_pass_rtl_dse2): ...new function.
2565 (pass_rtl_dse3): Replace declaration with that of...
2566 (make_pass_rtl_dse3): ...new function.
2567 (pass_rtl_cprop): Replace declaration with that of...
2568 (make_pass_rtl_cprop): ...new function.
2569 (pass_rtl_pre): Replace declaration with that of...
2570 (make_pass_rtl_pre): ...new function.
2571 (pass_rtl_hoist): Replace declaration with that of...
2572 (make_pass_rtl_hoist): ...new function.
2573 (pass_rtl_store_motion): Replace declaration with that of...
2574 (make_pass_rtl_store_motion): ...new function.
2575 (pass_cse_after_global_opts): Replace declaration with that of...
2576 (make_pass_cse_after_global_opts): ...new function.
2577 (pass_rtl_ifcvt): Replace declaration with that of...
2578 (make_pass_rtl_ifcvt): ...new function.
2579 (pass_into_cfg_layout_mode): Replace declaration with that of...
2580 (make_pass_into_cfg_layout_mode): ...new function.
2581 (pass_outof_cfg_layout_mode): Replace declaration with that of...
2582 (make_pass_outof_cfg_layout_mode): ...new function.
2583 (pass_loop2): Replace declaration with that of...
2584 (make_pass_loop2): ...new function.
2585 (pass_rtl_loop_init): Replace declaration with that of...
2586 (make_pass_rtl_loop_init): ...new function.
2587 (pass_rtl_move_loop_invariants): Replace declaration with that of...
2588 (make_pass_rtl_move_loop_invariants): ...new function.
2589 (pass_rtl_unswitch): Replace declaration with that of...
2590 (make_pass_rtl_unswitch): ...new function.
2591 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2592 (make_pass_rtl_unroll_and_peel_loops): ...new function.
2593 (pass_rtl_doloop): Replace declaration with that of...
2594 (make_pass_rtl_doloop): ...new function.
2595 (pass_rtl_loop_done): Replace declaration with that of...
2596 (make_pass_rtl_loop_done): ...new function.
2597 (pass_web): Replace declaration with that of...
2598 (make_pass_web): ...new function.
2599 (pass_cse2): Replace declaration with that of...
2600 (make_pass_cse2): ...new function.
2601 (pass_df_initialize_opt): Replace declaration with that of...
2602 (make_pass_df_initialize_opt): ...new function.
2603 (pass_df_initialize_no_opt): Replace declaration with that of...
2604 (make_pass_df_initialize_no_opt): ...new function.
2605 (pass_reginfo_init): Replace declaration with that of...
2606 (make_pass_reginfo_init): ...new function.
2607 (pass_inc_dec): Replace declaration with that of...
2608 (make_pass_inc_dec): ...new function.
2609 (pass_stack_ptr_mod): Replace declaration with that of...
2610 (make_pass_stack_ptr_mod): ...new function.
2611 (pass_initialize_regs): Replace declaration with that of...
2612 (make_pass_initialize_regs): ...new function.
2613 (pass_combine): Replace declaration with that of...
2614 (make_pass_combine): ...new function.
2615 (pass_if_after_combine): Replace declaration with that of...
2616 (make_pass_if_after_combine): ...new function.
2617 (pass_ree): Replace declaration with that of...
2618 (make_pass_ree): ...new function.
2619 (pass_partition_blocks): Replace declaration with that of...
2620 (make_pass_partition_blocks): ...new function.
2621 (pass_match_asm_constraints): Replace declaration with that of...
2622 (make_pass_match_asm_constraints): ...new function.
2623 (pass_regmove): Replace declaration with that of...
2624 (make_pass_regmove): ...new function.
2625 (pass_split_all_insns): Replace declaration with that of...
2626 (make_pass_split_all_insns): ...new function.
2627 (pass_fast_rtl_byte_dce): Replace declaration with that of...
2628 (make_pass_fast_rtl_byte_dce): ...new function.
2629 (pass_lower_subreg2): Replace declaration with that of...
2630 (make_pass_lower_subreg2): ...new function.
2631 (pass_mode_switching): Replace declaration with that of...
2632 (make_pass_mode_switching): ...new function.
2633 (pass_sms): Replace declaration with that of...
2634 (make_pass_sms): ...new function.
2635 (pass_sched): Replace declaration with that of...
2636 (make_pass_sched): ...new function.
2637 (pass_ira): Replace declaration with that of...
2638 (make_pass_ira): ...new function.
2639 (pass_reload): Replace declaration with that of...
2640 (make_pass_reload): ...new function.
2641 (pass_clean_state): Replace declaration with that of...
2642 (make_pass_clean_state): ...new function.
2643 (pass_branch_prob): Replace declaration with that of...
2644 (make_pass_branch_prob): ...new function.
2645 (pass_value_profile_transformations): Replace declaration with that
2647 (make_pass_value_profile_transformations): ...new function.
2648 (pass_postreload_cse): Replace declaration with that of...
2649 (make_pass_postreload_cse): ...new function.
2650 (pass_gcse2): Replace declaration with that of...
2651 (make_pass_gcse2): ...new function.
2652 (pass_split_after_reload): Replace declaration with that of...
2653 (make_pass_split_after_reload): ...new function.
2654 (pass_branch_target_load_optimize1): Replace declaration with that
2656 (make_pass_branch_target_load_optimize1): ...new function.
2657 (pass_thread_prologue_and_epilogue): Replace declaration with that
2659 (make_pass_thread_prologue_and_epilogue): ...new function.
2660 (pass_stack_adjustments): Replace declaration with that of...
2661 (make_pass_stack_adjustments): ...new function.
2662 (pass_peephole2): Replace declaration with that of...
2663 (make_pass_peephole2): ...new function.
2664 (pass_if_after_reload): Replace declaration with that of...
2665 (make_pass_if_after_reload): ...new function.
2666 (pass_regrename): Replace declaration with that of...
2667 (make_pass_regrename): ...new function.
2668 (pass_cprop_hardreg): Replace declaration with that of...
2669 (make_pass_cprop_hardreg): ...new function.
2670 (pass_reorder_blocks): Replace declaration with that of...
2671 (make_pass_reorder_blocks): ...new function.
2672 (pass_branch_target_load_optimize2): Replace declaration with that
2674 (make_pass_branch_target_load_optimize2): ...new function.
2675 (pass_leaf_regs): Replace declaration with that of...
2676 (make_pass_leaf_regs): ...new function.
2677 (pass_split_before_sched2): Replace declaration with that of...
2678 (make_pass_split_before_sched2): ...new function.
2679 (pass_compare_elim_after_reload): Replace declaration with that of...
2680 (make_pass_compare_elim_after_reload): ...new function.
2681 (pass_sched2): Replace declaration with that of...
2682 (make_pass_sched2): ...new function.
2683 (pass_stack_regs): Replace declaration with that of...
2684 (make_pass_stack_regs): ...new function.
2685 (pass_stack_regs_run): Replace declaration with that of...
2686 (make_pass_stack_regs_run): ...new function.
2687 (pass_df_finish): Replace declaration with that of...
2688 (make_pass_df_finish): ...new function.
2689 (pass_compute_alignments): Replace declaration with that of...
2690 (make_pass_compute_alignments): ...new function.
2691 (pass_duplicate_computed_gotos): Replace declaration with that of...
2692 (make_pass_duplicate_computed_gotos): ...new function.
2693 (pass_variable_tracking): Replace declaration with that of...
2694 (make_pass_variable_tracking): ...new function.
2695 (pass_free_cfg): Replace declaration with that of...
2696 (make_pass_free_cfg): ...new function.
2697 (pass_machine_reorg): Replace declaration with that of...
2698 (make_pass_machine_reorg): ...new function.
2699 (pass_cleanup_barriers): Replace declaration with that of...
2700 (make_pass_cleanup_barriers): ...new function.
2701 (pass_delay_slots): Replace declaration with that of...
2702 (make_pass_delay_slots): ...new function.
2703 (pass_split_for_shorten_branches): Replace declaration with that of...
2704 (make_pass_split_for_shorten_branches): ...new function.
2705 (pass_split_before_regstack): Replace declaration with that of...
2706 (make_pass_split_before_regstack): ...new function.
2707 (pass_convert_to_eh_region_ranges): Replace declaration with that
2709 (make_pass_convert_to_eh_region_ranges): ...new function.
2710 (pass_shorten_branches): Replace declaration with that of...
2711 (make_pass_shorten_branches): ...new function.
2712 (pass_set_nothrow_function_flags): Replace declaration with that of...
2713 (make_pass_set_nothrow_function_flags): ...new function.
2714 (pass_dwarf2_frame): Replace declaration with that of...
2715 (make_pass_dwarf2_frame): ...new function.
2716 (pass_final): Replace declaration with that of...
2717 (make_pass_final): ...new function.
2718 (pass_rtl_seqabstr): Replace declaration with that of...
2719 (make_pass_rtl_seqabstr): ...new function.
2720 (pass_release_ssa_names): Replace declaration with that of...
2721 (make_pass_release_ssa_names): ...new function.
2722 (pass_early_inline): Replace declaration with that of...
2723 (make_pass_early_inline): ...new function.
2724 (pass_inline_parameters): Replace declaration with that of...
2725 (make_pass_inline_parameters): ...new function.
2726 (pass_update_address_taken): Replace declaration with that of...
2727 (make_pass_update_address_taken): ...new function.
2728 (pass_convert_switch): Replace declaration with that of...
2729 (make_pass_convert_switch): ...new function.
2730 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
2731 to a subclass of simple_ipa_opt_pass along with...
2732 (pass_data_ipa_tree_profile): ...new pass_data instance and...
2733 (make_pass_ipa_tree_profile): ...new function.
2734 * tree-sra.c (pass_sra_early): Convert from a global struct to a
2735 subclass of gimple_opt_pass along with...
2736 (pass_data_sra_early): ...new pass_data instance and...
2737 (make_pass_sra_early): ...new function.
2738 (pass_sra): Convert from a global struct to a subclass of
2739 gimple_opt_pass along with...
2740 (pass_data_sra): ...new pass_data instance and...
2741 (make_pass_sra): ...new function.
2742 (pass_early_ipa_sra): Convert from a global struct to a subclass of
2743 gimple_opt_pass along with...
2744 (pass_data_early_ipa_sra): ...new pass_data instance and...
2745 (make_pass_early_ipa_sra): ...new function.
2746 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
2747 subclass of gimple_opt_pass along with...
2748 (pass_data_ccp): ...new pass_data instance and...
2749 (make_pass_ccp): ...new function.
2750 (pass_fold_builtins): Convert from a global struct to a subclass of
2751 gimple_opt_pass along with...
2752 (pass_data_fold_builtins): ...new pass_data instance and...
2753 (make_pass_fold_builtins): ...new function.
2754 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
2755 subclass of gimple_opt_pass along with...
2756 (pass_data_copy_prop): ...new pass_data instance and...
2757 (make_pass_copy_prop): ...new function.
2758 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
2759 global struct to a subclass of gimple_opt_pass along with...
2760 (pass_data_rename_ssa_copies): ...new pass_data instance and...
2761 (make_pass_rename_ssa_copies): ...new function.
2762 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
2763 subclass of gimple_opt_pass along with...
2764 (pass_data_dce): ...new pass_data instance and...
2765 (make_pass_dce): ...new function.
2766 (pass_dce_loop): Convert from a global struct to a subclass of
2767 gimple_opt_pass along with...
2768 (pass_data_dce_loop): ...new pass_data instance and...
2769 (make_pass_dce_loop): ...new function.
2770 (pass_cd_dce): Convert from a global struct to a subclass of
2771 gimple_opt_pass along with...
2772 (pass_data_cd_dce): ...new pass_data instance and...
2773 (make_pass_cd_dce): ...new function.
2774 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
2775 subclass of gimple_opt_pass along with...
2776 (pass_data_dominator): ...new pass_data instance and...
2777 (make_pass_dominator): ...new function.
2778 (pass_phi_only_cprop): Convert from a global struct to a subclass of
2779 gimple_opt_pass along with...
2780 (pass_data_phi_only_cprop): ...new pass_data instance and...
2781 (make_pass_phi_only_cprop): ...new function.
2782 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
2783 subclass of gimple_opt_pass along with...
2784 (pass_data_dse): ...new pass_data instance and...
2785 (make_pass_dse): ...new function.
2786 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
2787 a subclass of gimple_opt_pass along with...
2788 (pass_data_forwprop): ...new pass_data instance and...
2789 (make_pass_forwprop): ...new function.
2790 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
2791 struct to a subclass of gimple_opt_pass along with...
2792 (pass_data_tree_ifcombine): ...new pass_data instance and...
2793 (make_pass_tree_ifcombine): ...new function.
2794 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
2795 subclass of gimple_opt_pass along with...
2796 (pass_data_ch): ...new pass_data instance and...
2797 (make_pass_ch): ...new function.
2798 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
2799 subclass of gimple_opt_pass along with...
2800 (pass_data_tree_loop): ...new pass_data instance and...
2801 (make_pass_tree_loop): ...new function.
2802 (pass_tree_loop_init): Convert from a global struct to a subclass of
2803 gimple_opt_pass along with...
2804 (pass_data_tree_loop_init): ...new pass_data instance and...
2805 (make_pass_tree_loop_init): ...new function.
2806 (pass_lim): Convert from a global struct to a subclass of
2807 gimple_opt_pass along with...
2808 (pass_data_lim): ...new pass_data instance and...
2809 (make_pass_lim): ...new function.
2810 (pass_tree_unswitch): Convert from a global struct to a subclass of
2811 gimple_opt_pass along with...
2812 (pass_data_tree_unswitch): ...new pass_data instance and...
2813 (make_pass_tree_unswitch): ...new function.
2814 (pass_predcom): Convert from a global struct to a subclass of
2815 gimple_opt_pass along with...
2816 (pass_data_predcom): ...new pass_data instance and...
2817 (make_pass_predcom): ...new function.
2818 (pass_vectorize): Convert from a global struct to a subclass of
2819 gimple_opt_pass along with...
2820 (pass_data_vectorize): ...new pass_data instance and...
2821 (make_pass_vectorize): ...new function.
2822 (pass_graphite): Convert from a global struct to a subclass of
2823 gimple_opt_pass along with...
2824 (pass_data_graphite): ...new pass_data instance and...
2825 (make_pass_graphite): ...new function.
2826 (pass_graphite_transforms): Convert from a global struct to a subclass
2827 of gimple_opt_pass along with...
2828 (pass_data_graphite_transforms): ...new pass_data instance and...
2829 (make_pass_graphite_transforms): ...new function.
2830 (pass_check_data_deps): Convert from a global struct to a subclass of
2831 gimple_opt_pass along with...
2832 (pass_data_check_data_deps): ...new pass_data instance and...
2833 (make_pass_check_data_deps): ...new function.
2834 (pass_iv_canon): Convert from a global struct to a subclass of
2835 gimple_opt_pass along with...
2836 (pass_data_iv_canon): ...new pass_data instance and...
2837 (make_pass_iv_canon): ...new function.
2838 (pass_scev_cprop): Convert from a global struct to a subclass of
2839 gimple_opt_pass along with...
2840 (pass_data_scev_cprop): ...new pass_data instance and...
2841 (make_pass_scev_cprop): ...new function.
2842 (pass_record_bounds): Convert from a global struct to a subclass of
2843 gimple_opt_pass along with...
2844 (pass_data_record_bounds): ...new pass_data instance and...
2845 (make_pass_record_bounds): ...new function.
2846 (pass_complete_unroll): Convert from a global struct to a subclass of
2847 gimple_opt_pass along with...
2848 (pass_data_complete_unroll): ...new pass_data instance and...
2849 (make_pass_complete_unroll): ...new function.
2850 (pass_complete_unrolli): Convert from a global struct to a subclass of
2851 gimple_opt_pass along with...
2852 (pass_data_complete_unrolli): ...new pass_data instance and...
2853 (make_pass_complete_unrolli): ...new function.
2854 (pass_parallelize_loops): Convert from a global struct to a subclass
2855 of gimple_opt_pass along with...
2856 (pass_data_parallelize_loops): ...new pass_data instance and...
2857 (make_pass_parallelize_loops): ...new function.
2858 (pass_loop_prefetch): Convert from a global struct to a subclass of
2859 gimple_opt_pass along with...
2860 (pass_data_loop_prefetch): ...new pass_data instance and...
2861 (make_pass_loop_prefetch): ...new function.
2862 (pass_iv_optimize): Convert from a global struct to a subclass of
2863 gimple_opt_pass along with...
2864 (pass_data_iv_optimize): ...new pass_data instance and...
2865 (make_pass_iv_optimize): ...new function.
2866 (pass_tree_loop_done): Convert from a global struct to a subclass of
2867 gimple_opt_pass along with...
2868 (pass_data_tree_loop_done): ...new pass_data instance and...
2869 (make_pass_tree_loop_done): ...new function.
2870 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
2871 struct to a subclass of gimple_opt_pass along with...
2872 (pass_data_cse_reciprocals): ...new pass_data instance and...
2873 (make_pass_cse_reciprocals): ...new function.
2874 (pass_cse_sincos): Convert from a global struct to a subclass of
2875 gimple_opt_pass along with...
2876 (pass_data_cse_sincos): ...new pass_data instance and...
2877 (make_pass_cse_sincos): ...new function.
2878 (pass_optimize_bswap): Convert from a global struct to a subclass of
2879 gimple_opt_pass along with...
2880 (pass_data_optimize_bswap): ...new pass_data instance and...
2881 (make_pass_optimize_bswap): ...new function.
2882 (pass_optimize_widening_mul): Convert from a global struct to a
2883 subclass of gimple_opt_pass along with...
2884 (pass_data_optimize_widening_mul): ...new pass_data instance and...
2885 (make_pass_optimize_widening_mul): ...new function.
2886 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
2887 subclass of gimple_opt_pass along with...
2888 (pass_data_phiopt): ...new pass_data instance and...
2889 (make_pass_phiopt): ...new function.
2890 (pass_cselim): Convert from a global struct to a subclass of
2891 gimple_opt_pass along with...
2892 (pass_data_cselim): ...new pass_data instance and...
2893 (make_pass_cselim): ...new function.
2894 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
2895 subclass of gimple_opt_pass along with...
2896 (pass_data_phiprop): ...new pass_data instance and...
2897 (make_pass_phiprop): ...new function.
2898 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
2899 subclass of gimple_opt_pass along with...
2900 (pass_data_pre): ...new pass_data instance and...
2901 (make_pass_pre): ...new function.
2902 (pass_fre): Convert from a global struct to a subclass of
2903 gimple_opt_pass along with...
2904 (pass_data_fre): ...new pass_data instance and...
2905 (make_pass_fre): ...new function.
2906 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
2907 subclass of gimple_opt_pass along with...
2908 (pass_data_reassoc): ...new pass_data instance and...
2909 (make_pass_reassoc): ...new function.
2910 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
2911 subclass of gimple_opt_pass along with...
2912 (pass_data_sink_code): ...new pass_data instance and...
2913 (make_pass_sink_code): ...new function.
2914 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
2915 subclass of gimple_opt_pass along with...
2916 (pass_data_strlen): ...new pass_data instance and...
2917 (make_pass_strlen): ...new function.
2918 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
2919 struct to a subclass of gimple_opt_pass along with...
2920 (pass_data_build_alias): ...new pass_data instance and...
2921 (make_pass_build_alias): ...new function.
2922 (pass_build_ealias): Convert from a global struct to a subclass of
2923 gimple_opt_pass along with...
2924 (pass_data_build_ealias): ...new pass_data instance and...
2925 (make_pass_build_ealias): ...new function.
2926 (pass_ipa_pta): Convert from a global struct to a subclass of
2927 simple_ipa_opt_pass along with...
2928 (pass_data_ipa_pta): ...new pass_data instance and...
2929 (make_pass_ipa_pta): ...new function.
2930 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
2931 subclass of gimple_opt_pass along with...
2932 (pass_data_uncprop): ...new pass_data instance and...
2933 (make_pass_uncprop): ...new function.
2934 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
2935 global struct to a subclass of gimple_opt_pass along with...
2936 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
2937 (make_pass_late_warn_uninitialized): ...new function.
2938 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
2939 to a subclass of gimple_opt_pass along with...
2940 (pass_data_init_datastructures): ...new pass_data instance and...
2941 (make_pass_init_datastructures): ...new function.
2942 (pass_early_warn_uninitialized): Convert from a global struct to a
2943 subclass of gimple_opt_pass along with...
2944 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
2945 (make_pass_early_warn_uninitialized): ...new function.
2946 (pass_update_address_taken): Convert from a global struct to a
2947 subclass of gimple_opt_pass along with...
2948 (pass_data_update_address_taken): ...new pass_data instance and...
2949 (make_pass_update_address_taken): ...new function.
2950 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
2951 struct to a subclass of gimple_opt_pass along with...
2952 (pass_data_release_ssa_names): ...new pass_data instance and...
2953 (make_pass_release_ssa_names): ...new function.
2954 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
2955 subclass of gimple_opt_pass along with...
2956 (pass_data_stdarg): ...new pass_data instance and...
2957 (make_pass_stdarg): ...new function.
2958 * tree-switch-conversion.c (pass_convert_switch): Convert from a
2959 global struct to a subclass of gimple_opt_pass along with...
2960 (pass_data_convert_switch): ...new pass_data instance and...
2961 (make_pass_convert_switch): ...new function.
2962 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
2963 to a subclass of gimple_opt_pass along with...
2964 (pass_data_tail_recursion): ...new pass_data instance and...
2965 (make_pass_tail_recursion): ...new function.
2966 (pass_tail_calls): Convert from a global struct to a subclass of
2967 gimple_opt_pass along with...
2968 (pass_data_tail_calls): ...new pass_data instance and...
2969 (make_pass_tail_calls): ...new function.
2970 * tree-vect-generic.c (pass_lower_vector): Convert from a global
2971 struct to a subclass of gimple_opt_pass along with...
2972 (pass_data_lower_vector): ...new pass_data instance and...
2973 (make_pass_lower_vector): ...new function.
2974 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
2975 gimple_opt_pass along with...
2976 (pass_data_lower_vector_ssa): ...new pass_data instance and...
2977 (make_pass_lower_vector_ssa): ...new function.
2978 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
2979 to a subclass of gimple_opt_pass along with...
2980 (pass_data_slp_vectorize): ...new pass_data instance and...
2981 (make_pass_slp_vectorize): ...new function.
2982 (pass_ipa_increase_alignment): Convert from a global struct to a
2983 subclass of simple_ipa_opt_pass along with...
2984 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
2985 (make_pass_ipa_increase_alignment): ...new function.
2986 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
2987 gimple_opt_pass along with...
2988 (pass_data_vrp): ...new pass_data instance and...
2989 (make_pass_vrp): ...new function.
2990 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
2991 subclass of simple_ipa_opt_pass along with...
2992 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
2993 (make_pass_ipa_free_lang_data): ...new function.
2994 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
2995 gimple_opt_pass along with...
2996 (pass_data_tsan): ...new pass_data instance and...
2997 (make_pass_tsan): ...new function.
2998 (pass_tsan_O0): Convert from a global struct to a subclass of
2999 gimple_opt_pass along with...
3000 (pass_data_tsan_O0): ...new pass_data instance and...
3001 (make_pass_tsan_O0): ...new function.
3002 * var-tracking.c (pass_variable_tracking): Convert from a global
3003 struct to a subclass of rtl_opt_pass along with...
3004 (pass_data_variable_tracking): ...new pass_data instance and...
3005 (make_pass_variable_tracking): ...new function.
3006 * web.c (pass_web): Convert from a global struct to a subclass of
3007 rtl_opt_pass along with...
3008 (pass_data_web): ...new pass_data instance and...
3009 (make_pass_web): ...new function.
3010 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3011 declaration with that of...
3012 (make_pass_mode_switch_use): ...new function.
3013 (pass_resolve_sw_modes): Replace declaration with that of...
3014 (make_pass_resolve_sw_modes): ...new function.
3015 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3016 from a global struct to a subclass of rtl_opt_pass along with...
3017 (pass_data_mode_switch_use): ...new pass_data instance and...
3018 (make_pass_mode_switch_use): ...new function.
3019 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3020 from a global struct to a subclass of rtl_opt_pass along with...
3021 (pass_data_resolve_sw_modes): ...new pass_data instance and...
3022 (make_pass_resolve_sw_modes): ...new function.
3023 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3024 struct to a subclass of rtl_opt_pass along with...
3025 (pass_data_insert_vzeroupper): ...new pass_data instance and...
3026 (make_pass_insert_vzeroupper): ...new function.
3027 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3028 global struct to a subclass of rtl_opt_pass along with...
3029 (pass_data_work_around_errata): ...new pass_data instance and...
3030 (make_pass_work_around_errata): ...new function.
3031 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3032 struct to a subclass of rtl_opt_pass along with...
3033 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3034 (make_pass_mips_machine_reorg2): ...new function.
3036 2013-08-05 David Malcolm <dmalcolm@redhat.com>
3038 * passes.c (pass_manager::operator new): New.
3040 2013-08-05 David Malcolm <dmalcolm@redhat.com>
3042 Handwritten part of conversion of passes to C++ classes.
3044 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3045 (toplev.o): Add dep on PASS_MANAGER_H.
3046 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3047 of early local pases to reflect this moving from a global to a
3048 member of gcc::pass_manager.
3049 (cgraph_add_new_function): Likewise.
3050 * lto-cgraph.c (lto_output_node): Update for conversion of
3051 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3052 * passes.c (opt_pass::clone): New.
3053 (opt_pass::gate): New.
3054 (opt_pass::execute): New.
3055 (opt_pass::opt_pass): New.
3056 (pass_manager::execute_early_local_passes): New.
3057 (pass_manager::execute_pass_mode_switching): new.
3058 (finish_optimization_passes): Convert to...
3059 (pass_manager::finish_optimization_passes): ...this.
3060 (finish_optimization_passes): Update for conversion of passes to
3062 (register_dump_files_1): Use has_gate since we cannot portably
3063 check a vtable entry against NULL.
3064 (dump_one_pass): Likewise.
3065 (ipa_write_summaries_2): Likewise.
3066 (ipa_write_optimization_summaries_1): Likewise.
3067 (ipa_read_summaries_1): Likewise.
3068 (ipa_read_optimization_summaries_1): Likewise.
3069 (execute_ipa_stmt_fixups): Likewise.
3070 (pass_manager::pass_manager): Rewrite pass-creation, invoking
3071 pass-creation functions rather than wiring up globals, and
3072 storing the results in fields of pass_manager generated using
3074 (pass_manager::dump_profile_report): Update for conversion of
3075 passes to C++ classes.
3076 (pass_manager::execute_ipa_summary_passes): Likewise.
3077 (execute_one_ipa_transform_pass): Likewise.
3078 (execute_one_pass): Use has_gate and has_execute since we cannot
3079 portably check a vtable entry against NULL.
3080 * pass_manager.h (pass_manager::finish_optimization_passes): New.
3081 (pass_manager): Use pass-instances.def to add fields for the
3082 various pass instances.
3083 * toplev.c (finalize): Update for move of
3084 finish_optimization_passes to a method of gcc::pass_manager.
3085 * toplev.h (finish_optimization_passes): Move to method of class
3087 * tree-pass.h (struct pass_data): New.
3088 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3089 (opt_pass::gate): Convert to virtual function.
3090 (opt_pass::~opt_pass): New.
3091 (opt_pass::clone): New.
3092 (opt_pass::execute): Convert to virtual function.
3093 (opt_pass::opt_pass): New.
3094 (opt_pass::ctxt_): new.
3095 (gimple_opt_pass): Convert to subclass of opt_pass.
3096 (gimple_opt_pass::gimple_opt_pass): New.
3097 (rtl_opt_pass): Convert to subclass of opt_pass.
3098 (rtl_opt_pass::rtl_opt_pass): New.
3099 (ipa_opt_pass_d): Convert to subclass of opt_pass.
3100 (ipa_opt_pass_d::ipa_opt_pass_d): New.
3101 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3102 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3103 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3104 invocation of pass_mode_switching to reflect this moving from a
3105 global to a member of gcc::pass_manager.
3106 (ix86_option_override): Rework how pass_insert_vzeroupper is
3107 added to the pass_manager to reflect autogenerated changes.
3108 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3110 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
3112 PR rtl-optimization/57708
3113 * recog.c (peep2_find_free_register): Validate all regs in a
3116 2013-08-05 Jan Hubicka <jh@suse.cz>
3119 * cgraph.c (verify_cgraph_node): Accept local flags from other
3121 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3122 (function_and_variable_visibility): Likewise.
3123 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3125 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3127 * graph.c (init_graph_slim_pretty_print): Remove.
3128 (print_graph_cfg): Do not call it. Use local pretty printer.
3129 (start_graph_dump): Likewise.
3131 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3133 * gimple-pretty-print.c (buffer): Remove.
3134 (initialized): Likewise.
3135 (maybe_init_pretty_print): Likewise.
3136 (print_gimple_stmt): Do not call it. Use non-static local
3137 pretty_printer variable.
3138 (print_gimple_expr): Likewise.
3139 (print_gimple_seq): Likewise.
3140 (gimple_dump_bb): Likewise.
3142 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3144 * asan.c (asan_pp): Remove.
3145 (asan_pp_initialized): Likewise.
3146 (asan_pp_initialize): Likewise.
3147 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
3148 (asan_emit_stack_protection): Tidy. Use local pretty printer.
3149 (asan_add_global): Likewise.
3151 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
3153 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
3154 * diagnostic.h (diagnostic_flush_buffer): Adjust.
3155 * pretty-print.c (pp_formatted_text_data): Likewise.
3156 (pp_indent): Rename from pp_base_indent.
3157 (pp_format): Rename from pp_base_format.
3158 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3159 (pp_format_verbatim): Rename from pp_base_format_verbatim.
3160 (pp_flush): Rename from pp_base_flush.
3161 (pp_set_line_maximum_length): Rename from
3162 pp_base_set_line_maximum_length.
3163 (pp_clear_output_area): Rename from pp_base_clear_output_area.
3164 (pp_set_prefix): Rename from pp_base_set_prefix.
3165 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3166 (pp_emit_prefix): Rename from pp_base_emit_prefix.
3167 (pp_append_text): Rename from pp_base_append_text.
3168 (pp_formatted_text): Rename from pp_base_formatted_text.
3169 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3170 (pp_remaining_character_count_for_line): Rename from
3171 pp_base_remaining_character_count_for_line.
3172 (pp_newline): Rename from pp_base_newline.
3173 (pp_character): Rename from pp_base_character.
3174 (pp_string): Rename from pp_base_string.
3175 (pp_maybe_space): Rename from pp_base_maybe_space.
3176 * asan.c (asan_pp_string): Adjust.
3177 (asan_emit_stack_protection): Likewise.
3178 (asan_add_global): Likewise.
3179 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3180 * tree-mudflap.c (mf_varname_tree): Likewise.
3181 * tree-pretty-print.c (pp_tree_identifier): Rename from
3182 pp_base_tree_identifier.
3183 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3184 Declare as function.
3186 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3188 * pretty-print.h (pp_bar_bar): New.
3189 (pp_ampersand_ampersand): Likewise.
3190 (pp_less_equal): Likewise.
3191 (pp_greater_equal): Likewise.
3192 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3193 printer functions instead of pp_string or operators and punctuators.
3194 (dump_gimple_call): Likewise.
3195 (dump_gimple_omp_for): Likewise.
3196 (dump_gimple_transaction): Likewise.
3197 (dump_gimple_phi): Likewise.
3198 (pp_gimple_stmt_1): Likewise.
3199 * sched-vis.c (print_insn): Likewise.
3200 * tree-mudflap.c (mf_varname_tree): Likewise.
3201 * tree-pretty-print.c (dump_block_node): Likewise.
3202 (dump_generic_node): Likewise.
3204 2013-08-02 Jan Hubicka <jh@suse.cz>
3206 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3208 * lto-streamer-out.c (tree_is_indexable): Results decls and
3209 parm decls are not indexable.
3210 (DFS_write_tree_body): Do not follow args and results.
3211 (hash_tree): Likewise.
3212 (output_functions): Rearrange so struct function is needed
3213 only when real body is output; be able to also ouptut abstract
3214 functions; output DECL_ARGUMENTS and DECL_RESULT.
3215 (lto_output): When not in WPA, ale store abstract functions.
3216 (write_symbol): Do not care about RESULT_DECL.
3217 (output_symbol_p): Handle correctly sbtract decls.
3218 * lto-streamer-in.c (input_function): Rearrange so struct
3219 function can be NULL at entry; allow streaming of
3220 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3221 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3222 sanity check during LTO.
3223 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3224 RESULT_DECl and DECL_ARGUMENTS.
3225 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3228 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3230 * pretty-print.h (pp_underscore): New.
3232 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3233 printer functions instead of pp_character.
3234 (dump_binary_rhs): Likewise.
3235 (dump_ternary_rhs): Likewise.
3236 (dump_gimple_call_args): Likewise.
3237 (pp_points_to_solution): Likewise.
3238 (dump_gimple_call): Likewise.
3239 (dump_gimple_switch): Likewise.
3240 (dump_gimple_cond): Likewise.
3241 (dump_gimple_bind): Likewise.
3242 (dump_gimple_try): Likewise.
3243 (dump_gimple_omp_for): Likewise.
3244 (dump_gimple_omp_continue): Likewise.
3245 (dump_gimple_omp_single): Likewise.
3246 (dump_gimple_omp_sections): Likewise.
3247 (dump_gimple_omp_block): Likewise.
3248 (dump_gimple_omp_critical): Likewise.
3249 (dump_gimple_transaction): Likewise.
3250 (dump_gimple_asm): Likewise.
3251 (dump_gimple_phi): Likewise.
3252 (dump_gimple_omp_parallel): Likewise.
3253 (dump_gimple_omp_task): Likewise.
3254 (dump_gimple_omp_atomic_load): Likewise.
3255 (dump_gimple_omp_atomic_store): Likewise.
3256 (dump_gimple_mem_ops): Likewise.
3257 (pp_gimple_stmt_1): Likewise.
3258 (pp_cfg_jump): Likewise.
3259 (dump_implicit_edges): Likewise.
3260 (gimple_dump_bb_for_graph): Likewise.
3261 * graph.c (draw_cfg_node): Likewise.
3262 * langhooks.c (lhd_print_error_function): Likewise.
3263 * sched-vis.c (print_exp): Likewise.
3264 (print_value): Likewise.
3265 (print_pattern): Likewise.
3266 (print_insn): Likewise.
3267 (rtl_dump_bb_for_graph): Likewise.
3268 * tree-pretty-print.c (dump_function_declaration): Likewise.
3269 (dump_array_domain): Likewise.
3270 (dump_omp_clause): Likewise.
3271 (dump_location): Likewise.
3272 (dump_generic_node): Likewise.
3273 (print_struct_decl): Likewise.
3274 * diagnostic.c (diagnostic_show_locus): Use pp_space.
3276 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3278 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3279 candidate table when replacing a candidate statement.
3280 (replace_rhs_if_not_dup): Likewise.
3281 (replace_one_candidate): Likewise.
3283 2013-08-02 Jan Hubicka <jh@suse.cz>
3284 Martin Liska <marxin.liska@gmail.com>
3286 * cgraphunit.c (add_new_function): Fix logic when adding from
3288 (assemble_thunk): Rename to ...
3289 (expand_thunk); .. this one; export; get it working with
3290 general functions; make produced gimple valid.
3291 * cgraph.h (expand_thunk): Declare.
3293 2013-08-02 Jan Hubicka <jh@suse.cz>
3295 * ipa-cp.c (gather_context_independent_values): Use
3296 ipa_get_param_move_cost.
3297 (get_replacement_map): Remove PARAM; move parameter folding
3299 (create_specialized_node): Update.
3300 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3301 assert that we have gimple body; update move_cost.
3302 (count_formal_params): Assert that we have gimple body.
3303 (ipa_dump_param): New function.
3304 (ipa_alloc_node_params): Break out from ...
3305 (ipa_initialize_node_params): ... here.
3306 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3307 (ipa_write_node_info): Stream move costs.
3308 (ipa_read_node_info): Read move costs.
3309 (ipa_update_after_lto_read): Do not recompute node params.
3310 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3311 (ipa_get_param): Check we are not in WPA.
3312 (ipa_get_param_move_cost): New.
3313 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3314 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3315 parm numbers to be present.
3317 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
3319 PR rtl-optimization/58048
3320 * lra-constraints.c (process_alt_operands): Don't check asm
3321 operand on register.
3323 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
3325 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3326 the implied StoreLoad barrier for atomic operations if before.
3328 2013-08-02 Jan Hubicka <jh@suse.cz>
3329 Martin Liska <marxin.liska@gmail.com>
3331 * cgraph.c (cgraph_function_body_availability): Do not check
3333 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3334 symtab_node_availability): Declare.
3335 * ipa.c (can_replace_by_local_alias): New.
3336 (function_and_variable_visibility): Use it.
3337 * symtab.c (symtab_for_node_and_aliases,
3338 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3340 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
3342 PR rtl-optimization/57963
3343 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3344 (lra_constraints): Use them.
3346 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
3348 * config/arm/types.md (define_attr "type"): Add "load_acq"
3350 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3352 (cortex_a53_store1): Likewise.
3354 2013-08-01 Jan Hubicka <jh@suse.cz>
3356 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3357 partitions are not needed.
3359 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
3361 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3362 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3363 MAYBE_NON_Q_CLASS_P where appropriate.
3365 2013-08-01 Jan Hubicka <jh@suse.cz>
3367 * cgraph.h (release_function_body): Declare.
3368 * tree.c (free_lang_data_in_decl): Free, parameters and return values
3369 of unused delcarations.
3371 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3373 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3374 RTL form when subtracting a constant.
3376 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3378 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3379 Generate canonical plus rtx with negated immediate instead of minus
3381 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3383 2013-08-01 Jan Hubicka <jh@suse.cz>
3385 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3386 (cgraph_release_function_body): Likewise.
3387 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3388 * cgraph.h (cgrpah_node): Rename abstract_and_needed
3389 to used_as_abstract_origin.
3390 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3391 symbols used as abstract origins.
3392 * cgraphunit.c (analyze_functions): Update.
3393 * ipa.c (symtab_remove_unreachable_nodes): Recompute
3394 used_as_abstract_origin.
3395 * tree-inline.c (tree_function_versioning): Update
3396 used_as_abstract_origin; be ready for DECL_RESULT and
3397 DECL_ARGUMENTS to be NULL.
3399 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3400 for abstract functions.
3401 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3404 2013-08-01 Jan Hubicka <jh@suse.cz>
3406 * profile.c (compute_value_histograms): Fix thinko.
3408 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
3410 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
3411 aarch-common-protos.h to extra_headers.
3412 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3413 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3414 * config/aarch64/t-aarch64 (aarch-common.o): Define.
3416 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
3418 * config/aarch64/aarch64.md (define_attr "type"): Delete.
3419 Include "../arm/types.md". Define "type" attribute for all patterns.
3420 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3423 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3425 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3426 to support power8 load fusion.
3427 (fusion_gpr_mem_load): Likewise.
3429 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3431 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3432 declarations for power8 load fusion.
3433 (emit_fusion_gpr_load): Likewise.
3435 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3436 tuning for power8, turn on fusion mode by default. Turn on sign
3437 extending fusion mode if normal fusion mode is on, and we are at
3439 (fusion_gpr_load_p): New function, return true if we can fuse an
3440 addis instruction with a dependent load to a GPR.
3441 (emit_fusion_gpr_load): Emit the instructions for power8 load
3444 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3445 (VSX load fusion peepholes): New peepholes to fuse together an
3446 addi instruction with a VSX load instruction.
3448 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3449 peepholes to fuse an addis instruction with a load to a GPR base
3450 register. If we are supporting sign extending fusions, convert
3451 sign extending loads to zero extending loads and add an explicit
3454 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
3456 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
3457 aarch-common-protos.h to extra_headers.
3458 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3459 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3460 (arm_early_store_addr_dep): Likewise.
3461 (arm_no_early_alu_shift_dep): Likewise.
3462 (arm_no_early_alu_shift_value_dep): Likewise.
3463 (arm_no_early_mul_dep): Likewise.
3464 (arm_no_early_store_addr_dep): Likewise.
3465 (arm_mac_accumulator_is_mul_result): Likewise.
3466 (arm_mac_accumulator_is_result): Likewise.
3467 * config/arm/aarch-common.c: ... here. New file.
3468 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3470 (arm_early_store_addr_dep): Likewise.
3471 (arm_no_early_alu_shift_dep): Likewise.
3472 (arm_no_early_alu_shift_value_dep): Likewise.
3473 (arm_no_early_mul_dep): Likewise.
3474 (arm_no_early_store_addr_dep): Likewise.
3475 (arm_mac_accumulator_is_mul_result): Likewise.
3476 (arm_mac_accumulator_is_result): Likewise.
3477 * config/arm/aarch-common-protos.h: ... here. New file.
3478 * config/arm/t-arm (aarch-common.o): Define.
3480 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
3482 * config/arm/arm.md: Include new file "types.md".
3483 (define_attr "type"): Move from here to ...
3484 (define_attr "mul32"): Likewise.
3485 (define_attr "mul64"): Likewise.
3486 * config/arm/types.md: ... here. New file.
3488 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
3490 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3491 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3493 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3495 * gen-pass-instances.awk: Fix offset of substr().
3497 2013-07-31 David Malcolm <dmalcolm@redhat.com>
3499 * Makefile.in (pass-instances.def): New.
3500 (passes.o): Replace dependency on passes.def with one on
3503 * gen-pass-instances.awk: New.
3505 * passes.c (pass_manager::pass_manager): Use pass-instances.def
3506 rather than passes.def, updating local definition of NEXT_PASS
3507 macro to add an extra NUM parameter (currently unused).
3509 2013-07-30 David Malcolm <dmalcolm@redhat.com>
3511 * Makefile.in (PASS_MANAGER_H): New.
3512 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3513 (passes.o): Likewise.
3514 (statistics.o): Likewise.
3515 (cgraphunit.o): Likewise.
3516 (context.o): Depend on PASS_MANAGER_H.
3518 * pass_manager.h: New.
3520 * cgraphunit.c (cgraph_add_new_function): Update for moves
3521 of globals to fields of pass_manager.
3522 (analyze_function): Likewise.
3523 (expand_function): Likewise.
3524 (ipa_passes): Likewise.
3525 (compile): Likewise.
3527 * context.c (context::context): New.
3528 * context.h (context::context): New.
3529 (context::get_passes): New.
3530 (context::passes_): New.
3532 * lto-cgraph.c (input_node): Update for moves of globals to
3533 fields of pass_manager.
3535 * passes.c (all_passes): Remove, in favor of a field of the
3536 same name within the new class pass_manager.
3537 (all_small_ipa_passes): Likewise.
3538 (all_lowering_passes): Likewise.
3539 (all_regular_ipa_passes): Likewise.
3540 (all_late_ipa_passes): Likewise.
3541 (all_lto_gen_passes): Likewise.
3542 (passes_by_id): Likewise.
3543 (passes_by_id_size): Likewise.
3544 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3545 the new class pass_manager.
3546 (set_pass_for_id): Convert to...
3547 (pass_manager::set_pass_for_id): ...method.
3548 (get_pass_for_id): Convert to...
3549 (pass_manager::get_pass_for_id): ...method.
3550 (register_one_dump_file): Move body of implementation into...
3551 (pass_manager::register_one_dump_file): ...here.
3552 (register_dump_files_1): Convert to...
3553 (pass_manager::register_dump_files_1): ...method.
3554 (register_dump_files): Convert to...
3555 (pass_manager::register_dump_files): ...method.
3556 (create_pass_tab): Update for moves of globals to fields of
3558 (dump_passes): Move body of implementation into...
3559 (pass_manager::dump_passes): ...here.
3560 (register_pass): Move body of implementation into...
3561 (pass_manager::register_pass): ...here.
3562 (init_optimization_passes): Convert into...
3563 (pass_manager::pass_manager): ...constructor for new
3564 pass_manager class, and initialize the pass_lists array.
3565 (check_profile_consistency): Update for moves of globals to
3566 fields of pass_manager.
3567 (dump_profile_report): Move body of implementation into...
3568 (pass_manager::dump_profile_report): ...here.
3569 (ipa_write_summaries_1): Update for moves of pass lists from
3570 being globals to fields of pass_manager.
3571 (ipa_write_optimization_summaries): Likewise.
3572 (ipa_read_summaries): Likewise.
3573 (ipa_read_optimization_summaries): Likewise.
3574 (execute_all_ipa_stmt_fixups): Likewise.
3576 * statistics.c (statistics_fini): Update for moves of globals to
3577 fields of pass_manager.
3579 * toplev.c (general_init): Replace call to
3580 init_optimization_passes with construction of the pass_manager
3583 * tree-pass.h (all_passes): Remove, in favor of a field of the
3584 same name within the new class pass_manager.
3585 (all_small_ipa_passes): Likewise.
3586 (all_lowering_passes): Likewise.
3587 (all_regular_ipa_passes): Likewise.
3588 (all_lto_gen_passes): Likewise.
3589 (all_late_ipa_passes): Likewise.
3590 (passes_by_id): Likewise.
3591 (passes_by_id_size): Likewise.
3592 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3593 the new class pass_manager.
3594 (get_pass_for_id): Remove.
3596 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
3598 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
3601 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
3603 * arm.md (mulhi3): New expand pattern.
3605 2013-07-30 Jan Hubicka <jh@suse.cz>
3606 Martin Liska <marxin.liska@gmail.com>
3608 * profile.c (compute_value_histograms): Do not ICE when
3609 there is mismatch only on some counters.
3611 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
3613 PR rtl-optimization/57637
3614 * function.c (move_insn_for_shrink_wrap): Also check the
3615 GEN set of the LIVE problem for the liveness analysis
3616 if it exists, otherwise give up.
3618 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3620 PR tree-optimization/57993
3621 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
3622 replaced statement in the candidate table.
3623 (phi_add_costs): Return infinite cost when the hidden basis does
3624 not dominate all phis on which the candidate is dependent.
3625 (replace_one_candidate): Record replaced statement in the
3628 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
3630 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
3631 (ashlv2si3): New expander.
3632 (*ashlv2si3_i): New define_insn_and_split.
3633 * predicates.md (float_operation): Allow patterns with three
3636 PR rtl-optimization/58021
3637 * mode-switching.c (create_pre_exit): Always split off preceding
3638 insns if we are not at the basic block head.
3640 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
3642 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
3643 (UCLIBC_DYNAMIC_LINKER): New macro.
3644 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
3646 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
3647 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
3649 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
3650 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
3651 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
3652 for SF and DF modes. Use ieee_quad_format for TF mode.
3653 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
3654 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
3655 (mips_option_override): Handle `-mnan=legacy'.
3656 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
3657 `-mabs=2008' and `-mnan=2008'.
3658 (OPTION_DEFAULT_SPECS): Add "nan" default.
3659 (ASM_SPEC): Handle `-mnan='.
3660 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
3661 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
3662 comment accordingly.
3663 (neg<mode>2): Likewise.
3664 * config/mips/mips.opt (mabs, mnan): New options.
3665 * doc/install.texi (Configuration): Document `--with-nan=' option.
3666 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
3668 (MIPS Options): Document them.
3669 * config.gcc <mips*-*-*>: Handle `--with-nan='.
3670 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
3671 * configure: Regenerate.
3672 * config.in: Regenerate.
3674 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
3676 * config/i386/i386.md (float post-reload splitters): Do not check
3677 for subregs of SSE registers.
3679 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
3680 H.J. Lu <hongjiu.lu@intel.com>
3684 * config/i386/i386.md (post-reload splitter
3685 to avoid partial SSE reg dependency stalls): New pattern.
3687 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
3689 * config/s390/s390.md ("movcc"): Swap load and store instructions.
3691 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
3693 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
3694 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
3696 2013-07-26 Cary Coutant <ccoutant@google.com>
3698 * dwarf2out.c (die_checksum_ordered): Don't include template
3699 instantiations in signature.
3700 (is_template_parameter): New function.
3701 (is_template_instantiation): New function.
3702 (generate_skeleton_bottom_up): Don't include template instantiations
3704 (generate_skeleton): Likewise.
3705 (break_out_comdat_types): Move recursive call to break out nested
3707 (prune_unused_types_mark_generic_parms_dies): Call
3708 is_template_parameter.
3710 2013-07-26 Ian Bolton <ian.bolton@arm.com>
3712 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
3713 uses vector registers.
3714 * config/aarch64/iterators.md: Add attributes rtn and vas.
3716 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3717 Richard Earnshaw <richard.earnshaw@arm.com>
3719 * combine.c (simplify_comparison): Re-canonicalize operands
3721 * config/arm/arm.md (movcond_addsi): New splitter.
3723 2013-07-25 Sterling Augustine <saugustine@google.com>
3725 * dwarf2out.c (size_of_pubnames): Move code to...
3726 (include_pubname_in_output): ...here. New.
3727 (want_pubnames): Rearrange.
3728 (output_pubnames): Call include_pubname_in_output. Move assertion.
3730 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
3732 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
3734 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
3737 * doc/extend.texi: Remove obsolete builtins. Fix
3738 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
3740 2013-07-25 Jan Hubicka <jh@suse.cz>
3742 * cgraph.c (release_function_body): Break out from ...
3743 (cgraph_release_function_body): ... this one; also release DECL_RESULT
3745 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
3746 old_tree in the map.
3747 (create_specialized_node): Update.
3748 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
3750 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
3751 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
3752 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
3753 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
3756 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3758 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
3759 addsi3_carryin_alt2_<optab>): Correct output template.
3761 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3763 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
3764 Adjust for arm_restrict_it.
3765 Remove trailing whitespace.
3767 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
3769 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
3770 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
3772 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
3774 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
3776 PR rtl-optimization/57960
3777 * lra-constraints.c (process_alt_operands): Use the right mode
3778 when checking strict_low.
3780 2013-07-25 Jan Hubicka <jh@suse.cz>
3782 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
3783 * cgraph.c (cgraph_remove_node): Do not release function body
3784 when in cgraph streaming.
3785 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
3786 in other partitions are not considered reachable; fix handling of
3789 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3791 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
3793 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3798 * config/arm/arm.md ("*sibcall_insn): Replace use of
3799 Ss with US. Adjust output for v5 and v4t.
3800 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
3802 * config/arm/constraints.md ("Ss"): Rename to US.
3804 2013-07-25 Terry Guo <terry.guo@arm.com>
3806 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
3807 shift_add/shift_sub0/shift_sub1 RTXs.
3809 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
3810 Anton Blanchard <anton@au1.ibm.com>
3812 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
3813 (altivec_vpks<VI_char>ss): Likewise.
3814 (altivec_vpks<VI_char>us): Likewise.
3815 (altivec_vpku<VI_char>us): Likewise.
3816 (altivec_vpku<VI_char>um): Likewise.
3818 2013-07-24 David Malcolm <dmalcolm@redhat.com>
3820 Introduce context class.
3822 * Makefile.in (CONTEXT_H): New.
3823 (OBJS): Add context.o.
3824 (toplev.o): Add CONTEXT_H to dependencies.
3827 * toplev.c (general_init): Create the singleton gcc::context instance.
3833 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
3835 PR rtl-optimization/57968
3836 * mode-switching.c (create_pre_exit): Allow instructions that
3837 don't set a return register to need a non-exit mode.
3839 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3840 Anton Blanchard <anton@au1.ibm.com>
3842 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3843 operands to vperm for little endian.
3844 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3845 of lvsl to create the control mask for a vperm for little endian.
3847 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3848 Anton Blanchard <anton@au1.ibm.com>
3850 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3851 two operands for little-endian.
3853 2013-07-23 Steve Ellcey <sellcey@mips.com>
3855 * config/mips/mips.c (mips_case_values_threshold): New.
3856 (TARGET_CASE_VALUES_THRESHOLD): Define.
3858 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3859 Anton Blanchard <anton@au1.ibm.com>
3861 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
3862 selection of field for vector splat in little endian mode.
3864 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3866 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
3867 expanders to rs6000.md.
3868 (ior<mode>3): Likewise.
3869 (and<mode>3): Likewise.
3870 (one_cmpl<mode>2): Likewise.
3871 (nor<mode>3): Likewise.
3872 (andc<mode>3): Likewise.
3873 (eqv<mode>3): Likewise.
3874 (nand<mode>3): Likewise.
3875 (orc<mode>3): Likewise.
3877 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
3880 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
3881 to split multi-word logical operations.
3882 (rs6000_split_logical_di): Likewise.
3883 (rs6000_split_logical): Likewise.
3885 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
3886 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
3887 and allow TImode operations in 32-bit.
3888 (vsx_and<mode>3_64bit): Likewise.
3889 (vsx_ior<mode>3_32bit): Likewise.
3890 (vsx_ior<mode>3_64bit): Likewise.
3891 (vsx_xor<mode>3_32bit): Likewise.
3892 (vsx_xor<mode>3_64bit): Likewise.
3893 (vsx_one_cmpl<mode>2_32bit): Likewise.
3894 (vsx_one_cmpl<mode>2_64bit): Likewise.
3895 (vsx_nor<mode>3_32bit): Likewise.
3896 (vsx_nor<mode>3_64bit): Likewise.
3897 (vsx_andc<mode>3_32bit): Likewise.
3898 (vsx_andc<mode>3_64bit): Likewise.
3899 (vsx_eqv<mode>3_32bit): Likewise.
3900 (vsx_eqv<mode>3_64bit): Likewise.
3901 (vsx_nand<mode>3_32bit): Likewise.
3902 (vsx_nand<mode>3_64bit): Likewise.
3903 (vsx_orc<mode>3_32bit): Likewise.
3904 (vsx_orc<mode>3_64bit): Likewise.
3906 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
3907 logical types in GPRs.
3909 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
3910 logical insns to rs6000.md, and allow TImode operations in
3912 (altivec_ior<mode>3): Likewise.
3913 (altivec_xor<mode>3): Likewise.
3914 (altivec_one_cmpl<mode>2): Likewise.
3915 (altivec_nor<mode>3): Likewise.
3916 (altivec_andc<mode>3): Likewise.
3918 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
3919 attributes for moving the 128-bit logical operations into
3921 (BOOL_REGS_OUTPUT): Likewise.
3922 (BOOL_REGS_OP1): Likewise.
3923 (BOOL_REGS_OP2): Likewise.
3924 (BOOL_REGS_UNARY): Likewise.
3925 (BOOL_REGS_AND_CR0): Likewise.
3926 (one_cmpl<mode>2): Add support for DI logical operations on
3927 32-bit, splitting the operations to 32-bit.
3931 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
3932 changes to combine the 32/64-bit code, allow logical operations on
3933 TI mode in 32-bit, and to use similar match_operator patterns like
3934 scalar mode uses. Combine the Altivec and VSX code for logical
3935 operations, and move it here.
3936 (ior<mode>3, 128-bit types): Likewise.
3937 (xor<mode>3, 128-bit types): Likewise.
3938 (one_cmpl<mode>3, 128-bit types): Likewise.
3939 (nor<mode>3, 128-bit types): Likewise.
3940 (andc<mode>3, 128-bit types): Likewise.
3941 (eqv<mode>3, 128-bit types): Likewise.
3942 (nand<mode>3, 128-bit types): Likewise.
3943 (orc<mode>3, 128-bit types): Likewise.
3944 (and<mode>3_internal): Likewise.
3945 (bool<mode>3_internal): Likewise.
3946 (boolc<mode>3_internal1): Likewise.
3947 (boolc<mode>3_internal2): Likewise.
3948 (boolcc<mode>3_internal1): Likewise.
3949 (boolcc<mode>3_internal2): Likewise.
3950 (eqv<mode>3_internal1): Likewise.
3951 (eqv<mode>3_internal2): Likewise.
3952 (one_cmpl1<mode>3_internal): Likewise.
3954 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
3956 * config/microblaze/microblaze.c (microblaze_expand_prologue):
3957 Rename flag_stack_usage to flag_stack_usage_info.
3959 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
3961 * config/microblaze/sync.md: New file.
3962 * config/microblaze/microblaze.md: Include sync.md
3963 * config/microblaze/microblaze.c: Add print_operand 'y'.
3964 * config/microblaze/constraints.md: Add memory_contraint
3965 'Q' which is a single register.
3967 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
3969 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
3971 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
3973 * reload.c (find_reloads): Exit loop once we find this operand
3974 cannot be reloaded somehow for this alternative.
3976 * reload.c (find_reloads): Exit loop once we find a hard register.
3978 * rtlanal.c (computed_jump_p): Exit loop once we find label
3981 * i386.c (ix86_pad_returns): Exit loop after setting replace.
3983 * cfgloopmanip.c (remove_path): Exit loop after setting
3986 * gensupport.c (subst_dup): Avoid loop if code is not
3987 MATCH_DUP nor MATCH_OP_DUP.
3989 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
3991 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
3993 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
3995 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
3996 true for SP_REGNUM if mode == ptr_mode.
3997 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
3998 with value R0_REGNUM + 31.
4000 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4002 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4003 pad pointer-typed argument downward.
4005 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4007 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4008 and __ILP32__ when the ILP32 model is in use.
4010 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4012 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4013 (aarch64_load_symref_appropriately): In the case of
4014 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4015 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4016 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4017 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4018 if 'mode' doesn't equal to 'ptr_mode'.
4019 (aarch64_output_mi_thunk): Add an assertion on the alignment of
4020 'vcall_offset'; change to call aarch64_emit_move differently depending
4021 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4022 to calculate the upper bound of 'vcall_offset'.
4023 (aarch64_cannot_force_const_mem): Change to also return true if
4025 (aarch64_legitimize_reload_address): In the case of large
4026 displacements, add new local variable 'xmode' and an assertion
4027 based on it; change to use 'xmode' to generate the new rtx and
4029 (aarch64_asm_trampoline_template): Change to generate the template
4030 differently depending on TARGET_ILP32 or not; change to use
4031 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4032 (aarch64_trampoline_size): Removed.
4033 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4034 and replace immediate literals with it. Change to use 'ptr_mode'
4035 instead of 'DImode' and call convert_memory_address if the mode
4036 of 'fnaddr' doesn't equal to 'ptr_mode'.
4037 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4039 (aarch64_elf_asm_destructor): Likewise.
4040 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4041 on TARGET_ILP32 instead of aarch64_trampoline_size.
4042 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4043 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4044 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4045 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
4046 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4047 (storewb_pair<GPI:mode>_<P:mode>): ... this.
4048 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4049 depending on the value of 'mode'.
4050 (add_losym_<mode>): New.
4051 (ldr_got_small_<mode>): New, based on ldr_got_small.
4052 (ldr_got_small): Remove.
4053 (ldr_got_small_sidi): New.
4054 * config/aarch64/iterators.md (P): New.
4055 (PTR): Change to 'ptr_mode' in the condition.
4057 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4059 * config.gcc (aarch64*-*-*): Support --with-abi.
4060 (aarch64*-*-elf): Support --with-multilib-list.
4061 (aarch64*-*-linux*): Likewise.
4062 (supported_defaults): Add abi to aarch64*-*-*.
4063 * configure.ac: Mention AArch64 for --with-multilib-list.
4064 * configure: Re-generated.
4065 * config/aarch64/biarchilp32.h: New file.
4066 * config/aarch64/biarchlp64.h: New file.
4067 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4069 (MULTILIB_DEFAULTS): Ditto.
4070 (DRIVER_SELF_SPECS): Ditto.
4071 (ASM_SPEC): Update to also substitute -mabi.
4072 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4073 file whose name depends on -mabi= and -mbig-endian.
4074 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4076 (POINTER_SIZE): New define.
4077 (POINTERS_EXTEND_UNSIGNED): Ditto.
4078 (enum aarch64_abi_type): New enumeration tag.
4079 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4080 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4081 (TARGET_ILP32): New define.
4082 * config/aarch64/aarch64.opt (mabi): New.
4084 (ilp32, lp64): New values for -mabi.
4085 * config/aarch64/t-aarch64 (comma): New define.
4086 (MULTILIB_OPTIONS): Ditto.
4087 (MULTILIB_DIRNAMES): Ditto.
4088 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4089 * doc/invoke.texi: Document -mabi for AArch64.
4091 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
4093 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4095 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4096 Anton Blanchard <anton@au1.ibm.com>
4098 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4099 endianness when selecting field to splat.
4101 2013-07-22 Eric Christopher <echristo@gmail.com>
4103 * dwarf2out.c (die_odr_checksum): New function to use
4104 CHECKSUM_ macros and ULEB128 for DIE tag.
4105 (generate_type_signature): Use.
4107 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
4109 * config.gcc (sparc*-*-*): Accept leon3 processor.
4110 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4111 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4112 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4113 * config/sparc/sparc.opt (enum processor_type): Add leon3.
4114 (mfix-ut699): Adjust comment.
4115 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4116 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4117 (CPP_CPU_SPEC): Likewise.
4118 (ASM_CPU_SPEC): Likewise.
4119 * config/sparc/sparc.c (leon3_cost): New constant.
4120 (sparc_option_override): Add leon3 support.
4121 (mem_ref): New function.
4122 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4123 (sparc_do_work_around_errata): Look into the instruction in the delay
4124 slot and adjust accordingly. Add fix for the data cache nullify issues
4125 of the UT699. Change insertion position for the NOP.
4126 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4127 (leon3_load): New reservation.
4128 (leon_store): Bump latency to 2.
4129 (grfpu): New automaton.
4130 (grfpu_alu): New unit.
4131 (grfpu_ds): Likewise.
4132 (leon_fp_alu): Adjust.
4133 (leon_fp_mult): Delete.
4134 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4135 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4136 * config/sparc/sparc.md (cpu): Add leon3.
4137 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4139 (atomic_test_and_set): Likewise.
4142 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
4144 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
4145 default for R5900 targets.
4146 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4147 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4148 * config/mips/mips.c (mips_option_override): Report an error for
4149 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
4150 for -march=r5900 -mhard-float.
4152 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
4154 * df-problems.c (can_move_insns_across): Exit loop once we
4155 find a non-fixed, non-global register.
4157 * ipa-pure-const.c (propagate_nothrow): Exit loop after
4160 * omega.c (omega_eliminate_red): Break after setting red_found.
4161 (omega_problem_has_red_equations): Similarly after setting found.
4162 (omega_query_variable): Similarly after setting coupled.
4164 2013-07-22 Marek Polacek <polacek@redhat.com>
4166 * gimplify.c: Don't include gimple.h twice.
4168 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4170 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4171 instead of TARGET_THUMB1.
4172 (Pz): New constraint.
4173 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4175 (compare_negsi_si): Likewise.
4176 (compare_addsi2_op0): Likewise.
4177 (compare_addsi2_op1): Likewise.
4178 (addsi3_carryin_<optab>): Likewise.
4179 (addsi3_carryin_alt2_<optab>): Likewise.
4180 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4181 for arm_restrict_it.
4182 (subsi3_carryin): Likewise.
4183 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4184 (minmax_arithsi): Disable for arm_restrict_it.
4185 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4186 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4187 (satsi_<SAT:code>_shift): Likewise.
4188 (arm_shiftsi3): Add alternative for 16-bit encoding.
4189 (arm32_movhf): Disable for arm_restrict_it.
4190 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4191 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4193 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
4195 * config/arm/arm.md (attribute "insn"): Delete.
4196 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4197 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4198 (not_shiftsi): Update for attribute change.
4199 (not_shiftsi_compare0): Likewise.
4200 (not_shiftsi_compare0_scratch): Likewise.
4201 (arm_one_cmplsi2): Likewise.
4202 (thumb1_one_cmplsi2): Likewise.
4203 (notsi_compare0): Likewise.
4204 (notsi_compare0_scratch): Likewise.
4205 (thumb1_movdi_insn): Likewise.
4206 (arm_movsi_insn): Likewise.
4207 (movhi_insn_arch4): Likewise.
4208 (movhi_bytes): Likewise.
4209 (arm_movqi_insn): Likewise.
4210 (thumb1_movqi_insn): Likewise.
4211 (arm32_movhf): Likewise.
4212 (thumb1_movhf): Likewise.
4213 (arm_movsf_soft_insn): Likewise.
4214 (thumb1_movsf_insn): Likewise.
4215 (thumb_movdf_insn): Likewise.
4216 (movsicc_insn): Likewise.
4217 (movsfcc_soft_insn): Likewise.
4218 (and_scc): Likewise.
4219 (cond_move): Likewise.
4220 (if_move_not): Likewise.
4221 (if_not_move): Likewise.
4222 (if_shift_move): Likewise.
4223 (if_move_shift): Likewise.
4224 (if_shift_shift): Likewise.
4225 (if_not_arith): Likewise.
4226 (if_arith_not): Likewise.
4227 (cond_move_not): Likewise.
4228 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4229 (neon_mov<mode>): Likewise.
4230 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4231 (thumb2_movsi_vfp): Likewise.
4232 (movsf_vfp): Likewise.
4233 (thumb2_movsf_vfp): Likewise.
4234 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4236 (cortexa7_older_only): Likewise.
4237 (cortexa7_younger): Likewise.
4238 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4239 (1020alu_shift_op): Likewise.
4240 (1020alu_shift_reg_op): Likewise.
4241 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4242 (alu_shift_op): Likewise.
4243 (alu_shift_reg_op): Likewise.
4244 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4245 (11_alu_shift_op): Likewise.
4246 (11_alu_shift_reg_op): Likewise.
4247 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4248 (9_alu_shift_reg_op): Likewise.
4249 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4251 (cortex_a15_alu_shift): Likewise.
4252 (cortex_a15_alu_shift_reg): Likewise.
4253 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4254 (cortex_a5_alu_shift): Likewise.
4255 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4257 (cortex_a53_alu_shift): Likewise.
4258 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4260 (cortex_a7_alu_reg): Likewise.
4261 (cortex_a7_alu_shift): Likewise.
4262 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4263 (cortex_a8_alu_shift): Likewise.
4264 (cortex_a8_alu_shift_reg): Likewise.
4265 (cortex_a8_mov): Likewise.
4266 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4267 (cortex_a9_dp_shift): Likewise.
4268 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4269 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4270 (cortex_r4_mov): Likewise.
4271 (cortex_r4_alu_shift): Likewise.
4272 (cortex_r4_alu_shift_reg): Likewise.
4273 * config/arm/fa526.md (526_alu_op): Update for attribute change.
4274 (526_alu_shift_op): Likewise.
4275 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4276 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4277 (626te_alu_shift_op): Likewise.
4278 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4279 (726te_alu_op): Likewise.
4280 (726te_alu_shift_op): Likewise.
4281 (726te_alu_shift_reg_op): Likewise.
4282 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4283 (mp626_alu_shift_op): Likewise.
4284 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4285 (pj4_alu_e1_conds): Likewise.
4286 (pj4_alu): Likewise.
4287 (pj4_alu_conds): Likewise.
4288 (pj4_shift): Likewise.
4289 (pj4_shift_conds): Likewise.
4290 (pj4_alu_shift): Likewise.
4291 (pj4_alu_shift_conds): Likewise.
4293 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4295 * config/arm/predicates.md (shiftable_operator_strict_it):
4297 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4298 Disable cond_exec version for arm_restrict_it.
4299 (thumb2_smaxsi3): Convert to generate cond_exec.
4300 (thumb2_sminsi3): Likewise.
4301 (thumb32_umaxsi3): Likewise.
4302 (thumb2_uminsi3): Likewise.
4303 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4304 (thumb2_neg_abssi2): Likewise.
4305 (thumb2_mov_scc): Add alternative for 16-bit encoding.
4306 (thumb2_movsicc_insn): Adjust alternatives.
4307 (thumb2_mov_negscc): Disable for arm_restrict_it.
4308 (thumb2_mov_negscc_strict_it): New pattern.
4309 (thumb2_mov_notscc_strict_it): New pattern.
4310 (thumb2_mov_notscc): Disable for arm_restrict_it.
4311 (thumb2_ior_scc): Likewise.
4312 (thumb2_ior_scc_strict_it): New pattern.
4313 (thumb2_cond_move): Adjust for arm_restrict_it.
4314 (thumb2_cond_arith): Disable for arm_restrict_it.
4315 (thumb2_cond_arith_strict_it): New pattern.
4316 (thumb2_cond_sub): Adjust for arm_restrict_it.
4317 (thumb2_movcond): Likewise.
4318 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4319 (thumb2_zero_extendhisi2_v6): Likewise.
4320 (thumb2_zero_extendqisi2_v6): Likewise.
4321 (orsi_notsi_si): Likewise.
4322 (orsi_not_shiftsi_si): Likewise.
4324 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
4326 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4327 instruction sequence is 1 byte shorter.
4329 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
4331 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4332 it is not needed after split.
4334 2013-07-20 Iain Sandoe <iain@codesourcery.com>
4337 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4338 second label for nonlocal goto receivers. Don't output pic base labels
4339 unless we're producing PIC; mark that action unreachable().
4340 (ix86_save_reg): If the function contains a nonlocal label, save the
4342 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4343 * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4344 (update_pic_label_number_if_needed): New.
4345 (machopic_output_function_base_name): Adjust for nonlocal receiver
4347 (machopic_should_output_picbase_label): New.
4348 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4349 (nonlocal_goto_receiver): New insn and split.
4351 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
4353 * config/aarch64/aarch64-builtins.c
4354 (aarch64_fold_builtin): Fold abs in all modes.
4355 * config/aarch64/aarch64-simd-builtins.def
4356 (abs): Enable for all modes.
4357 * config/aarch64/arm_neon.h
4358 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4359 (vabs_f64): Add missing intrinsic.
4361 2013-07-19 Ian Bolton <ian.bolton@arm.com>
4363 * config/aarch64/arm_neon.h (vabs_s64): New function
4365 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
4368 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4369 * config/avr/avr.md (adjust_len): Add `round'.
4370 * config/avr/avr-protos.h (avr_out_round): New prototype.
4371 (avr_out_plus): Add `out_label' argument.
4372 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4373 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4374 Handle the case where `insn' is just a pattern.
4375 (avr_out_bitop): Handle the case where `insn' is just a pattern.
4376 (avr_out_round): New function.
4377 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4379 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
4381 * config/microblaze/microblaze.c (microblaze_expand_prologue):
4382 Add check for flag_stack_usage to handle -fstack-usage support
4384 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
4386 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4387 interaction for new Power8 flags and VSX.
4389 2013-07-18 Sriraman Tallam <tmsriram@google.com>
4392 * tree-inline.c (expand_call_inline): Emit errors during
4393 early_inlining only if optimization is not turned on.
4395 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4399 * passes.c (init_optimization_passes): Move the construction of
4400 the pass hierarchy into a new passes.def file.
4402 * Makefile.in (passes.o): Add dependency on passes.def.
4404 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4406 * passes.c (init_optimization_passes): Introduce macros for
4407 constructing the tree of passes (INSERT_PASSES_AFTER,
4408 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4409 TERMINATE_PASS_LIST).
4411 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
4412 Wei Mi <wmi@google.com>
4414 PR rtl-optimization/57878
4415 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4417 (reload_pseudo_compare_func): Check nregs first for reload
4420 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4422 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4424 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
4426 * read-rtl.c (validate_const_int): Once an invalid character is
4427 seen, quit the loop.
4429 * gengtype.c (write_roots): Similarly once we find the "deletable"
4430 or "if_marked" option.
4432 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
4434 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4435 "xtab" and "sat". Move value "clz" from here to ...
4436 (attriubte "type"): ... here.
4437 (satsi_<SAT:code>): Delete "insn" attribute.
4438 (satsi_<SAT:code>_shift): Likewise.
4439 (arm_zero_extendqisi2addsi): Likewise.
4440 (arm_extendqisi2addsi): Likewise.
4441 (clzsi2): Update for attribute changes.
4442 (rbitsi2): Likewise.
4443 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4445 (arm_usatsihi): Likewise.
4446 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4448 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
4450 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4451 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
4452 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
4453 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
4454 in alphabetical order.
4455 (attribute "core_cycles"): Update for attribute changes.
4456 (arm_addsi3): Likewise.
4457 (addsi3_compare0): Likewise.
4458 (addsi3_compare0_scratch): Likewise.
4459 (addsi3_compare_op1): Likewise.
4460 (addsi3_compare_op2): Likewise.
4461 (compare_addsi2_op0): Likewise.
4462 (compare_addsi2_op1): Likewise.
4463 (addsi3_carryin_shift_<optab>): Likewise.
4464 (subsi3_carryin_shift): Likewise.
4465 (rsbsi3_carryin_shift): Likewise.
4466 (arm_subsi3_insn): Likewise.
4467 (subsi3_compare0): Likewise.
4468 (subsi3_compare): Likewise.
4469 (arm_andsi3_insn): Likewise.
4470 (thumb1_andsi3_insn): Likewise.
4471 (andsi3_compare0): Likewise.
4472 (andsi3_compare0_scratch): Likewise.
4473 (zeroextractsi_compare0_scratch
4474 (andsi_not_shiftsi_si): Likewise.
4475 (iorsi3_insn): Likewise.
4476 (iorsi3_compare0): Likewise.
4477 (iorsi3_compare0_scratch): Likewise.
4478 (arm_xorsi3): Likewise.
4479 (thumb1_xorsi3_insn): Likewise.
4480 (xorsi3_compare0): Likewise.
4481 (xorsi3_compare0_scratch): Likewise.
4482 (satsi_<SAT:code>_shift): Likewise.
4484 (arm_shiftsi3): Likewise.
4485 (shiftsi3_compare0): Likewise.
4486 (not_shiftsi): Likewise.
4487 (not_shiftsi_compare0): Likewise.
4488 (not_shiftsi_compare0_scratch): Likewise.
4489 (arm_one_cmplsi2): Likewise.
4490 (thumb_one_complsi2): Likewise.
4491 (notsi_compare0): Likewise.
4492 (notsi_compare0_scratch): Likewise.
4493 (thumb1_zero_extendhisi2): Likewise.
4494 (arm_zero_extendhisi2): Likewise.
4495 (arm_zero_extendhisi2_v6): Likewise.
4496 (arm_zero_extendhisi2addsi): Likewise.
4497 (thumb1_zero_extendqisi2): Likewise.
4498 (thumb1_zero_extendqisi2_v6): Likewise.
4499 (arm_zero_extendqisi2): Likewise.
4500 (arm_zero_extendqisi2_v6): Likewise.
4501 (arm_zero_extendqisi2addsi): Likewise.
4502 (thumb1_extendhisi2): Likewise.
4503 (arm_extendhisi2): Likewise.
4504 (arm_extendhisi2_v6): Likewise.
4505 (arm_extendqisi): Likewise.
4506 (arm_extendqisi_v6): Likewise.
4507 (arm_extendqisi2addsi): Likewise.
4508 (thumb1_extendqisi2): Likewise.
4509 (thumb1_movdi_insn): Likewise.
4510 (arm_movsi_insn): Likewise.
4511 (movsi_compare0): Likewise.
4512 (movhi_insn_arch4): Likewise.
4513 (movhi_bytes): Likewise.
4514 (arm_movqi_insn): Likewise.
4515 (thumb1_movqi_insn): Likewise.
4516 (arm32_movhf): Likewise.
4517 (thumb1_movhf): Likewise.
4518 (arm_movsf_soft_insn): Likewise.
4519 (thumb1_movsf_insn): Likewise.
4520 (movdf_soft_insn): Likewise.
4521 (thumb_movdf_insn): Likewise.
4522 (arm_cmpsi_insn): Likewise.
4523 (cmpsi_shiftsi): Likewise.
4524 (cmpsi_shiftsi_swp): Likewise.
4525 (arm_cmpsi_negshiftsi_si): Likewise.
4526 (movsicc_insn): Likewise.
4527 (movsfcc_soft_insn): Likewise.
4528 (arith_shiftsi): Likewise.
4529 (arith_shiftsi_compare0
4530 (arith_shiftsi_compare0_scratch
4531 (sub_shiftsi): Likewise.
4532 (sub_shiftsi_compare0
4533 (sub_shiftsi_compare0_scratch
4534 (and_scc): Likewise.
4535 (cond_move): Likewise.
4536 (if_plus_move): Likewise.
4537 (if_move_plus): Likewise.
4538 (if_move_not): Likewise.
4539 (if_not_move): Likewise.
4540 (if_shift_move): Likewise.
4541 (if_move_shift): Likewise.
4542 (if_shift_shift): Likewise.
4543 (if_not_arith): Likewise.
4544 (if_arith_not): Likewise.
4545 (cond_move_not): Likewise.
4546 (thumb1_ashlsi3): Set type attribute.
4547 (thumb1_ashrsi3): Likewise.
4548 (thumb1_lshrsi3): Likewise.
4549 (thumb1_rotrsi3): Likewise.
4550 (shiftsi3_compare0_scratch): Likewise.
4551 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4552 (neon_mov<mode>): Likewise.
4553 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4555 (thumb2_movsi_insn): Likewise.
4556 (thumb2_cmpsi_neg_shiftsi): Likewise.
4557 (thumb2_extendqisi_v6): Likewise.
4558 (thumb2_zero_extendhisi2_v6): Likewise.
4559 (thumb2_zero_extendqisi2_v6): Likewise.
4560 (thumb2_shiftsi3_short): Likewise.
4561 (thumb2_addsi3_compare0_scratch): Likewise.
4562 (orsi_not_shiftsi_si): Likewise.
4563 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4564 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4566 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4567 (1020alu_shift_op): Likewise.
4568 (1020alu_shift_reg_op): Likewise.
4569 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4570 (alu_shift_op): Likewise.
4571 (alu_shift_reg_op): Likewise.
4572 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4573 (11_alu_shift_op): Likewise.
4574 (11_alu_shift_reg_op): Likewise.
4575 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4576 (9_alu_shift_reg_op): Likewise.
4577 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4579 (cortex_a15_alu_shift): Likewise.
4580 (cortex_a15_alu_shift_reg): Likewise.
4581 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4583 (cortex_a5_alu_shift): Likewise.
4584 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4586 (cortex_a53_alu_shift): Likewise.
4587 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4589 (cortex_a7_alu_reg): Likewise.
4590 (cortex_a7_alu_shift): Likewise.
4591 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4593 (cortex_a8_alu_shift): Likewise.
4594 (cortex_a8_alu_shift_reg): Likewise.
4595 (cortex_a8_mov): Likewise.
4596 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
4597 (cortex_a9_dp_shift): Likewise.
4598 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
4600 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
4602 (cortex_r4_mov): Likewise.
4603 (cortex_r4_alu_shift): Likewise.
4604 (cortex_r4_alu_shift_reg): Likewise.
4605 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
4606 (526_alu_shift_op): Likewise.
4607 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
4608 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
4609 (626te_alu_shift_op): Likewise.
4610 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
4611 (726te_alu_op): Likewise.
4612 (726te_alu_shift_op): Likewise.
4613 (726te_alu_shift_reg_op): Likewise.
4614 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
4615 (mp626_alu_shift_op): Likewise.
4616 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
4617 (pj4_alu_e1_conds): Likewise.
4618 (pj4_alu): Likewise.
4619 (pj4_alu_conds): Likewise.
4620 (pj4_shift): Likewise.
4621 (pj4_shift_conds): Likewise.
4622 (pj4_alu_shift): Likewise.
4623 (pj4_alu_shift_conds): Likewise.
4624 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4626 (cortexa7_older_only): Likewise.
4627 (cortexa7_younger): Likewise.
4629 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4631 * ipa-pure-const.c (generate_summary): Rename to...
4632 (pure_const_generate_summary): ... this.
4634 2013-07-17 Iain Sandoe <iain@codesourcery.com>
4636 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
4638 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
4641 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
4644 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4646 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
4647 enabled without -march=zEC12.
4648 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
4651 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
4653 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
4654 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
4655 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
4656 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
4657 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
4658 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
4659 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
4661 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
4662 and ISA_HAS_NMADD3_NMSUB3.
4663 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
4664 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
4665 (nmsub4<mode>, nmsub3<mode>): Likewise.
4666 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
4668 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
4670 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
4671 TARGET_MIPS5400 checking.
4673 2013-07-16 Jakub Jelinek <jakub@redhat.com>
4674 Peter Bergner <bergner@vnet.ibm.com>
4676 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
4677 registers in the comment.
4678 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
4679 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
4680 rather than FIRST_PSEUDO_REGISTERS.
4682 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
4684 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
4685 enable extra ISA flags with TARGET_HTM.
4687 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
4689 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
4692 2013-07-15 Cong Hou <congh@google.com>
4694 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
4695 in compare function for sorting.
4697 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
4699 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
4700 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
4701 * config/rs6000/rs6000.opt: Add -mhtm option.
4702 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
4703 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
4704 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
4705 __HTM__ if the HTM instructions are available.
4706 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
4707 htm_spr_reg_operand): New define_predicates.
4708 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
4709 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
4711 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
4712 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
4713 HTM builtin functions.
4714 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
4715 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
4716 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
4717 (rs6000_builtin_mask_calculate): Likewise.
4718 (rs6000_option_override_internal): Likewise.
4719 (bdesc_htm): Add new HTM builtin support.
4720 (htm_spr_num): New function.
4721 (htm_spr_regno): Likewise.
4722 (rs6000_htm_spr_icode): Likewise.
4723 (htm_expand_builtin): Likewise.
4724 (htm_init_builtins): Likewise.
4725 (rs6000_expand_builtin): Add support for HTM builtin functions.
4726 (rs6000_init_builtins): Likewise.
4727 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
4729 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
4730 (TARGET_HTM, MASK_HTM): Define macros.
4731 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
4732 (FIXED_REGISTERS): Likewise.
4733 (CALL_USED_REGISTERS): Likewise.
4734 (CALL_REALLY_USED_REGISTERS): Likewise.
4735 (REG_ALLOC_ORDER): Likewise.
4736 (enum reg_class): Likewise.
4737 (REG_CLASS_NAMES): Likewise.
4738 (REG_CLASS_CONTENTS): Likewise.
4739 (REGISTER_NAMES): Likewise.
4740 (ADDITIONAL_REGISTER_NAMES): Likewise.
4741 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
4742 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
4743 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
4744 * config/rs6000/htm.md: New file.
4745 * config/rs6000/htmintrin.h: New file.
4746 * config/rs6000/htmxlintrin.h: New file.
4748 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
4750 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4751 Define SYMBOL_TINY_GOT, update comment.
4752 * config/aarch64/aarch64.c
4753 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
4754 (aarch64_expand_mov_immediate): Likewise.
4755 (aarch64_print_operand): Likewise.
4756 (aarch64_classify_symbol): Likewise.
4757 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
4758 (ldr_got_tiny): Define.
4760 2013-07-13 Tobias Grosser <tobias@grosser.es>
4762 PR tree-optimization/54094
4763 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
4764 scheduling dimension for the parallelism check from the polyhedral
4765 information in the AST.
4766 * graphite-dependences.c (carries_deps): Do not assume the schedule is
4769 2013-07-13 Jason Merrill <jason@redhat.com>
4771 * print-tree.c (debug_vec_tree): Use debug_raw.
4772 (debug_raw (vec<tree, va_gc> &)): New.
4773 (debug_raw (vec<tree, va_gc> *)): New.
4774 * tree.h: Declare them.
4776 2013-07-13 Bin Cheng <bin.cheng@arm.com>
4778 * ifcvt.c (ifcvt_after_combine): New static variable.
4779 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
4781 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
4782 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
4783 rest_of_handle_if_after_reload): Pass new argument for if_convert.
4785 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
4787 * config/mips/mips.c (mips_expand_call): Remove empty statement.
4789 2013-07-12 Michael Matz <matz@suse.de>
4792 * convert.c (convert_to_real): Reject non-float inner types.
4794 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
4796 * config/aarch64/aarch64-protos.h
4797 (aarch64_simd_immediate_valid_for_move): Remove.
4798 * config/aarch64/aarch64.c (simd_immediate_info): New member.
4799 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
4801 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
4803 2013-07-11 Steve Ellcey <sellcey@mips.com>
4805 * config/mips/mips.c (mips_conditional_register_usage): Do not
4806 use t[0-7] registers in MIPS16 mode when optimizing for size.
4808 2013-07-11 Sriraman Tallam <tmsriram@google.com>
4810 * config/i386/i386.c (dispatch_function_versions): Fix array
4811 indexing of function_version_info to match actual_versions.
4813 2013-07-11 Teresa Johnson <tejohnson@google.com>
4815 * vec.h (struct va_gc): Move release out-of-line.
4816 (va_gc::release): Call ggc_free on released vec.
4818 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
4820 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
4821 Require GOT register as additional operand in UNSPEC.
4822 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
4823 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4824 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4825 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
4826 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
4827 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
4828 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
4830 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
4833 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
4834 name seen by assembler/linker rather if available.
4836 2013-07-11 Andreas Schwab <schwab@suse.de>
4838 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4840 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
4842 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4844 2013-07-10 Joseph Myers <joseph@codesourcery.com>
4846 * doc/tm.texi.in: Move hook documentation to ....
4847 * target.def: ... here.
4849 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4851 * doc/tm.texi: Regenerate.
4853 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
4856 * doc/invoke.texi: Document -Wconditionally-supported.
4858 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4861 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
4864 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4867 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
4868 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
4869 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
4870 Remove duplicate devices.
4871 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
4872 * config/avr/t-multilib: Regenerate.
4873 * config/avr/avr-tables.opt: Regenerate.
4874 * doc/avr-mmcu.texi: Regenerate.
4876 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
4879 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
4881 2013-07-10 Graham Stott <graham.stott@btinternet.com>
4883 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
4884 the cost of MULT when optimizing for size.
4886 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4888 * config/cr16/cr16-protos.h: Don't include target.h.
4890 2013-07-09 Joseph Myers <joseph@codesourcery.com>
4892 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
4893 adjust register size for TDmode and TFmode for VSX registers.
4895 2013-07-08 Kai Tietz <ktietz@redhat.com>
4898 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
4899 hook_bool_const_tree_true.
4901 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4903 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
4904 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
4905 * config/s390/s390.md: Define FPR*_REGNUM constants.
4906 Fix FPR2_REGNUM constant (18 -> 17).
4907 ("*trunc<BFP:mode><DFP_ALL:mode>2")
4908 ("*trunc<DFP_ALL:mode><BFP:mode>2")
4909 ("trunc<BFP:mode><DFP_ALL:mode>2")
4910 ("trunc<DFP_ALL:mode><BFP:mode>2")
4911 ("*extend<BFP:mode><DFP_ALL:mode>2")
4912 ("*extend<DFP_ALL:mode><BFP:mode>2")
4913 ("extend<BFP:mode><DFP_ALL:mode>2")
4914 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
4917 2013-07-08 Graham Stott <graham.stott@btinternet.com>
4919 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
4921 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4923 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
4924 and cfun_fpr_bit_p to cfun_fpr_save_p.
4925 (s390_frame_area, s390_register_info, s390_frame_info)
4926 (s390_emit_prologue, s390_emit_epilogue)
4927 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
4929 * config/s390/s390.h: Define *_REGNUM macros for floating point
4932 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
4934 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
4936 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
4938 PR rtl-optimization/57786
4939 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
4940 and break out of the loop when it is set to false.
4942 2013-07-08 Jakub Jelinek <jakub@redhat.com>
4945 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
4946 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
4947 (const_int 63)) 0)).
4948 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
4949 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
4950 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
4952 PR rtl-optimization/57829
4953 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
4954 mask bits outside of mode are just sign-extension from mode to HWI.
4956 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
4958 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
4959 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
4960 adjust_address instead of change_address to keep info about alignment.
4961 (emit_strmov): Remove.
4962 (emit_memmov): New function.
4963 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
4964 (expand_movmem_epilogue): Likewise and return updated rtx for
4966 (expand_constant_movmem_prologue): Likewise and return updated rtx for
4967 destination and source.
4968 (decide_alignment): Refactor, handle vector_loop.
4969 (ix86_expand_movmem): Likewise.
4970 (ix86_expand_setmem): Likewise.
4971 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
4973 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
4975 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
4976 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
4978 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
4980 * config/i386/sse.md (sse_movlhps): Change alternative 3
4981 of operand 2 to "m".
4983 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
4986 * config/i386/sse.md (iptr): New mode attribute.
4987 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
4988 (<sse>_vm<plusminus_insn><mode>3): Ditto.
4989 (<sse>_vmmul<mode>3): Ditto.
4990 (<sse>_vmdiv<mode>3): Ditto.
4991 (sse_vmrcpv4sf2): Ditto.
4992 (<sse>_vmsqrt<mode>2): Ditto.
4993 (sse_vmrsqrtv4sf2): Ditto.
4994 (<sse>_vm<code><mode>3): Ditto.
4995 (avx_vmcmp<mode>3): Ditto.
4996 (<sse>_vmmaskcmp<mode>3): Ditto.
4997 (<sse>_comi): Ditto.
4998 (<sse>_ucomi): Ditto.
4999 (*xop_vmfrcz_<mode>): Ditto.
5000 (*fmai_fmadd_<mode>): Ditto.
5001 (*fmai_fmsub_<mode>): Ditto.
5002 (*fmai_fnmadd_<mode>): Ditto.
5003 (*fmai_fnmsub_<mode>): Ditto.
5004 (*fma4i_vmfmadd_<mode>): Ditto.
5005 (*fma4i_vmfmsub_<mode>): Ditto.
5006 (*fma4i_vmfnmadd_<mode>): Ditto.
5007 (*fma4i_vmfnmsub_<mode>): Ditto.
5008 (*xop_vmfrcz_<mode>): Ditto.
5009 (sse_cvtps2pi): Ditto.
5010 (sse_cvttps2pi): Ditto.
5011 (sse_cvtss2si): Ditto.
5012 (sse_cvtss2si_2): Ditto.
5013 (sse_cvtss2siq_2): Ditto.
5014 (sse_cvttss2si): Ditto.
5015 (sse_cvttss2siq): Ditto.
5016 (sse_cvtsd2si): Ditto.
5017 (sse_cvtsd2si_2): Ditto.
5018 (sse_cvtsd2siq_2): Ditto.
5019 (sse_cvttsd2si): Ditto.
5020 (sse_cvttsd2siq): Ditto.
5021 (sse_cvtsd2ss): Ditto.
5022 (sse_cvtss2sd): Ditto.
5023 (avx2_pbroadcast<mode>): Ditto.
5024 (avx2_pbroadcast<mode>_1): Ditto.
5025 (*avx_vperm_broadcast_v4sf): Ditto.
5027 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5028 (sse_movlhps): Ditto.
5029 (sse_storehps): Ditto.
5030 (sse_loadhps): Ditto.
5031 (sse_storelps): Ditto.
5032 (sse_loadlps): Ditto.
5033 (*vec_concatv4sf): Ditto.
5034 (*vec_interleave_highv2df): Ditto.
5035 (*vec_interleave_lowv2df): Ditto.
5036 (*vec_extractv2df_1_sse): Ditto.
5037 (*vec_extractv2df_0_sse): Ditto.
5038 (sse2_storelpd): Ditto.
5039 (sse2_loadlpd): Ditto.
5040 (sse2_movsd): Ditto.
5041 (*vec_concatv4si): Ditto.
5042 (vec_concatv2di): Ditto.
5044 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5045 for Intel asm dialect.
5046 (mmx_punpcklwd): Ditto.
5047 (mmx_punpckldq): Ditto.
5049 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5050 for intel assembler dialect.
5052 2013-07-06 Jakub Jelinek <jakub@redhat.com>
5055 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5056 for BUILT_IN_C{LZ,LRSB}*.
5057 * tree.h (CASE_INT_FN): Add FN##IMAX case.
5058 * tree-vrp.c (extract_range_basic): Handle
5059 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
5060 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5061 fall thru to code calling set_value*.
5062 * builtins.c (expand_builtin): Remove *IMAX cases.
5063 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5064 if width is bigger than 2*HWI.
5066 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
5068 PR rtl-optimization/55342
5069 * lra-int.h (lra_subreg_reload_pseudos): New.
5070 * lra.c: Add undoing optional reloads to the block diagram.
5071 (lra_subreg_reload_pseudos): New.
5072 (lra_optional_reload_pseudos): Change comments.
5073 (lra): Init and clear lra_subreg_reload_pseudos. Clear
5074 lra_optional_reload_pseudos after undo transformations.
5075 * lra-assigns.c (pseudo_prefix_title): New.
5076 (lra_setup_reg_renumber): Use it.
5077 (spill_for): Ditto. Check subreg reload pseudos too.
5078 (assign_by_spills): Consider subreg reload pseudos too.
5079 * lra-constraints.c (simplify_operand_subreg): Use
5080 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5081 (curr_insn_transform): Recognize and do optional reloads.
5082 (undo_optional_reloads): New.
5083 (lra_undo_inheritance): Call undo_optional_reloads.
5085 2013-07-05 Thomas Quinot <quinot@adacore.com>
5087 * tree-complex.c (expand_complex_operations_1): Fix typo.
5089 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
5091 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5092 (tune_params): New member 'const vec_costs'.
5093 * config/aarch64/aarch64.c (generic_vector_cost): New.
5094 (generic_tunings): New member 'generic_vector_cost'.
5095 (aarch64_builtin_vectorization_cost): New.
5096 (aarch64_add_stmt_cost): New.
5097 (TARGET_VECTORIZE_ADD_STMT_COST): New.
5098 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5100 2013-07-03 Jakub Jelinek <jakub@redhat.com>
5103 * config/i386/predicates.md (vsib_address_operand): Disallow
5104 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5106 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
5109 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5110 expand_builtin_setjmp_receiver.
5111 (expand_label): Adjust, call expand_builtin_setjmp_receiver
5112 with NULL for the label parameter.
5113 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5114 the frame-pointer. Adjust comments.
5115 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5116 only if LABEL is non-NULL.
5118 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
5120 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5121 (ARM_ABI_AAPCS64): Ditto.
5123 (ARM_DEFAULT_ABI): Ditto.
5125 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
5127 * config/aarch64/aarch64-builtins.c
5128 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5129 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5131 * config/aarch64/aarch64-simd.md
5132 (aarch64_ld1<VALL:mode>): New.
5133 (aarch64_st1<VALL:mode>): Likewise.
5134 * config/aarch64/arm_neon.h
5135 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5137 2013-07-02 Sriraman Tallam <tmsriram@google.com>
5139 * config/i386/i386.c (gate_insert_vzeroupper): Check if
5141 (ix86_option_override_internal):Turn on all -mavx target flags by
5142 default as they are dependent on AVX anyway.
5144 2013-07-02 Cary Coutant <ccoutant@google.com>
5146 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5148 (loc_checksum_ordered): Likewise.
5149 (hash_loc_operands): Remove inline keyword.
5151 2013-07-02 Jakub Jelinek <jakub@redhat.com>
5153 PR tree-optimization/57741
5154 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5155 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5156 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5157 Allow REAL_CST step_exprs if flag_associative_math.
5158 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5160 2013-07-02 Ian Bolton <ian.bolton@arm.com>
5162 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5164 2013-07-02 Ian Bolton <ian.bolton@arm.com>
5166 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5168 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5170 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5172 (iorsi3_insn): Likewise.
5173 (arm_xorsi3): Likewise.
5175 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
5177 * arm.md (attribute "wtype"): Delete. Move attribute values from here
5179 (attribute "type"): ... here, and prefix with "wmmx_".
5180 (attribute "core_cycles"): Update for attribute changes.
5181 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5182 (tbcstv4hi): Likewise.
5183 (tbcstv2si): Likewise.
5184 (iwmmxt_iordi3): Likewise.
5185 (iwmmxt_xordi3): Likewise.
5186 (iwmmxt_anddi3): Likewise.
5187 (iwmmxt_nanddi3): Likewise.
5188 (iwmmxt_arm_movdi): Likewise.
5189 (iwmmxt_movsi_insn): Likewise.
5190 (mov<mode>_internal): Likewise.
5191 (and<mode>3_iwmmxt): Likewise.
5192 (ior<mode>3_iwmmxt): Likewise.
5193 (xor<mode>3_iwmmxt): Likewise.
5194 (add<mode>3_iwmmxt): Likewise.
5195 (ssaddv8qi3): Likewise.
5196 (ssaddv4hi3): Likewise.
5197 (ssaddv2si3): Likewise.
5198 (usaddv8qi3): Likewise.
5199 (usaddv4hi3): Likewise.
5200 (usaddv2si3): Likewise.
5201 (sub<mode>3_iwmmxt): Likewise.
5202 (sssubv8qi3): Likewise.
5203 (sssubv4hi3): Likewise.
5204 (sssubv2si3): Likewise.
5205 (ussubv8qi3): Likewise.
5206 (ussubv4hi3): Likewise.
5207 (ussubv2si3): Likewise.
5208 (mulv4hi3_iwmmxt): Likewise.
5209 (smulv4hi3_highpart): Likewise.
5210 (umulv4hi3_highpart): Likewise.
5211 (iwmmxt_wmacs): Likewise.
5212 (iwmmxt_wmacsz): Likewise.
5213 (iwmmxt_wmacu): Likewise.
5214 (iwmmxt_wmacuz): Likewise.
5215 (iwmmxt_clrdi): Likewise.
5216 (iwmmxt_clrv8qi): Likewise.
5217 (iwmmxt_clr4hi): Likewise.
5218 (iwmmxt_clr2si): Likewise.
5219 (iwmmxt_uavgrndv8qi3): Likewise.
5220 (iwmmxt_uavgrndv4hi3): Likewise.
5221 (iwmmxt_uavgv8qi3): Likewise.
5222 (iwmmxt_uavgv4hi3): Likewise.
5223 (iwmmxt_tinsrb): Likewise.
5224 (iwmmxt_tinsrh): Likewise.
5225 (iwmmxt_tinsrw): Likewise.
5226 (iwmmxt_textrmub): Likewise.
5227 (iwmmxt_textrmsb): Likewise.
5228 (iwmmxt_textrmuh): Likewise.
5229 (iwmmxt_textrmsh): Likewise.
5230 (iwmmxt_textrmw): Likewise.
5231 (iwmxxt_wshufh): Likewise.
5232 (eqv8qi3): Likewise.
5233 (eqv4hi3): Likewise.
5234 (eqv2si3): Likewise.
5235 (gtuv8qi3): Likewise.
5236 (gtuv4hi3): Likewise.
5237 (gtuv2si3): Likewise.
5238 (gtv8qi3): Likewise.
5239 (gtv4hi3): Likewise.
5240 (gtv2si3): Likewise.
5241 (smax<mode>3_iwmmxt): Likewise.
5242 (umax<mode>3_iwmmxt): Likewise.
5243 (smin<mode>3_iwmmxt): Likewise.
5244 (umin<mode>3_iwmmxt): Likewise.
5245 (iwmmxt_wpackhss): Likewise.
5246 (iwmmxt_wpackwss): Likewise.
5247 (iwmmxt_wpackdss): Likewise.
5248 (iwmmxt_wpackhus): Likewise.
5249 (iwmmxt_wpackwus): Likewise.
5250 (iwmmxt_wpackdus): Likewise.
5251 (iwmmxt_wunpckihb): Likewise.
5252 (iwmmxt_wunpckihh): Likewise.
5253 (iwmmxt_wunpckihw): Likewise.
5254 (iwmmxt_wunpckilb): Likewise.
5255 (iwmmxt_wunpckilh): Likewise.
5256 (iwmmxt_wunpckilw): Likewise.
5257 (iwmmxt_wunpckehub): Likewise.
5258 (iwmmxt_wunpckehuh): Likewise.
5259 (iwmmxt_wunpckehuw): Likewise.
5260 (iwmmxt_wunpckehsb): Likewise.
5261 (iwmmxt_wunpckehsh): Likewise.
5262 (iwmmxt_wunpckehsw): Likewise.
5263 (iwmmxt_wunpckelub): Likewise.
5264 (iwmmxt_wunpckeluh): Likewise.
5265 (iwmmxt_wunpckeluw): Likewise.
5266 (iwmmxt_wunpckelsb): Likewise.
5267 (iwmmxt_wunpckelsh): Likewise.
5268 (iwmmxt_wunpckelsw): Likewise.
5269 (ror<mode>3): Likewise.
5270 (ashr<mode>3_iwmmxt): Likewise.
5271 (lshr<mode>3_iwmmxt): Likewise.
5272 (ashl<mode>3_iwmmxt): Likewise.
5273 (ror<mode>3_di): Likewise.
5274 (ashr<mode>3_di): Likewise.
5275 (lshr<mode>3_di): Likewise.
5276 (ashl<mode>3_di): Likewise.
5277 (iwmmxt_wmadds): Likewise.
5278 (iwmmxt_wmaddu): Likewise.
5279 (iwmmxt_tmia): Likewise.
5280 (iwmmxt_tmiaph): Likewise.
5281 (iwmmxt_tmiabb): Likewise.
5282 (iwmmxt_tmiatb): Likewise.
5283 (iwmmxt_tmiabt): Likewise.
5284 (iwmmxt_tmiatt): Likewise.
5285 (iwmmxt_tmovmskb): Likewise.
5286 (iwmmxt_tmovmskh): Likewise.
5287 (iwmmxt_tmovmskw): Likewise.
5288 (iwmmxt_waccb): Likewise.
5289 (iwmmxt_wacch): Likewise.
5290 (iwmmxt_waccw): Likewise.
5291 (iwmmxt_waligni): Likewise.
5292 (iwmmxt_walignr): Likewise.
5293 (iwmmxt_walignr0): Likewise.
5294 (iwmmxt_walignr1): Likewise.
5295 (iwmmxt_walignr2): Likewise.
5296 (iwmmxt_walignr3): Likewise.
5297 (iwmmxt_wsadb): Likewise.
5298 (iwmmxt_wsadh): Likewise.
5299 (iwmmxt_wsadbz): Likewise.
5300 (iwmmxt_wsadhz): Likewise.
5301 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5302 (iwmmxt_wabsdiffb): Likewise.
5303 (iwmmxt_wabsdiffh): Likewise.
5304 (iwmmxt_wabsdiffw): Likewise.
5305 (iwmmxt_waddsubhx): Likewise
5306 (iwmmxt_wsubaddhx): Likewise.
5307 (addc<mode>3): Likewise.
5308 (iwmmxt_avg4): Likewise.
5309 (iwmmxt_avg4r): Likewise.
5310 (iwmmxt_wmaddsx): Likewise.
5311 (iwmmxt_wmaddux): Likewise.
5312 (iwmmxt_wmaddsn): Likewise.
5313 (iwmmxt_wmaddun): Likewise.
5314 (iwmmxt_wmulwsm): Likewise.
5315 (iwmmxt_wmulwum): Likewise.
5316 (iwmmxt_wmulsmr): Likewise.
5317 (iwmmxt_wmulumr): Likewise.
5318 (iwmmxt_wmulwsmr): Likewise.
5319 (iwmmxt_wmulwumr): Likewise.
5320 (iwmmxt_wmulwl): Likewise.
5321 (iwmmxt_wqmulm): Likewise.
5322 (iwmmxt_wqmulwm): Likewise.
5323 (iwmmxt_wqmulmr): Likewise.
5324 (iwmmxt_wqmulwmr): Likewise.
5325 (iwmmxt_waddbhusm): Likewise.
5326 (iwmmxt_waddbhusl): Likewise.
5327 (iwmmxt_wqmiabb): Likewise.
5328 (iwmmxt_wqmiabt): Likewise.
5329 (iwmmxt_wqmiatb): Likewise.
5330 (iwmmxt_wqmiatt): Likewise.
5331 (iwmmxt_wqmiabbn): Likewise.
5332 (iwmmxt_wqmiabtn): Likewise.
5333 (iwmmxt_wqmiatbn): Likewise.
5334 (iwmmxt_wqmiattn): Likewise.
5335 (iwmmxt_wmiabb): Likewise.
5336 (iwmmxt_wmiabt): Likewise.
5337 (iwmmxt_wmiatb): Likewise.
5338 (iwmmxt_wmiatt): Likewise.
5339 (iwmmxt_wmiabbn): Likewise.
5340 (iwmmxt_wmiabtn): Likewise.
5341 (iwmmxt_wmiatbn): Likewise.
5342 (iwmmxt_wmiattn): Likewise.
5343 (iwmmxt_wmiawbb): Likewise.
5344 (iwmmxt_wmiawbt): Likewise.
5345 (iwmmxt_wmiawtb): Likewise.
5346 (iwmmxt_wmiawtt): Likewise.
5347 (iwmmxt_wmiawbbn): Likewise.
5348 (iwmmxt_wmiawbtn): Likewise.
5349 (iwmmxt_wmiawtbn): Likewise.
5350 (iwmmxt_wmiawttn): Likewise.
5351 (iwmmxt_wmerge): Likewise.
5352 (iwmmxt_tandc<mode>3): Likewise.
5353 (iwmmxt_torc<mode>3): Likewise.
5354 (iwmmxt_torvsc<mode>3): Likewise.
5355 (iwmmxt_textrc<mode>3): Likewise.
5356 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5357 (wmmxt_pack): Likewise.
5358 (wmmxt_mult_c1): Likewise.
5359 (wmmxt_mult_c2): Likewise.
5360 (wmmxt_alu_c1): Likewise.
5361 (wmmxt_alu_c2): Likewise.
5362 (wmmxt_alu_c3): Likewise.
5363 (wmmxt_transfer_c1): Likewise.
5364 (wmmxt_transfer_c2): Likewise.
5365 (wmmxt_transfer_c3): Likewise.
5366 (marvell_f_iwmmxt_wstr): Likewise.
5367 (marvell_f_iwmmxt_wldr): Likewise.
5369 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
5371 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5373 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5376 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5377 * lra-constraints.c (need_for_split_p): Check call used hard regs
5378 living through calls.
5380 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5381 call used regs for call insn.
5383 2013-06-28 Jakub Jelinek <jakub@redhat.com>
5386 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5387 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5389 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
5391 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5393 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5395 * lra-constraints.c (need_for_split_p): Check call used hard regs
5396 living through calls.
5398 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
5401 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5402 to tie with any other modes. Eliminate Altivec vector mode tests,
5403 since these are a subset of ALTIVEC or VSX vector modes. Simplify
5404 code, to return 0 if testing MODE2 for a condition, if we've
5405 already tested MODE1 for the same condition.
5407 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5409 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5412 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5414 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5415 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5417 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5419 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5421 (aarch64_symbolic_constant_p): Remove.
5422 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5423 static. Fix line length and white space.
5424 (aarch64_symbolic_constant_p): Remove.
5425 * config/aarch64/predicates.md (aarch64_valid_symref):
5426 Use aarch64_classify_symbol_expression.
5428 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5430 * config/arm/constraints.md (Ts): New constraint.
5431 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5433 (compare_scc): Use "Ts" constraint for operand 0.
5434 (ior_scc_scc): Likewise.
5435 (and_scc_scc): Likewise.
5436 (and_scc_scc_nodom): Likewise.
5437 (ior_scc_scc_cmp): Likewise for operand 7.
5438 (and_scc_scc_cmp): Likewise.
5439 * config/arm/thumb2.md (thumb2_movsi_insn):
5440 Add alternatives for 16-bit encodings.
5441 (thumb2_movhi_insn): Likewise.
5442 (thumb2_movsicc_insn): Likewise.
5443 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
5444 (thumb2_negscc): Use "Ts" constraint.
5445 Move mvn instruction outside cond_exec block.
5446 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5447 for 16-bit encodings.
5449 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5451 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5453 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5454 (mulsi3subsi): Likewise.
5455 (mulsidi3adddi): Likewise.
5456 (mulsidi3_v6): Likewise.
5457 (umulsidi3_v6): Likewise.
5458 (umulsidi3adddi_v6): Likewise.
5459 (smulsi3_highpart_v6): Likewise.
5460 (umulsi3_highpart_v6): Likewise.
5461 (mulhisi3tb): Likewise.
5462 (mulhisi3bt): Likewise.
5463 (mulhisi3tt): Likewise.
5464 (maddhisi4): Likewise.
5465 (maddhisi4tb): Likewise.
5466 (maddhisi4tt): Likewise.
5467 (maddhidi4): Likewise.
5468 (maddhidi4tb): Likewise.
5469 (maddhidi4tt): Likewise.
5470 (zeroextractsi_compare0_scratch): Likewise.
5471 (insv_zero): Likewise.
5472 (insv_t2): Likewise.
5473 (anddi_notzesidi_di): Likewise.
5474 (anddi_notsesidi_di): Likewise.
5475 (andsi_notsi_si): Likewise.
5476 (iordi_zesidi_di): Likewise.
5477 (xordi_zesidi_di): Likewise.
5478 (andsi_iorsi3_notsi): Likewise.
5480 (smax_m1): Likewise.
5482 (not_shiftsi): Likewise.
5483 (unaligned_loadsi): Likewise.
5484 (unaligned_loadhis): Likewise.
5485 (unaligned_loadhiu): Likewise.
5486 (unaligned_storesi): Likewise.
5487 (unaligned_storehi): Likewise.
5488 (extv_reg): Likewise.
5489 (extzv_t2): Likewise.
5491 (udivsi3): Likewise.
5492 (arm_zero_extendhisi2addsi): Likewise.
5493 (arm_zero_extendqisi2addsi): Likewise.
5494 (compareqi_eq0): Likewise.
5495 (arm_extendhisi2_v6): Likewise.
5496 (arm_extendqisi2addsi): Likewise.
5497 (arm_movt): Likewise.
5498 (thumb2_ldrd): Likewise.
5499 (thumb2_ldrd_base): Likewise.
5500 (thumb2_ldrd_base_neg): Likewise.
5501 (thumb2_strd): Likewise.
5502 (thumb2_strd_base): Likewise.
5503 (thumb2_strd_base_neg): Likewise.
5504 (arm_negsi2): Add alternative for 16-bit encoding.
5505 (arm_one_cmplsi2): Likewise.
5507 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5509 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5510 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5511 (movdfcc): Likewise.
5512 * config/arm/vfp.md (*thumb2_movsf_vfp):
5513 Disable predication for arm_restrict_it.
5514 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5515 (*thumb2_movdfcc_vfp): Likewise.
5516 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5517 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5518 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5519 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5520 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5521 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5522 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5523 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5524 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5525 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5526 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5527 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5528 Disable predication for arm_restrict_it.
5530 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
5532 * config/i386/bmiintrin.h (_bextr_u32): New.
5533 (_bextr_u64): Ditto.
5535 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
5537 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5538 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5539 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5540 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5541 * config/mips/n32-elf.h: ...this new file.
5543 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
5546 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5547 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5549 2013-06-27 Catherine Moore <clm@codesourcery.com>
5551 * config/mips/mips-tables.opt: Regenerate.
5552 * config/mips/mips-cpus.def: Add m14ke and m14kec.
5553 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5554 * doc/invoke.texi: Add -m14kc.
5556 2013-06-27 Jakub Jelinek <jakub@redhat.com>
5559 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5560 constraints of operand 1 and 2.
5563 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5564 to match RTL canonicalization. Swap predicates and
5565 constraints of operand 1 and 2.
5567 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
5569 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5570 Process OP_INOUT regs for splitting too.
5572 2013-06-27 Jakub Jelinek <jakub@redhat.com>
5574 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5575 decl before the loop, initialize to NULL.
5576 (vectorizable_load): Initialize ptr_incr to NULL.
5578 2013-06-27 Martin Jambor <mjambor@suse.cz>
5581 * ipa-ref.h (ipa_maybe_record_reference): Declare.
5582 * ipa-ref.c (ipa_maybe_record_reference): New function.
5583 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5584 * ipa-cp.c (create_specialized_node): Record potential references from
5586 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5588 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
5590 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5591 parameter 'mode' of type 'enum machine_mode mode'; change to pass
5592 'mode' to force_reg.
5593 (aarch64_add_offset): Update calls to aarch64_force_temporary.
5594 (aarch64_expand_mov_immediate): Likewise.
5596 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
5598 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
5599 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
5601 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5603 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
5604 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5605 (struct machine_function): Add tbegin_p.
5606 (s390_canonicalize_comparison): Fold CC mode compares to
5607 conditional jump if possible.
5608 (s390_emit_jump): Return the emitted jump.
5609 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
5610 Handle CCRAWmode compares.
5611 (s390_option_override): Default to -mhtm if available.
5612 (s390_reg_clobbered_rtx): Handle floating point regs as well.
5613 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
5614 FPRs instead of df_regs_ever_live_p.
5615 (s390_optimize_nonescaping_tx): New function.
5616 (s390_init_frame_layout): Extend clobbered_regs array to cover
5618 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
5619 (s390_expand_tbegin): New function.
5620 (enum s390_builtin): New enum definition.
5621 (code_for_builtin): New array definition.
5622 (s390_init_builtins): New function.
5623 (s390_expand_builtin): New function.
5624 (TARGET_INIT_BUILTINS): Define.
5625 (TARGET_EXPAND_BUILTIN): Define.
5626 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
5627 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
5628 (s390_alc_comparison): Likewise.
5629 * config/s390/s390-modes.def: Add CCRAWmode.
5630 * config/s390/s390.h (processor_flags): Add PF_TX.
5631 (TARGET_CPU_HTM): Define macro.
5632 (TARGET_HTM): Define macro.
5633 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
5634 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
5635 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5636 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
5637 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
5638 (TBEGIN_MASK, TBEGINC_MASK): New constants.
5639 ("*cc_to_int"): Move up.
5640 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
5641 constants other than 0.
5642 ("*ccraw_to_int"): New insn and splitter definition.
5643 ("tbegin", "tbegin_nofloat", "tbegin_retry")
5644 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
5645 ("tx_assist"): New expander.
5646 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
5647 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
5648 * config/s390/s390.opt: Add -mhtm option.
5649 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
5650 * config/s390/htmxlintrin.h: New file.
5651 * config/s390/htmintrin.h: New file.
5652 * config/s390/s390intrin.h: New file.
5653 * doc/extend.texi: Document htm builtins.
5654 * config.gcc: Add the new header files to extra_headers.
5656 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
5658 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
5659 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
5661 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5662 Pat Haugen <pthaugen@us.ibm.com>
5663 Peter Bergner <bergner@vnet.ibm.com>
5665 * config/rs6000/power8.md: New.
5666 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
5667 setting for power8 entry.
5668 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
5669 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
5670 test for Power4/Power5 only.
5671 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
5673 (force_new_group): Adjust comment.
5674 * config/rs6000/rs6000.md: Include power8.md.
5676 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
5678 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
5679 * config/arm/arm-protos.h (arm_max_conditional_execute): New
5681 (tune_params): Update comment.
5682 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
5683 (arm_max_conditional_execute): New function.
5684 (thumb2_final_prescan_insn): Use max_insn_skipped and
5685 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
5687 2013-06-25 Jakub Jelinek <jakub@redhat.com>
5689 PR tree-optimization/57705
5690 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
5691 SSA_NAME step, provided that it is not defined inside the loop.
5692 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
5693 (get_initial_def_for_induction): Handle SSA_NAME IV step.
5695 2013-06-25 Martin Jambor <mjambor@suse.cz>
5698 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
5699 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
5701 (ipa_note_param_call): Initialize member_ptr flag.
5702 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
5703 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
5704 (ipa_write_indirect_edge_info): Stream member_ptr flag.
5705 (ipa_read_indirect_edge_info): Likewise.
5707 2013-06-25 Richard Biener <rguenther@suse.de>
5710 * passes.c (init_optimization_passes): Move pass_fold_builtins
5711 and pass_dce earlier with -Og.
5713 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
5715 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
5716 <BIT_FIELD_REF>: Remove trailing TAB.
5717 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
5720 2013-06-24 Martin Jambor <mjambor@suse.cz>
5722 PR tree-optimization/57358
5723 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
5724 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
5725 (ipa_analyze_params_uses): Generate pessimistic info when true.
5727 2013-06-24 Martin Jambor <mjambor@suse.cz>
5729 PR tree-optimization/57539
5730 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
5731 global.inlined_to of the new node to it. All callers changed.
5732 * ipa-inline-transform.c (clone_inlined_nodes): New variable
5733 inlining_into, pass it to cgraph_clone_node.
5734 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
5735 ipa_free_edge_args_substructures.
5736 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
5737 rdesc linked list. Do not assert rdesc edges have inlined caller.
5738 Assert we have found an rdesc in the rdesc list.
5740 2013-06-24 Richard Biener <rguenther@suse.de>
5742 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
5743 (pointer_set_lookup): Declare.
5744 (class pointer_map): New template class implementing a
5745 generic pointer to T map.
5746 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
5747 pointer_map<T>::contains, pointer_map<T>::insert,
5748 pointer_map<T>::traverse): New functions.
5749 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
5750 (pointer_set_lookup): New function.
5751 (pointer_set_contains): Use pointer_set_lookup.
5752 (pointer_set_insert): Likewise.
5753 (insert_aux): Remove.
5754 (struct pointer_map_t): Embed a pointer_set_t.
5755 (pointer_map_create): Adjust.
5756 (pointer_map_destroy): Likewise.
5757 (pointer_map_contains): Likewise.
5758 (pointer_map_insert): Likewise.
5759 (pointer_map_traverse): Likewise.
5760 * tree-streamer.h (struct streamer_tree_cache_d): Use a
5761 pointer_map<unsigned> instead of a pointer_map_t.
5762 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
5763 (streamer_tree_cache_lookup): Likewise.
5764 (streamer_tree_cache_create): Likewise.
5765 (streamer_tree_cache_delete): Likewise.
5766 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
5767 pointer_map<unsigned> instead of a pointer_map_t.
5768 (lto_init_tree_ref_encoder): Adjust.
5769 (lto_destroy_tree_ref_encoder): Likewise.
5770 * lto-section-out.c (lto_output_decl_index): Likewise.
5771 (lto_record_function_out_decl_state): Likewise.
5772 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
5774 2013-06-24 Richard Biener <rguenther@suse.de>
5776 PR tree-optimization/57488
5777 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
5779 2013-06-24 Alan Modra <amodra@gmail.com>
5781 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
5782 (gen_easy_altivec_constant): Likewise.
5783 * config/rs6000/predicates.md (easy_vector_constant_add_self,
5784 easy_vector_constant_msb): Likewise.
5786 2013-06-23 Jakub Jelinek <jakub@redhat.com>
5789 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
5790 add missing return true.
5792 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
5795 * config/sh/predicates.md (general_extend_operand): Invoke
5796 general_movsrc_operand for memory operands.
5797 (general_movsrc_operand): Allow reg+reg addressing, do not use
5798 general_operand for memory operands.
5800 2013-06-23 Sriraman Tallam <tmsriram@google.com>
5802 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
5803 when current target options does not apply.
5804 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
5805 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
5806 * config/i386/bmiintrin.h: Pass appropriate target
5807 attributes to header.
5808 * config/i386/mmintrin.h: Ditto.
5809 * config/i386/nmmintrin.h: Ditto.
5810 * config/i386/avx2intrin.h: Ditto.
5811 * config/i386/fxsrintrin.h: Ditto.
5812 * config/i386/tbmintrin.h: Ditto.
5813 * config/i386/xsaveintrin.h: Ditto.
5814 * config/i386/f16cintrin.h: Ditto.
5815 * config/i386/xtestintrin.h: Ditto.
5816 * config/i386/xsaveoptintrin.h: Ditto.
5817 * config/i386/bmi2intrin.h: Ditto.
5818 * config/i386/lzcntintrin.h: Ditto.
5819 * config/i386/smmintrin.h: Ditto.
5820 * config/i386/wmmintrin.h: Ditto.
5821 * config/i386/x86intrin.h: Remove all header include guards.
5822 * config/i386/prfchwintrin.h: Ditto.
5823 * config/i386/pmmintrin.h: Ditto.
5824 * config/i386/tmmintrin.h: Ditto.
5825 * config/i386/xmmintrin.h: Ditto.
5826 * config/i386/popcntintrin.h: Ditto.
5827 * config/i386/rdseedintrin.h: Ditto.
5828 * config/i386/ammintrin.h: Ditto.
5829 * config/i386/emmintrin.h: Ditto.
5830 * config/i386/immintrin.h: Remove all header include guards.
5831 * config/i386/fma4intrin.h: Ditto.
5832 * config/i386/lwpintrin.h: Ditto.
5833 * config/i386/xopintrin.h: Ditto.
5834 * config/i386/ia32intrin.h: Ditto.
5835 * config/i386/avxintrin.h: Ditto.
5836 * config/i386/rtmintrin.h: Ditto.
5837 * config/i386/fmaintrin.h: Ditto.
5838 * config/i386/mm3dnow.h: Ditto.
5840 2013-06-22 Sriraman Tallam <tmsriram@google.com>
5842 * common/config/i386/i386-common.c: Handle LZCNT.
5844 2013-06-22 Andi Kleen <ak@linux.intel.com>
5846 * doc/extend.texi: Use __atomic_store_n instead of
5847 __atomic_store in HLE example.
5849 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
5851 * config/sh/sh.c: Remove <cstdlib> workaround.
5853 2013-06-21 Andi Kleen <ak@linux.intel.com>
5855 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
5857 2013-06-21 Andi Kleen <ak@linux.intel.com>
5859 * doc/extend.texi: Document that __atomic_clear and
5860 __atomic_test_and_set should only be used with bool.
5862 2013-06-20 Jan Hubicka <jh@suse.cz>
5864 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
5866 * tree.c (decls_same_for_odr): New function.
5867 (same_for_edr): New function.
5868 (types_same_for_odr): New function.
5869 (get_binfo_at_offset): Use it.
5870 * tree.h (types_same_for_odr): Declare.
5872 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
5873 Jason Merrill <jason@redhat.com>
5875 * system.h: Include <cstdlib> as well as <stdlib.h>.
5877 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
5880 * config/i386/i386.c (construct_container): Report error if
5881 long double is used with disabled x87 float returns.
5883 2013-06-20 Jan Hubicka <jh@suse.cz>
5885 * lto-cgraph.c (input_symtab): Do not set cgraph state.
5887 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
5889 PR rtl-optimization/57425
5890 PR rtl-optimization/57569
5891 * alias.c (write_dependence_p): Remove parameters mem_mode and
5892 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
5893 Changed all callers.
5894 (canon_anti_dependence): Get comments and semantics in sync.
5895 Add parameter mem_canonicalized. Changed all callers.
5896 * rtl.h (canon_anti_dependence): Update prototype.
5898 2013-06-20 Richard Biener <rguenther@suse.de>
5900 * data-streamer-in.c (streamer_read_uhwi): Optimize single
5901 byte case, inline streamer_read_uchar and defer section
5904 2013-06-20 Richard Biener <rguenther@suse.de>
5906 PR tree-optimization/57584
5907 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
5908 SSA names into the expanded expression that take part in
5909 abnormal coalescing.
5911 2013-06-19 Sharad Singhai <singhai@google.com>
5913 * gcov.c (print_usage): Handle new option.
5914 (process_args): Ditto.
5915 (get_gcov_intermediate_filename): New function.
5916 (output_intermediate_file): New function.
5917 (output_gcov_file): New function
5918 (generate_results): Handle new option.
5919 (release_function): Relase demangled name.
5920 (read_graph_file): Handle demangled name.
5921 (output_lines): Ditto.
5922 * doc/gcov.texi: Document gcov intermediate format.
5924 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
5927 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
5928 (lra_emit_add): Use the functions. Add comment about Y as an
5931 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
5934 * collect2.c (collect_atexit): New.
5935 (collect_exit): Delete.
5936 (main): Register collect_atexit with atexit.
5937 (collect_wait): Change collect_exit to exit.
5939 * collect2.h (collect_exit): Delete.
5940 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
5942 2013-06-19 Wei Mi <wmi@google.com>
5944 PR rtl-optimization/57518
5945 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
5946 if regno is used in paradoxical subreg.
5947 (update_equiv_regs): Check pdx_subregs[regno] before
5948 set a reg to be equivalent with a mem.
5950 2013-06-19 Matthias Klose <doko@ubuntu.com>
5953 * file-find.h (find_a_file): Add a mode parameter.
5954 * file-find.c (find_a_file): Likewise.
5955 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
5956 with X_OK for the executables.
5957 * collect2.c (main): Call find_a_file with X_OK.
5959 2013-06-19 Steve Ellcey <sellcey@mips.com>
5962 * config/mips/mips.md (casesi_internal_mips16_<mode>):
5963 Use NEXT_INSN instead of next_real_insn.
5965 2013-06-19 Jan Hubicka <jh@suse.cz>
5967 * cgraph.h (const_value_known_p): Replace by ...
5968 (ctor_for_folding): .. this one.
5969 * cgraphunit.c (process_function_and_variable_attributes): Use it.
5970 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
5971 * expr.c (expand_expr_real_1): Likewise.
5972 (string_constant): Likewise.
5973 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
5974 * ipa.c (process_references): Likewise.
5975 (symtab_remove_unreachable_nodes): Likewise.
5976 * ipa-inline-analysis.c (param_change_prob): Likewise.
5977 * gimple-fold.c (canonicalize_constructor_val): Likewise.
5978 (get_base_constructor): Likwise.
5979 * varpool.c (varpool_remove_node): Likewise.
5980 (varpool_remove_initializer): LIkewise.
5981 (dump_varpool_node): LIkwise.
5982 (const_value_known_p): Rewrite to ...
5983 (ctor_for_folding): ... this one.
5985 2013-06-19 Jakub Jelinek <jakub@redhat.com>
5988 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
5989 PERSONALITY in $PATH derived prefixes.
5991 2013-06-19 Jeff Law <law@redhat.com>
5993 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
5996 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
5997 (simplify_bitwise_binary): Use it to simpify certain binary ops on
6000 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
6002 * config/arm/vfp.md: Move VFP instruction classification documentation
6004 * config/arm/arm.md: ... here. Update instruction classification
6007 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
6009 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6010 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6011 pattern. Use more efficient sequences on ARMv5 and Thumb2.
6013 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
6016 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6017 with NEXT_INSN. Use tablejump_p to check for jump table data
6020 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
6023 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6024 that now in C++ the value is correct per the C++ standards.
6026 2013-06-19 Richard Biener <rguenther@suse.de>
6028 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6031 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6034 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6037 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6038 with next_active_insn.
6040 2013-06-18 Sriraman Tallam <tmsriram@google.com>
6042 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6043 functions are inlined during failures to flag an error.
6044 * tree-inline.c (expand_call_inline): Allow the error to be flagged
6045 in early inline pass.
6047 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
6049 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6052 2013-06-18 Julian Brown <julian@codesourcery.com>
6054 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6055 Permit virtual register pre-reload if !strict.
6056 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6058 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6060 * config/arm/neon.md (movmisalign<mode>): Use
6061 neon_perm_struct_or_reg_operand instead of
6062 neon_struct_or_register_operand.
6063 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6064 neon_permissive_struct_operand instead of neon_struct_operand.
6065 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6066 neon_vector_mem_operand.
6067 * config/arm/predicates.md (neon_struct_operand): Adjust call to
6068 neon_vector_mem_operand.
6069 (neon_permissive_struct_operand): New.
6070 (neon_struct_or_register_operand): Rename to...
6071 (neon_perm_struct_or_reg_operand): This. Adjust call to
6072 neon_vector_mem_operand.
6074 2013-06-18 Richard Biener <rguenther@suse.de>
6076 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6077 * lto-streamer.h: Include pointer-set.h.
6078 (struct lto_decl_slot): Remove.
6079 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6080 Remove next_index entry.
6081 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6082 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6083 (lto_init_tree_ref_encoder): Adjust.
6084 (lto_destroy_tree_ref_encoder): Likewise.
6085 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6086 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6087 (lto_output_decl_index): Adjust.
6088 (lto_new_out_decl_state): Likewise.
6089 (lto_record_function_out_decl_state): Likewise.
6090 * lto-streamer-out.c (copy_function): Likewise.
6092 2013-06-18 Richard Biener <rguenther@suse.de>
6094 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6095 * cgraphunit.c: Include cfgloop.h.
6096 (init_lowered_empty_function): Initialize the loop tree.
6097 (assemble_thunk): Insert new BBs into loops.
6099 2013-06-18 Richard Biener <rguenther@suse.de>
6101 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6102 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6103 the map from cache entry to cache index optional.
6104 (streamer_tree_cache_replace_tree): Adjust accordingly.
6105 (streamer_tree_cache_append): Likewise.
6106 (streamer_tree_cache_delete): Likewise.
6107 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6108 streamer cache map from cache entry to cache index.
6109 * lto-streamer-out.c (create_output_block): Adjust.
6111 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
6113 * config/arm/arm.md (attribute "insn"): Move multiplication and
6114 division attributes to...
6115 (attribute "type"): ... here. Remove mult.
6116 (attribute "mul32"): New attribute.
6117 (attribute "mul64"): Add umaal.
6118 (*arm_mulsi3): Update attributes.
6119 (*arm_mulsi3_v6): Likewise.
6120 (*thumb_mulsi3): Likewise.
6121 (*thumb_mulsi3_v6): Likewise.
6122 (*mulsi3_compare0): Likewise.
6123 (*mulsi3_compare0_v6): Likewise.
6124 (*mulsi_compare0_scratch): Likewise.
6125 (*mulsi_compare0_scratch_v6): Likewise.
6126 (*mulsi3addsi): Likewise.
6127 (*mulsi3addsi_v6): Likewise.
6128 (*mulsi3addsi_compare0): Likewise.
6129 (*mulsi3addsi_compare0_v6): Likewise.
6130 (*mulsi3addsi_compare0_scratch): Likewise.
6131 (*mulsi3addsi_compare0_scratch_v6): Likewise.
6132 (*mulsi3subsi): Likewise.
6133 (*mulsidi3adddi): Likewise.
6134 (*mulsi3addsi_v6): Likewise.
6135 (*mulsidi3adddi_v6): Likewise.
6136 (*mulsidi3_nov6): Likewise.
6137 (*mulsidi3_v6): Likewise.
6138 (*umulsidi3_nov6): Likewise.
6139 (*umulsidi3_v6): Likewise.
6140 (*umulsidi3adddi): Likewise.
6141 (*umulsidi3adddi_v6): Likewise.
6142 (*smulsi3_highpart_nov6): Likewise.
6143 (*smulsi3_highpart_v6): Likewise.
6144 (*umulsi3_highpart_nov6): Likewise.
6145 (*umulsi3_highpart_v6): Likewise.
6146 (mulhisi3): Likewise.
6147 (*mulhisi3tb): Likewise.
6148 (*mulhisi3bt): Likewise.
6149 (*mulhisi3tt): Likewise.
6150 (maddhisi4): Likewise.
6151 (*maddhisi4tb): Likewise.
6152 (*maddhisi4tt): Likewise.
6153 (maddhidi4): Likewise.
6154 (*maddhidi4tb): Likewise.
6155 (*maddhidi4tt): Likewise.
6157 (udivsi3): Likewise.
6158 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6159 (thumb2_mulsi_short_compare0): Likewise.
6160 (thumb2_mulsi_short_compare0_scratch): Likewise.
6161 * config/arm/arm1020e.md (1020mult1): Update attribute change.
6162 (1020mult2): Likewise.
6163 (1020mult3): Likewise.
6164 (1020mult4): Likewise.
6165 (1020mult5): Likewise.
6166 (1020mult6): Likewise.
6167 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6169 (cortex_a15_mult64): Likewise.
6170 (cortex_a15_sdiv): Likewise.
6171 (cortex_a15_udiv): Likewise.
6172 * config/arm/arm1026ejs.md (mult1): Update attribute change.
6178 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6179 (pj4_ir_div): Likewise.
6180 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6181 (11_mult2): Likewise.
6182 (11_mult3): Likewise.
6183 (11_mult4): Likewise.
6184 (11_mult5): Likewise.
6185 (11_mult6): Likewise.
6186 (11_mult7): Likewise.
6187 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6188 (cortex_a8_mla): Likewise.
6189 (cortex_a8_mull): Likewise.
6190 (cortex_a8_smulwy): Likewise.
6191 (cortex_a8_smlald): Likewise.
6192 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6193 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6194 (cortex_r4_mul_3): Likewise.
6195 (cortex_r4_mla_4): Likewise.
6196 (cortex_r4_mla_3): Likewise.
6197 (cortex_r4_smlald): Likewise.
6198 (cortex_r4_mull): Likewise.
6199 (cortex_r4_sdiv): Likewise.
6200 (cortex_r4_udiv): Likewise.
6201 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6202 (cortex_a7_idiv): Likewise.
6203 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6204 (9_mult2): Likewise.
6205 (9_mult3): Likewise.
6206 (9_mult4): Likewise.
6207 (9_mult5): Likewise.
6208 (9_mult6): Likewise.
6209 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6210 (cortex_a53_sdiv): Likewise.
6211 (cortex_a53_udiv): Likewise.
6212 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6213 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6214 (mp626_mult2): Likewise.
6215 (mp626_mult3): Likewise.
6216 (mp626_mult4): Likewise.
6217 * config/arm/fa526.md (526_mult1): Update attribute change.
6218 (526_mult2): Likewise.
6219 * config/arm/arm-generic.md (mult): Update attribute change.
6220 (mult_ldsched_strongarm): Likewise.
6221 (mult_ldsched): Likewise.
6222 (multi_cycle): Likewise.
6223 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6224 * config/arm/fa606te.md (606te_mult1): Update attribute change.
6225 (606te_mult2): Likewise.
6226 (606te_mult3): Likewise.
6227 (606te_mult4): Likewise.
6228 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6229 (cortex_a9_mac16): Likewise.
6230 (cortex_a9_multiply): Likewise.
6231 (cortex_a9_mac): Likewise.
6232 (cortex_a9_multiply_long): Likewise.
6233 * config/arm/fa626te.md (626te_mult1): Update attribute change.
6234 (626te_mult2): Likewise.
6235 (626te_mult3): Likewise.
6236 (626te_mult4): Likewise.
6238 2013-06-18 Richard Biener <rguenther@suse.de>
6241 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6243 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6246 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6247 with next_active_insn.
6249 2013-06-18 Alan Modra <amodra@gmail.com>
6251 * config/rs6000/rs6000.h (enum data_align): New.
6252 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6253 (DATA_ABI_ALIGNMENT): Define.
6254 (CONSTANT_ALIGNMENT): Correct comment.
6255 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6256 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6258 2013-06-17 David Malcolm <dmalcolm@redhat.com>
6260 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6261 ATTRIBUTE_UNUSED marking.
6263 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
6265 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6266 alternative and update.
6267 (aarch64_dup_lanedi): Delete.
6268 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6269 * config/aarch64/aarch64-simd-builtins.def: Update.
6271 2013-06-17 Richard Biener <rguenther@suse.de>
6273 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6274 (lto_input_scc): Declare.
6275 (lto_input_tree_1): Likewise.
6276 (struct lto_stats_d): Add num_tree_bodies_output and
6277 num_pickle_refs_output.
6278 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6279 (lto_read_tree_1): Split out from ...
6280 (lto_read_tree): ... this.
6281 (lto_input_scc): New function.
6282 (lto_input_tree_1): Split out from ...
6283 (lto_input_tree): ... this. Handle LTO_tree_scc.
6284 (lto_data_in_create): Create the streamer cache without hashes.
6285 * lto-streamer-out.c (create_output_block): Create the streamer
6286 cache with hashes when not doing WPA.
6287 (lto_write_tree_1): Split out from ...
6288 (lto_write_tree): ... this.
6289 (get_symbol_initial_value): New function.
6290 (lto_output_tree_1): Split out from ...
6291 (lto_output_tree): ... this. Write trees as series of SCCs
6292 using a DFS walk via DFS_write_tree.
6293 (struct sccs, struct scc_entry): New types.
6294 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6295 (DFS_write_tree_body): New function.
6296 (DFS_write_tree): Likewise.
6297 (hash_tree): Likewise.
6298 (scc_entry_compare): Likewise.
6299 (hash_scc): Likewise.
6300 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6301 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6302 TREE_CHAIN as regular reference.
6303 (streamer_read_integer_cst): Remove.
6304 (streamer_get_pickled_tree): Adjust.
6305 * tree-streamer-out.c (streamer_write_chain): Disable streaming
6306 of DECL_EXTERNALs in BLOCK_VARS for now.
6307 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6309 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6310 Add hash value argument and record that if hashes are recorded
6312 (streamer_tree_cache_insert_1): Adjust.
6313 (streamer_tree_cache_insert): Likewise.
6314 (streamer_tree_cache_insert_at): Rename to ...
6315 (streamer_tree_cache_replace_tree): ... this and adjust.
6316 (streamer_tree_cache_append): Adjust.
6317 (record_common_node): Likewise.
6318 (streamer_tree_cache_create): Add argument whether to
6319 record hash values together with trees.
6320 (streamer_tree_cache_delete): Adjust.
6321 * tree-streamer.h (struct streamer_tree_cache_d): Add
6323 (streamer_read_integer_cst): Remove.
6324 (streamer_tree_cache_insert): Adjust.
6325 (streamer_tree_cache_append): Likewise.
6326 (streamer_tree_cache_insert_at): Rename to ...
6327 (streamer_tree_cache_replace_tree): ... this and adjust.
6328 (streamer_tree_cache_create): Add argument whether to record hashes.
6329 (streamer_tree_cache_get): Rename to ...
6330 (streamer_tree_cache_get_tree): ... this.
6331 (streamer_tree_cache_get_hash): New function.
6332 * tree.c (cache_integer_cst): New function.
6333 * tree.h (cache_integer_cst): Declare.
6334 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6335 * lto-symtab.c (lto_varpool_replace_node): Only release
6336 DECL_INITIAL of non-prevailing decls.
6337 * varpool.c (varpool_remove_initializer): Do not release
6338 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6340 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
6342 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6343 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6344 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6345 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6346 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6347 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6348 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6349 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6350 instead of TARGET_64BIT.
6351 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6352 Require ISA_HAS_<D>DIV.
6354 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
6356 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6357 (mips*-*-linux*): Move default with_llsc setting to where other
6359 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6361 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6362 Likewise. Remove default with_tune setting. Move default float
6363 setting to its own block. Handle with_llsc in the same block as above.
6365 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
6367 PR rtl-optimization/57425
6368 PR rtl-optimization/57569
6369 * alias.c (write_dependence_p): Add new parameters mem_mode,
6370 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
6371 Changed all callers.
6372 (canon_anti_dependence): New function.
6373 * cse.c (check_dependence): Use canon_anti_dependence.
6374 * cselib.c (cselib_invalidate_mem): Likewise.
6375 * rtl.h (canon_anti_dependence): Declare.
6377 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
6379 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6380 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6381 ".set mips3" for 64-bit targets.
6383 2013-06-15 Dehao Chen <dehao@google.com>
6385 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6386 * gimple-low.c (gimple_check_call_matching_types): Likewise.
6387 (gimple_check_call_args): Likewise.
6388 * value-prof.c (check_ic_target): Likewise.
6389 * ipa-inline.c (early_inliner): Likewise.
6390 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6391 * cgraph.c (cgraph_create_edge_1): Likewise.
6392 (cgraph_make_edge_direct): Likewise.
6394 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6397 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6398 rs6000_output_move_128bit to handle emitting quad memory
6399 operations. Set attribute length to 8 bytes.
6401 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
6403 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6405 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6406 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6407 (aarch64_<su>mlsl<mode>): Likewise.
6409 2013-06-14 Mike Stump <mikestump@comcast.net>
6411 * Makefile.in (TARGET_H): Add insn-codes.h.
6413 2013-06-14 Alan Modra <amodra@gmail.com>
6417 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6418 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
6419 bitfield expansion when EXPAND_MEMORY.
6420 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6422 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6424 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6425 test for clearing quad memory on 32-bit later.
6427 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
6429 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6430 (fold_negate_expr): Likewise.
6431 (fold_real_zero_addition_p): Handle vectors.
6432 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6434 2013-06-14 Alan Modra <amodra@gmail.com>
6436 * varasm.c (force_const_mem): Revert 2013-06-07 change.
6438 2013-06-13 Jan Hubicka <jh@suse.cz>
6440 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6441 Local comdats are not externally visible.
6442 * symtab.c (dump_symtab_base): Dump externally visible.
6443 (verify_symtab_base): Verify back links in the symtab hash.
6445 2013-06-13 Bin Cheng <bin.cheng@arm.com>
6447 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6448 CONVERT_EXPR as equal nodes.
6450 2013-06-13 Bin Cheng <bin.cheng@arm.com>
6452 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6454 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
6456 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6457 Generalize to complex and vector.
6458 * tree.c (build_all_ones_cst): New function.
6459 * tree.h (build_all_ones_cst): Declare it.
6461 2013-06-13 Alan Modra <amodra@gmail.com>
6463 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6464 * config/rs6000/rs6000.md (signbittf2): New insn.
6465 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6466 (abstf2_internal, cmptf_internal2): Likewise.
6467 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6469 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6470 Pat Haugen <pthaugen@us.ibm.com>
6471 Peter Bergner <bergner@vnet.ibm.com>
6473 * config/rs6000/rs6000.c (emit_load_locked): Add support for
6474 power8 byte, half-word, and quad-word atomic instructions.
6475 (emit_store_conditional): Likewise.
6476 (rs6000_expand_atomic_compare_and_swap): Likewise.
6477 (rs6000_expand_atomic_op): Likewise.
6479 * config/rs6000/sync.md (larx): Add new modes for power8.
6481 (AINT): New mode iterator to include TImode as well as normal
6482 integer modes on power8.
6483 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6484 that VSX registers are not considered. Use AINT mode iterator
6485 instead of INT1 to allow inclusion of quad word atomic operations
6487 (load_locked<mode>): Likewise.
6488 (store_conditional<mode>): Likewise.
6489 (atomic_compare_and_swap<mode>): Likewise.
6490 (atomic_exchange<mode>): Likewise.
6491 (atomic_nand<mode>): Likewise.
6492 (atomic_fetch_<fetchop_name><mode>): Likewise.
6493 (atomic_nand_fetch<mode>): Likewise.
6494 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6496 (ATOMIC): On power8, add QImode, HImode modes.
6497 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6498 modes that promote to SImode.
6499 (load_lockedti): Convert TImode arguments to PTImode, so that we
6500 get a guaranteed even/odd register pair.
6501 (load_lockedpti): Likewise.
6502 (store_conditionalti): Likewise.
6503 (store_conditionalpti): Likewise.
6505 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6506 atomic load/store instructions.
6509 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
6511 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6513 (insn_count): New attribute, with most cases extracted from...
6514 (length): ...here. Redefine most cases in terms of insn_count.
6515 (single_insn): Delete.
6516 (can_delay): Use insn_count to check for single instructions.
6517 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6518 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6519 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6520 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6521 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6522 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6523 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6524 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6525 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6526 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6527 rather than "length".
6528 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
6529 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6530 Use "insn_count" rather than "length".
6531 * config/mips/mips-dsp.md
6532 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6533 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6536 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
6538 PR tree-optimization/57361
6539 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6541 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
6543 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6545 (aarch64_simd_combine<mode>): New instruction expansion.
6546 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6548 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6549 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6551 2013-06-12 Jan Hubicka <jh@suse.cz>
6553 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6554 decl has when in streaming stage.
6555 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6556 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6558 2013-06-12 Roland Stigge <stigge@antcom.de>
6561 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6563 2013-06-12 Jakub Jelinek <jakub@redhat.com>
6565 PR tree-optimization/57537
6566 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6567 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6569 2013-06-12 Richard Biener <rguenther@suse.de>
6571 * data-streamer.h (streamer_write_char_stream): CSE
6572 obs->current_pointer.
6573 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6574 streamer_write_char_stream manually and optimize the resulting loop.
6575 (streamer_write_hwi_stream): Likewise.
6577 2013-06-12 Jan Hubicka <jh@suse.cz>
6579 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6580 * cgraph.h (varpool_create_empty_node): Declare.
6581 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6583 * symtab.c (symtab_unregister_node): Be lax about missin entries
6585 (symtab_get_node): Update comment.
6586 * varpool.c (varpool_create_empty_node): Break out from ...
6587 (varpool_node_for_decl): ... here.
6588 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6590 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
6592 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6593 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6594 part. Use straight-line flow at the end.
6595 <COMPONENT_REF>: Remove superfluous else.
6596 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
6598 2013-06-12 Jakub Jelinek <jakub@redhat.com>
6601 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
6602 target hook even for !TREE_PUBLIC decls. If no resolution info
6603 is available, return false for common and external decls.
6605 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
6607 * config/rl78/constraints.md (U): New constraint.
6608 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
6611 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
6614 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
6615 to allow returning address to AT_PLATFORM name.
6617 2013-06-11 Jan Hubicka <jh@suse.cz>
6619 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
6620 * cgraph.h (symtab_node_base): Add weakref flag.
6621 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
6622 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
6623 (output_weakrefs): Use weakref flag.
6624 * fold-const.c (simple_operand_p): Handle WEAK.
6625 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
6626 * ipa.c (varpool_externally_visible_p): Drop weakref.
6627 (function_and_variable_visibility): Update comment; fix weakref
6628 sanity checks; do not clear DECL_WEAK on them.
6629 * lto-cgraph.c (lto_output_node): update.
6630 (lto_output_varpool_node): Update.
6631 (input_overwrite_node): Update.
6632 (input_node): Update.
6633 (input_varpool_node): Update.
6634 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
6635 (lto_symtab_merge_symbols): Add sanity check.
6636 (lto_symtab_prevailing_decl): Do not special case weakrefs.
6637 * passes.c (rest_of_decl_compilation): Set static flag, too.
6638 * symtab.c (dump_symtab_base): Dump weakref.
6639 (verify_symtab_base): Sanity check weakrefs.
6640 (symtab_make_decl_local): Remove duplicated code.
6641 (symtab_alias_ultimate_target): Simplify.
6642 * varpool.c (varpool_create_variable_alias): Set weakref flag.
6644 2013-06-11 DJ Delorie <dj@redhat.com>
6646 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
6647 (rl78_unwind_word_mode): New.
6649 2013-06-11 David Malcolm <dmalcolm@redhat.com>
6651 * final.c (debug_prefix_maps): Make static.
6653 2013-06-11 David Malcolm <dmalcolm@redhat.com>
6655 * function.c (initial_trampoline): Remove stray copy.
6657 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
6659 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
6661 2013-06-11 Martin Jambor <mjambor@suse.cz>
6663 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
6664 within bounds at the beginning of the function.
6666 2013-06-11 Alan Modra <amodra@gmail.com>
6668 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
6670 (get_named_section): Don't NULL !DECL_P decl.
6672 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
6674 * doc/invoke.texi (core-avx2): Document.
6676 (atom): Updated with MOVBE.
6678 2013-06-11 Richard Biener <rguenther@suse.de>
6680 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
6682 2013-06-11 Anton Blanchard <anton@samba.org>
6684 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
6685 correct shift value in little-endian mode.
6687 2013-06-11 Jakub Jelinek <jakub@redhat.com>
6690 * varasm.c (get_variable_align): Move #endif to the right place.
6692 2013-06-10 Cary Coutant <ccoutant@google.com>
6694 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
6695 for hash so that hash table traversal order is deterministic.
6697 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6698 Pat Haugen <pthaugen@us.ibm.com>
6699 Peter Bergner <bergner@vnet.ibm.com>
6701 * config/rs6000/vector.md (GPR move splitter): Do not split moves
6702 of vectors in GPRS if they are direct moves or quad word load or
6705 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
6707 (direct_move_p): Likewise.
6708 (quad_load_store_p): Likewise.
6710 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
6711 classes into bins based on the physical register type.
6712 (reg_class_to_reg_type): Likewise.
6713 (IS_STD_REG_TYPE): Likewise.
6714 (IS_FP_VECT_REG_TYPE): Likewise.
6715 (reload_fpr_gpr): Arrays to determine what insn to use if we can
6716 use direct move instructions.
6717 (reload_gpr_vsx): Likewise.
6718 (reload_vsx_gpr): Likewise.
6719 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
6720 information that is a simplification of register classes. Also
6721 precalculate direct move reload helpers.
6722 (direct_move_p): New function to return true if the operation can
6723 be done as a direct move instruciton.
6724 (quad_load_store_p): New function to return true if the operation
6725 is a quad memory operation.
6726 (rs6000_legitimize_address): If quad memory, only allow register
6727 indirect for TImode addresses.
6728 (rs6000_legitimate_address_p): Likewise.
6729 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
6730 (rs6000_reload_register_type): Likewise.
6731 (register_to_reg_type): Return register type.
6732 (rs6000_secondary_reload_simple_move): New helper function for
6733 secondary reload and secondary memory needed to identify anything
6734 that is a simple move, and does not need reloading.
6735 (rs6000_secondary_reload_direct_move): New helper function for
6736 secondary reload to identify cases that can be done with several
6737 instructions via the direct move instructions.
6738 (rs6000_secondary_reload_move): New helper function for secondary
6739 reload to identify moves between register types that can be done.
6740 (rs6000_secondary_reload): Add support for quad memory operations
6741 and for direct move.
6742 (rs6000_secondary_memory_needed): Likewise.
6743 (rs6000_debug_secondary_memory_needed): Change argument names.
6744 (rs6000_output_move_128bit): New function to return the move to
6745 use for 128-bit moves, including knowing about the various
6746 limitations of quad memory operations.
6748 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
6749 memory operations. call rs6000_output_move_128bit for the actual
6750 instruciton(s) to generate.
6751 (vsx_movti_64bit): Likewise.
6753 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
6754 (UNSPEC_P8V_MTVSRWZ): Likewise.
6755 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
6756 (UNSPEC_P8V_MTVSRD): Likewise.
6757 (UNSPEC_P8V_XXPERMDI): Likewise.
6758 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
6759 (UNSPEC_FUSION_GPR): Likewise.
6760 (FMOVE128_GPR): New iterator for direct move.
6761 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
6764 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
6765 loads and direct move instructions.
6766 (zero_extendsidi2_lfiwzx): Likewise.
6767 (extendsidi2_lfiwax): Likewise.
6768 (extendsidi2_nocell): Likewise.
6769 (floatsi<mode>2_lfiwax): Likewise.
6771 (floatunssi<mode>2_lfiwzx): Likewise.
6773 (fix_trunc<mode>_stfiwx): Likewise.
6774 (fixuns_trunc<mode>_stfiwx): Likewise.
6775 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
6776 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
6777 (parity<mode>2_cmpb): Set length/type attr.
6778 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
6779 for 'mr.' to fast_compare.
6780 (bpermd_<mode>): Change type attr to popcnt.
6781 (p8_fmrgow_<mode>): New insns for power8 direct move support.
6782 (p8_mtvsrwz_1): Likewise.
6783 (p8_mtvsrwz_2): Likewise.
6784 (reload_fpr_from_gpr<mode>): Likewise.
6785 (p8_mtvsrd_1): Likewise.
6786 (p8_mtvsrd_2): Likewise.
6787 (p8_xxpermdi_<mode>): Likewise.
6788 (reload_vsx_from_gpr<mode>): Likewise.
6789 (reload_vsx_from_gprsf): Likewise.
6790 (p8_mfvsrd_3_<mode>): LIkewise.
6791 (reload_gpr_from_vsx<mode>): Likewise.
6792 (reload_gpr_from_vsxsf): Likewise.
6793 (p8_mfvsrd_4_disf): Likewise.
6794 (multi-word GPR splits): Do not split direct moves or quad memory
6797 2013-06-10 David Malcolm <dmalcolm@redhat.com>
6799 * tree-into-ssa.c (interesting_blocks): Make static.
6801 2013-06-10 Jakub Jelinek <jakub@redhat.com>
6804 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
6805 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
6806 Use DATA_ABI_ALIGNMENT for that case instead if defined.
6807 (get_variable_align): New function.
6808 (get_variable_section, emit_bss, emit_common,
6809 assemble_variable_contents, place_block_symbol): Use
6810 get_variable_align instead of DECL_ALIGN.
6811 (assemble_noswitch_variable): Add align argument, use it
6812 instead of DECL_ALIGN.
6813 (assemble_variable): Adjust caller. Use get_variable_align
6814 instead of DECL_ALIGN.
6815 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
6817 (DATA_ABI_ALIGNMENT): Define.
6818 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
6819 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
6820 opt is false, only return the psABI mandated alignment increase.
6821 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
6822 (DATA_ABI_ALIGNMENT): ... this.
6823 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
6824 (DATA_ABI_ALIGNMENT): ... this.
6825 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
6826 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
6827 (DATA_ABI_ALIGNMENT): ... this.
6828 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
6829 * doc/tm.texi: Regenerated.
6831 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
6833 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
6834 cmp_code to construct REG_EQUAL note.
6836 2013-06-09 Jakub Jelinek <jakub@redhat.com>
6839 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6840 that operands[2] doesn't overlap with operands[0].
6842 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
6843 Jan Hubicka <jh@suse.cz>
6845 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6846 hack to mark symbols as used.
6848 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
6850 PR rtl-optimization/57559
6851 * lra-constraints.c (process_alt_operands): Don't discourage
6852 memory with known offset for offsetable memory constraint.
6853 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
6855 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
6857 * varasm.c (struct oc_local_state): Reorder fields.
6858 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
6859 and adjust accordingly.
6860 (output_constructor): Reorder initialization code and adjust call to
6861 output_constructor_bitfield.
6863 2013-06-07 Jan Hubicka <jh@suse.cz>
6865 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
6867 2013-06-07 David Malcolm <dmalcolm@redhat.com>
6869 * tree-object-size.c (unknown): Make const.
6871 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6873 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
6874 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
6875 for last alternative in the cpu_facility attribute.
6877 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6880 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
6881 (xordi3): Change operand 2 constraint to arm_xordi_operand.
6882 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
6883 * config/arm/constraints.md (Dg): New constraint.
6884 * config/arm/neon.md (xordi3_neon): Remove.
6885 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
6886 * config/arm/predicates.md (arm_xordi_operand): New predicate.
6888 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6890 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
6891 Clean up alternatives.
6893 2013-06-07 Alan Modra <amodra@gmail.com>
6895 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
6898 2013-06-07 Alan Modra <amodra@gmail.com>
6900 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
6902 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6904 * config/arm/constraints.md (Df): New constraint.
6905 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
6906 Correct length attribute for last two alternatives.
6908 2013-06-07 Alan Modra <amodra@gmail.com>
6910 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6911 override user -mfp-in-toc.
6912 (offsettable_ok_by_alignment): Consider just the current access
6913 rather than the whole object, unless BLKmode. Handle
6914 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
6915 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
6916 for -mcmodel=medium.
6917 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
6918 override user -mfp-in-toc or -msum-in-toc. Default to
6919 -mno-fp-in-toc for -mcmodel=medium.
6921 2013-06-06 DJ Delorie <dj@redhat.com>
6923 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
6924 TARGET_VALID_POINTER_MODE.
6926 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6927 Pat Haugen <pthaugen@us.ibm.com>
6928 Peter Bergner <bergner@vnet.ibm.com>
6930 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
6931 Document new power8 builtins.
6933 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
6934 condition code register, to allow 128-bit logical operations to be
6935 done in the VSX or GPR registers.
6936 (nor<mode>3): Use the canonical form for nor.
6937 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
6938 vclz*, and vpopcnt* vector instructions.
6939 (nand<mode>3): Likewise.
6940 (orc<mode>3): Likewise.
6941 (clz<mode>2): LIkewise.
6942 (popcount<mode>2): Likewise.
6944 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
6945 that only the GPRs are recognized.
6947 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6948 support for new power8 builtins.
6950 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
6952 (xscvdpspn): Likewise.
6958 (vpopcnt): Likewise.
6959 (vpopcntb): Likewise.
6960 (vpopcnth): Likewise.
6961 (vpopcntw): Likewise.
6962 (vpopcntd): Likewise.
6967 (eqv_v16qi3): Likewise.
6968 (eqv_v8hi3): Likewise.
6969 (eqv_v4si3): Likewise.
6970 (eqv_v2di3): Likewise.
6971 (eqv_v4sf3): Likewise.
6972 (eqv_v2df3): Likewise.
6974 (nand_v16qi3): Likewise.
6975 (nand_v8hi3): Likewise.
6976 (nand_v4si3): Likewise.
6977 (nand_v2di3): Likewise.
6978 (nand_v4sf3): Likewise.
6979 (nand_v2df3): Likewise.
6981 (orc_v16qi3): Likewise.
6982 (orc_v8hi3): Likewise.
6983 (orc_v4si3): Likewise.
6984 (orc_v2di3): Likewise.
6985 (orc_v4sf3): Likewise.
6986 (orc_v2df3): Likewise.
6988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
6989 allow power8 quad mode in 64-bit.
6990 (rs6000_builtin_vectorized_function): Add support to vectorize
6991 ISA 2.07 count leading zeros, population count builtins.
6992 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
6993 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
6994 (builtin_function_type): Add vgbbd builtin function which takes an
6996 (altivec_expand_vec_perm_const): Add support for new power8 merge
6999 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7000 that does not include TImdoe for use with 32-bit.
7001 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7003 (UNSPEC_VSX_CVDPSPN): Likewise.
7004 (vsx_xscvdpspn): Likewise.
7005 (vsx_xscvspdpn): Likewise.
7006 (vsx_xscvdpspn_scalar): Likewise.
7007 (vsx_xscvspdpn_directmove): Likewise.
7008 (vsx_and<mode>3): Split logical operations into 32-bit and
7009 64-bit. Add support to do logical operations on TImode as well as
7010 VSX vector types. Allow logical operations to be done in either
7011 VSX registers or in general purpose registers in 64-bit mode. Add
7012 splitters if GPRs were used. For AND, add clobber of CCmode to
7013 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
7015 (vsx_and<mode>3_32bit): Likewise.
7016 (vsx_and<mode>3_64bit): Likewise.
7017 (vsx_ior<mode>3): Likewise.
7018 (vsx_ior<mode>3_32bit): Likewise.
7019 (vsx_ior<mode>3_64bit): Likewise.
7020 (vsx_xor<mode>3): Likewise.
7021 (vsx_xor<mode>3_32bit): Likewise.
7022 (vsx_xor<mode>3_64bit): Likewise.
7023 (vsx_one_cmpl<mode>2): Likewise.
7024 (vsx_one_cmpl<mode>2_32bit): Likewise.
7025 (vsx_one_cmpl<mode>2_64bit): Likewise.
7026 (vsx_nor<mode>3): Likewise.
7027 (vsx_nor<mode>3_32bit): Likewise.
7028 (vsx_nor<mode>3_64bit): Likewise.
7029 (vsx_andc<mode>3): Likewise.
7030 (vsx_andc<mode>3_32bit): Likewise.
7031 (vsx_andc<mode>3_64bit): Likewise.
7032 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7033 and xxlorc instructions.
7034 (vsx_eqv<mode>3_64bit): Likewise.
7035 (vsx_nand<mode>3_32bit): Likewise.
7036 (vsx_nand<mode>3_64bit): Likewise.
7037 (vsx_orc<mode>3_32bit): Likewise.
7038 (vsx_orc<mode>3_64bit): Likewise.
7040 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7042 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7044 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7045 (p8_vmrgow): Likewise.
7046 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7047 GPRs to be split under VSX.
7048 (p8v_clz<mode>2): Add power8 count leading zero support.
7049 (p8v_popcount<mode>2): Add power8 population count support.
7050 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7053 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7056 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7058 (vec_nand): Likewise.
7059 (vec_vclz): Likewise.
7060 (vec_vclzb): Likewise.
7061 (vec_vclzd): Likewise.
7062 (vec_vclzh): Likewise.
7063 (vec_vclzw): Likewise.
7064 (vec_vgbbd): Likewise.
7065 (vec_vmrgew): Likewise.
7066 (vec_vmrgow): Likewise.
7067 (vec_vpopcnt): Likewise.
7068 (vec_vpopcntb): Likewise.
7069 (vec_vpopcntd): Likewise.
7070 (vec_vpopcnth): Likewise.
7071 (vec_vpopcntw): Likewise.
7073 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7075 PR rtl-optimization/57468
7076 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7079 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7081 PR rtl-optimization/57459
7082 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7083 type when setting live regs.
7085 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7087 * config/s390/s390.opt (mlra): New option.
7088 * config/s390/s390.c (s390_decompose_address): Check displacement
7089 for all registers for LRA.
7090 (s390_secondary_reload): Don't used secondary reloads for LRA.
7091 (s390_lra_p): New function.
7092 (TARGET_LRA_P): Define.
7093 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7094 of attribute cpu_facility to zarch for the last alternative.
7095 (*cmpmem_short): Ditto.
7097 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
7099 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7100 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
7101 (arm_expand_prologue): Likewise.
7103 2013-06-06 Teresa Johnson <tejohnson@google.com>
7106 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7107 as this is now done by redirect_edge_and_branch_force.
7108 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7109 barriers, and fix interaction with splitting.
7110 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7111 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7112 reflect changes made in the routine.
7113 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7114 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7115 since this is called in cfglayout mode, and replace partition fixup
7116 with assert as that is now done by force_nonfallthru_and_redirect.
7117 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7118 already be marked with region crossing note.
7119 (insert_section_boundary_note): Make non-static, gate on flag
7120 has_bb_partition, rewrite to also check for multiple partitions.
7121 (rest_of_handle_reorder_blocks): Remove call to
7122 insert_section_boundary_note, now done later during free_cfg.
7123 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7124 * bb-reorder.h (insert_section_boundary_note): Declare.
7125 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7126 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7127 invoke insert_section_boundary_note.
7128 (try_redirect_by_replacing_jump): Remove unnecessary
7129 check for region crossing note.
7130 (fixup_partition_crossing): New function.
7131 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7132 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7133 in non-cfglayout mode.
7134 (force_nonfallthru_and_redirect): Fixup partition boundaries,
7135 remove old code that tried to do this. Emit barrier correctly
7136 when we are in cfglayout mode.
7137 (last_bb_in_partition): New function.
7138 (rtl_split_edge): Correctly fixup partition boundaries.
7139 (commit_one_edge_insertion): Remove old code that tried to
7140 fixup region crossing edge since this is now handled in
7141 split_block, and set up insertion point correctly since
7142 block may now end in a jump.
7143 (verify_hot_cold_block_grouping): Guard against checking when not in
7144 linearized RTL mode.
7145 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7147 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7148 rtl_verify_flow_info, so not called in cfglayout mode.
7149 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7150 (fixup_reorder_chain): Remove old code that attempted to fixup region
7151 crossing note as this is now handled in force_nonfallthru_and_redirect.
7152 (duplicate_insn_chain): Don't duplicate switch section notes.
7153 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7155 * basic-block.h (emit_barrier_after_bb): Declare.
7157 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7159 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7160 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7161 arm_usatsihi): Adjust alternatives for arm_restrict_it.
7163 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7165 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7167 * config/arm/ldmstm.md: Regenerate.
7169 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7171 * config/arm/sync.md (atomic_loaddi_1):
7172 Disable predication for arm_restrict_it.
7173 (arm_load_exclusive<mode>): Likewise.
7174 (arm_load_exclusivesi): Likewise.
7175 (arm_load_exclusivedi): Likewise.
7176 (arm_load_acquire_exclusive<mode>): Likewise.
7177 (arm_load_acquire_exclusivesi): Likewise.
7178 (arm_load_acquire_exclusivedi): Likewise.
7179 (arm_store_exclusive<mode>): Likewise.
7180 (arm_store_exclusive<mode>): Likewise.
7181 (arm_store_release_exclusivedi): Likewise.
7182 (arm_store_release_exclusive<mode>): Likewise.
7184 2013-06-06 Richard Biener <rguenther@suse.de>
7186 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7188 (lto_tag_is_tree_code_p): Adjust.
7189 (lto_tag_is_gimple_code_p): Likewise.
7190 (lto_gimple_code_to_tag): Likewise.
7191 (lto_tag_to_gimple_code): Likewise.
7192 (lto_tree_code_to_tag): Likewise.
7193 (lto_tag_to_tree_code): Likewise.
7194 * data-streamer.h (streamer_write_hwi_in_range): Use
7195 uhwi streaming to stream the normalized range.
7196 (streamer_read_hwi_in_range): Likewise.
7198 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7200 * config/arm/arm.md (enabled_for_depr_it): New attribute.
7201 (predicable_short_it): Likewise.
7202 (predicated): Likewise.
7203 (enabled): Handle above.
7204 (define_cond_exec): Set predicated attribute to yes.
7206 2013-06-05 Mike Stump <mikestump@comcast.net>
7208 * gdbinit.in (__FUNCTION__): Add.
7210 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
7212 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7213 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7215 2013-06-05 Jan Hubicka <jh@suse.cz>
7217 * varasm.c (mark_decl_referenced): Revert the removal until targets
7220 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
7222 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7223 instead of mark_decl_referenced.
7225 2013-06-05 Jan Hubicka <jh@suse.cz>
7227 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7228 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7229 and symtab_used_from_object_file_p.
7230 (cgraph_make_node_local_1): Clear forced_by_abi.
7231 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7232 * cgraph.h (symtab_node_base): Add forced_by_abi.
7233 (decide_is_variable_needed): Remove.
7234 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7235 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7236 (decide_is_symbol_needed): ... this one; handle symbols in general;
7237 always analyze virtuals; honnor forced_by_abi.
7238 (cgraph_finalize_function): Update.
7239 (varpool_finalize_decl): Update.
7240 (symbol_defined_and_needed): Remove.
7241 (analyze_functions): Update.
7242 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7243 output_refs, input_overwrite_node): Handle forced_by_abi.
7244 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7245 (address_taken_from_non_vtable_p): ... this one.
7246 (comdat_can_be_unshared_p_1): New function.
7247 (cgraph_comdat_can_be_unshared_p): Rename to ...
7248 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7249 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7250 (function_and_variable_visibility): Clear forced_by_abi as needed.
7251 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7252 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7253 * symtab.c (dump_symtab_base): Dump forced_by_abi.
7254 * varpool.c (decide_is_variable_needed): Remove.
7256 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7258 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7259 (arm_option_override): Override arm_restrict_it where appropriate.
7260 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7261 * config/arm/arm.opt (mrestrict-it): New command-line option.
7262 * doc/invoke.texi: Document -mrestrict-it.
7264 2013-06-05 David Malcolm <dmalcolm@redhat.com>
7266 * tsan.c (tsan_atomic_table): Make const.
7268 2013-06-05 Richard Biener <rguenther@suse.de>
7270 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7271 index associated with the tree we are supposed to replace.
7272 * tree-streamer-out.c (pack_ts_base_value_fields): Output
7273 TREE_ASM_WRITTEN as zero for everything but SSA names.
7275 2013-06-05 David Malcolm <dmalcolm@redhat.com>
7277 * tree-ssa-structalias.c (call_stmt_vars): Make static.
7279 2013-06-04 Jan Hubicka <jh@suse.cz>
7281 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7282 (input_node, input_varpool_node): Handle correctly external same
7284 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7285 nodes at ltrans stage.
7287 2013-06-04 Jan Hubicka <jh@suse.cz>
7289 * ipa-inline.c (update_caller_keys): Fix availability test.
7290 (update_callee_keys): Likewise.
7291 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7292 to follow ELF standard.
7294 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
7296 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7297 (mips64r5900el-*-elf*): New configurations.
7298 * config/mips/mips-cpus.def (r5900): New processor.
7299 * config/mips/mips-tables.opt: Regenerate.
7300 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7301 (mips_issue_rate): Handle PROCESSOR_R5900.
7302 (mips_reorg_process_insns): Force reorder mode for the R5900.
7303 * config/mips/mips.h (TARGET_MIPS5900): Define.
7304 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7306 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7308 * config/mips/mips.md (processor): Add r5900.
7309 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7311 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7313 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7314 into function to generate MOVI instruction.
7315 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7316 (aarch64_preferred_simd_mode): Turn into wrapper.
7317 (aarch64_output_scalar_simd_mov_immediate): New function.
7318 * config/aarch64/aarch64-protos.h: Add prototype for above.
7320 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7322 * config/aarch64/aarch64.c (simd_immediate_info): Remove
7323 element_char member.
7324 (sizetochar): Return signed char.
7325 (aarch64_simd_valid_immediate): Remove elchar and other
7326 unnecessary variables.
7327 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7328 Calculate element_char as required.
7329 * config/aarch64/aarch64-protos.h: Update and move prototype
7330 for aarch64_output_simd_mov_immediate.
7331 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7334 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7336 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7337 information completed by aarch64_simd_valid_immediate.
7338 (aarch64_legitimate_constant_p): Update arguments.
7339 (aarch64_simd_valid_immediate): Work with struct rather than many
7341 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7342 (aarch64_simd_make_constant): Update arguments.
7343 (aarch64_output_simd_mov_immediate): Work with struct rather than
7344 many pointers. Output immediate directly rather than as operand.
7345 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7347 * config/aarch64/constraints.md (Dn): Update arguments.
7349 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7351 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7353 (aarch64_simd_immediate_valid_for_move): Remove.
7354 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7355 (aarch64_simd_make_constant): Update call.
7356 (aarch64_output_simd_mov_immediate): Update call.
7357 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7359 * config/aarch64/constraints.md (Dn): Update call.
7361 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7363 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7364 return type to bool for prototype.
7365 (aarch64_legitimate_constant_p): Check for true instead of not -1.
7366 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7367 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7369 2013-06-04 Catherine Moore <clm@codesourcery.com>
7371 * config/mips/mips.opt (meva): New.
7372 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7373 (ASM_SPEC): Handle -meva.
7374 * doc/invoke.texi (meva): Document.
7376 2013-06-04 Alan Modra <amodra@gmail.com>
7378 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7381 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7383 * rtl.def: Add extra fourth optional field to define_cond_exec.
7384 * gensupport.c (process_one_cond_exec): Process attributes from
7386 * doc/md.texi: Document fourth field in define_cond_exec.
7388 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
7390 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7391 out the processing order as in store_bit_field_1.
7393 2013-06-04 Jan Hubicka <jh@suse.cz>
7396 * cgraphunit.c (cgraph_process_same_body_aliases): Create
7397 non-VAR_DECL node if it does not exist yet.
7399 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
7401 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7402 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7403 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7404 target_cpu_default setting.
7406 2013-06-03 Teresa Johnson <tejohnson@google.com>
7408 * dumpfile.c (opt_info_switch_p): Change -fopt-info
7409 default to -fopt-info=optimized instead of all.
7410 * doc/invoke.texi: Ditto.
7411 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7412 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7413 (execute_vect_slp): Emit BB vectorization success under
7414 MSG_OPTIMIZED_LOCATIONS.
7415 * tree-vect-slp.c (vect_slp_transform_bb): Change
7416 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7417 * tree-vect-loop.c (vect_transform_loop): Ditto.
7419 2013-06-03 Jason Merrill <jason@redhat.com>
7422 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7423 Use TARGET_EXPR for C++.
7425 2013-06-03 Jakub Jelinek <jakub@redhat.com>
7427 PR rtl-optimization/57268
7428 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7429 if DEBUG_INSN_P (insn).
7432 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
7434 PR rtl-optimization/57268
7435 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7436 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7438 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
7440 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7441 (ix86_avoid_lea_for_addr): Likewise.
7442 (exact_dependency_1): Likewise.
7443 (ix86_adjust_cost): Likewise.
7444 (swap_top_of_ready_list): Fix formatting and !reload_completed check
7446 (do_reorder_for_imul): Fix typo, formatting and
7447 !reload_completed check removed.
7448 (ix86_sched_reorder): Fix typo and formatting.
7449 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7452 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
7454 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7456 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
7458 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7459 <STRING_CST>: Likewise.
7460 <VECTOR_CST>: Likewise.
7462 2013-06-01 Janus Weil <janus@gcc.gnu.org>
7463 Mikael Morin <mikael@gcc.gnu.org>
7465 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7466 * config.in: Regenerated.
7467 * configure: Regenerated.
7469 2013-06-01 Jan Hubicka <jh@suse.cz>
7472 * cgraphunit.c (compile): When weakref is not supported,
7473 set up transparent aliases before final output pass.
7474 * varasm.c (assemble_alias): Do not try to do it here.
7476 2013-06-01 Jan Hubicka <jh@suse.cz>
7479 * passes.c (for_per_function): Skip unanalyzed functions.
7481 2013-06-01 Jan Hubicka <jh@suse.cz>
7483 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7484 (lto_symtab_merge_symbols_1): ... this one.
7485 (lto_symtab_merge_cgraph_nodes): Rename to ...
7486 (lto_symtab_merge_symbols): ... this one; simplify.
7487 * cgraph.c (same_body_aliases_done): Rename to ...
7488 (cpp_implicit_aliases_done): ... this one.
7489 (cgraph_create_function_alias): Update.
7490 (cgraph_same_body_alias): Update.
7491 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7492 (verify_edge_corresponds_to_fndecl): Simplify.
7493 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7494 (cgraph_node): Remove same_body_alias.
7495 (varpool_node): Remove alias_of and extra_name_alias.
7496 (same_body_aliases_done): Rename to ..
7497 (cpp_implicit_aliases_done): ... this one.
7498 (symtab_alias_ultimate_target): Add default parameter.
7499 (symtab_resolve_alias): New function.
7500 (fixup_same_cpp_alias_visibility): Declare.
7501 (cgraph_function_node): Add default parameter.
7502 (cgraph_node_asm_name): Likewise.
7503 (cgraph_function_or_thunk_node): Add default parameter; do
7504 not ICE when it is NULL.
7505 (varpool_variable_node): Likewise.
7506 * tree-emutls.c (create_emultls_var): Update.
7507 (ipa_lower_emutls): Update.
7508 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7509 (cgraph_reset_node): Reset alias info.
7510 (cgraph_finalize_function): Update.
7511 (fixup_same_cpp_alias_visibility): Move to symtab.c.
7512 (analyze_function): Simplify.
7513 (cgraph_process_same_body_aliases): Simplify.
7514 (analyze_functions): Fixup same body aliases.
7515 (handle_alias_pairs): Simplify.
7516 (assemble_thunk): Update.
7517 (assemble_thunks_and_aliases): Update.
7518 (output_weakrefs): Rewrite.
7519 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7520 (lto_output_varpool_node): Likewise.
7521 (compute_ltrans_boundary): Remve assert.
7522 (get_alias_symbol): New functoin.
7523 (input_node): Rewrite alias handling.
7524 (input_varpool_node): Likewise.
7525 * ipa-pure-const.c (propagate_pure_const): Fix formating.
7526 * ipa.c (process_references): Handle weakrefs correctly.
7527 (symtab_remove_unreachable_nodes): Likewise.
7528 * trans-mem.c (get_cg_data): Update.
7529 (ipa_tm_create_version_alias): Update.
7530 (ipa_tm_execute): Update.
7531 * symtab.c (dump_symtab_base): Dump aliases.
7532 (verify_symtab_base): Verify aliases.
7533 (symtab_node_availability): New function.
7534 (symtab_alias_ultimate_target): Simplify.
7535 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7536 handle all the fixup cases.
7537 (symtab_resolve_alias): New function.
7538 * passes.c (ipa_write_summaries): Handle weakrefs.
7539 * varpool.c (varpool_analyze_node): Simplify.
7540 (assemble_aliases): Update.
7541 (varpool_create_variable_alias): Simplify.
7542 (varpool_extra_name_alias): Simplify.
7543 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7544 (lto_symtab_merge_symbols): ... this one.
7546 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
7549 PR rtl-optimization/57268
7550 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7551 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7553 2013-06-01 Tobias Burnus <burnus@net-b.de>
7556 2013-05-31 Tobias Burnus <burnus@net-b.de>
7559 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7562 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
7564 PR rtl-optimization/57268
7565 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7566 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7568 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
7570 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7571 unordered comparison operators when -fno-trapping-math is in effect
7573 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7574 and implement unordered comparison operators properly on the e500.
7576 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
7578 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7579 for constant scalar integers.
7580 (simplify_relational_operation_1): Likewise.
7582 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7584 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7585 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
7588 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
7589 Igor Zamyatin <igor.zamyatin@intel.com>
7591 Silvermont (SLM) architecture performance tuning.
7592 * config/i386/i386.h (enum ix86_tune_indices): Add
7593 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7594 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
7596 * config/i386/i386.c (initial_ix86_tune_features)
7597 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
7598 (ix86_lea_outperforms): Handle Silvermont tuning.
7599 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
7601 (ix86_use_lea_for_mov): Likewise.
7602 (ix86_avoid_lea_for_addr): Likewise.
7603 (ix86_lea_for_add_ok): Likewise.
7604 (exact_dependency_1): New function.
7605 (exact_store_load_dependency): Likewise.
7606 (ix86_adjust_cost): Handle Silvermont tuning.
7607 (do_reoder_for_imul): Likewise.
7608 (swap_top_of_ready_list): New function.
7609 (ix86_sched_reorder): Changed to handle Silvermont tuning.
7611 * config/i386/i386.md (peepholes that split memory operand in fp
7614 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
7616 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7617 Remove un-necessary braces.
7619 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
7621 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7622 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
7624 2013-05-31 Tobias Burnus <burnus@net-b.de>
7627 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
7629 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7632 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
7633 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
7634 * config/arm/neon.md (iordi3_neon): Remove.
7635 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
7636 * config/arm/predicates.md (imm_for_neon_logic_operand):
7637 Move to earlier in the file.
7638 (neon_logic_op2): Likewise.
7639 (arm_iordi_operand_neon): New predicate.
7641 2013-05-31 Richard Biener <rguenther@suse.de>
7643 PR tree-optimization/57478
7644 PR tree-optimization/57453
7645 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
7648 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
7650 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
7651 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
7653 2013-05-30 Tobias Burnus <burnus@net-b.de>
7654 Thomas Koenig <tkoenig@gcc.gnu.org>
7657 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
7658 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
7660 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
7662 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
7664 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
7666 * target.def (register_usage_leveling_p): New hook.
7667 * targhooks.c (default_register_usage_leveling_p): New.
7668 * targhooks.h (default_register_usage_leveling_p): New prototype.
7669 * lra-assigns.c (register_usage_leveling_p): Use the hook.
7670 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
7671 * doc/tm.texi: Update.
7672 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
7674 2013-05-30 Ian Bolton <ian.bolton@arm.com>
7676 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
7677 (*insv_reg<mode>): New define_insn.
7679 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
7681 PR rtl-optimization/57439
7682 * postreload.c (move2add_valid_value_p): Check that we have
7683 a zero subreg_regno_offset when accessing the register in
7686 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
7687 Igor Zamyatin <igor.zamyatin@intel.com>
7689 Silvermont (SLM) architecture pipeline model, tuning and
7691 * config.gcc: Add slm config options and target.
7693 * config/i386/slm.md: New.
7695 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
7697 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): New case
7699 (ix86_target_macros_internal): Likewise.
7701 * gcc/config/i386/i386.c (slm_cost): New cost.
7702 (m_SLM): New macro flag.
7703 (initial_ix86_tune_features): Set m_SLM.
7704 (x86_accumulate_outgoing_args): Likewise.
7705 (x86_arch_always_fancy_math_387): Likewise.
7706 (processor_target_table): Add slm cost.
7707 (cpu_names): Add slm cpu name.
7708 (x86_option_override_internal): Set SLM ISA.
7709 (ix86_issue_rate): New case PROCESSOR_SLM.
7710 (ia32_multipass_dfa_lookahead): Likewise.
7711 (fold_builtin_cpu): Add slm.
7713 * config/i386/i386.h (TARGET_SLM): New target macro.
7714 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
7715 (processor_type): Add PROCESSOR_SLM.
7717 * config/i386/i386.md (cpu): Add new value "slm".
7718 (slm.md): Include slm.md.
7720 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
7721 Zhenqiang Chen <zhenqiang.chen@linaro.org>
7723 * config/arm/arm-protos.h: Add and update function protos.
7724 * config/arm/arm.c (use_simple_return_p): New added.
7725 (thumb2_expand_return): Check simple_return flag.
7726 * config/arm/arm.md: Add simple_return and conditional simple_return.
7727 * config/arm/iterators.md: Add iterator for return and simple_return.
7729 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
7731 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
7732 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
7733 (arm_emit_vfp_multi_reg_pop): Likewise.
7734 (thumb2_emit_ldrd_pop): Likewise.
7735 (arm_expand_epilogue): Add misc REG_CFA notes.
7736 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
7738 2013-05-29 Lawrence Crowl <crowl@google.com>
7740 * config/arm/t-arm: Update for below.
7742 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
7743 Change type to hash_table. Update dependent calls and types.
7745 * config/i386/t-cygming: Update for below.
7747 * config/i386/t-interix: Update for below.
7749 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
7750 Change type to hash_table. Update dependent calls and types.
7751 (i386_find_on_wrapper_list::wrappers): Likewise.
7753 * config/ia64/t-ia64: Update for below.
7755 * config/ia64/ia64.c (bundle_state_table):
7756 Change type to hash_table. Update dependent calls and types.
7758 * config/mips/mips.c (mips_reorg_process_insns::htab):
7759 Change type to hash_table. Update dependent calls and types.
7761 * config/sol2.c (solaris_comdat_htab):
7762 Change type to hash_table. Update dependent calls and types.
7764 * config/t-sol2: Update for above.
7766 2013-05-29 Teresa Johnson <tejohnson@google.com>
7768 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
7769 functions are not yet marked as defined.
7771 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7772 Pat Haugen <pthaugen@us.ibm.com>
7773 Peter Bergner <bergner@vnet.ibm.com>
7775 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
7779 (vrotl<mode>3): Likewise.
7780 (vashl<mode>3): Likewise.
7781 (vlshr<mode>3): Likewise.
7782 (vashr<mode>3): Likewise.
7784 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7785 support for power8 V2DI builtins.
7787 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
7788 power8 V2DI builtins.
7789 (vupkhsw): Likewise.
7790 (vupklsw): Likewise.
7791 (vaddudm): Likewise.
7796 (vpkudum): Likewise.
7797 (vpksdss): Likewise.
7798 (vpkudus): Likewise.
7799 (vpksdus): Likewise.
7804 (vsubudm): Likewise.
7805 (vcmpequd): Likewise.
7806 (vcmpgtsd): Likewise.
7807 (vcmpgtud): Likewise.
7808 (vcmpequd_p): Likewise.
7809 (vcmpgtsd_p): Likewise.
7810 (vcmpgtud_p): Likewise.
7811 (vupkhsw): Likewise.
7812 (vupklsw): Likewise.
7813 (vaddudm): Likewise.
7818 (vpksdss): Likewise.
7819 (vpksdus): Likewise.
7820 (vpkudum): Likewise.
7821 (vpkudus): Likewise.
7826 (vsubudm): Likewise.
7828 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
7829 support for power8 V2DI instructions.
7831 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
7832 power8 V2DI instructions. Combine pack and unpack insns to use an
7833 iterator for each mode. Check whether a particular mode supports
7834 Altivec instructions instead of just checking TARGET_ALTIVEC.
7835 (UNSPEC_VPKUWUM): Likewise.
7836 (UNSPEC_VPKSHSS): Likewise.
7837 (UNSPEC_VPKSWSS): Likewise.
7838 (UNSPEC_VPKUHUS): Likewise.
7839 (UNSPEC_VPKSHUS): Likewise.
7840 (UNSPEC_VPKUWUS): Likewise.
7841 (UNSPEC_VPKSWUS): Likewise.
7842 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7843 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7844 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7845 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7846 (UNSPEC_VUPKHSB): Likewise.
7847 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7848 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7849 (UNSPEC_VUPKHSH): Likewise.
7850 (UNSPEC_VUPKLSB): Likewise.
7851 (UNSPEC_VUPKLSH): Likewise.
7853 (VI_char): Likewise.
7854 (VI_scalar): Likewise.
7855 (VI_unit): Likewise.
7857 (VP_small): Likewise.
7858 (VP_small_lc): Likewise.
7859 (VU_char): Likewise.
7860 (add<mode>3): Likewise.
7861 (altivec_vaddcuw): Likewise.
7862 (altivec_vaddu<VI_char>s): Likewise.
7863 (altivec_vadds<VI_char>s): Likewise.
7864 (sub<mode>3): Likewise.
7865 (altivec_vsubcuw): Likewise.
7866 (altivec_vsubu<VI_char>s): Likewise.
7867 (altivec_vsubs<VI_char>s): Likewise.
7868 (altivec_vavgs<VI_char>): Likewise.
7869 (altivec_vcmpbfp): Likewise.
7870 (altivec_eq<mode>): Likewise.
7871 (altivec_gt<mode>): Likewise.
7872 (altivec_gtu<mode>): Likewise.
7873 (umax<mode>3): Likewise.
7874 (smax<mode>3): Likewise.
7875 (umin<mode>3): Likewise.
7876 (smin<mode>3): Likewise.
7877 (altivec_vpkuhum): Likewise.
7878 (altivec_vpkuwum): Likewise.
7879 (altivec_vpkshss): Likewise.
7880 (altivec_vpkswss): Likewise.
7881 (altivec_vpkuhus): Likewise.
7882 (altivec_vpkshus): Likewise.
7883 (altivec_vpkuwus): Likewise.
7884 (altivec_vpkswus): Likewise.
7885 (altivec_vpks<VI_char>ss): Likewise.
7886 (altivec_vpks<VI_char>us): Likewise.
7887 (altivec_vpku<VI_char>us): Likewise.
7888 (altivec_vpku<VI_char>um): Likewise.
7889 (altivec_vrl<VI_char>): Likewise.
7890 (altivec_vsl<VI_char>): Likewise.
7891 (altivec_vsr<VI_char>): Likewise.
7892 (altivec_vsra<VI_char>): Likewise.
7893 (altivec_vsldoi_<mode>): Likewise.
7894 (altivec_vupkhsb): Likewise.
7895 (altivec_vupkhs<VU_char>): Likewise.
7896 (altivec_vupkls<VU_char>): Likewise.
7897 (altivec_vupkhsh): Likewise.
7898 (altivec_vupklsb): Likewise.
7899 (altivec_vupklsh): Likewise.
7900 (altivec_vcmpequ<VI_char>_p): Likewise.
7901 (altivec_vcmpgts<VI_char>_p): Likewise.
7902 (altivec_vcmpgtu<VI_char>_p): Likewise.
7903 (abs<mode>2): Likewise.
7904 (vec_unpacks_hi_v16qi): Likewise.
7905 (vec_unpacks_hi_v8hi): Likewise.
7906 (vec_unpacks_lo_v16qi): Likewise.
7907 (vec_unpacks_hi_<VP_small_lc>): Likewise.
7908 (vec_unpacks_lo_v8hi): Likewise.
7909 (vec_unpacks_lo_<VP_small_lc>): Likewise.
7910 (vec_pack_trunc_v8h): Likewise.
7911 (vec_pack_trunc_v4si): Likewise.
7912 (vec_pack_trunc_<mode>): Likewise.
7914 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
7916 (vec_vmaxsd): Likewise.
7917 (vec_vmaxud): Likewise.
7918 (vec_vminsd): Likewise.
7919 (vec_vminud): Likewise.
7920 (vec_vpksdss): Likewise.
7921 (vec_vpksdus): Likewise.
7922 (vec_vpkudum): Likewise.
7923 (vec_vpkudus): Likewise.
7924 (vec_vrld): Likewise.
7925 (vec_vsld): Likewise.
7926 (vec_vsrad): Likewise.
7927 (vec_vsrd): Likewise.
7928 (vec_vsubudm): Likewise.
7929 (vec_vupkhsw): Likewise.
7930 (vec_vupklsw): Likewise.
7932 2013-05-29 Jan Hubicka <jh@suse.cz>
7934 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
7935 flags; reorder rest of fields in more consistent way.
7936 (varpool_node): Remove analyzed, finalized and alias.
7937 (cgraph_ndoe): Likewise.
7938 (symtab_alias_ultimate_target): New function.
7939 (cgraph_function_node): Move offline.
7940 (cgraph_reset_node): Declare.
7941 (cgraph_comdat_can_be_unshared_p): Remove.
7942 (varpool_remove_initializer): Declare.
7943 (varpool_first_defined_variable, varpool_next_defined_variable
7944 cgraph_first_defined_function, cgraph_next_defined_function): Update.
7945 (cgraph_function_with_gimple_body_p): Update.
7946 (varpool_all_refs_explicit_p): Update.
7947 (symtab_alias_target): New function.
7948 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
7949 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
7950 (cgraph_function_or_thunk_node): Simplify using
7951 symtab_alias_ultimate_target.
7952 (varpool_variable_node): Likewise.
7953 * cgraph.c (cgraph_create_function_alias): Update.
7954 (cgraph_add_thunk): Update.
7955 (cgraph_remove_node): Update.
7956 (dump_cgraph_node): Do not dump removed flags.
7957 (cgraph_function_body_availability): Update.
7958 (cgraph_propagate_frequency): Update.
7959 (verify_cgraph_node): Check sanity of local flag.
7960 (cgraph_function_node): Move here from cgraph.h; revamp for
7961 cgraph_function_or_thunk_node.
7962 * lto-symtab.c (lto_varpool_replace_node): Update.
7963 (lto_symtab_resolve_can_prevail_p): Update.
7964 (lto_symtab_merge_cgraph_nodes): Update.
7965 * ipa-cp.c (determine_versionability, initialize_node_lattices,
7966 propagate_constants_accross_call, devirtualization_time_bonus,
7967 ipcp_propagate_stage): Update.
7968 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
7969 * ipa-inline-transform.c (clone_inlined_nodes,
7970 preserve_function_body_p): Update.
7971 * ipa-reference.c (propagate): Update.
7972 (write_node_summary_p): Update.
7973 * toplev.c (wrapup_global_declaration_2): Update.
7974 * cgraphunit.c (cgraph_analyze_function): Rename to ...
7975 (analyze_function) ... this one.
7976 (cgraph_process_new_functions): Update.
7977 (cgraph_reset_node): Export.
7978 (cgraph_finalize_function): Update.
7979 (cgraph_add_new_function): Update.
7980 (process_function_and_variable_attributes): Update.
7981 (varpool_finalize_decl): Update.
7982 (symbol_finalized): Remove.
7983 (symbol_finalized_and_needed): Rename to ...
7984 (symbol_defined_and_needed): ... update.
7985 (cgraph_analyze_functions): Update.
7986 (handle_alias_pairs): Update.
7987 (mark_functions_to_output): Update.
7988 (assemble_thunk): Update.
7989 (output_in_order): Update.
7990 (output_weakrefs): Update.
7991 (finalize_compilation_unit): Update.
7992 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
7993 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
7994 input_node, input_varpool_node): Update.
7995 * dbxout.c (dbxout_expand_expr): Update.
7996 * cgraphclones.c (cgraph_clone_node): Update.
7997 (cgraph_copy_node_for_versioning): Update.
7998 (cgraph_materialize_clone): Update.
7999 (cgraph_materialize_all_clones): Update.
8000 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8001 propagate_pure_const, propagate_nothrow): Update.
8002 * lto-streamer-out.c (lto_output, write_symbol): Update.
8003 * ipa-utils.c (ipa_reverse_postorder): Update.
8004 * ipa-inline.c (can_inline_edge_p): Update.
8005 (update_caller_keys, ipa_inline): Update.
8006 * dwarf2out.c (reference_to_unused,
8007 premark_types_used_by_global_vars_helper): Update.
8008 * tree-eh.c (tree_could_trap_p): Update.
8009 * ipa-split.c (consider_split, execute_split_functions): Update.
8010 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8011 has_addr_references_p): Update; move ahead in file for better
8013 (process_references): Simplify.
8014 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8016 (cgraph_comdat_can_be_unshared_p): Make static.
8017 (cgraph_externally_visible_p): Update.
8018 (varpool_externally_visible_p): Update.
8019 (function_and_variable_visibility): Update.
8020 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8021 ipa_tm_mark_force_output_node): Update.
8022 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8023 estimate_edge_devirt_benefit, inline_generate_summary,
8024 inline_write_summary): Update.
8025 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8026 * ipa-prop.c (ipa_compute_jump_functions): Update.
8027 (ipa_print_node_params, ipa_prop_read_section,
8028 ipa_update_after_lto_read, read_replacements_section): Update.
8029 * varasm.c (mark_decl_referenced): Update.
8030 (assemble_alias, dump_tm_clone_pairs): Update.
8031 * tree-inline.c (copy_bb): Update.
8032 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8034 * symtab.c (dump_symtab_base): Print new flags.
8035 (verify_symtab_base): Verify new flags.
8036 (symtab_alias_ultimate_target): New function.
8037 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8038 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8040 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8042 * i386.c (ix86_get_function_versions_dispatcher,
8043 ix86_generate_version_dispatcher_body): Update.
8044 (fold_builtin_cpu): Use varpool_add_new_variable.
8045 * varpool.c (varpool_remove_initializer): Break out from ...
8046 (varpool_remove_node): ... this one.
8047 (dump_varpool_node, varpool_node_for_asm,
8048 cgraph_variable_initializer_availability, varpool_analyze_node,
8049 varpool_assemble_decl, varpool_remove_unreferenced_decls,
8050 varpool_finalize_named_section_flags, varpool_create_variable_alias):
8053 2013-05-29 Jan Hubicka <jh@suse.cz>
8055 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8057 2013-05-29 Easwaran Raman <eraman@google.com>
8059 PR tree-optimization/57442
8060 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8061 when control exits the main loop.
8063 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
8065 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8067 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8068 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8069 * rx/t-rx: Add rx100 under multi library matches option for nofpu
8072 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8074 PR tree-optimization/57441
8075 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8076 Don't limit size of incr_vec to number of candidates.
8078 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
8080 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8081 and mips16 directories.
8082 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8083 (MULTILIB_DIRNAMES): Ditto.
8084 (MULTILIB_EXCEPTIONS): Add new exceptions.
8085 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8086 (MULTILIB_DIRNAMES): Ditto.
8087 (MULTILIB_EXCEPTIONS): Add new exceptions.
8089 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8090 Marcus Shawcroft <marcus.shawcroft@arm.com>
8092 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8093 SYMBOL_TINY_ABSOLUTE.
8094 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8095 SYMBOL_TINY_ABSOLUTE.
8096 (aarch64_expand_mov_immediate): Likewise.
8097 (aarch64_classify_symbol): Likewise.
8098 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8099 Permit SYMBOL_TINY_ABSOLUTE.
8100 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8102 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8103 Marcus Shawcroft <marcus.shawcroft@arm.com>
8105 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8106 Refactor if/switch. Replace gcc_assert with if.
8108 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8110 * config/i386/i386.c (initial_ix86_tune_features): Enable
8111 FP Reassociation for AMD bdver1 and bdver2.
8113 2013-05-29 Martin Jambor <mjambor@suse.cz>
8115 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8116 and IMAGPART_EXPR do not occur within other handled_components.
8118 2013-05-29 Richard Biener <rguenther@suse.de>
8120 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8121 access on whether the use is in the BB we currently try to
8123 (vect_bb_vectorization_profitable_p): Pass the BB we currently
8124 vectorize to vect_bb_slp_scalar_cost.
8126 2013-05-29 Richard Biener <rguenther@suse.de>
8128 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8129 computing scalar cost offsetted by stmts that are kept live
8131 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8132 for computation of scalar cost.
8134 2013-05-28 Steve Ellcey <sellcey@mips.com>
8136 * config/mips/mips-cpus.def (mips32r2): Change processor type.
8138 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
8140 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8141 array notation built-in reduction functions.
8142 * doc/passes.texi (Passes): Added documentation about changes done
8144 * doc/invoke.texi (C Dialect Options): Added documentation about
8145 the -fcilkplus flag.
8146 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8147 (BUILTINS_DEF): Depend on cilkplus.def.
8148 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
8149 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8150 * cilkplus.def: New file.
8152 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
8154 PR rtl-optimization/57439
8155 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8157 2013-05-28 Easwaran Raman <eraman@google.com>
8159 PR tree-optimization/57337
8160 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8161 (find_insert_point): Correctly identify the insertion point
8162 when two statements with the same UID is compared.
8164 2013-05-28 Richard Biener <rguenther@suse.de>
8166 PR tree-optimization/56787
8167 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8168 from the list of data references.
8169 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8171 (vect_analyze_loop_operations): Likewise.
8172 (vect_transform_loop): Remove clobbers.
8174 2013-05-28 Martin Jambor <mjambor@suse.cz>
8176 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8177 and REALPART_EXPRs have scalar type.
8179 2013-05-28 Richard Biener <rguenther@suse.de>
8181 PR tree-optimization/57411
8182 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8184 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8185 virtual operand propagation.
8187 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
8189 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8190 destination register for bmasksi_vis.
8191 (vector_init_bshuffle): Likewise.
8192 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8194 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
8196 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8197 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8198 mode if the instruction isn't available in the original mode.
8199 * config/sparc/sparc.opt (mfix-ut699): New option.
8200 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8201 (divdf3): Turn into expander.
8202 (divdf3_nofix): New insn.
8203 (divdf3_fix): Likewise.
8204 (divsf3): Disable if -mfix-ut699.
8205 (sqrtdf2): Turn into expander.
8206 (sqrtdf2_nofix): New insn.
8207 (sqrtdf2_fix): Likewise.
8208 (sqrtsf2): Disable if -mfix-ut699.
8210 2013-05-27 Richard Biener <rguenther@suse.de>
8213 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8214 block for the new loop.
8216 2013-05-27 Richard Biener <rguenther@suse.de>
8218 PR tree-optimization/57343
8219 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8220 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8221 (number_of_iterations_cond): Do not build the folded tree.
8223 2013-05-27 Richard Biener <rguenther@suse.de>
8227 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8228 OEP_CONSTANT_ADDRESS_OF retained.
8230 PR tree-optimization/57417
8231 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8233 (set_ssa_val_to): Compare addresses using
8234 get_addr_base_and_unit_offset.
8236 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
8238 PR rtl-optimization/56833
8239 * postreload.c (move2add_record_mode): New function.
8240 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8241 (move2add_use_add2_insn): Use move2add_record_sym_value.
8242 (move2add_use_add3_insn): Likewise.
8243 (reload_cse_move2add): Use move2add_valid_value_p and
8244 move2add_record_mode. Invalidate call-clobbered and REG_INC
8245 affected regs by setting reg_mode to VOIDmode.
8246 (move2add_note_store): Don't pretend the inside of a SUBREG is
8247 the actual destination. Invalidate single/leading registers by
8248 setting reg_mode to VOIDmode.
8249 Use move2add_record_sym_value, move2add_valid_value_p and
8250 move2add_record_mode.
8252 2013-05-27 Richard Biener <rguenther@suse.de>
8254 PR tree-optimization/57396
8255 * tree-affine.c (double_int_constant_multiple_p): Properly
8256 return false for val == 0 and div != 0.
8258 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8260 * config/mips/mips.h: Use #elif in preprocessor conditions.
8262 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8265 * config/mips/constraints.md (kl): New constraint.
8266 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8267 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
8268 constraint for operand 0. Split after CSE for MIPS16. Emit a move
8270 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8272 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8275 * config/mips/mips.c (mips_can_inline_p): New function.
8276 (TARGET_CAN_INLINE_P): Define.
8278 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
8280 * sched-int.h (ds_t, dw_t): Make unsigned int.
8281 Fix documentation that describes how all the ds_t bits are used.
8282 Reserve the last bit for delayed-branch scheduling.
8283 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8284 (BITS_PER_DEP_WEAK): Fix definition and documentation.
8285 (gen_dep_weak_1): Remove prototype.
8286 * sched-deps.c (get_dep_weak_1): Make static.
8287 * target.def (speculate_insn, needs_block_p, gen_spec_check,
8288 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8289 * doc/tm.texi: Regenerate.
8290 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8292 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
8295 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8297 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
8298 Sandra Loosemore <sandra@codesourcery.com>
8300 * config.gcc (powerpc-*): Allow native for with-cpu.
8302 2013-05-24 Jeff Law <law@redhat.com>
8304 PR tree-optimization/57124
8305 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8306 conversion feeding a condition if the range has an overflow
8307 if -fstrict-overflow. Add warnings for when we do make the
8310 2013-05-24 Dehao Chen <dehao@google.com>
8312 * tree-cfg.c (locus_discrim_map): Fix the typo.
8313 (locus_discrim_hasher): Likewise.
8314 (locus_discrim_hasher::hash): Likewise.
8315 (locus_discrim_hasher::equal): Likewise.
8317 2013-05-24 Martin Jambor <mjambor@suse.cz>
8319 PR tree-optimization/57294
8320 * cgraph.h (ipa_record_stmt_references): Declare.
8321 * cgraphbuild.c (ipa_record_stmt_references): New function.
8322 (build_cgraph_edges): Use ipa_record_stmt_references.
8323 (rebuild_cgraph_edges): Likewise.
8324 (cgraph_rebuild_references): Likewise.
8325 * ipa-prop.c (ipa_modify_call_arguments): Discard references
8326 associated with the old statement and build references from the
8327 newly built statements.
8328 * ipa-ref.c (ipa_remove_stmt_references): New function.
8329 * ipa-ref.h (ipa_remove_stmt_references): Declare.
8331 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
8333 * lra-constraints.c (emit_spill_move): Use smaller mode for
8335 (check_and_process_move): Consider mem-reg moves for secondary
8337 (curr_insn_transform): Don't lose insns emitted before for
8338 secondary memory moves.
8339 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
8340 reg set up in the current insn.
8342 2013-05-24 Dehao Chen <dehao@google.com>
8344 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8346 (locus_descrim_hasher::equal): Likewise.
8347 (build_gimple_cfg): New discriminator assignment algorithm.
8348 (make_edges): Likewise.
8349 (next_discriminator_for_locus): Likewise.
8350 (same_line_p): Likewise.
8351 (assign_discriminators): Likewise.
8352 (make_cond_expr_edges): Likewise.
8353 (make_gimple_switch_edges): Likewise.
8354 (make_goto_expr_edges): Likewise.
8355 (make_gimple_asm_edges): Likewise.
8357 2013-05-24 Ian Bolton <ian.bolton@arm.com>
8359 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8360 X format specifier to only display bottom 16 bits.
8361 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8362 immediate to match for operand 2, since it will be masked.
8364 2013-05-24 Richard Biener <rguenther@suse.de>
8366 PR tree-optimization/57287
8367 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8368 all SSA names that occur in abnormal PHIs.
8370 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
8373 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8374 that index is not negative.
8376 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
8378 PR rtl-optimization/55177
8379 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8380 (simplify_byte_swapping_operation): New.
8381 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8382 (simplify_relational_operation_1): Deal with BSWAP.
8384 2013-05-23 Richard Henderson <rth@redhat.com>
8387 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8388 (ix86_reorg): Call it.
8390 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
8393 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8394 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8395 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8397 2013-05-23 Christian Bruel <christian.bruel@st.com>
8400 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8402 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8403 Marcus Shawcroft <marcus.shawcroft@arm.com>
8405 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8406 * config/aarch64/constraints.md (Usa): Remove.
8407 * doc/md.texi (AArch64 Usa): Remove.
8409 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8410 Marcus Shawcroft <marcus.shawcroft@arm.com>
8412 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8413 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8414 * config/aarch64/predicates.md (aarch64_const_address): Remove.
8415 (aarch64_mov_operand): Use aarch64_mov_operand_p.
8417 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
8419 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8420 instruction (AdvSIMD).
8421 * config/aarch64/aarch64-builtins.c
8422 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8423 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8425 2013-05-23 Martin Jambor <mjambor@suse.cz>
8428 * tree.h (contains_bitfld_component_ref_p): Declare.
8429 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8430 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
8432 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8433 not access a bit-field. Assert all final offsets are byte-aligned.
8435 2013-05-23 Richard Biener <rguenther@suse.de>
8437 PR tree-optimization/57380
8438 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8439 least one invariant or re-used load.
8440 * passes.c (init_optimization_passes): Move pass_phiprop before
8443 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
8445 * config/aarch64/aarch64-simd.md
8446 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8448 2013-05-23 Richard Biener <rguenther@suse.de>
8451 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8452 OEP_CONSTANT_ADDRESS_OF retained.
8454 2013-05-23 Jakub Jelinek <jakub@redhat.com>
8457 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8458 don't lower unit. Handle unit not being always BITS_PER_WORD.
8460 2013-05-23 Richard Biener <rguenther@suse.de>
8462 PR rtl-optimization/57341
8463 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8464 instead of true_dependence.
8466 2013-05-22 David Malcolm <dmalcolm@redhat.com>
8468 * bb-reorder.c (branch_threshold): Make const.
8469 (exec_threshold): Ditto.
8471 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8472 Pat Haugen <pthaugen@us.ibm.com>
8473 Peter Bergner <bergner@vnet.ibm.com>
8475 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8476 documentation for the power8 crypto builtins.
8478 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8480 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8481 macros for defining power8 builtin functions.
8482 (BU_P8V_AV_2): Likewise.
8483 (BU_P8V_AV_P): Likewise.
8484 (BU_P8V_VSX_1): Likewise.
8485 (BU_P8V_OVERLOAD_1): Likewise.
8486 (BU_P8V_OVERLOAD_2): Likewise.
8487 (BU_CRYPTO_1): Likewise.
8488 (BU_CRYPTO_2): Likewise.
8489 (BU_CRYPTO_3): Likewise.
8490 (BU_CRYPTO_OVERLOAD_1): Likewise.
8491 (BU_CRYPTO_OVERLOAD_2): Likewise.
8492 (XSCVSPDP): Fix typo, point to the correct instruction.
8493 (VCIPHER): Add power8 crypto builtins.
8494 (VCIPHERLAST): Likewise.
8495 (VNCIPHER): Likewise.
8496 (VNCIPHERLAST): Likewise.
8497 (VPMSUMB): Likewise.
8498 (VPMSUMH): Likewise.
8499 (VPMSUMW): Likewise.
8500 (VPERMXOR_V2DI): Likewise.
8501 (VPERMXOR_V4SI: Likewise.
8502 (VPERMXOR_V8HI: Likewise.
8503 (VPERMXOR_V16QI: Likewise.
8504 (VSHASIGMAW): Likewise.
8505 (VSHASIGMAD): Likewise.
8507 (VPERMXOR): Likewise.
8508 (VSHASIGMA): Likewise.
8510 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8511 __CRYPTO__ if the crypto instructions are available.
8512 (altivec_overloaded_builtins): Add support for overloaded power8
8515 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8516 support for power8 crypto builtins.
8517 (builtin_function_type): Likewise.
8518 (altivec_init_builtins): Add support for builtins that take vector
8519 long long (V2DI) arguments.
8521 * config/rs6000/crypto.md: New file, define power8 crypto
8524 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8525 Pat Haugen <pthaugen@us.ibm.com>
8526 Peter Bergner <bergner@vnet.ibm.com>
8528 * doc/invoke.texi (Option Summary): Add power8 options.
8529 (RS/6000 and PowerPC Options): Likewise.
8531 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8532 constraints.md instead of rs6000.h. Reorder w* constraints. Add
8533 wm, wn, wr documentation.
8535 * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8536 registers if direct move instructions are enabled.
8537 (wn): New constraint for no registers.
8538 (wq): New constraint for quad word even GPR registers.
8539 (wr): New constraint if 64-bit instructions are enabled.
8540 (wv): New constraint if power8 vector instructions are enabled.
8541 (wQ): New constraint for quad word memory locations.
8543 * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8544 constraint for 0..15 for crypto instructions.
8545 (gpc_reg_operand): If VSX allow registers in VSX registers as well
8546 as GPR and floating point registers.
8547 (int_reg_operand): New predicate to match only GPR registers.
8548 (base_reg_operand): New predicate to match base registers.
8549 (quad_int_reg_operand): New predicate to match even GPR registers
8550 for quad memory operations.
8551 (vsx_reg_or_cint_operand): New predicate to allow vector logical
8552 operations in both GPR and VSX registers.
8553 (quad_memory_operand): New predicate for quad memory operations.
8554 (reg_or_indexed_operand): New predicate for direct move support.
8556 * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8557 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8558 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8559 (POWERPC_MASKS): Add power8 options.
8560 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8563 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8564 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8566 * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8567 (-mpower8-fusion): New power8 options.
8568 (-mpower8-fusion-sign): Likewise.
8569 (-mpower8-vector): Likewise.
8570 (-mcrypto): Likewise.
8571 (-mdirect-move): Likewise.
8572 (-mquad-memory): Likewise.
8574 * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8576 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8578 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8579 (rs6000_debug_vector_unit): Add p8_vector.
8580 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8581 definitions. Also print fusion state.
8582 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8583 (rs6000_builtin_mask_calculate): Add power8 builtin support.
8584 (rs6000_option_override_internal): Add support for power8.
8585 (rs6000_common_init_builtins): Add debugging for skipped builtins
8587 (rs6000_adjust_cost): Add power8 support.
8588 (rs6000_issue_rate): Likewise.
8589 (insn_must_be_first_in_group): Likewise.
8590 (insn_must_be_last_in_group): Likewise.
8591 (force_new_group): Likewise.
8592 (rs6000_register_move_cost): Likewise.
8593 (rs6000_opt_masks): Likewise.
8595 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
8596 power8 capable assembler, default to power7 options.
8597 (TARGET_DIRECT_MOVE): Likewise.
8598 (TARGET_CRYPTO): Likewise.
8599 (TARGET_P8_VECTOR): Likewise.
8600 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
8601 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
8602 (VECTOR_MEM_P8_VECTOR_P): Likewise.
8603 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
8604 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
8605 (TARGET_XSCVDPSPN): Likewise.
8606 (TARGET_XSCVSPDPN): Likewsie.
8607 (TARGET_SYNC_HI_QI): Likewise.
8608 (TARGET_SYNC_TI): Likewise.
8609 (MASK_CRYPTO): Likewise.
8610 (MASK_DIRECT_MOVE): Likewise.
8611 (MASK_P8_FUSION): Likewise.
8612 (MASK_P8_VECTOR): Likewise.
8613 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
8614 temporary used by some of the direct move instructions to get two FP
8615 temporary registers does not force creation of a stack frame.
8616 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
8617 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
8618 that any VSX registers are tieable, even if they are also an
8619 Altivec vector mode.
8620 (r6000_reg_class_enum): Add wm, wr, wv constraints.
8621 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
8622 (RS6000_BTM_CRYPTO): Likewise.
8623 (RS6000_BTM_COMMON): Likewise.
8625 * config/rs6000/rs6000.md (cpu attribute): Add power8.
8626 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
8627 (enum rs6000_vector): Add power8 vector support.
8629 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8633 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
8634 (any_sibcall_could_use_r3): this and handle indirect calls.
8635 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
8637 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8639 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
8641 2013-05-22 Richard Biener <rguenther@suse.de>
8644 * profile.c (branch_prob): Do not split blocks that are
8645 abnormally receiving from ECF_RETURNS_TWICE functions.
8647 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
8649 * recog.c (offsettable_address_addr_space_p): Fix calculation of
8650 address mode. Move pointer mode initialization to the same place.
8652 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
8654 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
8655 while it has any effect.
8657 2013-05-21 Easwaran Raman <eraman@google.com>
8659 PR tree-optimization/57322
8660 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
8661 UID of the statement added to the BB to be 1.
8663 2013-05-21 Jakub Jelinek <jakub@redhat.com>
8665 PR tree-optimization/57331
8666 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
8667 of conversion from pointer type to integral type with integer.
8669 2013-05-21 Martin Jambor <mjambor@suse.cz>
8672 * ipa-prop.c (ipa_read_node_info): Process param_used and
8673 controlled_uses in the same order as when writing.
8675 2013-05-21 Magnus Granberg <baldrick@free.fr>
8678 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
8680 2013-05-21 Richard Biener <rguenther@suse.de>
8682 PR tree-optimization/57318
8683 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
8684 estimate stmts with side-effects as likely eliminated.
8686 2013-05-21 Richard Biener <rguenther@suse.de>
8688 PR tree-optimization/57330
8689 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
8690 preserve the call stmts fntype.
8692 2013-05-21 Richard Biener <rguenther@suse.de>
8694 PR tree-optimization/57303
8695 * tree-ssa-sink.c (statement_sink_location): Improve killing
8696 stmt detection and properly handle self-assignments.
8698 2013-05-21 Christian Bruel <christian.bruel@st.com>
8700 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
8701 spanning registers. LEAF_REG_REMAP is supported only for contiguous
8702 registers. Set register size out of the PARALLEL loop.
8704 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
8707 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
8708 (*fmasf4, *fmasf4_media): New insns.
8710 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
8712 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
8713 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
8714 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
8715 (mips_idiv_insns): Update the comments to say that the returned
8716 instruction counts are in units of BASE_INSN_LENGTH.
8717 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
8718 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
8719 using 2 rather than 4 as the length of indirect MIPS16 and
8720 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
8721 length of a NOP. Don't divide MIPS16 lengths by 2.
8722 (mips16_split_long_branches): Assume a branch is long if the
8723 length is greater than 4 rather than 8.
8724 * config/mips/mips.md (length): Give MIPS16 lengths directly,
8725 rather than multiplying them by 2. Multiply instruction counts
8726 by BASE_INSN_LENGTH rather than 4.
8727 (*jump_mips16, tls_get_tp_mips16_<mode>)
8728 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
8730 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
8732 * config/mips/mips.md (extended_mips16): Remove branch case.
8733 (length): Remove duplicated extended_mips16 test.
8735 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
8737 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
8739 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
8741 * recog.h (Recog_data): Rename to...
8742 (recog_data_d): ...this.
8743 (recog_data): Update accordingly.
8744 * recog.c (recog_data): Likewise.
8745 * reload.c (save_recog_data): Likewise.
8746 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
8747 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
8749 2013-05-17 Julian Brown <julian@codesourcery.com>
8751 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
8752 found in a REG_EQUAL note, invalidate it.
8754 2013-05-17 Easwaran Raman <eraman@google.com>
8756 * tree-ssa-reassoc.c (find_insert_point): New function.
8757 (insert_stmt_after): Likewise.
8758 (get_def_stmt): Likewise.
8759 (ensure_ops_are_available): Likewise.
8760 (not_dominated_by): Likewise.
8761 (rewrite_expr_tree): Do not move statements beyond what is
8762 necessary. Remove call to swap_ops_for_binary_stmt...
8763 (reassociate_bb): ... and move it here.
8764 (build_and_add_sum): Assign UIDs for new statements.
8765 (linearize_expr): Likewise.
8766 (do_reassoc): Renumber gimple statement UIDs.
8768 2013-05-17 Jan Hubicka <jh@suse.cz>
8770 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
8772 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8773 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8775 (output_weakrefs): Update.
8777 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
8778 Martin Jambor <mjambor@suse.cz>
8781 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
8782 value that corresponds to the given aggval is found in values vector.
8784 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
8786 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
8787 sse, sse2, sse3, ssse3 and sse4a flags to options.
8789 2013-05-17 David Malcolm <dmalcolm@redhat.com>
8791 * gengtype-state.c: (s_expr_writer): New class, to handle
8792 prettifying of output layout of s-expressions.
8793 (state_writer): New class, to write out gtype.state.
8794 (state_written_type_count): Move this variable into member data of
8796 (s_expr_writer::s_expr_writer): New code: constructor for new class
8797 (state_writer::state_writer(): ditto
8798 (s_expr_writer::write_new_line): New function
8799 (s_expr_writer::write_any_indent): ditto
8800 (s_expr_writer::begin_s_expr): ditto
8801 (s_expr_writer::end_s_expr): ditto
8802 (write_state_fileloc): convert to method of state_writer...
8803 (state_writer:: write_state_fileloc): ...and use methods of
8804 s_expr_writer to write indentation into the gtype.state output file
8805 to visually represent the hierarchical structure of the list
8807 (write_state_fields): ditto, renaming to...
8808 (state_writer::write_state_fields)
8809 (write_state_a_string): ditto, renaming to...
8810 (state_writer::write_state_a_string)
8811 (write_state_string_option): ditto, renaming to...
8812 (state_writer::write_state_string_option)
8813 (write_state_type_option): ditto, renaming to...
8814 (state_writer::write_state_type_option)
8815 (write_state_nested_option): ditto, renaming to...
8816 (state_writer::write_state_nested_option)
8817 (write_state_option): ditto, renaming to...
8818 (state_writer::write_state_option)
8819 (write_state_options): ditto, renaming to...
8820 (state_writer::write_state_options)
8821 (write_state_lang_bitmap): ditto, renaming to...
8822 (state_writer::write_state_lang_bitmap)
8823 (write_state_version): ditto, renaming to...
8824 (state_writer::write_state_version)
8825 (write_state_scalar_type): ditto, renaming to...
8826 (state_writer::write_state_scalar_type)
8827 (write_state_string_type): ditto, renaming to...
8828 (state_writer::write_state_string_type)
8829 (write_state_undefined_type): ditto, renaming to...
8830 (state_writer::write_state_undefined_type)
8831 (write_state_struct_union_type): ditto, renaming to...
8832 (state_writer::write_state_struct_union_type)
8833 (write_state_struct_type): ditto, renaming to...
8834 (state_writer::write_state_struct_type)
8835 (write_state_user_struct_type): ditto, renaming to...
8836 (state_writer::write_state_user_struct_type)
8837 (write_state_lang_struct_type): ditto, renaming to...
8838 (state_writer::write_state_lang_struct_type)
8839 (write_state_param_struct_type): ditto, renaming to...
8840 (state_writer::write_state_param_struct_type)
8841 (write_state_pointer_type): ditto, renaming to...
8842 (state_writer::write_state_pointer_type)
8843 (write_state_array_type): ditto, renaming to...
8844 (state_writer::write_state_array_type)
8845 (write_state_gc_used): ditto, renaming to...
8846 (state_writer::write_state_gc_used)
8847 (write_state_common_type_content): ditto, renaming to...
8848 (state_writer::write_state_common_type_content)
8849 (write_state_type): ditto, renaming to...
8850 (state_writer::write_state_type)
8851 (write_state_pair_list): ditto, renaming to...
8852 (state_writer::write_state_pair_list)
8853 (write_state_pair): ditto, renaming to...
8854 (state_writer::write_state_pair)
8855 (write_state_typedefs): ditto, renaming to...
8856 (state_writer::write_state_typedefs)
8857 (write_state_structures): ditto, renaming to...
8858 (state_writer::write_state_structures)
8859 (write_state_param_structs): ditto, renaming to...
8860 (state_writer::write_state_param_structs)
8861 (write_state_variables): ditto, renaming to...
8862 (state_writer::write_state_variables)
8863 (write_state_srcdir): ditto, renaming to...
8864 (state_writer::write_state_srcdir)
8865 (write_state_files_list): ditto, renaming to...
8866 (state_writer::write_state_files_list)
8867 (write_state_languages): ditto, renaming to...
8868 (state_writer::write_state_languages)
8869 (write_state): create a state_writer instance and use it when
8870 writing out the state file
8872 2013-05-17 Mike Stump <mikestump@comcast.net>
8874 PR rtl-optimization/57304
8875 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
8876 accessing DF_REF_REAL_LOC.
8878 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8880 PR rtl-optimization/57281
8881 PR rtl-optimization/57300
8882 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
8883 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
8884 what the other splitter did if the registers are dead.
8886 2013-05-17 Richard Biener <rguenther@suse.de>
8888 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
8891 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8893 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
8896 2013-05-17 Marek Polacek <polacek@redhat.com>
8898 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
8899 length when doing non-zero store of storing '\0' to '\0'.
8901 2013-05-17 Jakub Jelinek <jakub@redhat.com>
8903 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
8904 vect_external_def oprnd1 with loop_vinfo, try to emit
8905 optional cast, negation and and stmts on the loop preheader
8906 edge instead of into the pattern def seq.
8908 PR tree-optimization/57051
8909 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
8910 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
8912 2013-05-16 Nick Clifton <nickc@redhat.com>
8914 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
8915 (rl78_is_naked_func): New function.
8916 (rl78_expand_prologue): Skip prologue generation for naked functions.
8917 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
8918 * doc/extend.texi (naked): Add RL78 to the list of processors
8919 that supports this attribute.
8921 2013-05-16 Jeff Law <law@redhat.com>
8923 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
8925 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
8927 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
8928 cache parameters using detect_caches_amd also for CYRIX,
8929 NSC and TM2 signatures.
8931 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
8932 Dzianis Kahanovich <mahatma@eu.by>
8936 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8937 VIA/Centaur processors and determine their cache parameters
8938 using detect_caches_amd.
8940 2013-05-16 Teresa Johnson <tejohnson@google.com>
8942 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
8943 (rtl_verify_edges): New function.
8944 (rtl_verify_bb_insns): Ditto.
8945 (rtl_verify_bb_pointers): Ditto.
8946 (rtl_verify_bb_insn_chain): Ditto.
8947 (rtl_verify_fallthru): Ditto.
8948 (rtl_verify_bb_layout): Ditto.
8949 (rtl_verify_flow_info_1): Outline checks into new functions.
8950 (rtl_verify_flow_info): Ditto.
8952 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
8954 * cfghooks.c (copy_bbs): Add update_dominance argument.
8955 * cfghooks.h (copy_bbs): Update prototype.
8956 * tree-cfg.c (gimple_duplicate_sese_region):
8957 Add update_dominance argument.
8958 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
8959 * tree-ssa-loop-ch.c (copy_loop_headers): Update
8960 gimple_duplicate_sese_region call.
8961 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
8962 Update copy_bbs call.
8963 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
8964 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
8966 2013-05-16 Jakub Jelinek <jakub@redhat.com>
8968 * tree-vectorizer.h (NUM_PATTERNS): Increment.
8969 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
8970 vect_recog_rotate_pattern.
8971 (vect_recog_rotate_pattern): New function.
8973 2013-05-16 Jason Merrill <jason@redhat.com>
8975 * Makefile.in (LLINKER): New variable.
8976 (mostlyclean): Remove link mutex.
8977 * configure.ac: Handle --enable-link-mutex.
8978 * lock-and-run.sh: New script.
8980 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8983 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
8986 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8988 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
8990 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
8992 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
8993 * config/arm/arm.c (next_consecutive_mem): New function.
8994 (gen_movmem_ldrd_strd): Likewise.
8995 * config/arm/arm.md (movmemqi): Update condition and code.
8996 (unaligned_loaddi, unaligned_storedi): New patterns.
8998 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9000 * config.gcc: Obsolete *-*-solaris2.9*.
9001 * doc/install.texi (Specific, *-*-solaris2*): Document it.
9003 2013-05-16 Richard Biener <rguenther@suse.de>
9005 * passes.c (init_optimization_passes): Move pass_parallelize_loops
9006 earlier, after GRAPHITE transforms and IV canonicalization.
9008 2013-05-16 Jakub Jelinek <jakub@redhat.com>
9010 * omp-low.c (extract_omp_for_data): For collapsed loops,
9011 if at least one of the loops is known at compile time to
9012 iterate zero times, set count to 0.
9013 (expand_omp_regimplify_p): New function.
9014 (expand_omp_for_generic): For collapsed loops, if at least
9015 one of the loops isn't known to iterate at least once,
9016 add runtime check with setting count to 0.
9017 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9018 For unsigned types if it isn't known at compile time that
9019 the loop will iterate at least once, add runtime check to bypass
9020 the whole loop if initial condition isn't true.
9022 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
9024 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9026 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
9029 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9030 transformations to avoid an infinite loop.
9032 2013-05-16 Marek Polacek <polacek@redhat.com>
9034 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9036 2013-05-15 Leif Ekblad <leif@rdos.net>
9038 * config/i386/i386.c (ix86_decompose_address): Use
9039 DEFAULT_TLS_SEG_REG to access TLS segment register.
9040 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9041 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9042 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9044 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
9047 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9048 sibling calls to functions that would normally be lazily bound,
9049 unless $gp is call-clobbered.
9051 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
9053 * config/i386/i386.c (ix86_option_override_internal): Update
9054 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
9055 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9056 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9057 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9059 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9061 2013-05-15 Andreas Schwab <schwab@suse.de>
9063 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9064 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
9067 2013-05-15 Teresa Johnson <tejohnson@google.com>
9069 * loop-unroll.c (report_unroll_peel): Check decision before
9070 emitting unroll/peel message.
9072 2013-05-15 Teresa Johnson <tejohnson@google.com>
9074 * function.h (has_bb_partition): New rtl_data flag.
9075 (bb_reorder_complete): Ditto.
9076 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9077 instead of flag_reorder_blocks_and_partition.
9078 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9079 with some enhancements.
9080 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9081 * bb-reorder.c (connect_traces): Check for has_bb_partition
9082 instead of flag_reorder_blocks_and_partition.
9083 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9084 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9085 verify_hot_cold_block_grouping.
9086 (partition_hot_cold_basic_blocks): Set has_bb_partition.
9088 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9091 * config/arm/predicates.md (call_insn_operand): New predicate.
9092 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
9093 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9094 if insn is not a tail call.
9095 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9097 * config/arm/arm.h (enum reg_class): New caller save register class.
9098 (REG_CLASS_NAMES): Likewise.
9099 (REG_CLASS_CONTENTS): Likewise.
9100 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9103 2013-05-15 Richard Biener <rguenther@suse.de>
9105 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9106 of MSG_OPTIMIZED_LOCATIONS.
9107 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9108 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9110 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9111 of MSG_OPTIMIZED_LOCATIONS.
9112 (execute_vect_slp): Likewise.
9113 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9114 (vect_create_cond_for_alias_checks): Likewise.
9115 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9116 (vect_recog_widen_mult_pattern): Likewise.
9117 (vect_recog_widen_sum_pattern): Likewise.
9118 (vect_recog_over_widening_pattern): Likewise.
9119 (vect_recog_widen_shift_pattern): Likewise.
9120 (vect_recog_vector_vector_shift_pattern): Likewise.
9121 (vect_recog_divmod_pattern): Likewise.
9122 (vect_recog_mixed_size_cond_pattern): Likewise.
9123 (vect_recog_bool_pattern): Likewise.
9124 (vect_pattern_recog_1): Likewise.
9126 2013-05-15 Martin Jambor <mjambor@suse.cz>
9128 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9129 non-functions to builtin_unreachable.
9130 * ipa-inline-transform.c (inline_call): Do not assert estimates were
9131 correct when new direct edges were discovered.
9133 2013-05-15 Martin Jambor <mjambor@suse.cz>
9135 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9136 header, print symbol order instead of node uid, print more information
9137 about indirect edge targets.
9138 (ipa_make_edge_direct_to_target): Print symbol order instead of node
9140 (ipa_make_edge_direct_to_target): Likewise.
9141 (remove_described_reference): Likewise.
9142 (propagate_controlled_uses): Likewise.
9143 (ipa_print_node_params): Also print symbol order.
9144 (ipcp_transform_function): Print symbol order instead of node uids.
9145 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9146 (cgraph_get_create_real_symbol_node): Likewise.
9147 * ipa-cp.c (print_lattice): Likewise.
9148 (print_all_lattices): Likewise.
9149 (determine_versionability): Likewise.
9150 (initialize_node_lattices): Likewise.
9151 (estimate_local_effects): Likewise.
9152 (update_profiling_info): Likewise.
9153 (create_specialized_node): Likewise.
9154 (perhaps_add_new_callers): Likewise.
9155 (decide_about_value): Likewise.
9156 (decide_whether_version_node): Likewise.
9157 (identify_dead_nodes): Likewise.
9158 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9159 (dump_inline_summary): Likewise.
9160 (estimate_node_size_and_time): Likewise.
9161 (inline_analyze_function): Likewise.
9162 * ipa-inline.c (report_inline_failed_reason): Likewise.
9163 (want_early_inline_function_p): Likewise.
9164 (edge_badness): Likewise.
9165 (update_edge_key): Likewise.
9166 (inline_small_functions): Likewise. Add dumping of order to two other
9168 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9169 instead of node uids.
9170 (propagate_pure_const): Likewise.
9171 (propagate_pure_const): Likewise.
9172 * ipa-utils.c (dump_cgraph_node_set): Likewise.
9173 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9174 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9176 * tree-pretty-print.c (dump_function_header): Likewise.
9177 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9178 Print symbol order instead of node uids.
9180 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9182 * config/s390/s390.c (s390_register_move_cost): Don't impose the
9183 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9185 2013-05-15 Richard Biener <rguenther@suse.de>
9187 PR tree-optimization/57275
9188 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9189 return value for fail to do runtime alias checks for gather loads.
9191 2013-05-15 Jan Hubicka <jh@suse.cz>
9195 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9196 weakrefs are not external.
9197 (lto_symtab_merge_decls): Fix thinko when dealing with
9198 non-lto_symtab decls.
9199 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9200 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9201 * varpool.c (dump_varpool_node): Dump more flags.
9203 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
9205 * config/i386/i386.c (processor_alias_table): Add instruction
9206 FSGSBASE for AMD bdver3 architecture.
9208 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9210 * tree.c (warn_deprecated_use): Print file:line using locus color.
9211 * diagnostic.c (diagnostic_report_current_module): Print file:line
9212 and file:line:column using locus color.
9214 2013-05-14 Mike Stump <mikestump@comcast.net>
9216 * gdbinit.in: Add __null.
9218 2013-05-14 Mike Stump <mikestump@comcast.net>
9220 * recog.h: Rename struct recog_data to Recog_data.
9221 * recog.c: Likewise.
9222 * reload.c (can_reload_into): Likewise.
9223 * config/picochip/picochip.c: Likewise.
9225 2013-05-14 Mike Stump <mikestump@comcast.net>
9227 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9229 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
9231 * resource.h (struct resources): Remove unch_memory member.
9232 (CLEAR_RESOURCE): Don't clear unch_memory.
9233 * resource.c (mark_referenced_resources): Don't set it.
9234 (mark_set_resources): Likewise.
9235 (mark_target_live_regs): Don't clear it.
9236 (init_resource_info): Likewise.
9237 * reorg.c (resource_conflicts_p): Don't compare it.
9238 (redundant_insn): Don't set it.
9240 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9242 * emit-rtl.c (next_label): Remove unused function.
9243 (skip_consecutive_labels, link_cc0_insns): Move to ...
9244 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9245 only place where these functions are used, and make them static.
9247 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
9249 * fold-const.c (fold_negate_expr): Handle vectors.
9250 (fold_truth_not_expr): Make it static.
9251 (fold_invert_truthvalue): New static function.
9252 (invert_truthvalue_loc): Handle vectors. Do not call
9253 fold_truth_not_expr directly.
9254 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9255 <TRUTH_NOT_EXPR>: Do not cast to boolean.
9256 (fold_comparison): Handle vector constants.
9257 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9258 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9259 * tree.h (fold_truth_not_expr): Remove declaration.
9261 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
9263 * config/aarch64/aarch64-simd.md
9264 (aarch64_vcond_internal<mode>): Rename to...
9265 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9266 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9267 float modes. Clarify all iterator modes.
9268 (vcond<mode><mode>): Use new name for vcond expanders.
9269 (vcond<v_cmp_result><mode>): Likewise.
9270 (vcondu<mode><mode>: Likewise.
9271 * config/aarch64/iterators.md (VDQF_COND): New.
9273 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
9276 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9277 variable for the shift amount. Check that we shift by non-negative
9280 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
9283 * config/arm/arm.h (EPILOGUE_USES): Only return true
9284 for LR_REGNUM after epilogue_completed.
9286 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
9288 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9291 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9294 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9295 and Solaris 11+/x86 with gld.
9296 * configure: Regenerate.
9298 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9300 * expmed.c (expand_shift_1): Canonicalize rotates by
9301 constant bitsize / 2 to bitsize - 1.
9302 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9303 case ROTATERT>: Likewise.
9306 2013-05-10 Jakub Jelinek <jakub@redhat.com>
9308 * config/i386/i386.md (rotateinv): New code attr.
9309 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9310 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9311 roll $31, %eax, etc.
9313 2013-05-14 Richard Biener <rguenther@suse.de>
9316 * tree-eh.c (sink_clobbers): Give up for successors with
9317 multiple predecessors and no virtual uses.
9319 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
9321 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9322 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9324 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9327 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9328 the case when both op0 and op1 have VOIDmode.
9330 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
9332 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9333 in multiply-accumulate mode.
9335 2013-05-13 Guozhi Wei <carrot@google.com>
9337 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9339 2013-05-13 Kai Tietz <ktietz@redhat.com>
9342 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9343 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9344 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9345 * config/i386/i386.c (ix86_option_override_internal): Likewise.
9346 (ix86_expand_prologue): Likewise.
9347 (ix86_expand_split_stack_prologue): Likewise.
9348 (legitimate_pic_address_disp_p): Likewise.
9349 (legitimize_pic_address): Likewise.
9350 (legitimize_tls_address): Likewise.
9351 (legitimize_pe_coff_symbol): Likewise.
9352 (output_pic_addr_const): Likewise.
9353 (construct_plt_address): Likewise.
9354 (ix86_expand_call): Likewise.
9355 (x86_output_mi_thunk): Likewise.
9356 (x86_function_profiler): Likewise.
9358 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
9360 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9361 similar switch cases.
9362 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9363 (aarch64_simd_mov_to_<mode>low): Delete.
9364 (aarch64_simd_mov_to_<mode>high): Delete.
9365 (move_lo_quad_<mode>): Add w<-r alternative.
9366 (aarch64_simd_move_hi_quad_<mode>): Likewise.
9367 (aarch64_simd_mov_from_*): Update type attribute.
9368 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9371 2013-05-13 Jan Hubicka <jh@suse.cz>
9373 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9374 * config/i386/i386.c (ix86_compute_frame_layout,
9375 ix86_expand_epilogue, emit_i387_cw_initialization,
9376 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9377 ix86_local_alignment): Fix use of size/speed predicates.
9379 2013-05-13 Jakub Jelinek <jakub@redhat.com>
9381 PR tree-optimization/45216
9382 PR tree-optimization/57157
9383 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9384 the (-Y) & (B - 1) variant if OP is |.
9385 * expmed.c (expand_shift_1): For rotations by const0_rtx just
9386 return shifted. Use (-op1) & (prec - 1) as other_amount
9387 instead of prec - op1.
9389 2013-05-13 Martin Jambor <mjambor@suse.cz>
9392 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9393 (ipa_constant_data): New type.
9394 (ipa_jump_func): Use ipa_constant_data to hold information about
9395 constant jump functions.
9396 (ipa_get_jf_constant): Adjust to jump function type changes.
9397 (ipa_get_jf_constant_rdesc): New function.
9398 (ipa_param_descriptor): New field controlled_uses.
9399 (ipa_get_controlled_uses): New function.
9400 (ipa_set_controlled_uses): Likewise.
9401 * ipa-ref.h (ipa_find_reference): Declare.
9402 * ipa-prop.c (ipa_cst_ref_desc): New type.
9403 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9405 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
9406 New parameter cs. Adjust all callers.
9407 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9408 (remove_described_reference): New function.
9409 (jfunc_rdesc_usable): Likewise.
9410 (try_make_edge_direct_simple_call): Decrement controlled use count,
9411 attempt to remove reference if it hits zero.
9412 (combine_controlled_uses_counters): New function.
9413 (propagate_controlled_uses): Likewise.
9414 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9415 (ipa_edge_duplication_hook): Duplicate reference descriptions.
9416 (ipa_print_node_params): Print described use counter.
9417 (ipa_write_jump_function): Adjust to jump function type changes.
9418 (ipa_read_jump_function): New parameter CS, pass it to
9419 ipa_set_jf_constant. Adjust caller.
9420 (ipa_write_node_info): Stream controlled use count
9421 (ipa_read_node_info): Likewise.
9422 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9424 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9425 count. Remove cloning-added reference if it reaches zero.
9426 * ipa-ref.c (ipa_find_reference): New function.
9428 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
9430 * config/i386/i386.c (processor_target_table): Modified default
9431 alignment values for AMD BD and BT architectures.
9433 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
9435 * tree-vect-generic.c (uniform_vector_p): Move ...
9436 * tree.c (uniform_vector_p): ... here.
9437 * tree.h (uniform_vector_p): Declare it.
9438 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9441 2013-05-13 Jakub Jelinek <jakub@redhat.com>
9443 PR tree-optimization/57230
9444 * tree-ssa-strlen.c (handle_char_store): Record length for
9445 array store from STRING_CST.
9447 PR tree-optimization/57230
9448 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9451 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
9453 * config/epiphany/epiphany.c (epiphany_init): Check size of
9454 NUM_MODES_FOR_MODE_SWITCHING.
9455 (epiphany_expand_prologue):
9456 Remove CONFIG_REGNUM initial value handling code.
9457 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9458 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9459 (emit_set_fp_mode, epiphany_mode_after): Likewise.
9460 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9461 Don't return 1 for FP_MODE_NONE.
9462 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9463 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9464 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9465 * config/epiphany/epiphany.md (save_config): New pattern.
9467 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
9469 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9471 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
9473 * config/i386/i386.md (memory): Handle sseishft1.
9474 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9475 (*vec_extractv2di_1): Ditto.
9477 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
9479 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9482 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
9484 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9485 Add mthumb/march=armv7-a multilib.
9486 Add mthumb/march=armv7-r multilib.
9487 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9489 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
9491 * config/v850/t-rtems: Add more multilibs.
9493 2013-05-10 Richard Biener <rguenther@suse.de>
9495 PR tree-optimization/57214
9496 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9497 not propagate from SSA names that occur in abnormal PHI nodes.
9499 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
9501 * stor-layout.c (element_precision): New function.
9502 * machmode.h (element_precision): Declare it.
9503 * tree.c (build_minus_one_cst): New function.
9504 (element_precision): Likewise.
9505 * tree.h (build_minus_one_cst): Declare new function.
9506 (element_precision): Likewise.
9507 * fold-const.c (operand_equal_p): Use element_precision.
9508 (fold_binary_loc): Handle vector types.
9509 * convert.c (convert_to_integer): Use element_precision.
9510 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9513 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
9515 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9516 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9517 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9518 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9519 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9520 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9521 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9522 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9523 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9524 (Uuw8): New constraints.
9525 (Usb4): Move into alphabetical order.
9526 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9527 (sd8_operand, ub8_operand, uw8_operand): New predicates.
9528 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9529 previously unnamed patterns.
9530 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9531 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9532 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9533 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9534 of set_attr_alternative/if_then_else. Use extended_mips16 instead
9535 of specific lengths.
9537 2013-05-10 Jakub Jelinek <jakub@redhat.com>
9539 * config/i386/i386.md (rotateinv): New code attr.
9540 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9541 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9542 roll $31, %eax, etc.
9544 PR tree-optimization/45216
9545 PR tree-optimization/57157
9546 * tree-ssa-forwprop.c (simplify_rotate): New function.
9547 (ssa_forward_propagate_and_combine): Call it.
9549 2013-05-10 Richard Biener <rguenther@suse.de>
9551 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9552 disable peeling when we version for aliasing.
9553 (vector_alignment_reachable_p): Honor explicit user alignment.
9554 (vect_supportable_dr_alignment): Likewise.
9555 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9556 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9557 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9558 then peeling to arrange for the cost-model check to come first.
9560 2013-05-10 Alan Modra <amodra@gmail.com>
9562 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9563 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9564 * configure: Regenerate.
9566 2013-05-10 Alan Modra <amodra@gmail.com>
9569 * varasm.c (default_elf_select_section): Move !DECL_P check..
9570 (get_named_section): ..to here before calling get_section_name.
9572 (default_section_type_flags): Add DECL_P check.
9573 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9574 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9576 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
9578 * config/epiphany/epiphany.c (epiphany_expand_prologue):
9579 When using gen_stack_adjust_str with a register offset, add a
9580 REG_FRAME_RELATED_EXPR note.
9582 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
9584 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9585 (*vec_extractv4si_zext_mem): Ditto.
9586 (*vec_extractv2di): Add 0->x and x->x alternatives.
9587 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9588 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9590 2013-05-09 Jason Merrill <jason@redhat.com>
9592 N3639 C++1y VLA support
9593 * gimplify.c (gimplify_vla_decl): Don't touch an existing
9596 * tree.c (build_constructor_va): New.
9597 * tree.h: Declare it.
9599 2013-05-09 Martin Jambor <mjambor@suse.cz>
9602 * gimple-fold.c (canonicalize_constructor_val): Call
9603 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
9605 2013-05-09 Jan Hubicka <jh@suse.cz>
9606 Richard Biener <rguenther@suse.de>
9609 * symtab.c (symtab_make_decl_local): Do not add private names.
9611 2013-05-09 Jan Hubicka <jh@suse.cz>
9614 * symtab.c (insert_to_assembler_name_hash): Handle clones.
9615 (unlink_from_assembler_name_hash): Likewise.
9616 (symtab_prevail_in_asm_name_hash, symtab_register_node,
9617 symtab_unregister_node, symtab_initialize_asm_name_hash,
9618 change_decl_assembler_name): Update.
9620 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
9622 * config/aarch64/aarch64.md: New movtf split.
9623 (*movtf_aarch64): Update.
9624 (aarch64_movdi_tilow): Handle TF modes and rename to
9625 aarch64_movdi_<mode>low.
9626 (aarch64_movdi_tihigh): Handle TF modes and rename to
9627 aarch64_movdi_<mode>high
9628 (aarch64_movtihigh_di): Handle TF modes and rename to
9629 aarch64_mov<mode>high_di
9630 (aarch64_movtilow_di): Handle TF modes and rename to
9631 aarch64_mov<mode>low_di
9632 (aarch64_movtilow_tilow): Remove spurious whitespace.
9633 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
9635 (aarch64_print_operand): Update.
9637 2013-05-09 Alan Modra <amodra@gmail.com>
9639 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
9641 * configure: Regenerate.
9643 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
9645 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
9646 splitter preparation statements.
9647 * config/i386/sse.md (*vec_extract* splitters): Ditto.
9648 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
9651 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9653 * gimple-ssa-strength-reduction.c (count_candidates): Change
9654 return value to int.
9655 (analyze_candidates_and_replace): Change type of length to int.
9657 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
9659 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
9660 (*vec_extract<mode>): Use VI12_128 mode iterator.
9661 (*vec_extract<mode>_mem): Ditto.
9662 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
9665 2013-05-08 Diego Novillo <dnovillo@google.com>
9670 2012-08-17 Diego Novillo <dnovillo@google.com>
9673 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
9674 * config.in: Regenerate.
9675 * configure: Regenerate.
9676 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
9678 2013-05-08 Jan Hubicka <jh@suse.cz>
9681 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
9682 * cgraph.h (symtab_node_base): Add unique_name.
9683 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9684 input_overwrite_node, input_varpool_node): Stream unique_name.
9685 * cgraphclones.c (cgraph_create_virtual_clone,
9686 cgraph_function_versioning): Set unique_name.
9687 * ipa.c (function_and_variable_visibility): Set unique_name.
9689 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9691 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
9692 (alloc_cand_and_find_basis): Restrict conditional candidate
9693 processing to CAND_MULTs.
9695 2013-05-08 Jan Hubicka <jh@suse.cz>
9698 lto-symtab.c (lto_symtab_symbol_p): New function.
9699 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
9700 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
9701 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
9702 Skip static symbols.
9704 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
9706 PR tree-optimization/57200
9707 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
9708 Only call inform if the preceding warning_at returns true.
9710 2013-05-07 Han Shen <shenhan@google.com>
9712 * cfgexpand.c (record_or_union_type_has_array_p): New function.
9713 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
9714 * common.opt (fstack-protector-strong): New option.
9715 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
9716 * doc/invoke.texi (Optimization Options): Document
9717 "-fstack-protector-strong".
9718 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
9720 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
9722 * config/mips/mips.c (mips_machine_reorg2): Return 0.
9724 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
9726 * ira.c (update_equiv_regs): Add insn having equiv memory even if
9727 it is not lhs of the insn.
9728 (setup_reg_equiv): Remove insn having equiv memory which it is not
9730 * lra-constraints.c (process_address): Try to improve generation
9731 code for address base + disp.
9732 (lra_constraints): Make correct the code for checking insn setting
9733 up backward equivalence. Remove insn only if it is in the init
9735 * lra-eliminations.c (update_reg_eliminate): Change return value.
9736 (lra_eliminate): Use the result.
9738 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
9740 * config/i386/sse.md (ssescalarnummask): New mode attribute.
9741 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
9742 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
9743 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
9744 register target operands.
9745 (*vec_extractv8hi_sse2): New pattern.
9746 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
9747 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
9748 (*vec_extract<mode>_mem): New insn and split pattern.
9750 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
9752 * config/arm/arm.c (arm_asan_shadow_offset): New function.
9753 (TARGET_ASAN_SHADOW_OFFSET): Define.
9754 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
9755 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
9757 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9759 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
9760 (incr_vec_index): Return -1 if increment not found.
9761 (create_add_on_incoming_edge): Assert if increment not found.
9762 (record_increment): Limit number of increments recorded.
9763 (all_phi_incrs_profitable): Return false if an increment not found.
9764 (replace_profitable_candidates): Don't process increments that were
9766 (analyze_candidates_and_replace): Limit size of incr_vec.
9768 2013-05-07 Richard Biener <rguenther@suse.de>
9770 * calls.c (special_function_p): setjmp-like functions are leaf.
9771 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
9772 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
9774 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
9776 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
9777 (aarch64_simd_mov<mode>): New expander.
9778 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
9779 (aarch64_simd_mov_to_<mode>high): Likewise.
9780 (aarch64_simd_mov_from_<mode>low): Likewise.
9781 (aarch64_simd_mov_from_<mode>high): Likewise.
9782 (aarch64_dup_lane<mode>): Update.
9783 (aarch64_dup_lanedi): New instruction pattern.
9784 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
9785 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
9787 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9789 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
9790 (replace_mult_candidate): Remove unnecessary argument; remove
9791 unnecessary parameter from call to introduce_cast_before_cand.
9792 (replace_unconditional_candidate): Remove unnecessary parameter
9793 from call to replace_mult_candidate.
9794 (replace_conditional_candidate): Likewise.
9795 (insert_initializers): Use make_temp_ssa_name.
9796 (introduce_cast_before_cand): Remove unnecessary argument; use
9798 (replace_one_candidate): Remove unnecessary argument; remove
9799 unnecessary parameter from calls to introduce_cast_before_cand.
9800 (replace_profitable_candidates): Remove unnecessary parameters
9801 from calls to replace_one_candidate.
9803 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9805 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9806 phi def as possibly hiding a basis for a CAND_ADD whose operands
9807 have been commuted in the analysis.
9808 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
9810 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
9812 * config/aarch64/aarch64.md
9813 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
9814 shift value between 0-4.
9816 2013-05-07 Richard Biener <rguenther@suse.de>
9818 * double-int.h (rshift): New overload.
9819 * double-int.c (rshift): New function.
9820 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
9821 (create_reference_ops_from_ref): Remove.
9822 (vn_reference_insert): Use shared ops for constructing the
9823 reference and copy it.
9825 2013-05-07 Richard Biener <rguenther@suse.de>
9828 * tree-eh.c (sink_clobbers): Properly propagate
9829 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
9831 2013-05-07 Jakub Jelinek <jakub@redhat.com>
9833 PR tree-optimization/57149
9834 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9835 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9836 collect_phi_def_edges, execute_late_warn_uninitialized): Use
9837 uninit_undefined_value_p instead of ssa_undefined_value_p.
9840 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9841 for modifier == EXPAND_INITIALIZER.
9843 2013-05-07 Anton Blanchard <anton@samba.org>
9845 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9846 for powerpc64 little endian.
9847 * configure: Regenerate.
9849 2013-05-06 Graham Stott <grahams@btinternet.com>
9851 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9852 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
9853 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
9854 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
9856 2013-05-06 Graham Stott <grahams@btinternet.com>
9858 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
9859 codes which allow non-lvalues.
9861 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
9863 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
9864 components are all 1s.
9865 (integer_minus_onep): New function.
9866 * tree.h (integer_minus_onep): Declare it.
9867 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
9868 integer_minus_onep instead of integer_all_onesp.
9870 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9873 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
9874 variations of these patterns.
9876 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
9878 * config/i386/i386.md (isa): Add x64_sse4 member.
9879 (enabled): Handle x64_sse4.
9880 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
9881 instruction for 64bit SSE4_1 targets. Update insn attributes.
9882 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
9883 instruction for SSE4_1 targets. Update insn attributes.
9884 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
9885 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
9886 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
9888 (*vec_extractv4si): Rename from *sse4_1_pextrd.
9889 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
9890 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
9892 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9895 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
9897 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
9899 * final.c (do_assembler_dialects): Don't handle curly braces and
9900 vertical bar escaped by % as dialect delimiters.
9901 (output_asm_insn): Print curly braces and vertical bar if escaped
9902 by % and ASSEMBLER_DIALECT defined.
9903 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
9904 * doc/tm.texi: Regenerated.
9906 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
9908 * config/mips/mips.c: Include tree-pass.h.
9909 (mips_reorg): Split in pre- and post-dbr_schedule parts.
9910 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
9911 (pass_mips_machine_reorg2): New machine specific pass.
9912 (insert_pass_mips_machine_reorg2): New pass plugin definition.
9913 (mips_option_override): Register the new pass.
9914 * rtl.h (cleanup_barriers): Remove prototype.
9915 (dbr_schedule): Likewise.
9916 * jump.c (cleanup_barriers): Make static.
9917 * reorg.c (dbr_schedule): Likewise.
9919 2013-05-06 Richard Biener <rguenther@suse.de>
9921 PR tree-optimization/57185
9922 * tree-parloops.c (add_field_for_reduction): Handle anonymous
9925 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
9928 * config/i386/i386.c (add_parameter_dependencies): Add dependence
9929 between "first_arg" and "insn", not "last" and "insn".
9931 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9933 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
9934 (find_candidates_in_block): Re-enable slsr_process_phi.
9935 (create_phi_basis): Fix double counting of candidate adjustment.
9937 2013-05-06 Richard Biener <rguenther@suse.de>
9940 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
9941 the edge is also fallthru, preserve it and just clear the
9943 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
9944 also complex, preserve that and just clear the fallthru flag.
9945 * tree-inline.c (update_ssa_across_abnormal_edges): Also
9946 update virtual operands.
9948 2013-05-06 Alan Modra <amodra@gmail.com>
9950 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
9951 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
9952 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
9953 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
9954 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9955 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
9957 2013-05-06 Alan Modra <amodra@gmail.com>
9959 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
9960 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
9961 (DEFAULT_ASM_ENDIAN): Define.
9962 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
9963 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
9964 Update -K PIC clause from sysv4.h.
9965 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
9966 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
9968 2013-05-06 Alan Modra <amodra@gmail.com>
9970 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
9971 twice for little-endian.
9972 (ashrdi3_no_power, ashrdi3): Support little-endian.
9974 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
9977 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
9978 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
9980 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
9981 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
9982 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
9985 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
9986 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
9988 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
9989 * config/pa/pa.opt: Make mbig-switch a no-op.
9990 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
9991 (CASE_VECTOR_MODE): Always return SImode.
9992 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
9993 for the !TARGET_BIG_SWITCH case.
9994 * config/pa/pa-linux.h: Likewise.
9995 * config/pa/pa-openbsd.h: Likewise.
9996 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
9997 * config/pa/pa.md (short_jump): Remove define_insn.
9998 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
9999 (casesi0): Remove define_insn.
10000 (type): Remove btable_branch.
10001 (pa_combine_type): Likewise.
10002 (in_nullified_branch_delay): Likewise.
10003 (in_call_delay): Likewise.
10004 (define_delay): Likewise.
10005 (define_insn_reservation "Z3"): Likewise.
10006 (define_insn_reservation "Z4"): Likewise.
10007 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10008 (pa_adjust_insn_length): Remove adjustment for btable branches.
10009 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10012 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
10014 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10015 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10016 Add m->r,x alternatives.
10017 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10018 splitters using SWI48x mode iterator.
10019 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
10020 TARGET_64BIT. Add m->x alternative.
10021 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10022 Add o->x alternative. Enable for TARGET_SSE.
10023 (sse_storeq): Remove expander.
10024 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10025 with memory input operand.
10026 (*vec_extractv2di_1 splitter): New.
10027 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10028 * config/i386/i386.md (ssevecmodelower): New mode attribute.
10030 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
10032 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
10033 (INT_LOWPART): Delete.
10034 (extract_MB): Adjust.
10035 (extract_ME): Adjust.
10036 (print_operand): Adjust.
10038 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
10040 * config/rs6000/predicates.md (reg_or_add_cint_operand,
10041 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10042 (reg_or_logical_cint_operand, easy_fp_constant,
10043 logical_const_operand): Delete "CONST_DOUBLE" case.
10044 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10045 "HOST_BITS_PER_WIDE_INT == 64" test.
10046 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
10047 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10048 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10049 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10051 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10052 CONST_DOUBLE DImode/VOIDmode case.
10053 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10054 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
10055 CONST_DOUBLE VOIDmode case.
10056 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10057 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10058 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10059 Delete CONST_DOUBLE case.
10060 (splitters for mov FMOVE64 const_double): Delete
10061 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
10062 "HOST_BITS_PER_WIDE_INT >= 64" test.
10063 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10065 (mov DI const_double): Delete.
10067 2013-05-04 Jakub Jelinek <jakub@redhat.com>
10069 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10070 on op shows all bits zero in mode of a lowpart subreg, return zero.
10072 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10075 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10076 to save TFmode registers and DImode to save TImode registers for
10077 caller save operations.
10078 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10079 mark being partially clobbered since they only use the first
10082 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10083 and TDmode only use the upper 64-bits of each VSX register.
10085 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10087 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10088 (find_candidates_in_block): Disable slsr_process_phi.
10090 2013-05-03 Guozhi Wei <carrot@google.com>
10092 * coverage.c (coverage_obj_init): Move the construction of gcov
10094 (build_init_ctor): ... here.
10096 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10098 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10099 (slsr_cand_d): Redefine def_phi.
10100 (stride_status, phi_adjust_status, count_phis_status): New enums.
10101 (find_phi_def): New.
10102 (find_basis_for_base_expr): New.
10103 (find_basis_for_candidate): Handle hidden bases.
10104 (alloc_cand_and_find_basis): Handle phi candidates.
10105 (slsr_process_phi): New.
10106 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10107 (create_mul_imm_cand): Likewise.
10108 (create_add_ssa_cand): Exclude phi base candidates.
10109 (create_add_imm_cand): Likewise.
10110 (slsr_process_cast): Likewise.
10111 (slsr_process_copy): Likewise.
10112 (find_candidates_in_block): Handle phi candidates.
10113 (dump_candidate): Likewise.
10114 (unconditional_cands): Delete.
10115 (unconditional_cands_with_known_stride_p): Delete.
10116 (phi_dependent_cand_p): New.
10117 (cand_increment): Handle phi-dependent candidates.
10118 (replace_dependent): Delete.
10119 (replace_mult_candidate): New.
10120 (replace_unconditional_candidate): New.
10121 (incr_vec_index): Move to avoid forward reference.
10122 (create_add_on_incoming_edge): New.
10123 (create_phi_basis): New.
10124 (replace_dependents): Delete.
10125 (replace_conditional_candidate): New.
10126 (phi_add_costs): New.
10127 (replace_uncond_cands_and_profitable_phis): New.
10128 (record_increment): Handle phi adjustments.
10129 (record_phi_increments): New.
10130 (record_increments): Handle phi adjustments.
10131 (phi_incr_cost): New.
10132 (lowest_cost_path): Handle phis.
10133 (total_savings): Likewise.
10134 (analyze_increments): Likewise.
10135 (ncd_with_phi): New.
10136 (ncd_of_cand_and_phis): New.
10137 (nearest_common_dominator_for_cands): Handle phi increments.
10138 (all_phi_incrs_profitable): New.
10139 (replace_profitable_candidates): Handle phi-dependent candidates.
10140 (analyze_candidates_and_replace): Likewise.
10142 2013-05-03 Teresa Johnson <tejohnson@google.com>
10145 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10146 do not exceed REG_BR_PROB_BASE.
10148 2013-05-03 Jeff Law <law@redhat.com>
10150 PR tree-optimization/57144
10151 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10152 operand of the condition will bit into the new type when eliminating
10153 a cast feeding a condition.
10155 2013-05-03 Jakub Jelinek <jakub@redhat.com>
10157 PR rtl-optimization/57130
10158 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10159 of COMPARE as in_code to the recursive call if needed.
10161 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
10163 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10164 (enabled): Handle new members.
10165 * config/i386/sse.md (*vec_concatv2si): Merge from
10166 *vec_concatv2si_sse2 and vec_concatv2si_sse.
10167 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10169 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
10171 PR tree-optimization/57027
10172 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10173 for fnms opportunity, check we got the prerequisite kind
10174 of tree / gimple before using accessor functions.
10176 2013-05-03 Richard Biener <rguenther@suse.de>
10178 * double-int.h (lshift): New overload without precision
10179 and arith argument.
10180 (operator *=, operator +=, operator -=): Move ...
10181 * double-int.c (operator *=, operator +=, operator -=): ... here
10182 and implement more efficiently.
10183 (mul_double_with_sign): Remove.
10184 (lshift_double): Adjust to take unsinged shift argument, push
10185 dispatching code to callers.
10186 (mul_double_wide_with_sign): Add early out for callers that
10187 are not interested in high parts or overflow.
10188 (lshift): New function.
10189 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10190 dispatch code here.
10191 (lrotate, rrotate): Use logical shifts.
10192 * expr.c (get_inner_reference): Use lshift.
10193 * fixed-value.c (do_fixed_divide): Likewise.
10194 * tree-dfa.c (get_ref_base_and_extent): Likewise.
10195 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10196 (indirect_refs_may_alias_p): Likewise.
10197 (stmt_kills_ref_p_1): Likewise.
10199 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
10201 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10203 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
10205 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10206 scalar form of FABD instruction.
10208 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
10210 * lra-constraints.c (process_alt_operands): Add checking alt
10211 number to choose the best alternative.
10213 2013-05-02 Richard Biener <rguenther@suse.de>
10215 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10216 bitmap and its handling.
10217 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10219 2013-05-02 Richard Biener <rguenther@suse.de>
10221 PR middle-end/57140
10222 * tree-inline.c (copy_loops): Properly handle removed loops.
10223 (copy_cfg_body): Mark destination loops for fixup if source
10224 loops needed fixup.
10226 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
10229 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10230 generating simple_return for naked functions.
10232 2013-05-02 Martin Jambor <mjambor@suse.cz>
10234 PR middle-end/56988
10235 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10236 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10238 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10239 ipa_agg_replacement_value structures.
10240 (known_aggs_to_agg_replacement_list): Likewise.
10241 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10242 (read_agg_replacement_chain): Likewise.
10243 (ipcp_transform_function): Also check that by_ref flags match.
10245 2013-05-02 Richard Biener <rguenther@suse.de>
10247 * graphds.h (struct graph): Add obstack member.
10248 * graphds.c (new_graph): Initialize obstack and allocate
10250 (add_edge): Allocate edge from the obstack.
10251 (free_graph): Free the obstack instead of all edges and vertices.
10253 2013-05-02 Teresa Johnson <tejohnson@google.com>
10255 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10257 * cfg.c (update_bb_profile_for_threading): Ditto.
10258 * tree-inline.c (copy_bb): Ditto.
10259 (copy_edges_for_bb): Ditto.
10260 (initialize_cfun): Ditto.
10261 (copy_cfg_body): Ditto.
10262 (expand_call_inline): Ditto.
10263 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10264 (estimate_node_size_and_time): Ditto.
10265 (inline_merge_summary): Ditto.
10266 * cgraphclones.c (cgraph_clone_edge): Ditto.
10267 (cgraph_clone_node): Ditto.
10268 * sched-rgn.c (compute_dom_prob_ps): Ditto.
10269 (compute_trg_info): Ditto.
10271 2013-05-02 Ian Bolton <ian.bolton@arm.com>
10273 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10274 S reg when fp attribute set.
10275 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10277 2013-05-02 Ian Bolton <ian.bolton@arm.com>
10279 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10281 (*and_one_cmplsi3_compare0_uxtw): Likewise.
10282 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10283 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10285 2013-05-02 Richard Biener <rguenther@suse.de>
10287 * tree-scalar-evolution.c (scev_info_hasher): Remove.
10288 (struct instantiate_cache_entry): New type.
10289 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10290 (struct instantiate_cache_type): New type.
10291 (set_instantiated_value, get_instantiated_value): Remove.
10292 (get_instantiated_value_entry): New function.
10293 (instantiate_scev_name): Use the new cache and adjust.
10294 (instantiate_scev_poly): Adjust.
10295 (instantiate_scev_binary): Likewise.
10296 (instantiate_array_ref): Likewise.
10297 (instantiate_scev_convert): Likewise.
10298 (instantiate_scev_not): Likewise.
10299 (instantiate_scev_3): Likewise.
10300 (instantiate_scev_2): Likewise.
10301 (instantiate_scev_r): Likewise.
10302 (instantiate_scev): Likewise.
10303 (resolve_mixers): Likewise.
10305 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
10308 * lra-constraints.c (best_small_class_operands_num): Remove.
10309 (process_alt_operands): Remove small_class_operands_num. Take
10310 small classes operands into losers and only if the operand is not
10311 matched. Modify debugging output.
10312 (curr_insn_transform): Remove best_small_class_operands_num.
10315 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10317 * config/aarch64/aarch64-builtins.c
10318 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10319 * config/aarch64/aarch64-simd-builtins.def
10320 (reduc_splus_): Add new modes.
10321 (reduc_uplus_): New.
10322 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10323 (reduc_uplus_v4sf): Likewise.
10324 (reduc_splus_v4sf): Likewise.
10325 (aarch64_addv<mode>): Likewise.
10326 (reduc_uplus_<mode>): Likewise.
10327 (reduc_splus_<mode>): Likewise.
10328 (aarch64_addvv2di): Likewise.
10329 (reduc_uplus_v2di): Likewise.
10330 (reduc_splus_v2di): Likewise.
10331 (aarch64_addvv2si): Likewise.
10332 (reduc_uplus_v2si): Likewise.
10333 (reduc_splus_v2si): Likewise.
10334 (reduc_<sur>plus_<mode>): New.
10335 (reduc_<sur>plus_v2di): Likewise.
10336 (reduc_<sur>plus_v2si): Likewise.
10337 (reduc_<sur>plus_v4sf): Likewise.
10338 (aarch64_addpv4sf): Likewise.
10339 * config/aarch64/arm_neon.h
10340 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10341 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10342 add UNSPEC_SADDV, UNSPEC_UADDV.
10344 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10346 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10348 * config/aarch64/arm_neon.h
10349 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10351 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10353 * config/aarch64/aarch64-builtins
10354 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10356 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10358 * config/aarch64/aarch64-simd-builtins.def
10359 (reduc_smax_): New.
10360 (reduc_smin_): Likewise.
10361 (reduc_umax_): Likewise.
10362 (reduc_umin_): Likewise.
10363 (reduc_smax_nan_): Likewise.
10364 (reduc_smin_nan_): Likewise.
10367 (smax): Update for V2SF, V4SF and V2DF modes.
10370 (smin_nan): Likewise.
10371 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10372 (<su><maxmin><mode>3): ...This, refactor.
10373 (s<maxmin><mode>3): New.
10374 (<maxmin_uns><mode>3): Likewise.
10375 (reduc_<maxmin_uns>_<mode>): Refactor.
10376 (reduc_<maxmin_uns>_v4sf): Likewise.
10377 (reduc_<maxmin_uns>_v2si): Likewise.
10378 (aarch64_<fmaxmin><mode>: Remove.
10379 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10381 (vmin<q>_f<32,64>): Likewise.
10382 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10384 (su): Add mappings for smax, smin, umax, umin.
10386 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10387 (FMAXMIN): Rename as...
10388 (FMAXMIN_UNS): ...This.
10390 (fmaxminv): Likewise.
10391 (fmaxmin): Likewise.
10393 (maxmin_uns_op): Likewise.
10395 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10397 * config/aarch64/arm_neon.h
10398 (vac<ge, gt><sd>_f<32, 64>): Rename to...
10399 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10400 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10402 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10404 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10405 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10407 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10409 * config/aarch64/aarch64-simd.md
10410 (vcond<mode>_internal): Handle special cases for constant masks.
10411 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10412 (vcondu<mode><mode>): Likewise.
10413 (vcond<v_cmp_result><mode>): New.
10415 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10417 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10418 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10419 * config/aarch64/aarch64-simd-builtins.def
10420 (cmeq): Update to BUILTIN_VALLDI.
10425 * config/aarch64/arm_neon.h
10426 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10427 to builtins or C as appropriate.
10429 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10431 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10433 (cmhi): Rename to...
10435 * config/aarch64/aarch64-simd.md
10436 (simd_mode): Add SF.
10437 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10438 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10439 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10440 (cstore<mode>_neg): ...This.
10441 * config/aarch64/iterators.md
10443 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10444 (COMPARISONS): New.
10445 (UCOMPARISONS): Likewise.
10446 (optab): Add missing comparisons.
10452 (VCMP_S): Likewise.
10453 (VCMP_U): Likewise.
10454 (V_cmp_result): Add DF, SF modes.
10455 (v_cmp_result): Likewise.
10457 (vmtype): Likewise.
10458 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10460 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
10462 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10463 define_insn to define_insn_and_split.
10464 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10465 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10466 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10467 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10468 (thumb2_negscc): Likewise.
10470 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
10472 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10474 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
10476 * config/arm/thumb2.md: Remove trailing whitespaces.
10478 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
10480 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10481 Use gen_int_mode rather than GEN_INT.
10483 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
10485 * value-prof.c (stream_in_histogram_value): Remove the strayed
10488 2013-04-30 Richard Biener <rguenther@suse.de>
10490 PR middle-end/57122
10491 * cfghooks.c (split_edge): Properly check for the loop latch edge.
10493 2013-04-30 Richard Biener <rguenther@suse.de>
10495 PR middle-end/57107
10496 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10498 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
10500 PR rtl-optimization/56957
10501 PR rtl-optimization/57105
10502 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10503 variable. Use just INSN_UID for determining whether an insn
10504 should be only disconnected from the insn stream.
10505 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10507 2013-04-30 Jakub Jelinek <jakub@redhat.com>
10509 PR tree-optimization/57104
10510 * tsan.c (instrument_expr): Don't instrument accesses to
10511 DECL_HARD_REGISTER VAR_DECLs.
10513 2013-04-30 Richard Biener <rguenther@suse.de>
10515 * function.h (loops_for_fn): New inline function.
10516 (set_loops_for_fn): Likewise.
10517 * cfgloop.h (place_new_loop): Add struct function parameter.
10518 (get_loop): Likewise.
10519 (get_loops): Likewise.
10520 (number_of_loops): Likewise.
10521 (fel_next): Adjust.
10522 (fel_init): Likewise.
10523 * cfg.c (get_loop_copy): Adjust.
10524 * cfgloop.c (flow_loops_dump): Likewise.
10525 (record_loop_exits): Likewise.
10526 (verify_loop_structure): Likewise.
10527 * cfgloopanal.c (mark_irreducible_loops): Likewise.
10528 (estimate_reg_pressure_cost): Likewise.
10529 (mark_loop_exit_edges): Likewise.
10530 * cfgloopmanip.c (place_new_loop): Likewise.
10531 (add_loop): Likewise.
10532 (duplicate_loop): Likewise.
10533 * graph.c (draw_cfg_nodes): Likewise.
10534 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10535 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10536 (extract_affine_chrec): Likewise.
10537 (build_scop_iteration_domain): Likewise.
10538 * graphite.c (graphite_initialize): Likewise.
10539 * ira-build.c (create_loop_tree_nodes): Likewise.
10540 (more_one_region_p): Likewise.
10541 (rebuild_regno_allocno_maps): Likewise.
10542 (mark_loops_for_removal): Likewise.
10543 (mark_all_loops_for_removal): Likewise.
10544 (remove_unnecessary_regions): Likewise.
10545 (ira_build): Likewise.
10546 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10547 * loop-init.c (fix_loop_structure): Likewise.
10548 (gate_rtl_move_loop_invariants): Likewise.
10549 (gate_rtl_unswitch): Likewise.
10550 (gate_rtl_unroll_and_peel_loops): Likewise.
10551 (rtl_doloop): Likewise.
10552 * lto-streamer-in.c (input_cfg): Likewise.
10553 * lto-streamer-out.c (output_cfg): Likewise.
10554 * modulo-sched.c (sms_schedule): Likewise.
10555 * predict.c (tree_estimate_probability): Likewise.
10556 (tree_estimate_probability_driver): Likewise.
10557 (estimate_loops): Likewise.
10558 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10559 (move_sese_region_to_fn): Likewise.
10560 (debug_loop_num): Likewise.
10561 * tree-chrec.c (chrec_evaluate): Likewise.
10562 (hide_evolution_in_other_loops_than_loop): Likewise.
10563 (chrec_component_in_loop_num): Likewise.
10564 (reset_evolution_in_loop): Likewise.
10565 (evolution_function_is_invariant_rec_p): Likewise.
10566 * tree-if-conv.c (main_tree_if_conversion): Likewise.
10567 * tree-inline.c (copy_loops): Likewise.
10568 (copy_cfg_body): Likewise.
10569 (tree_function_versioning): Likewise.
10570 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10571 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10573 (add_to_evolution_1): Likewise.
10574 (scev_const_prop): Likewise.
10575 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10576 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10577 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10578 (tree_ssa_lim_initialize): Likewise.
10579 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10580 (verify_loop_closed_ssa): Likewise.
10581 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
10582 (tree_ssa_loop_im): Likewise.
10583 (tree_ssa_loop_unswitch): Likewise.
10584 (tree_vectorize): Likewise.
10585 (check_data_deps): Likewise.
10586 (tree_ssa_loop_ivcanon): Likewise.
10587 (tree_ssa_loop_bounds): Likewise.
10588 (tree_complete_unroll): Likewise.
10589 (tree_complete_unroll_inner): Likewise.
10590 (tree_parallelize_loops): Likewise.
10591 (tree_ssa_loop_prefetch): Likewise.
10592 (tree_ssa_loop_ivopts): Likewise.
10593 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10594 * tree-vectorizer.c (vectorize_loops): Likewise.
10596 2013-04-29 Mike Frysinger <vapier@gentoo.org>
10598 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
10599 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
10600 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
10601 with EABI_LINK_SPEC.
10603 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
10606 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
10609 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
10612 * lra-constraints.c (process_alt_operands): Discourage a bit more
10613 using memory for pseudos. Print cost dump for alternatives.
10614 Modify cost values for conflicts with early clobbers.
10615 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
10617 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
10620 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
10622 2013-04-29 Ian Bolton <ian.bolton@arm.com>
10624 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
10625 from/to S register.
10626 (movdi_aarch64): Support LDR/STR from/to D register.
10628 2013-04-29 Ian Bolton <ian.bolton@arm.com>
10630 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
10631 or higher by default.
10633 2013-04-29 Richard Biener <rguenther@suse.de>
10635 PR middle-end/57075
10636 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
10637 even if not adding abnormal edges for calls that can make
10640 2013-04-29 Richard Biener <rguenther@suse.de>
10642 PR middle-end/57103
10643 * tree-cfg.c (move_stmt_op): Fix condition under which to update
10645 (move_stmt_r): Remove redundant checking.
10647 2013-04-29 Teresa Johnson <tejohnson@google.com>
10650 * basic-block.h (apply_scale): New function.
10651 (apply_probability): Use apply_scale.
10652 * gimple-streamer-in.c (input_bb): Ditto.
10653 * lto-streamer-in.c (input_cfg): Ditto.
10654 * lto-cgraph.c (merge_profile_summaries): Ditto.
10655 * tree-optimize.c (execute_fixup_cfg): Ditto.
10656 * tree-inline.c (copy_bb): Update comment to use apply_scale.
10657 (copy_edges_for_bb): Ditto.
10658 (copy_cfg_body): Ditto.
10660 2013-04-29 Tom de Vries <tom@codesourcery.com>
10662 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
10663 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
10664 (tail_merge_optimize): Handle current_loops == NULL.
10666 2013-04-26 Jeff Law <law@redhat.com>
10668 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
10669 (simplify_cond_using_ranges): Generalize code to simplify
10670 COND_EXPRs where one argument is a constant and the other
10671 is an SSA_NAME created by an integral type conversion.
10673 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10675 * config/arm/arm.md (store_minmaxsi): Use only when
10676 optimize_insn_for_size_p.
10678 2013-04-29 Christian Bruel <christian.bruel@st.com>
10681 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
10683 2013-04-29 Richard Biener <rguenther@suse.de>
10685 PR middle-end/57089
10686 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
10687 loop tree make sure to schedule a fixup for the child as well.
10688 (expand_omp_for_generic): Properly add loops.
10689 (expand_omp_for_static_nochunk): Likewise.
10690 (expand_omp_for_static_chunk): Likewise.
10691 (expand_omp_for): For the degenerate case fixup loops.
10692 (expand_omp_sections): Fix default bb placement in loops.
10693 (expand_omp_atomic_pipeline): Properly add loops.
10695 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10697 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
10699 2013-04-29 Tom de Vries <tom@codesourcery.com>
10701 * tree-ssa-tail-merge.c: Update header comment.
10703 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10705 * config/aarch64/arm_neon.h
10706 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
10707 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
10708 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
10709 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
10710 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
10711 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
10712 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
10713 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
10715 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10717 * config/aarch64/aarch64-simd.md
10718 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
10719 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
10720 fix_trunc, fixuns_trunc.
10721 (ftrunc<VDQF:mode>2): New.
10722 * config/aarch64/iterators.md (optab): Add fix, fixuns.
10723 (fix_trunc_optab): New.
10725 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10727 * config/aarch64/aarch64-builtins.c
10728 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
10729 iceilf, lround, iroundf.
10731 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
10734 * config/i386/i386.h (enum ix86_tune_indices)
10735 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
10736 New, split from X86_TUNE_INTER_UNIT_MOVES.
10737 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
10738 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
10739 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
10740 (TARGET_INTER_UNIT_MOVES): Remove.
10741 * config/i386/i386.c (initial_ix86_tune_features): Update.
10742 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
10743 (ix86_expand_convert_uns_didf_sse): Use
10744 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10745 (ix86_expand_vector_init_one_nonzero): Ditto.
10746 (ix86_expand_vector_init_interleave): Ditto.
10747 (inline_secondary_memory_needed): Return true for moves from SSE class
10748 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
10749 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
10750 * config/i386/constraints.md (Yi, Ym): Depend on
10751 TARGET_INTER_UNIT_MOVES_TO_VEC.
10752 (Yj, Yn): New constraints.
10753 * config/i386/i386.md (*movdi_internal): Change constraints of
10754 operand 1 from Yi to Yj and from Ym to Yn.
10755 (*movsi_internal): Ditto.
10756 (*movdf_internal): Ditto.
10757 (*movsf_internal): Ditto.
10758 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
10759 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10760 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
10761 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
10762 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
10763 * config/i386/sse.md (movdi_to_sse): Ditto.
10764 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
10765 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
10766 TARGET_INTER_UNIT_MOVES.
10767 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
10768 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
10769 instead of TARGET_INTER_UNIT_MOVES.
10770 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
10771 operand 1 from Yi to Yj and from Ym to Yn.
10773 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10775 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
10776 (float_truncate_hi_): Likewise.
10777 (float_extend_lo_): Likewise.
10778 (float_truncate_lo_): Likewise.
10779 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
10780 (aarch64_float_extend_lo_v2df): Likewise.
10781 (vec_unpacks_hi_v4sf): Likewise.
10782 (aarch64_float_truncate_lo_v2sf): Likewise.
10783 (aarch64_float_truncate_hi_v4sf): Likewise.
10784 (vec_pack_trunc_v2df): Likewise.
10785 (vec_pack_trunc_df): Likewise.
10787 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10789 * config/aarch64/aarch64-builtins.c
10790 (aarch64_fold_builtin): Fold float conversions.
10791 * config/aarch64/aarch64-simd-builtins.def
10792 (floatv2si, floatv4si, floatv2di): New.
10793 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
10794 * config/aarch64/aarch64-simd.md
10795 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
10796 * config/aarch64/iterators.md (FLOATUORS): New.
10797 (optab): Add float, floatuns.
10798 (su_optab): Likewise.
10800 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10802 * config/aarch64/aarch64-builtins.c
10803 (aarch64_builtin_vectorized_function): Use new names for
10805 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
10806 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
10807 (fcvtzu): Split as...
10808 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
10809 (fcvtas): Split as...
10810 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
10811 (fcvtau): Split as...
10812 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
10813 (fcvtps): Split as...
10814 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
10815 (fcvtpu): Split as...
10816 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
10817 (fcvtms): Split as...
10818 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
10819 (fcvtmu): Split as...
10820 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
10821 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
10822 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
10823 (lfrintnusf, lfrintnudf): Likewise.
10824 * config/aarch64/aarch64-simd.md
10825 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
10827 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
10828 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
10829 (fcvt_pattern): Likewise.
10831 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10833 * config/aarch64/aarch64-simd.md
10834 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10835 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10837 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10839 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10840 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10841 (vrnd<a,m,n,p>_f32): Implement using builtins.
10842 (vrnd<i,x><q>_f<32, 64>): New.
10844 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
10846 * config/aarch64/aarch64-builtins.c
10847 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10848 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10849 (frintz): Rename to...
10851 (frintp): Rename to...
10853 (frintm): Rename to...
10855 (frinti): Rename to...
10856 (nearbyint): ...this.
10857 (frintx): Rename to...
10859 (frinta): Rename to...
10861 * config/aarch64/aarch64-simd.md
10862 (aarch64_frint<frint_suffix><mode>): Delete.
10863 (<frint_pattern><mode>2): Convert to insn.
10864 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
10865 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
10866 (frint_pattern): Likewise.
10867 (frint_suffix): Likewise.
10869 2013-04-29 Richard Biener <rguenther@suse.de>
10871 PR tree-optimization/57081
10872 * loop-init.c: Include tree-flow.h.
10873 (loop_optimizer_finalize): Free number of iteration estimates.
10874 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
10876 2013-04-29 Jakub Jelinek <jakub@redhat.com>
10878 PR tree-optimization/57083
10879 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
10880 non-singleton shift count range, zero extend low_bound for uns case.
10882 * config/i386/predicates.md (general_vector_operand): New predicate.
10883 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
10884 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
10885 if they aren't nonimmediate operands. If their original values
10886 satisfy const_vector_equal_evenodd_p, don't shift them.
10887 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
10888 predicates. For the SSE4.1 case force operands[{1,2}] into registers
10889 if not nonimmediate_operand.
10890 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
10891 instead of register_operand.
10892 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
10894 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
10896 * stor-layout.c (finalize_size_functions): Allocate a structure and
10897 reset cfun before dumping the functions.
10899 2013-04-27 Jakub Jelinek <jakub@redhat.com>
10901 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
10904 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
10905 use xop_pmacsdqh if uns_p.
10906 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
10907 the immediate rotate count.
10909 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
10911 * rtl.h (struct rtx_def): Add comment for field jump.
10912 (LRA_SUBREG_P): New macro.
10913 * recog.c (register_operand): Check LRA_SUBREG_P.
10914 * lra.c (lra): Add note at the end of RTL code. Align non-empty
10916 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
10917 (lra_final_code_change): Skip subreg change for operators.
10918 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
10919 if there are no operand changes.
10920 * lra-constraints.c (curr_insn_set): New.
10921 (match_reload): Set LRA_SUBREG_P.
10922 (emit_spill_move): Ditto.
10923 (check_and_process_move): Use curr_insn_set. Process only single
10924 set insns. Don't initialize sec_mem_p and change_p.
10925 (simplify_operand_subreg): Use LRA_SUBREG_P.
10926 (reg_in_class_p): New function.
10927 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
10928 of #ifdef. Add code to remove cycling.
10929 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
10930 non-null disp. Reload inner instead of disp when base and index
10931 are null. Try to put lo_sum into register.
10932 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
10933 (check_and_process_move): Move code for move cost check to
10934 simple_move_p. Remove equiv_substitution.
10935 (simple_move_p): New function.
10936 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
10937 curr_insn_set. Call check_and_process_move only for single set
10938 insns. Use the new function. Move call of check_and_process_move
10939 after operand equiv substitution and address process.
10941 2013-04-26 Jakub Jelinek <jakub@redhat.com>
10944 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
10945 with nonlocal goto receivers or returns twice calls, ignore
10946 unininitialized values from abnormal edges to nl goto receiver
10947 or returns twice call.
10949 2013-04-26 Jakub Jelinek <jakub@redhat.com>
10951 PR tree-optimization/57051
10952 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
10953 and VEC_RSHIFT_EXPR if shift count is a multiple of element
10956 2013-04-26 Richard Biener <rguenther@suse.de>
10958 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
10959 (expand_omp_taskreg): Likewise. Mark loops for fixup.
10960 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
10961 (fixup_loop_arrays_after_move): New function.
10962 (move_sese_region_to_fn): Properly outline the loop tree parts
10963 of the SESE region.
10965 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
10967 * config/i386/i386.md (type, unit): Fix long lines.
10969 2013-04-26 Richard Biener <rguenther@suse.de>
10971 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
10972 (lto-streamer-out.o): Likewise.
10973 * cfgloop.c (init_loops_structure): Export, add struct function
10974 argument and adjust.
10975 (flow_loops_find): Adjust.
10976 * cfgloop.h (enum loop_estimation): Add EST_LAST.
10977 (init_loops_structure): Declare.
10978 * lto-streamer-in.c: Include cfgloop.h.
10979 (input_cfg): Input the loop tree.
10980 * lto-streamer-out.c: Include cfgloop.h.
10981 (output_cfg): Output the loop tree.
10982 (output_struct_function_base): Do not drop PROP_loops.
10984 2013-03-26 Richard Biener <rguenther@suse.de>
10986 * tree-cfg.c (execute_build_cfg): Build the loop tree.
10987 (pass_build_cfg): Provide PROP_loops.
10988 (move_sese_region_to_fn): Remove loops that are outlined into fn
10990 * tree-inline.c: Include cfgloop.h.
10991 (initialize_cfun): Do not drop PROP_loops.
10992 (copy_loops): New function.
10993 (copy_cfg_body): Copy loop structure.
10994 (tree_function_versioning): Initialize destination loop tree.
10995 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
10996 (pass_parallelize_loops): Do IL verification.
10997 * loop-init.c (loop_optimizer_init): Fixup loops if required.
10998 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
10999 the CFG make sure we fixup loops as well.
11000 * tree-ssa-tail-merge.c: Include cfgloop.h.
11001 (replace_block_by): When merging loop latches mark loops for fixup.
11002 * lto-streamer-out.c (output_struct_function_base): Drop
11003 PROP_loops for now.
11004 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11005 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11006 * ipa-split.c: Include cfgloop.h.
11007 (split_function): Add the new return block to the loop tree root.
11008 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11009 whether we have removed the forwarder block.
11010 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11011 * cfgloop.h (place_new_loop): Declare.
11012 * cfgloopmanip.c (place_new_loop): Export.
11013 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11014 (tree-switch-conversion.o): Likewise.
11015 (tree-complex.o): Likewise.
11016 (tree-inline.o): Likewise.
11017 (tree-ssa-tailmerge.o): Likewise.
11018 (ipa-split.o): Likewise.
11019 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11020 (tree-ssa-copy.o): Likewise.
11021 * tree-switch-conversion.c: Include cfgloop.h
11022 (process_switch): If we emit a bit-test cascade, schedule loops
11024 * tree-complex.c: Include cfgloop.h.
11025 (expand_complex_div_wide): Properly add new basic-blocks to loops.
11026 * asan.c: Include cfgloop.h.
11027 (create_cond_insert_point): Properly add new basic-blocks to
11028 loops, schedule loop fixup.
11029 * cfgloop.c (verify_loop_structure): Check that looks are not
11031 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11033 (expand_omp_for_generic): Likewise.
11034 (expand_omp_sections): Likewise.
11035 (expand_omp_atomic_pipeline): Schedule loops for fixup.
11036 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11037 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11038 is initialized, not when loops are present.
11039 * tree-parloops.c (parallelize_loops): Remove checking here.
11040 * passes.c (init_optimization_passes): Schedule a copy-propagation
11041 pass before complete unrolling of inner loops.
11043 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11045 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11046 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11047 (colorize_init): Add argument to _WIN32 version.
11048 * toplev.c: Include diagnostic-color.h.
11049 (process_options): Default to -fdiagnostics-color=auto if
11050 GCC_COLORS env var is in the environment.
11051 * common.opt (fdiagnostics-color=): Add Var and Init.
11052 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11053 env var is in the environment, the default is auto rather than never.
11055 * diagnostic.h (file_name_as_prefix): Add context argument.
11056 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
11057 the string as locus.
11058 * langhooks.c (lhd_print_error_function): Adjust caller.
11060 2013-04-25 Lawrence Crowl <crowl@google.com>
11062 * var-tracking.c (shared_hash_def::htab):
11063 Change type to hash_table. Update dependent calls and types.
11065 2013-04-25 Lawrence Crowl <crowl@google.com>
11067 * Makefile.in: Update as needed below.
11069 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11070 Move declaration to after the type's method definitons.
11072 * attribs.c (htab_t scoped_attributes::attribute_hash):
11073 Change type to hash_table. Update dependent calls and types.
11075 * bitmap.c (htab_t bitmap_desc_hash):
11076 Change type to hash_table. Update dependent calls and types.
11078 * cselib.c (htab_t cselib_hash_table):
11079 Change type to hash_table. Update dependent calls and types.
11081 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11082 (hash_string_slot_node): Move implementation into lto-streamer.h
11083 struct string_slot_hasher.
11084 (eq_string_slot_node): Likewise.
11086 * data-streamer-out.c: Update output_block::string_hash_table
11087 dependent calls and types.
11089 * dwarf2cfi.c (htab_t trace_index):
11090 Change type to hash_table. Update dependent calls and types.
11092 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11093 Change type to hash_table. Update dependent calls and types.
11094 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11095 (htab_t optimize_external_refs::map): Likewise.
11096 (htab_t output_comp_unit::extern_map): Likewise.
11097 (htab_t output_comdat_type_unit::extern_map): Likewise.
11098 (htab_t output_macinfo::macinfo_htab): Likewise.
11099 (htab_t optimize_location_lists::htab): Likewise.
11100 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11102 * except.c (htab_t ehspec_hash_type):
11103 Change type to hash_table. Update dependent calls and types.
11104 (assign_filter_values::ttypes): Likewise.
11105 (assign_filter_values::ehspec): Likewise.
11106 (sjlj_assign_call_site_values::ar_hash): Likewise.
11107 (convert_to_eh_region_ranges::ar_hash): Likewise.
11109 * gcse.c (htab_t pre_ldst_table):
11110 Change type to hash_table. Update dependent calls and types.
11112 * ggc-common.c (htab_t saving_htab):
11113 Change type to hash_table. Update dependent calls and types.
11114 (htab_t loc_hash): Likewise.
11115 (htab_t ptr_hash): Likewise.
11116 (call_count): Rename ggc_call_count.
11117 (call_alloc): Rename ggc_call_alloc.
11118 (loc_descriptor): Rename make_loc_descriptor.
11119 (add_statistics): Rename ggc_add_statistics.
11121 * ggc-common.c (saving_htab):
11122 Change type to hash_table. Update dependent calls and types.
11124 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11125 (push_gimplify_context): Likewise.
11126 (pop_gimplify_context): Likewise.
11127 (struct gimple_temp_hash_elt): Added.
11128 (struct gimplify_hasher): Likewise.
11129 (struct gimplify_ctx.temp_htab):
11130 Change type to hash_table. Update dependent calls and types.
11132 * gimple-fold.c: Include gimplify-ctx.h.
11134 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11135 Change type to hash_table. Update dependent calls and types.
11136 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11137 avoid potential global name collision.
11139 * gimplify.c: Include gimplify-ctx.h.
11140 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11141 (htab_t gimplify_ctx::temp_htab):
11142 Update dependent calls and types for new type hash_table.
11143 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11144 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11146 * gimplify-ctx.h: New.
11147 (struct gimple_temp_hash_elt): Move from gimplify.c.
11148 (class gimplify_hasher): New.
11149 (struct gimplify_ctx): Move from gimple.h.
11150 (htab_t gimplify_ctx::temp_htab):
11151 Change type to hash_table. Update dependent calls and types.
11153 * graphite-clast-to-gimple.c: Include graphite-htab.h.
11154 (htab_t ivs_params::newivs_index):
11155 Change type to hash_table. Update dependent calls and types.
11156 (htab_t ivs_params::params_index): Likewise.
11157 (htab_t print_generated_program::params_index): Likewise.
11158 (htab_t gloog::newivs_index): Likewise.
11159 (htab_t gloog::params_index): Likewise.
11161 * graphite.c: Include graphite-htab.h.
11162 4htab_t graphite_transform_loops::bb_pbb_mapping):
11163 Change type to hash_table. Update dependent calls and types.
11165 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11166 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11167 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11169 * graphite-dependences.c: Include graphite-htab.h.
11170 (loop_is_parallel_p): Change hash table type of parameter.
11172 * graphite-htab.h: New.
11173 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11174 (extern find_pbb_via_hash): Move from graphite-poly.h.
11175 (extern loop_is_parallel_p): Move from graphite-poly.h.
11176 (extern get_loop_body_pbbs): Move from graphite-poly.h.
11178 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11179 (extern loop_is_parallel_p): Move to graphite-htab.h.
11180 (extern get_loop_body_pbbs): Move to graphite-htab.h.
11182 * haifa-sched.c (htab_t delay_htab):
11183 Change type to hash_table. Update dependent calls and types.
11184 (htab_t delay_htab_i2): Likewise.
11186 * ira-color.c (htab_t allocno_hard_regs_htab):
11187 Change type to hash_table. Update dependent calls and types.
11189 * ira-costs.c (htab_t cost_classes_htab):
11190 Change type to hash_table. Update dependent calls and types.
11192 * loop-invariant.c (htab_t merge_identical_invariants::eq):
11193 Change type to hash_table. Update dependent calls and types.
11195 * loop-iv.c (htab_t bivs):
11196 Change type to hash_table. Update dependent calls and types.
11198 * loop-unroll.c (htab_t opt_info::insns_to_split):
11199 Change type to hash_table. Update dependent calls and types.
11200 (htab_t opt_info::insns_with_var_to_expand): Likewise.
11202 * lto-streamer.h (struct string_slot): Move from data-streamer.h
11203 (struct string_slot_hasher): New.
11204 (htab_t output_block::string_hash_table):
11205 Change type to hash_table. Update dependent calls and types.
11207 * lto-streamer-in.c (freeing_string_slot_hasher): New.
11208 (htab_t file_name_hash_table):
11209 Change type to hash_table. Update dependent calls and types.
11211 * lto-streamer-out.c: Update output_block::string_hash_table dependent
11214 * lto-streamer.c (htab_t tree_htab):
11215 Change type to hash_table. Update dependent calls and types.
11217 * omp-low.c: Include gimplify-ctx.h.
11219 * passes.c (htab_t name_to_pass_map):
11220 Change type to hash_table. Update dependent calls and types.
11221 (pass_traverse): Rename to passes_pass_traverse.
11223 * plugin.c (htab_t event_tab):
11224 Change type to hash_table. Update dependent calls and types.
11226 * postreload-gcse.c (htab_t expr_table):
11227 Change type to hash_table. Update dependent calls and types.
11228 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11230 * sese.c (debug_rename_map_1): Make extern.
11231 (htab_t copy_bb_and_scalar_dependences::rename_map):
11232 Change type to hash_table. Update dependent calls and types.
11234 * sese.h (extern debug_rename_map): Move to .c file.
11236 * store-motion.c (htab_t store_motion_mems_table):
11237 Change type to hash_table. Update dependent calls and types.
11239 * trans-mem.c (htab_t tm_new_mem_hash):
11240 Change type to hash_table. Update dependent calls and types.
11242 * tree-browser.c (htab_t TB_up_ht):
11243 Change type to hash_table. Update dependent calls and types.
11245 * tree-cfg.c (htab_t discriminator_per_locus):
11246 Change type to hash_table. Update dependent calls and types.
11248 * tree-complex.c: Include tree-hasher.h
11249 (htab_t complex_variable_components):
11250 Change type to hash_table. Update dependent calls and types.
11252 * tree-eh.c (htab_t finally_tree):
11253 Change type to hash_table. Update dependent calls and types.
11255 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11256 struct int_tree_hasher.
11257 (extern int_tree_map_eq): Likewise.
11258 (uid_decl_map_hash): Removed.
11259 (extern decl_tree_map_eq): Likewise.
11261 * tree-hasher.h: New.
11262 (struct int_tree_hasher): New.
11263 (typedef int_tree_htab_type): New.
11265 * tree-inline.c: Include gimplify-ctx.h.
11267 * tree-mudflap.c: Include gimplify-ctx.h.
11269 * tree-parloops.c: Include tree-hasher.h.
11270 (htab_t eliminate_local_variables_stmt::decl_address):
11271 Change type to hash_table. Update dependent calls and types.
11272 (htab_t separate_decls_in_region::decl_copies): Likewise.
11274 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11275 Change type to hash_table. Update dependent calls and types.
11277 * tree-sra.c (candidates):
11278 Change type to hash_table. Update dependent calls and types.
11280 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11282 (int_tree_map_hash): Likewise.
11284 * tree-ssa-dom.c (htab_t avail_exprs):
11285 Change type to hash_table. Update dependent calls and types.
11287 * tree-ssa-live.c (var_map_base_init::tree_to_index):
11288 Change type to hash_table. Update dependent calls and types.
11290 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11291 Change type to hash_table. Update dependent calls and types.
11293 * tree-ssa-phiopt.c (seen_ssa_names):
11294 Change type to hash_table. Update dependent calls and types.
11296 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11297 Change type to hash_table. Update dependent calls and types.
11299 * tree-ssa-uncprop.c (equiv):
11300 Change type to hash_table. Update dependent calls and types.
11302 2013-04-25 Jakub Jelinek <jakub@redhat.com>
11304 PR rtl-optimization/57003
11305 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11306 call note_stores with kill_clobbered_value callback again after
11307 killing regs_invalidated_by_call.
11309 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11311 * config/aarch64/aarch64-simd.md
11312 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11313 (aarch64_simd_bsl<mode>): Likewise.
11314 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11316 2013-04-25 Marek Polacek <polacek@redhat.com>
11318 PR tree-optimization/57066
11319 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11321 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
11323 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11325 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11327 * config/aarch64/aarch64-builtins.c
11328 (aarch64_fold_builtin): New.
11329 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11330 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11331 * config/aarch64/aarch64-simd-builtins.def (abs): New.
11332 * config/aarch64/arm_neon.h
11333 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11335 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11336 Tejas Belagod <tejas.belagod@arm.com>
11338 * config/aarch64/aarch64-builtins.c
11339 (aarch64_gimple_fold_builtin): New.
11340 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11341 * config/aarch64/aarch64-simd-builtins.def (addv): New.
11342 * config/aarch64/aarch64-simd.md (addpv4sf): New.
11343 (addvv4sf): Update.
11344 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11346 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11348 * config/aarch64/aarch64.md
11349 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11351 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11353 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11354 (*ngcsi_uxtw): New pattern.
11356 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11357 Julian Brown <julian@codesourcery.com>
11359 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11360 (TB_DREG): Add T_V4HF.
11361 (v4hf_UP): New macro.
11362 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11363 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11364 Handle initialisation of V4HF. Adjust initialisation of reinterpret
11366 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11367 (arm_vector_mode_supported_p): Handle V4HF.
11368 (arm_mangle_map): Handle V4HFmode.
11369 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11370 * config/arm/arm_neon_builtins.def: Add entries for
11371 vcvtv4hfv4sf, vcvtv4sfv4hf.
11372 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11373 (neon_vcvtv4hfv4sf): Likewise.
11374 * config/arm/neon-gen.ml: Handle half-precision floating point
11376 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11377 * config/arm/arm_neon.h: Regenerate.
11378 * config/arm/neon.ml (type elts): Add F16.
11379 (type vectype): Add T_float16x4, T_floatHF.
11380 (type vecmode): Add V4HF.
11381 (type features): Add Requires_FP_bit feature.
11382 (elt_width): Handle F16.
11383 (elt_class): Likewise.
11384 (elt_of_class_width): Likewise.
11385 (mode_of_elt): Refactor.
11386 (type_for_elt): Handle F16, fix error messages.
11387 (vectype_size): Handle T_float16x4.
11388 (vcvt_sh): New function.
11389 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11390 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11391 (string_of_mode): Handle V4HF.
11392 * doc/arm-neon-intrinsics.texi: Regenerate.
11394 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11396 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11397 format specifier in 'X' case.
11399 2013-04-25 Alan Modra <amodra@gmail.com>
11402 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11403 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11404 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11405 Repeat for many other rotate/shift and mask patterns using subregs.
11406 Name lshiftrt insns.
11407 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11408 on WORDS_BIG_ENDIAN.
11410 2013-04-25 Alan Modra <amodra@gmail.com>
11412 * config.gcc: Support little-endian powerpc-linux targets.
11413 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11414 (LINK_OS_LINUX_SPEC): Define.
11415 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11416 Preserve MASK_LITTLE_ENDIAN.
11417 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11418 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11419 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11420 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11421 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11422 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11423 Correct fp word order for little-endian. Don't shift toc entries
11424 smaller than a word for little-endian.
11425 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11426 (bswapdi2 splits): Correct low-part subreg for little-endian.
11427 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11428 low/high where such is correct only for be.
11429 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11430 little-endian for -mcall-aixdesc.
11432 2013-04-25 Alan Modra <amodra@gmail.com>
11434 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11435 replace_equiv_address_nv.
11437 2013-04-25 Alan Modra <amodra@gmail.com>
11439 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11441 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11444 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11445 * rtl.h (struct rtx_def): ...
11447 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11449 PR rtl-optimizations/57046
11450 * lra-constraints (split_reg): Set up lra_risky_transformations_p
11451 for multi-reg splits.
11453 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
11455 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11457 2013-04-24 Sterling Augustine <saugustine@google.com>
11459 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11460 (comp_dir_string, debug_str_dwo_section): New.
11461 (DEBUG_STR_DWO_SECTION): Rename to ...
11462 (DEBUG_DWO_STR_SECTION): ... this.
11463 (DEBUG_NORM_STR_SECTION): Delete.
11464 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11465 (DEBUG_STR_DWO_SECTION_FLAGS): New.
11466 (find_AT_string): Move most logic to ...
11467 (find_AT_string_in_table): ... here. New.
11468 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11469 add_skeleton_AT_string. Delete logic.
11470 (output_skeleton_debug_sections): Remove call to
11471 add_top_level_skeleton_die_attrs.
11472 (add_comp_dir_attribute): Move logic to comp_dir_string.
11473 (dwarf2out_init): Initialize debug_str_dwo_section.
11474 (output_indirect_string): Call find_string_form.
11475 (output_indirect_strings): Rewrite.
11476 (prune_unused_types): Empty skeleton_debug_str_hash.
11477 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11478 (dwarf2out_finish): Call output_indirect_strings.
11480 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
11482 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11484 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11486 * rtl.h (struct rtx_def): Add comment for field jump.
11487 (LRA_SUBREG_P): New macro.
11488 * recog.c (register_operand): Check LRA_SUBREG_P.
11489 * lra.c (lra): Add note at the end of RTL code. Align non-empty
11491 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11492 (lra_final_code_change): Skip subreg change for operators.
11493 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11494 if there are no operand changes.
11495 * lra-constraints.c (curr_insn_set): New.
11496 (match_reload): Set LRA_SUBREG_P.
11497 (emit_spill_move): Ditto.
11498 (check_and_process_move): Use curr_insn_set. Process only single
11499 set insns. Don't initialize sec_mem_p and change_p.
11500 (simplify_operand_subreg): Use LRA_SUBREG_P.
11501 (reg_in_class_p): New function.
11502 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
11503 of #ifdef. Add code to remove cycling.
11504 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11505 non-null disp. Reload inner instead of disp when base and index
11506 are null. Try to put lo_sum into register.
11507 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11508 (check_and_process_move): Move code for move cost check to
11509 simple_move_p. Remove equiv_substitution.
11510 (simple_move_p): New function.
11511 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
11512 curr_insn_set. Call check_and_process_move only for single set
11513 insns. Use the new function. Move call of check_and_process_move
11514 after operand equiv substitution and address process.
11516 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
11518 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11519 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11520 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11522 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
11524 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11526 2013-04-24 Marek Polacek <polacek@redhat.com>
11528 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11529 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11530 (select_loops_exit_conditions): Likewise.
11531 (number_of_iterations_for_all_loops): Likewise.
11532 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11533 (scev_analysis): Likewise.
11535 2013-04-02 Catherine Moore <clm@codesourcery.com>
11536 Chao-ying Fu <fu@mips.com>
11538 * config/mips/micromips.md (jraddiusp): New pattern.
11539 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11540 instruction if possible.
11542 2013-04-24 Alan Modra <amodra@gmail.com>
11544 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11546 2013-04-24 Julian Brown <julian@codesourcery.com>
11547 Chung-Lin Tang <cltang@codesourcery.com>
11549 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11550 dependency behavior in enumeration type DIE generation. Add TODO note
11551 to comments about future DW_FORM_sdata/udata re-work of related code.
11553 2013-04-23 Lawrence Crowl <crowl@google.com>
11555 * Makefile.in: Update as needed below.
11557 * hash-table.h (class hash_table):
11558 Correct many methods with parameter types compare_type to the correct
11559 value_type. (Correct code was unlikely to notice the change.)
11560 (hash_table::elements_with_deleted) New.
11561 (class hashtable::iterator): New.
11562 (hashtable::begin()): New.
11563 (hashtable::end()): New.
11564 (FOR_EACH_HASH_TABLE_ELEMENT): New.
11566 * statistics.c (statistics_hashes):
11567 Change type to hash_table. Update dependent calls and types.
11569 * tree-into-ssa.c (var_infos):
11570 Change type to hash_table. Update dependent calls and types.
11572 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11573 Change type to hash_table. Update dependent calls and types.
11575 * tree-ssa-loop-im.c (struct mem_ref.refs):
11576 Change type to hash_table. Update dependent calls and types.
11578 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11579 Change type to hash_table. Update dependent calls and types.
11581 * tree-ssa-sccvn.c (vn_tables_s::nary):
11582 Change type to hash_table. Update dependent calls and types.
11583 (vn_tables_s::phis): Likewise.
11584 (vn_tables_s::references): Likewise.
11586 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11587 (vn_reference_eq): Update parameter and return types.
11589 * tree-ssa-structalias.c (pointer_equiv_class_table):
11590 Change type to hash_table. Update dependent calls and types.
11591 (location_equiv_class_table): Likewise.
11593 * tree-vect-data-refs.c: Consequential changes for making
11594 peeling a hash_table.
11596 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
11597 (destroy_loop_vec_info): Dependent hash_table update.
11599 * tree-vectorizer.h (peeling_htab):
11600 Change type to hash_table. Update dependent calls and types.
11602 2013-04-23 Shiva Chen <shiva0217@gmail.com>
11604 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
11605 to check the register content is equal or not.
11606 * lra-constraints.c (match_reload): Use lra_assign_reg_val
11607 to assign register content record.
11608 * lra-eliminations.c (update_reg_eliminate): Use
11609 lra_update_reg_val_offset to update register content offset.
11610 * lra-int.h (struct lra_reg): Add offset member.
11611 (lra_reg_val_equal_p): New static inline function.
11612 (lra_update_reg_val_offset): New static inline function.
11613 (lra_assign_reg_val): New static inline function.
11614 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
11615 to assign register content record.
11616 (initialize_lra_reg_info_element): Initial offset to zero.
11618 2013-04-23 Catherine Moore <clm@codesourcery.com>
11620 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
11621 operands. Record compression.
11623 2013-04-23 Xinliang David Li <davidxl@google.com>
11625 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
11627 2013-04-23 Richard Biener <rguenther@suse.de>
11629 PR middle-end/57036
11630 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
11631 parameter, only add abnormal goto edges from the copied body
11632 if the call could perform abnormal gotos.
11633 (copy_cfg_body): Adjust.
11635 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
11637 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
11639 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
11641 * coretypes.h (gimple_stmt_iterator): Add struct to make
11644 2013-04-23 Richard Biener <rguenther@suse.de>
11646 PR tree-optimization/57026
11647 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
11648 from SSA names occuring in abnormal PHI nodes.
11650 2013-04-22 Andi Kleen <ak@linux.intel.com>
11652 * lto/lto.c (print_lto_report_1): Fix LTO report names.
11654 2013-04-22 Andi Kleen <ak@linux.intel.com>
11656 * lto/lto.c (print_lto_report_1): Declare early.
11657 (read_cgraph_and_symbols): Call print_lto_report_1 early.
11659 2013-04-22 Andi Kleen <ak@linux.intel.com>
11661 * common.opt (-flto-report-wpa): Add.
11662 * doc/invoke.texi (-flto-report-wpa): Add.
11663 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
11666 2013-04-22 Xinliang David Li <davidxl@google.com>
11668 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
11669 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
11670 * Makefile.in: New dependency
11672 David Daney <ddaney.cavm@gmail.com>
11674 * configure.ac (gcc_cv_as_micromips_support): Use the
11675 --fatal-warnings option.
11676 * configure: Regenerate.
11678 2013-04-22 Marek Polacek <polacek@redhat.com>
11681 * tsan.c (instrument_expr): Don't instrument expression
11682 in case its size is zero.
11684 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
11688 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
11690 * config/alpha/alpha.c (TARGET_LRA_P): New define.
11692 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
11694 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
11695 (gimple_stmt_iterator): New typedef.
11696 * gimple.h (gimple_stmt_iterator): Rename to...
11697 (gimple_stmt_iterator_d): ... This.
11698 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
11699 trees be valid for GIMPLE and GENERIC.
11700 (TARGET_GIMPLE_FOLD_BUILTIN): New.
11701 * gimple-fold.c (gimple_fold_call): Call target hook
11702 gimple_fold_builtin.
11703 * hooks.c (hook_bool_gsiptr_false): New.
11704 * hooks.h (hook_bool_gsiptr_false): New.
11705 * target.def (fold_stmt): New.
11706 * doc/tm.texi: Regenerate.
11708 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
11711 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
11712 a set sp if no stack realignment.
11714 2013-04-22 Nick Clifton <nickc@redhat.com>
11716 * config.gcc (tilegx-linux): Extend extra_objs rather than
11718 (tilepro-linux): Likewise.
11720 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
11722 * config/aarch64/aarch64-builtins.c
11724 (CF0, CF1, CF2, CF3, CF4, CF10): New.
11725 (VAR<1-12>): Add MAP parameter.
11726 (BUILTIN_*): Likewise.
11727 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
11728 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
11729 (aarch64_ushl_n<mode>): Likewise.
11730 (aarch64_sshr_n<mode>): Likewise.
11731 (aarch64_ushr_n<mode>): Likewise.
11732 (aarch64_<maxmin><mode>): Likewise.
11733 (aarch64_sqrt<mode>): Likewise.
11734 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
11735 (vshr<q>_n_*): Likewise.
11737 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
11739 * config/aarch64/aarch64-builtins.c
11740 (aarch64_simd_builtin_type_mode): Handle SF types.
11742 (BUILTIN_GPF): Define.
11743 (aarch64_init_simd_builtins): Handle SF types.
11744 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
11745 (frecps): Likewise.
11746 (frecpx): Likewise.
11747 * config/aarch64/aarch64-simd.md
11748 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
11749 (aarch64_frecpe<mode>): New.
11750 (aarch64_frecps<mode>): Likewise.
11751 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
11752 (v8type): Add frecp<esx>.
11753 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
11754 (aarch64_frecps<mode>): Likewise.
11755 * config/aarch64/iterators.md (FRECP): New.
11756 (frecp_suffix): Likewise.
11757 * config/aarch64/arm_neon.h
11758 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
11760 2013-04-22 Christian Bruel <christian.bruel@st.com>
11763 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
11764 (REG_CLASS_NAMES): Idem.
11765 (REG_CLASS_CONTENTS): Idem.
11766 (REGCLASS_HAS_FP_REG): Idem.
11767 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
11768 (sh_conditional_register_usage): Idem.
11770 2013-04-21 Jeff Law <law@redhat.com>
11772 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
11773 (ssa_forward_propagate_and_combine): Use it.
11775 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
11777 * lra.c: Update the flow chart diagram.
11779 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
11781 PR rtl-optimization/56847
11782 * lra-constraints.c (process_alt_operands): Discourage alternative
11783 with non-matche doffsettable memory constraint fro memory with
11786 2013-04-19 Richard Biener <rguenther@suse.de>
11788 PR tree-optimization/56982
11789 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
11791 * gimplify.c (gimplify_call_expr): Notice special calls.
11792 (gimplify_modify_expr): Likewise.
11793 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
11794 abnormal control flow receivers.
11795 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
11796 in the same way as cfun->has_nonlocal_labels.
11797 (gimple_purge_dead_abnormal_call_edges): Likewise.
11798 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
11799 receivers start a basic-block.
11801 2013-04-19 Richard Biener <rguenther@suse.de>
11803 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
11805 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
11806 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
11807 (SLP_TREE_LOAD_PERMUTATION): Add.
11808 (vect_transform_slp_perm_load): Adjust prototype.
11809 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
11810 (vect_free_slp_instance): Likewise.
11811 (vect_create_new_slp_node): Likewise.
11812 (vect_supported_slp_permutation_p): Remove.
11813 (vect_slp_rearrange_stmts): Adjust.
11814 (vect_supported_load_permutation_p): Likewise. Inline
11815 vect_supported_slp_permutation_p here.
11816 (vect_analyze_slp_instance): Compute load permutations per
11817 slp node instead of per instance.
11818 (vect_get_slp_defs): Adjust.
11819 (vect_transform_slp_perm_load): Likewise.
11820 (vect_schedule_slp_instance): Remove redundant code.
11821 (vect_schedule_slp): Remove hack for PR56270, add it ...
11822 * tree-vect-stmts.c (vectorizable_load): ... here, do not
11823 CSE loads for SLP. Adjust.
11825 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
11827 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
11828 spelling in two comments.
11830 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
11833 * config/arm/arm.c (load_multiple_sequence): Require SP
11834 as base register for loads if SP is in the register list.
11836 2013-04-19 Martin Jambor <mjambor@suse.cz>
11838 PR tree-optimization/56718
11839 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11840 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11841 and made public. Adjusted all callers.
11842 (ipa_intraprocedural_devirtualization): New function.
11843 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11844 (ipa_intraprocedural_devirtualization): Likewise.
11845 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11847 2013-04-19 Richard Biener <rguenther@suse.de>
11849 PR tree-optimization/57000
11850 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11852 2013-04-19 Terry Guo <terry.guo@arm.com>
11854 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
11856 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
11857 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
11858 (cortex_m4_fmacs): Use new reservations.
11859 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
11861 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
11863 PR rtl-optimization/56999
11864 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
11866 (lra_coalesce): Remove split_origin_bitmap and related code.
11867 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
11868 ranges if necessary.
11870 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
11872 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
11874 (ix86_expand_call): Remove clobbered_registers array and use
11875 x86_64_ms_sysv_extra_clobbered_registers instead.
11876 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
11878 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
11880 * config/i386/i386.md (*call_rex64_ms_sysv): Use
11881 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
11882 (*call_value_rex64_ms_sysv): Ditto.
11884 2013-04-18 Cary Coutant <ccoutant@google.com>
11886 * dwarf2out.c (output_pubnames): Check die_perennial_p of
11887 parent instead of die_mark.
11889 2013-04-18 Diego Novillo <dnovillo@google.com>
11891 * gimple.c (create_gimple_tmp): New.
11892 (get_expr_type): New.
11893 (build_assign): New.
11894 (build_type_cast): New.
11895 * gimple.h (enum ssa_mode): Define.
11896 (gimple_seq_set_location): New.
11897 * asan.c (build_check_stmt): Change some gimple_build_* calls
11898 to use build_assign and build_type_cast.
11900 2013-04-18 Richard Biener <rguenther@suse.de>
11902 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
11903 handle negative step. Remove redundant checks.
11904 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
11905 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
11906 for negative step and grouped loads fail to vectorize.
11908 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
11910 * emit-rtl.c (reset_insn_used_flags): New function.
11911 (reset_all_used_flags): Use it.
11912 (verify_insn_sharing): New function.
11913 (verify_rtl_sharing): Fix verification for SEQUENCEs.
11915 2013-04-18 Jakub Jelinek <jakub@redhat.com>
11917 PR tree-optimization/56984
11918 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
11919 and (x >> M) >= N don't register any assertion if N << M is the
11922 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
11924 * lower-subreg.c (resolve_simple_move): If called self-recursive,
11925 do not delete_insn insns that have not yet been emitted, only
11926 unlink them with remove_insn.
11927 * df-scan.c (df_insn_delete): Revert r197492.
11929 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
11931 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
11932 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
11934 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
11936 * config/arm/arm.md (movsicc_insn): Convert define_insn into
11937 define_insn_and_split.
11938 (and_scc,ior_scc,negscc): Likewise.
11939 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
11941 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
11943 * config/arm/arm.c (use_return_insn): Return 0 for targets that
11944 can benefit from using a sequence of LDRD instructions in epilogue
11945 instead of a single LDM instruction.
11947 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
11950 * doc/extend.texi: Fix typo.
11952 2013-04-17 Richard Biener <rguenther@suse.de>
11954 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
11955 (vect_build_slp_tree): ... here.
11956 (vect_build_slp_tree_1): Compute which stmts of the SLP group
11957 match. Remove special-casing of mismatched complex loads.
11958 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
11959 re-try the match with swapped commutative operands.
11960 (vect_supported_load_permutation_p): Remove special-casing of
11961 mismatched complex loads.
11962 (vect_analyze_slp_instance): Adjust.
11964 2013-04-17 Richard Biener <rguenther@suse.de>
11966 PR rtl-optimization/56921
11967 * cfgloop.h (struct loop): Add simple_loop_desc member.
11968 (struct niter_desc): Mark with GTY(()).
11969 (simple_loop_desc): Do not use aux field but simple_loop_desc.
11970 * loop-iv.c (get_simple_loop_desc): Likewise.
11971 (free_simple_loop_desc): Likewise.
11974 2013-04-16 Richard Biener <rguenther@suse.de>
11976 PR rtl-optimization/56921
11977 * loop-init.c (pass_rtl_move_loop_invariants): Add
11978 TODO_do_not_ggc_collect to todo_flags_finish.
11979 (pass_rtl_unswitch): Same.
11980 (pass_rtl_unroll_and_peel_loops): Same.
11981 (pass_rtl_doloop): Same.
11983 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
11985 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
11986 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
11987 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
11989 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
11990 * tree-streamer.c (record_common_node): Adjust reference in comment.
11992 2013-04-17 Terry Guo <terry.guo@arm.com>
11994 * config/arm/cortex-m4.md: Add a new bypass.
11996 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11998 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12000 (*subs_<optab><mode>_multp2): New pattern.
12001 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12002 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12004 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12006 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12007 (*subs_mul_imm_<mode>): New pattern.
12009 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
12012 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12013 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
12014 (vsx_movti_32bit): Change j->wa to O->wa.
12016 2013-04-16 Richard Biener <rguenther@suse.de>
12018 PR rtl-optimization/56921
12019 * loop-init.c (pass_rtl_move_loop_invariants): Add
12020 TODO_do_not_ggc_collect to todo_flags_finish.
12021 (pass_rtl_unswitch): Same.
12022 (pass_rtl_unroll_and_peel_loops): Same.
12023 (pass_rtl_doloop): Same.
12025 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
12027 * config/arm/arm.c (emit_multi_reg_push): New declaration
12028 for an existing function.
12029 (arm_emit_strd_push): New function.
12030 (arm_expand_prologue): Used here.
12031 (arm_emit_ldrd_pop): New function.
12032 (arm_expand_epilogue): Used here.
12033 (arm_get_frame_offsets): Update condition.
12034 (arm_emit_multi_reg_pop): Add a special case for load of a single
12035 register with writeback.
12037 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
12039 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12042 2013-04-16 Richard Biener <rguenther@suse.de>
12044 PR tree-optimization/56756
12045 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12046 (first_mem_ref_loc): New.
12047 (execute_sm): Place the load temporarily before a previous
12048 access instead of in the latch edge to ensure its SSA dependencies
12049 are defined at points dominating the load.
12051 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
12053 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12054 correct fix by moving header and footer insn to the footer of
12055 the merged basic block. Clear BB_END of the merged-away block.
12057 PR middle-end/43631
12058 * emit-rtl.c (make_note_raw): New function.
12059 (link_insn_into_chain): New static inline function.
12060 (add_insn): Use it.
12061 (add_insn_before, add_insn_after): Factor insn chain linking code...
12062 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12063 using link_insn_into_chain.
12064 (note_outside_basic_block_p): New helper function for emit_note_after
12065 and emit_note_before.
12066 (emit_note_after): Use nobb variant of add_insn_after if the note
12067 should not be contained in a basic block.
12068 (emit_note_before): Use nobb variant of add_insn_before if the note
12069 should not be contained in a basic block.
12070 (emit_note_copy): Use make_note_raw.
12071 (emit_note): Likewise.
12072 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12073 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12074 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12075 the moved barrier the tail of the basic block it follows.
12076 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12078 2013-04-15 Jakub Jelinek <jakub@redhat.com>
12080 PR tree-optimization/56962
12081 * gimple-ssa-strength-reduction.c (record_increment): Only set
12082 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12083 either rhs1 or rhs2 is equal to c->base_expr.
12085 2013-04-15 Richard Biener <rguenther@suse.de>
12087 PR tree-optimization/56933
12088 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12090 (GROUP_READ_WRITE_DEPENDENCE): Remove.
12091 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12092 * tree-vect-data-refs.c (vect_analyze_group_access): Move
12093 dependence check ...
12094 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12096 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12097 GROUP_READ_WRITE_DEPENDENCE.
12099 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12101 * emit-rtl.c (reset_all_used_flags): New function.
12102 (verify_rtl_sharing): Call reset_all_used_flags before and after
12103 performing the checks.
12105 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12107 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12108 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12109 * config/arm/constraints.md (De): New constraint.
12110 * config/arm/neon.md (anddi3_neon): Delete.
12111 (neon_vand<mode>): Expand to standard anddi3 pattern.
12112 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12113 Move earlier in the file.
12114 (neon_inv_logic_op2): Likewise.
12115 (arm_anddi_operand_neon): New predicate.
12117 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12119 * configure.ac (gcc_cv_ld_as_needed): Set
12120 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12121 Use -z ignore, -z record on *-*-solaris2*.
12122 (HAVE_LD_AS_NEEDED): Update comment.
12123 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12124 * configure: Regenerate.
12125 * config.in: Regenerate.
12126 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12127 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12128 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12129 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12130 equivalents. Fix markup.
12131 * doc/tm.texi: Regenerate.
12133 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
12135 * config/i386/i386.opt: New option mstack-protector-guard=.
12136 * config/i386/i386-opts.h: Add enum stack_protector_guard.
12137 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12138 TARGET_SSP_TLS_GUARD.
12139 * config/i386/i386.c (ix86_option_override_internal): Set
12140 ix86_stack_protector_guard.
12141 * config/i386/i386.md (stack_protect_set): Enable for
12142 TARGET_SSP_TLS_GUARD only.
12143 (stack_protect_set_<mode>): Ditto.
12144 (stack_protect_test): Ditto.
12145 (stack_protect_test_<mode>): Ditto.
12146 * doc/invoke.texi (i386 Option): Document.
12148 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
12151 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12152 (S_MODES): Set H_MODE bit.
12153 (SF_MODES): Set only S_MODE and SF_MODE bits.
12154 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12155 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12156 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12157 <MODE_FLOAT>: Likewise.
12159 2013-04-15 Joey Ye <joey.ye@arm.com>
12161 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12163 2013-04-15 Joey Ye <joey.ye@arm.com>
12165 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12167 (thumb_far_jump_used_p): Count instruction size and set
12170 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
12172 * reorg.c (fill_simple_delay_slots): Reindent block of code.
12173 * resource.c (mark_target_live_regs): Reformat conditional block.
12175 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
12177 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12178 notes, they are emitted only just before final.
12179 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12181 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
12183 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12184 * cfgrtl.c (delete_insn): Call it here instead.
12185 * lra-spills.c (lra_final_code_change): Use delete_insn.
12186 * haifa-sched.c (sched_remove_insn): Likewise.
12187 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12188 returning to the nop pool.
12189 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12190 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
12192 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
12194 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12195 * doc/tm.texi: Regenerated.
12197 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
12199 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12202 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
12204 * df-core.c (df_find_def): Compare register numbers.
12205 (df_find_use): Likewise.
12207 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
12210 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12211 lra_in_progress for return.
12213 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
12215 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12216 define_insn into define_insn_and_split and emit movsicc patterns.
12218 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
12220 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12222 2013-04-12 Richard Biener <rguenther@suse.de>
12224 * tree-pass.h (TODO_do_not_ggc_collect): New.
12225 * passes.c (execute_one_ipa_transform_pass): Honor
12226 TODO_do_not_ggc_collect.
12227 (execute_one_pass): Likewise.
12230 2013-04-10 Richard Biener <rguenther@suse.de>
12232 * passes.c (init_optimization_passes): Remove reload pass.
12233 * ira.c (do_reload): Merge into ...
12235 (rest_of_handle_reload): Remove.
12236 (pass_reload): Likewise.
12237 * config/i386/i386.c (ix86_option_override): Refer to ira instead
12238 of reload for vzeroupper pass placement.
12240 2013-04-12 Jakub Jelinek <jakub@redhat.com>
12242 PR tree-optimization/56918
12243 PR tree-optimization/56920
12244 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12245 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
12246 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
12247 use wide_mul_with_sign method.
12249 2013-04-12 Richard Biener <rguenther@suse.de>
12251 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12252 not be considered a gimple constant.
12254 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
12256 * fold-const.c (const_binop): Handle vector shifts by a scalar.
12257 (fold_binary_loc): Call const_binop also for mixed vector-scalar
12260 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
12261 Jakub Jelinek <jakub@redhat.com>
12263 * opts.c: Include diagnostic-color.h.
12264 (common_handle_option): Handle OPT_fdiagnostics_color_.
12265 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12266 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12267 (diagnostic-color.o): New.
12268 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12269 (diagnostic_color_rule): New enum.
12270 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12271 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12272 the location string.
12273 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12274 either NULL, or color kind.
12275 * diagnostic-color.c: New file.
12276 * diagnostic-color.h: New file.
12277 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12279 * doc/invoke.texi (-fdiagnostics-color): Document.
12280 * pretty-print.h (pp_show_color): Define.
12281 (struct pretty_print_info): Add show_color field.
12282 * diagnostic.c: Include diagnostic-color.h.
12283 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12284 macros. Colorize error:, warning: etc. strings and also the location
12286 (diagnostic_show_locus): Colorize the caret line.
12287 * pretty-print.c: Include diagnostic-color.h.
12288 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
12289 inside of %< %> quotes or quoted through q format modifier.
12291 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12293 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12295 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12297 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12298 code in CC_NZ mode.
12299 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12302 2013-04-11 Marek Polacek <polacek@redhat.com>
12304 PR tree-optimization/48184
12305 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12307 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
12309 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12310 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12311 (skip_simple_arithmetic): Tidy up.
12312 * tree.h (skip_simple_constant_arithmetic): Declare.
12314 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12316 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12318 2013-04-11 Richard Biener <rguenther@suse.de>
12320 * tree-vect-loop.c (get_initial_def_for_induction): Properly
12321 generate vector constants.
12323 2013-04-11 Richard Biener <rguenther@suse.de>
12325 PR tree-optimization/56878
12326 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12327 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12329 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12330 Prefer to align the DR with the most invariant base address.
12332 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
12334 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12337 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
12339 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12340 floating-point vector comparisons against 0.
12342 2013-04-11 Jakub Jelinek <jakub@redhat.com>
12344 PR tree-optimization/56899
12345 * fold-const.c (extract_muldiv_1): Apply distributive law
12346 only if TYPE_OVERFLOW_WRAPS (ctype).
12348 2013-04-11 Bin Cheng <bin.cheng@arm.com>
12351 * ira-costs.c (scan_one_insn): Check whether the source rtx of
12352 loading has side effect.
12354 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
12356 * config/sparc/sparc.c: Include tree-pass.h.
12357 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12358 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
12359 head of file. Change return type. Split off gate function.
12360 (sparc_gate_work_around_errata): New function.
12361 (pass_work_around_errata): New pass definition.
12362 (insert_pass_work_around_errata) New pass insert definition to
12363 insert pass_work_around_errata just after delayed-branch scheduling.
12364 (sparc_option_override): Insert the pass.
12365 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12367 2013-04-10 David S. Miller <davem@davemloft.net>
12369 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12370 or -mcpu=hypersparc.
12372 * target.def (cstore_mode): New hook.
12373 * target.h: Include insn-codes.h
12374 * targhooks.c: Likewise.
12375 (default_cstore_mode): New function.
12376 * targhooks.h: Declare it.
12377 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12378 * doc/tm.texi: Rebuild.
12379 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12380 target hook, rather than inspecting the insn_data.
12381 * config/sparc/sparc.c (sparc_cstore_mode): New function.
12382 (TARGET_CSTORE_MODE): Redefine.
12383 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12385 * config/sparc/predicates.md (cstore_result_operand): New special
12387 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12388 Use it for operand 0.
12389 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12390 (*snesi_special): Likewise.
12391 (*snesi_zero): Likewise.
12392 (*seqsi_zero): Likewise.
12393 (*sltu_insn): Likewise.
12394 (*sgeu_insn): Likewise.
12395 (*seqdi_special): Make operand 0 and comparison operation be of
12397 (*snedi_special): Likewise.
12398 (*snedi_special_vis3): Likewise.
12399 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12400 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12401 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12402 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12403 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12404 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12405 (*sltu_extend_sp64): Likewise.
12406 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12407 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12408 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12409 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12410 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12412 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
12414 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12415 (aarch64_start_file): Use the new function.
12417 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
12418 Jason Merrill <jason@redhat.com>
12420 * common.opt: Add -gdwarf.
12421 * opts.c (common_handle_option): Handle it.
12422 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12424 2013-04-10 Richard Biener <rguenther@suse.de>
12426 * passes.c (execute_todo): Do not call ggc_collect conditional here.
12427 (execute_one_ipa_transform_pass): But unconditionally here.
12428 (execute_one_pass): And here.
12429 (init_optimization_passes): Remove reload pass.
12430 * tree-pass.h (TODO_ggc_collect): Remove.
12431 (pass_reload): Likewise.
12432 * ira.c (do_reload): Merge into ...
12434 (rest_of_handle_reload): Remove.
12435 (pass_reload): Likewise.
12436 * config/i386/i386.c (ix86_option_override): Refer to ira instead
12437 of reload for vzeroupper pass placement.
12438 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12439 and todo_flags_finish of all passes.
12441 2013-04-10 Richard Biener <rguenther@suse.de>
12443 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12444 first_const_oprnd field, rename first_def_type to first_op_type.
12445 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12446 (vect_get_and_check_slp_defs): Always use the type of the
12447 operand. Allow mixed vect_external_def, vect_constant_def types.
12448 (vect_get_constant_vectors): Handle mixed vect_external_def,
12449 vect_constant_def types.
12451 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
12453 PR tree-optimization/55524
12454 * tree-ssa-math-opts.c
12455 (convert_mult_to_fma): Don't use an fms construct
12456 when we don't have an fms operation, but fnma, and it looks
12457 likely that we'll be able to use the latter.
12459 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
12461 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12463 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12464 inline fail caused by overwritable functions.
12466 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
12468 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12469 unnecessary bits in the constant power of two case.
12471 2013-04-10 Richard Biener <rguenther@suse.de>
12473 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12474 broken code swapping operands.
12475 (vect_build_slp_tree): Do not compute load permutations here.
12476 (vect_analyze_slp_instance): Compute load permutations here,
12477 after building the SLP tree.
12479 2013-04-09 Christian Bruel <christian.bruel@st.com>
12481 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12482 of next/prev_real_insn.
12484 2013-04-09 Jan Hubicka <jh@suse.cz>
12486 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12487 Drop aliased parameter.
12488 (function_and_variable_visibility): Do not handle alias pairs.
12489 * cgraph.c (varpool_externally_visible_p): Update prototype.
12490 * varpool.c (varpool_add_new_variable): Update.
12492 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12494 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12496 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
12498 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12500 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12502 2013-04-09 Marek Polacek <polacek@redhat.com>
12504 PR tree-optimization/48762
12505 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12507 2013-04-09 Richard Biener <rguenther@suse.de>
12509 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12511 (vect_build_slp_tree): Likewise.
12512 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12513 calculating the cost of a SLP instance.
12514 (vect_analyze_slp_instance): Use it from here, after building
12517 2013-04-09 Jakub Jelinek <jakub@redhat.com>
12519 PR middle-end/56883
12520 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12521 expand_omp_for_static_chunk): Use simple_p = true in
12522 force_gimple_operand_gsi calls when assigning to addressable decls.
12524 2013-04-09 Jeff Law <law@redhat.com>
12526 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12527 when the boolean was created by converting a wider object which
12528 had a boolean range.
12530 2013-04-09 Richard Biener <rguenther@suse.de>
12532 * tree-vectorizer.h (slp_void_p): Remove.
12533 (slp_tree): Typedef before _slp_tree declaration.
12534 (struct _slp_tree): Use a vector of slp_tree as children.
12535 (vect_get_place_in_interleaving_chain): Remove.
12536 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12538 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12540 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12541 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12542 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12543 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12544 Use slp_node instead of slp_void_p and adjust.
12546 2013-04-09 Richard Biener <rguenther@suse.de>
12548 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12549 work that is not necessary.
12551 2013-04-09 Jakub Jelinek <jakub@redhat.com>
12553 PR tree-optimization/56854
12554 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12555 forward into clobber stmts if it would change MEM_REF lhs into
12558 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
12560 * tree.c (type_hash_lookup, type_hash_add): Make static.
12561 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12563 2013-04-09 Richard Biener <rguenther@suse.de>
12565 * tree.h (unsave_expr_now): Remove.
12566 * tree-inline.c (mark_local_for_remap_r): Remove.
12567 (unsave_expr_1): Likewise.
12568 (unsave_r): Likewise.
12569 (unsave_expr_now): Likewise.
12570 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12571 (propagate_tree_value): Likewise.
12573 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
12575 * doc/rtl.texi (sequence): Rewrite documentation to match the
12576 current use of SEQUENCE rtl objects.
12577 * rtl.def (SEQUENCE): Likewise.
12579 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12580 Update documentation.
12581 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12582 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12584 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12586 2013-04-08 Teresa Johnson <tejohnson@google.com>
12588 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12589 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12591 (estimate_edge_size_and_time): Add comment to suggest using rounding
12593 (estimate_node_size_and_time): Ditto.
12594 (remap_edge_change_prob): Use helper rounding divide methods.
12595 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
12596 (gimple_mod_pow2_value_transform): Ditto.
12597 (gimple_mod_subtract_transform): Ditto.
12598 (gimple_ic_transform): Ditto.
12599 (gimple_stringops_transform): Ditto.
12600 * stmt.c (conditional_probability): Ditto.
12601 (emit_case_dispatch_table): Ditto.
12602 * lto-cgraph.c (merge_profile_summaries): Ditto.
12603 * tree-optimize.c (execute_fixup_cfg): Ditto.
12604 * cfgcleanup.c (try_forward_edges): Ditto.
12605 * cfgloopmanip.c (scale_loop_profile): Ditto.
12607 (duplicate_loop_to_header_edge): Ditto.
12608 (lv_adjust_loop_entry_edge): Ditto.
12609 * tree-vect-loop.c (vect_transform_loop): Ditto.
12610 * profile.c (compute_branch_probabilities): Ditto.
12611 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
12612 * lto-streamer-in.c (input_cfg): Ditto.
12613 * gimple-streamer-in.c (input_bb): Ditto.
12614 * ipa-cp.c (update_profiling_info): Ditto.
12615 (update_specialized_profile): Ditto.
12616 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
12617 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
12619 * sched-rgn.c (compute_dom_prob_ps): Ditto.
12620 (compute_trg_info): Ditto.
12621 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
12622 (purge_dead_edges): Ditto.
12623 * loop-unswitch.c (unswitch_loop): Ditto.
12624 * cgraphclones.c (cgraph_clone_edge): Ditto.
12625 (cgraph_clone_node): Ditto.
12626 * tree-inline.c (copy_bb): Ditto.
12627 (copy_edges_for_bb): Ditto.
12628 (initialize_cfun): Ditto.
12629 (copy_cfg_body): Ditto.
12630 (expand_call_inline): Ditto.
12632 2013-04-08 Kai Tietz <ktietz@redhat.com>
12634 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
12635 TARGET_CYGWIN64 by TARGET_64BIT.
12637 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
12639 * config/epiphany/epiphany.md (GPR_1): New constant.
12640 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
12641 * config/epiphany/epiphany.c (gen_compare_reg):
12642 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
12643 is already in place.
12644 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
12645 Don't require being called during rtl expansion; If y operlaps r0,
12647 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
12648 (epiphany_expand_epilogue): Likewise.
12650 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
12651 Don't use CC_FPmode for ORDERED / UNORDERED.
12652 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
12654 * config/epiphany/constraints.md (CnL): New constraint.
12655 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
12656 * config/epiphany/predicates.md (add_operand): Allow 1024.
12658 * config/epiphany/epiphany.md (logical_op): New code iterator.
12659 (op_mnc): New code attribute.
12660 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
12661 (mov_f+1, mov_f+2): New peephole2 patterns.
12663 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
12664 (cstoresi4): Also allow re-use of zero result when doing a NE
12665 comparison to a non-zero operand.
12666 Use (clobber (scratch)) for first insn if the gpr output is not needed.
12668 * config/epiphany/epiphany.md (<insn_opname>v2si3):
12669 Use gen_addsi3_i / gen_subsi3_i.
12671 2013-04-08 Jakub Jelinek <jakub@redhat.com>
12675 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
12676 contain anything but clobbers, at most one __builtin_stack_restore,
12677 optionally debug stmts and final resx, and if it has at least one
12678 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
12679 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
12680 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
12681 which isn't defaut definition, remove them.
12682 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
12683 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
12684 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
12685 with MEM_REF LHS with SSA_NAME address.
12687 2013-04-08 Jeff Law <law@redhat.com>
12689 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
12691 2013-04-08 Richard Biener <rguenther@suse.de>
12693 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
12695 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
12696 determined vector type.
12697 (vect_analyze_data_refs): Likewise.
12698 (vect_get_new_vect_var): Adjust.
12699 (vect_create_destination_var): Preserve SSA name versions.
12700 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
12701 not dump anything here.
12703 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
12705 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
12706 Add member lr_slot_known.
12707 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
12709 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12710 Remove code that sets lr_slot_offset according to what a previous
12711 version of epiphany_emit_save_restore used to do.
12712 (epiphany_emit_save_restore): When doing an lr save or restore,
12713 set/verify lr_slot_known and lr_slot_offset.
12715 2013-04-08 Xinyu Qi <xyqi@marvell.com>
12718 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
12721 2013-04-08 Richard Biener <rguenther@suse.de>
12723 * alias.c (find_base_term): Fix thinko in previous change.
12725 2013-04-08 Jakub Jelinek <jakub@redhat.com>
12727 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
12728 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
12729 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
12730 if possible to compute val.
12731 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
12732 For QImode integers don't require anything about precision. Use
12733 const_with_all_bytes_same to find out if the constant doesn't have
12734 repeated bytes in it.
12736 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12738 * config/s390/s390.c (s390_expand_insv): Only accept insertions
12741 2013-04-08 Marek Polacek <polacek@redhat.com>
12743 PR rtl-optimization/48182
12744 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
12747 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
12750 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
12751 nuses, make sure we have a label.
12753 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12756 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
12757 (rs6000_emit_swdiv_low_precision): Remove.
12758 (rs6000_emit_swdiv): Rewrite to handle between one and four
12759 iterations of Newton-Raphson generally; modify required number of
12760 iterations for some cases.
12761 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
12763 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
12765 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
12766 set-but-unused variable.
12768 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
12769 basic blocks of released function bodies garbage-collectable.
12771 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
12772 (struct rtl_opt_pass): Add TODO_df_finish.
12774 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
12776 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12778 * config/arm/constraints.md (q): New constraint.
12779 * config/arm/ldrdstrd.md: New file.
12780 * config/arm/arm.md (ldrdstrd.md) New include.
12781 (arm_movdi): Use "q" instead of "r" constraint
12782 for double-word memory access.
12783 (movdf_soft_insn): Likewise.
12784 * config/arm/vfp.md (movdi_vfp): Likewise.
12785 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
12786 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
12787 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
12788 (mem_ok_for_ldrd_strd): Likewise.
12789 (output_move_double): Update assertion.
12791 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12793 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
12795 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12797 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
12798 define_insn_and_split.
12799 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
12801 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12803 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
12804 define_insn_and_split.
12805 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
12806 (shiftsi3_compare): New pattern.
12807 (rrx): New pattern.
12808 * config/arm/unspecs.md (UNSPEC_RRX): New.
12810 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12812 * config/arm/arm.md (negdi_extendsidi): New pattern.
12813 (negdi_zero_extendsidi): Likewise.
12815 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12817 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
12818 define_insn_and_split.
12819 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
12820 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
12822 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12824 * config/arm/arm.md (arm_subdi3): Convert define_insn into
12825 define_insn_and_split.
12826 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
12827 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
12829 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12831 * config/arm/arm.md (subsi3_carryin): New pattern.
12832 (subsi3_carryin_const): Likewise.
12833 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
12834 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12836 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12838 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12840 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
12842 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12843 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12845 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12847 * config/arm/arm.c (arm_expand_builtin): Change fcode
12848 type to unsigned int.
12850 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12852 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
12854 2013-04-04 Ian Lance Taylor <iant@google.com>
12856 * doc/standards.texi (Standards): The Go frontend supports the Go 1
12859 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
12861 PR middle-end/56729
12862 * df-scan.c (df_insn_delete): Disable failing assert.
12864 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12866 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
12867 New function prototype.
12868 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
12869 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
12870 (arm_builtin_vectorized_function): New function.
12872 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12874 * config/arm/arm_neon_builtins.def: New file.
12875 * config/arm/arm.c (neon_builtin_data): Move contents to
12876 arm_neon_builtins.def.
12877 (enum arm_builtins): Include neon builtin definitions.
12878 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
12879 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
12881 2013-04-04 Marek Polacek <polacek@redhat.com>
12883 PR tree-optimization/48186
12884 * predict.c (maybe_hot_frequency_p): Return false if
12885 HOT_BB_FREQUENCY_FRACTION is 0.
12886 (cgraph_maybe_hot_edge_p): Likewise.
12888 2013-04-04 Richard Biener <rguenther@suse.de>
12890 PR tree-optimization/56826
12891 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
12894 2013-04-04 Richard Biener <rguenther@suse.de>
12896 PR tree-optimization/56213
12897 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
12898 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
12900 2013-04-04 Richard Biener <rguenther@suse.de>
12902 PR tree-optimization/56837
12903 * tree-loop-distribution.c (classify_partition): For non-zero
12904 values require that the value has the same precision as its
12905 mode to be useful as memset value.
12907 2013-04-03 Nick Clifton <nickc@redhat.com>
12909 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
12910 (fmssf4): Use fmsf.s on E3V5 architectures.
12911 (fnmasf4): Use fnmaf.s on E3V5 architectures.
12912 (fnmssf4): Use fnmsf.s on E3V5 architectures.
12914 2013-04-03 Jeff Law <law@redhat.com>
12916 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
12917 (lra-eliminations.o): Likewise.
12919 2013-04-03 Teresa Johnson <tejohnson@google.com>
12921 * gcov-io.c (compute_working_sets): Moved most of body of old
12922 compute_working_sets here from profile.c.
12923 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
12924 (gcov_working_set_t): Moved typedef here from basic-block.h
12925 (compute_working_set): Declare.
12926 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
12927 (get_working_sets): Renamed from compute_working_set,
12928 replace most of body with call to new compute_working_sets.
12929 (get_exec_counts): Replace call to compute_working_sets
12930 to get_working_sets.
12931 * profile.h (get_working_sets): Renamed from compute_working_set.
12932 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
12933 to get_working_sets.
12934 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
12935 * gcov-dump.c (dump_working_sets): New function.
12937 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
12939 * hwint.c (sext_hwi, zext_hwi): New functions.
12940 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
12941 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
12942 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
12943 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
12944 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
12945 (sext_hwi, zext_hwi): New functions.
12947 2013-04-03 Jeff Law <law@redhat.com>
12949 PR tree-optimization/56799
12950 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
12951 back test for widening conversion erroneously dropped in prior change.
12953 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12956 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
12957 instead of next_real_insn.
12959 2013-04-03 Marek Polacek <polacek@redhat.com>
12962 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
12964 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12967 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
12969 (thumb1_output_casesi): Likewise.
12970 (thumb2_output_casesi): Likewise.
12972 2013-04-03 Richard Biener <rguenther@suse.de>
12974 PR tree-optimization/56817
12975 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
12977 (tree_unroll_loops_completely_1): ... new function to manually
12978 walk the loop tree, properly defering outer loops of unrolled
12979 loops to later iterations.
12981 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12983 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
12984 (vectorizable_load): Likewise.
12985 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
12986 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
12988 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
12990 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
12993 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
12995 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
12997 2013-04-03 Bin Cheng <bin.cheng@arm.com>
12999 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13001 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13003 PR tree-optimization/56790
13004 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13007 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13009 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13011 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13012 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13015 2013-04-03 Jakub Jelinek <jakub@redhat.com>
13018 * tree.h (force_folding_builtin_constant_p): New decl.
13019 * builtins.c (force_folding_builtin_constant_p): New variable.
13020 (fold_builtin_constant_p): Fold immediately also if
13021 force_folding_builtin_constant_p.
13023 2013-04-03 Richard Biener <rguenther@suse.de>
13025 PR tree-optimization/56812
13026 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13027 DRs of the same interleaving chain are independent.
13029 2013-04-02 Jason Merrill <jason@redhat.com>
13031 * gdbinit.in (pbb): Use debug fn.
13033 2013-04-02 Lawrence Crowl <crowl@google.com>
13035 * sese.h (struct ivtype_map_elt_s): Remove unused.
13036 (extern debug_ivtype_map): Remove unused.
13037 (extern eq_ivtype_map_elts): Remove unused.
13038 * sese.c (debug_ivtype_map): Removed unused.
13039 (debug_ivtype_map_1): Removed unused.
13040 (debug_ivtype_elt): Remove unused.
13041 (eq_ivtype_map_elts): Remove unused.
13044 2013-04-02 Kai Tietz <ktietz@redhat.com>
13047 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13048 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13049 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13051 (legitimize_pe_coff_symbol): Likewise.
13052 (is_imported_p): New helper-function.
13053 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13054 for Windows x64 targets.
13055 (ix86_expand_prologue): Optimize for pe-coff targets.
13056 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13057 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13058 medium/large code-model.
13059 (legitimize_pic_address): Likewise.
13060 (legitimize_tls_address): Likewise.
13061 (ix86_expand_call): Likewise.
13062 (x86_output_mi_thunk): Likewise.
13063 (get_dllimport_decl): Add new beimport argument.
13064 (construct_plt_address): Don't assert for x64 pe-coff targets.
13065 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13067 (SYMBOL_FLAG_STUBVAR): New macro.
13068 (SYMBOL_REF_STUBVAR_P): Likewise.
13069 * config/i386/winnt.c (stub_list): New structure.
13070 (stub_head): New local variable.
13071 (i386_pe_record_stub): New function.
13072 (i386_pe_file_end): Emit refptr-stubs.
13074 2013-04-02 Jakub Jelinek <jakub@redhat.com>
13076 PR rtl-optimization/56745
13077 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13078 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13081 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13082 and both of them are MEM_REFs, just compare first argument for
13083 equality and attempt to deal even with differing offsets.
13086 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13087 of gimple_clobber_p to be MEM_REF.
13088 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13089 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
13090 after gimplification.
13091 * asan.c (get_mem_ref_of_assignment): Don't instrument
13092 gimple_clobber_p stmts.
13093 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13094 gimple_clobber_p stmt if they have MEM_REF lhs and
13095 are dead because of another gimple_clobber_p stmt.
13096 * tree-ssa-live.c (clear_unused_block_pointer): Treat
13097 gimple_clobber_p stmts like debug stmts.
13098 (remove_unused_locals): Remove clobbers with MEM_REF lhs
13099 that refer to unused VAR_DECLs or uninitialized values.
13100 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13101 gimple_clobber_p stmts if they refer to removed parameters.
13102 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13105 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
13107 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13108 using SWI48 mode attribute.
13110 2013-04-02 Wei Mi <wmi@google.com>
13112 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13113 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13114 *<rotate_insn><mode>3_mask in i386.md.
13116 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
13118 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13120 2013-04-02 Richard Biener <rguenther@suse.de>
13122 PR tree-optimization/56778
13123 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13124 Runtime alias tests are not supported for gather loads.
13125 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13126 stmts referenced from SSA operands before updating SSA form.
13128 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
13129 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13131 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13132 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13133 * config/arm/cortex-a53.md: New file.
13134 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13135 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13136 * config/arm/arm.c (arm_issue_rate): Likewise.
13137 * config/arm/arm-tune.md: Regenerate
13138 * config/arm/arm-tables.opt: Regenerate.
13139 * config/arm/arm-cores.def: Add cortex-a53.
13141 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
13143 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13146 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
13148 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13149 scalar load/store operations using B/H registers.
13150 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13152 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
13154 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13156 * config/aarch64/aarch64.c
13157 (aarch64_simd_scalar_immediate_valid_for_move): New.
13158 * config/aarch64/aarch64-protos.h
13159 (aarch64_simd_scalar_immediate_valid_for_move): New.
13160 * config/aarch64/constraints.md (Dh, Dq): New.
13161 * config/aarch64/iterators.md (hq): New.
13163 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
13165 * reorg.c (get_branch_condition): Deal with conditional returns.
13166 (fill_simple_delay_slots): Remove dead code dealing with jumps.
13168 2013-04-01 Wei Mi <wmi@google.com>
13170 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13171 Truncate operand 2 using %b asm operand modifier.
13172 (*<shift_insn><mode>3_mask): Ditto.
13173 (*<rotate_insn><mode>3_mask): Ditto.
13175 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
13177 PR middle-end/56798
13178 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13180 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
13182 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13184 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13186 2013-03-30 Lawrence Crowl <crowl@google.com>
13188 * dse.c (clear_alias_sets): Remove never set.
13189 (disqualified_clear_alias_sets): Remove never set.
13190 (clear_alias_mode_pool): Remove never set.
13191 (dse_step0): Remove condition that is never true.
13192 (canon_address): Remove condition that is never true.
13193 (dse_step7): Remove condition that is never true.
13194 (rest_of_handle_dse): Remove condition that is never true.
13195 (rest_of_handle_dse::did_global): Remove never read from above.
13196 (dse_step2_spill): Remove never called from above.
13197 (dse_step5_spill): Remove never called from above.
13199 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
13201 * doc/md.texi (Standard Names) <casesi>: Update documentation for
13202 JUMP_TABLE_DATA changes.
13203 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13204 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13205 (Insns) <jump_table_data>: New entry.
13206 * doc/tm.texi: Regenerate.
13208 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13210 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13211 for table jump at the end of a basic block using tablejump_p.
13212 * targhooks.c (default_invalid_within_doloop): Likewise.
13213 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13214 target hook implementation that is identical to the default hook.
13215 (rs6000_invalid_within_doloop): Remove.
13217 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13218 unused variable from tablejump_p call.
13220 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13221 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13222 (INSN_DELETED_P): Likewise.
13223 (emit_jump_table_data): New prototype.
13224 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13225 after 4th as unused.
13226 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13227 * sched-vis.c (print_insn): Likewise.
13228 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13229 insn for compatibility with back ends that use next_active_insn to
13230 identify jump table data.
13231 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13232 (remove_insn): Likewise.
13233 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13235 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13236 (emit_jump_table_data): New function.
13238 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13239 basic block, a JUMP_TABLE_DATA never is.
13240 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13241 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13242 off from code handling real insns.
13243 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13244 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13245 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13246 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13247 is not a NONDEBUG_INSN_P.
13248 * ira-costs.c (scan_one_insn): Likewise.
13249 * jump.c (mark_all_labels): Likewise.
13250 (mark_jump_label_1): Likewise.
13251 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13252 * lra.c (get_insn_freq): Expect all insns reaching here to be in
13254 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13255 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13256 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13257 JUMP_TABLE_DATA_P insns.
13258 (calculate_elim_costs_all_insns): Likewise.
13259 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13260 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13261 (delete_output_reload): Code style fixups.
13262 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13263 insn flags on this non-insn.
13264 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13265 as scheduling barriers, for pre-change compatibility.
13266 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13267 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13269 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13270 redundant JUMP_TABLE_DATA_P test.
13271 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13272 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13273 (frv_for_each_packet): Likewise.
13274 * config/i386/i386.c (min_insn_size): Likewise.
13275 (ix86_avoid_jump_mispredicts): Likewise.
13276 * config/m32r/m32r.c (m32r_is_insn): Likewise.
13277 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13278 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13279 (mips16_insn_length): Robustify.
13280 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13281 (mips16_split_long_branches): Likewise.
13282 * config/pa/pa.c (pa_combine_instructions): Likewise.
13283 * config/rs6000/rs6000.c (get_next_active_insn): Treat
13284 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13285 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13286 as contributing to pool range lengths.
13287 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13288 Remove redundant JUMP_TABLE_DATA_P test.
13289 (sh_loop_align): Likewise.
13290 (split_branches): Likewise.
13291 (sh_insn_length_adjustment): Likewise.
13292 * config/spu/spu.c (get_branch_target): Likewise.
13294 2013-03-29 Jan Hubicka <jh@suse.cz>
13296 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13297 gcov streaming; stream hot bb threshold to ltrans.
13298 * predict.c (get_hot_bb_threshold): Break out from ....
13299 (maybe_hot_count_p): ... here.
13300 (set_hot_bb_threshold): New function.
13301 * lto-section-in.c (lto_section_name): Add profile.
13302 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13303 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13304 and data-streamer.h
13305 (histogram_entry): New structure.
13306 (histogram, histogram_pool): New global vars.
13307 (histogram_hash): New structure.
13308 (histogram_hash::hash): New method.
13309 (histogram_hash::equal): Likewise.
13310 (account_time_size): New function.
13311 (cmp_counts): New function.
13312 (dump_histogram): New function.
13313 (ipa_profile_generate_summary): New function.
13314 (ipa_profile_write_summary): New function.
13315 (ipa_profile_read_summary): New function.
13316 (ipa_profile): Decide on threshold.
13317 (pass_ipa_profile): Add ipa_profile_write_summary and
13318 ipa_profile_read_summary.
13319 * Makefile.in (ipa.o): Update dependencies.
13320 * lto-streamer.h (LTO_section_ipa_profile): New section.
13322 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
13324 * tree.h (VAR_P): New.
13326 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
13329 * doc/invoke.texi ([-fwhole-program]): Fix typo.
13331 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
13333 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13334 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13335 (control_flow_insn_p): Likewise.
13336 * cfgrtl.c (duplicate_insn_chain): Likewise.
13337 * final.c (get_attr_length_1): Likewise.
13338 (shorten_branches): Likewise.
13339 (final_scan_insn): Likewise.
13340 * function.c (instantiate_virtual_regs): Likewise.
13341 * gcse.c (insert_insn_end_basic_block): Likewise.
13342 * ira-costs.c (scan_one_insn): Likewise.
13343 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13344 * lra.c (check_rtl): Likewise.
13345 * reload1.c (elimination_costs_in_insn): Likewise.
13346 * reorg.c (follow_jumps): Likewise.
13348 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13349 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13350 (thumb_far_jump_used_p): Likewise.
13351 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13352 (workaround_speculation): Likewise.
13353 (add_sched_insns_for_speculation): Likewise.
13354 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13355 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13356 (frv_for_each_packet): Likewise.
13357 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13358 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13359 (final_emit_insn_group_barriers): Likewise.
13360 * config/m32r/m32r.c (m32r_is_insn): Likewise.
13361 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13362 (mips16_insn_length): Likewise.
13363 * config/pa/pa.c (pa_reorg): Likewise.
13364 (pa_combine_instructions): Likewise.
13365 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13366 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13367 (sh_reorg): Likewise.
13368 (split_branches): Likewise.
13369 * config/spu/spu.c (get_branch_target): Likewise.
13371 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13374 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
13376 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13377 Fix declaration name.
13379 2013-03-28 Lawrence Crowl <crowl@google.com>
13381 * graphds.h (struct graph.indicies): Remove unused.
13382 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13383 (SCOP_ORIGINAL_PDDRS): Remove unused.
13384 * sese.h (extern insert_loop_close_phis): Removed unused.
13385 (extern insert_guard_phis): Removed unused.
13386 (extern ivtype_map_elt_info): Removed unused.
13387 (new_ivtype_map_elt): Removed unused.
13388 * sese.c (ivtype_map_elt_info): Removed unused.
13390 2013-03-28 Lawrence Crowl <crowl@google.com>
13392 * Makefile.in: Add several missing include dependences.
13394 (test-dump.o): New. This object is not added to any executable,
13395 but is present for ad-hoc testing.
13397 (debug (const bitmap_head_def &)): New.
13398 (debug (const bitmap_head_def *)): New.
13400 (extern debug (const bitmap_head_def &)): New.
13401 (extern debug (const bitmap_head_def *)): New.
13403 (debug (edge_def &)): New.
13404 (debug (edge_def *)): New.
13406 (debug (basic_block_def &)): New.
13407 (debug (basic_block_def *)): New.
13409 (dump_node (const_tree, int, FILE *)): Correct source file.
13411 (debug (die_struct &)): New.
13412 (debug (die_struct *)): New.
13414 (extern debug (die_struct &)): New.
13415 (extern debug (die_struct *)): New.
13416 * gimple-pretty-print.c
13417 (debug (gimple_statement_d &)): New.
13418 (debug (gimple_statement_d *)): New.
13419 * gimple-pretty-print.h
13420 (extern debug (gimple_statement_d &)): New.
13421 (extern debug (gimple_statement_d *)): New.
13423 (debug (ira_allocno_copy &)): New.
13424 (debug (ira_allocno_copy *)): New.
13425 (debug (ira_allocno &)): New.
13426 (debug (ira_allocno *)): New.
13428 (extern debug (ira_allocno_copy &)): New.
13429 (extern debug (ira_allocno_copy *)): New.
13430 (extern debug (ira_allocno &)): New.
13431 (extern debug (ira_allocno *)): New.
13433 (debug (live_range &)): New.
13434 (debug (live_range *)): New.
13436 (debug (lra_live_range &)): New.
13437 (debug (lra_live_range *)): New.
13439 (debug (lra_live_range &)): New.
13440 (debug (lra_live_range *)): New.
13442 (debug (omega_pb_d &)): New.
13443 (debug (omega_pb_d *)): New.
13445 (extern debug (omega_pb_d &)): New.
13446 (extern debug (omega_pb_d *)): New.
13448 (debug (const rtx_def &)): New.
13449 (debug (const rtx_def *)): New.
13451 (debug_tree (tree): Move within file.
13452 (debug_raw (const tree_node &)): New.
13453 (debug_raw (const tree_node *)): New.
13454 (dump_tree_via_hooks (const tree_node *, int)): New.
13455 (debug (const tree_node &)): New.
13456 (debug (const tree_node *)): New.
13457 (debug_verbose (const tree_node &)): New.
13458 (debug_verbose (const tree_node *)): New.
13459 (debug_head (const tree_node &)): New.
13460 (debug_head (const tree_node *)): New.
13461 (debug_body (const tree_node &)): New.
13462 (debug_body (const tree_node *)): New.
13463 (debug_vec_tree (tree): Move and reimplement in terms of dump.
13464 (debug (vec<tree, va_gc> &)): New.
13465 (debug (vec<tree, va_gc> *)): New.
13467 (extern debug (const rtx_def &)): New.
13468 (extern debug (const rtx_def *)): New.
13470 (debug_raw (simple_bitmap_def &)): New.
13471 (debug_raw (simple_bitmap_def *)): New.
13472 (debug (simple_bitmap_def &)): New.
13473 (debug (simple_bitmap_def *)): New.
13475 (extern debug (simple_bitmap_def &)): New.
13476 (extern debug (simple_bitmap_def *)): New.
13477 (extern debug_raw (simple_bitmap_def &)): New.
13478 (extern debug_raw (simple_bitmap_def *)): New.
13480 (debug (vinsn_def &)): New.
13481 (debug (vinsn_def *)): New.
13482 (debug_verbose (vinsn_def &)): New.
13483 (debug_verbose (vinsn_def *)): New.
13484 (debug (expr_def &)): New.
13485 (debug (expr_def *)): New.
13486 (debug_verbose (expr_def &)): New.
13487 (debug_verbose (expr_def *)): New.
13488 (debug (vec<rtx> &)): New.
13489 (debug (vec<rtx> *)): New.
13491 (extern debug (vinsn_def &)): New.
13492 (extern debug (vinsn_def *)): New.
13493 (extern debug_verbose (vinsn_def &)): New.
13494 (extern debug_verbose (vinsn_def *)): New.
13495 (extern debug (expr_def &)): New.
13496 (extern debug (expr_def *)): New.
13497 (extern debug_verbose (expr_def &)): New.
13498 (extern debug_verbose (expr_def *)): New.
13499 (extern debug (vec<rtx> &)): New.
13500 (extern debug (vec<rtx> *)): New.
13502 (_list_iter_cond_expr): Make inline instead of static.
13504 (debug (sreal &)): New.
13505 (debug (sreal *)): New.
13507 (extern debug (sreal &)): New.
13508 (extern debug (sreal *)): New.
13510 (extern debug_raw (const tree_node &)): New.
13511 (extern debug_raw (const tree_node *)): New.
13512 (extern debug (const tree_node &)): New.
13513 (extern debug (const tree_node *)): New.
13514 (extern debug_verbose (const tree_node &)): New.
13515 (extern debug_verbose (const tree_node *)): New.
13516 (extern debug_head (const tree_node &)): New.
13517 (extern debug_head (const tree_node *)): New.
13518 (extern debug_body (const tree_node &)): New.
13519 (extern debug_body (const tree_node *)): New.
13520 (extern debug (vec<tree, va_gc> &)): New.
13521 (extern debug (vec<tree, va_gc> *)): New.
13523 (debug (struct loop &)): New.
13524 (debug (struct loop *)): New.
13525 (debug_verbose (struct loop &)): New.
13526 (debug_verbose (struct loop *)): New.
13527 * tree-dump.c: Add header dependence.
13529 (extern debug (struct loop &)): New.
13530 (extern debug (struct loop *)): New.
13531 (extern debug_verbose (struct loop &)): New.
13532 (extern debug_verbose (struct loop *)): New.
13534 (debug (data_reference &)): New.
13535 (debug (data_reference *)): New.
13536 (debug (vec<data_reference_p> &)): New.
13537 (debug (vec<data_reference_p> *)): New.
13538 (debug (vec<ddr_p> &)): New.
13539 (debug (vec<ddr_p> *)): New.
13541 (extern debug (data_reference &)): New.
13542 (extern debug (data_reference *)): New.
13543 (extern debug (vec<data_reference_p> &)): New.
13544 (extern debug (vec<data_reference_p> *)): New.
13545 (extern debug (vec<ddr_p> &)): New.
13546 (extern debug (vec<ddr_p> *)): New.
13548 (debug (pt_solution &)): New.
13549 (debug (pt_solution *)): New.
13551 (extern debug (pt_solution &)): New.
13552 (extern debug (pt_solution *)): New.
13554 (debug (_var_map &)): New.
13555 (debug (_var_map *)): New.
13556 (debug (tree_live_info_d &)): New.
13557 (debug (tree_live_info_d *)): New.
13559 (extern debug (_var_map &)): New.
13560 (extern debug (_var_map *)): New.
13561 (extern debug (tree_live_info_d &)): New.
13562 (extern debug (tree_live_info_d *)): New.
13564 2013-03-28 Jan Hubicka <jh@suse.cz>
13566 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13568 2013-03-28 Ian Bolton <ian.bolton@arm.com>
13570 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13571 record only when desired or required.
13573 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
13575 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13576 *vec_extractv2di_1_rex64. Use x64 isa attribute.
13578 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13580 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13581 (*andsi3_compare0_uxtw): New pattern.
13582 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13583 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13585 2013-03-28 Jan Hubicka <jh@suse.cz>
13587 * data-streamer-in.c (streamer_read_gcov_count): New function.
13588 * gimple-streamer-out.c: Include value-prof.h.
13589 (output_gimple_stmt): Output histogram.
13590 (output_bb): Use streamer_write_gcov_count.
13591 * value-prof.c: Include data-streamer.h
13592 (dump_histogram_value): Add HIST_TYPE_MAX.
13593 (stream_out_histogram_value): New function.
13594 (stream_in_histogram_value): New function.
13595 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
13596 (stream_out_histogram_value, stream_in_histogram_value): Declare.
13597 * data-streamer-out.c (streamer_write_gcov_count): New function.
13598 (streamer_write_gcov_count_stream): New function.
13599 * lto-cgraph.c (lto_output_edge): Update counter streaming.
13600 (lto_output_node): Likewise.
13601 (input_node, input_edge): Likewise.
13602 * lto-streamer-out.c (output_cfg): Update streaming.
13603 * lto-streamer-in.c (input_cfg): Likewise.
13604 * data-streamer.h (streamer_write_gcov_count,
13605 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
13606 * gimple-streamer-in.c: Include value-prof.h
13607 (input_gimple_stmt): Input histograms.
13608 (input_bb): Update profile streaming.
13610 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
13612 * genmodes.c (emit_max_int): New function.
13613 (emit_insn_modes_h): Added call to emit_max_function.
13614 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
13616 * machmode.def: Fixed comment.
13618 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
13620 * combine.c (try_combine): Removed useless assert.
13621 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
13623 2013-03-28 Marek Polacek <polacek@redhat.com>
13624 Richard Biener <rguenther@suse.de>
13626 PR tree-optimization/56695
13627 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
13628 build signed result of a vector comparison.
13629 * tree-cfg.c (verify_gimple_comparison): Check that a result
13630 of a vector comparison has signed type.
13632 2013-03-28 Richard Biener <rguenther@suse.de>
13634 PR tree-optimization/37021
13635 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
13636 do not restrict gaps between groups.
13637 * tree-vect-stmts.c (vectorizable_load): Properly account for
13638 a gap between groups.
13640 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
13642 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
13643 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
13646 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
13648 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
13649 * doc/extend.texi (Named Address Spaces): Ditto.
13650 (Variable Attributes): Ditto.
13652 2013-03-27 Kai Tietz <ktietz@redhat.com>
13654 * config.build: Add support for cygwin x64 target.
13655 * config.gcc: Likewise.
13656 * config.host: Likewise.
13657 * configure.ac: Likewise
13658 * configure: Regenerated.
13660 2013-03-27 Kai Tietz <ktietz@redhat.com>
13662 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
13663 * config/i386/t-cygwin-w64: New file.
13664 * config/i386/cygwin-w64.h: New file.
13665 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
13666 and add support for x64-cygwin target.
13667 (CPP_SPEC): Likewise.
13668 (CXX_WRAP_SPEC_LIST): Undefine before define.
13669 (LIBGCJ_SONAME): Use 15 as version.
13671 2013-03-27 Richard Biener <rguenther@suse.de>
13673 PR tree-optimization/56716
13674 * tree-ssa-structalias.c (perform_var_substitution): Adjust
13675 dumping for ref nodes.
13677 2013-03-27 Martin Jambor <mjambor@suse.cz>
13679 PR tree-optimization/55334
13680 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
13681 restricted pointers to arrays.
13683 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
13685 * Makefile.in (.SUFFIXES): Add .cc.
13686 (.c.o): Apply same recipe for implicit rule .cc.o.
13688 2013-03-27 Richard Biener <rguenther@suse.de>
13690 PR tree-optimization/37021
13691 * tree-vect-data-refs.c (vect_check_strided_load): Allow
13692 REALPART/IMAGPART_EXPRs around the supported refs.
13693 * tree-ssa-structalias.c (find_func_aliases): Assume that
13694 floating-point values are not used to transfer pointers.
13696 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
13698 * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
13699 * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
13700 * gcc/doc/tm.texi: Regenerate.
13701 * gcc/targhooks.h (default_has_ifunc_p): New.
13702 * gcc/targhooks.c (default_has_ifunc_p): Ditto.
13703 * gcc/config/linux-protos.h: New file.
13704 * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
13705 this hook for linux which disables support of indirect functions in
13707 * gcc/config/linux-android.c: New file.
13708 * gcc/config/t-linux-android.c: Ditto.
13709 * gcc/config.gcc: Added new object file linux-android.o.
13710 * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
13711 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
13712 * gcc/varasm.c (do_assemble_alias): Likewise.
13713 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
13714 doesn't support indirect functions.
13715 * configure: Regenerate.
13717 2013-03-27 Bin Cheng <bin.cheng@arm.com>
13720 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
13721 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
13724 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
13726 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
13728 2013-03-27 Terry Guo <terry.guo@arm.com>
13730 * config/arm/arm-cores.def: Added core cortex-r7.
13731 * config/arm/arm-tune.md: Regenerated.
13732 * config/arm/arm-tables.opt: Regenerated.
13733 * doc/invoke.texi: Added entry for core cortex-r7.
13735 2013-03-27 Walter Lee <walt@tilera.com>
13737 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
13738 double-decrement of next_scratch_regno.
13740 2013-03-27 Walter Lee <walt@tilera.com>
13742 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
13744 (insn_v1mulus): Ditto.
13745 (insn_v2muls): Ditto.
13747 2013-03-27 Walter Lee <walt@tilera.com>
13749 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
13750 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
13752 2013-03-27 Walter Lee <walt@tilera.com>
13754 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
13755 (*sibcall_value): Ditto.
13757 2013-03-27 Walter Lee <walt@tilera.com>
13759 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
13760 (insn_mnz_v8qi): ... this ...
13761 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
13763 (insn_v<n>mnz): Replaced by ...
13764 (insn_v1mnz): ... this ...
13765 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
13767 (insn_mz_<mode>): Replaced by ...
13768 (insn_mz_v8qi): ... this ...
13769 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
13771 (insn_v<n>mz): Replaced by ...
13772 (insn_v1mz): ... this ...
13773 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
13776 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
13778 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
13780 2013-03-26 Roland McGrath <mcgrathr@google.com>
13782 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
13783 than fprintf with a non-constant, non-format string.
13785 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
13787 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
13788 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
13789 operand 0 predicate.
13790 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
13791 attribute. Use general_x64nomem_operand as operand 1 predicate.
13792 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
13793 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
13794 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
13795 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
13796 (mov<mode>_insv_1): Remove expander. Merge insn with
13797 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
13798 Use general_x64nomem_operand as operand 1 predicate.
13799 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
13800 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
13801 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
13802 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
13803 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
13804 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
13805 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
13806 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
13807 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
13808 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
13809 (general_x64nomem_operand): Ditto.
13811 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13813 * config/rtems.opt: Add -pthread option.
13815 2013-03-26 Richard Biener <rguenther@suse.de>
13817 * alias.c (find_base_term): Avoid redundant and not used recursion.
13818 (base_alias_check): Get the initial base term from the caller.
13819 (true_dependence_1): Compute and pass base terms to base_alias_check.
13820 (write_dependence_p): Likewise.
13821 (may_alias_p): Likewise.
13823 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
13825 * config/aarch64/aarch64.c (aarch64_classify_address): Support
13826 PC-relative load in SI modes and above only.
13828 2013-03-26 Xinyu Qi <xyqi@marvell.com>
13830 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
13831 * config/arm/iwmmxt.md (WCGR0): Update.
13832 (WCGR1, WCGR2, WCGR3): Likewise.
13834 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
13836 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13837 Use x64 and nox64 isa attributes.
13839 2013-03-26 Richard Biener <rguenther@suse.de>
13841 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13842 alignment computations and rely on get_object_alignment_1
13843 for the !TYPE_P case.
13844 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13846 2013-03-26 Walter Lee <walt@tilera.com>
13848 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13849 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13851 2013-03-25 Jeff Law <law@redhat.com>
13853 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
13854 check for INTEGRAL_TYPE_P that was missing due to checking in
13855 wrong version of prior patch.
13857 2013-03-25 Walter Lee <walt@tilera.com>
13859 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
13860 TILEGX_INSN_SHUFFLEBYTES1.
13861 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
13863 (tilegx_builtins): Ditto.
13864 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
13866 2013-03-25 Walter Lee <walt@tilera.com>
13868 * config/tilegx/tilegx.md (floatsisf2): New pattern.
13869 (floatunssisf2): New pattern.
13870 (floatsidf2): New pattern.
13871 (floatunssidf2): New pattern.
13873 2013-03-25 Walter Lee <walt@tilera.com>
13875 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
13876 tests for constraint J, K, N, P.
13878 2013-03-25 Walter Lee <walt@tilera.com>
13880 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
13881 Use indirect/pcrel encoding.
13882 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
13885 2013-03-25 Steve Ellcey <sellcey@mips.com>
13887 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
13888 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
13889 * config/mips/mips.c (mips_option_override): Set IMADD default.
13890 * config/mips/mips.h (PTF_AVOID_IMADD): New.
13891 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
13892 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
13893 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
13894 * doc/invoke.texi (-mimadd/-mno-imadd): New.
13896 2013-03-25 Jeff Law <law@redhat.com>
13898 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
13899 slightly to avoid creating and folding useless trees. Simplify
13900 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
13902 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
13904 * config/i386/i386.md (*zero_extendsidi2): Merge with
13905 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
13906 * config/i386/predicates.md (x86_64_zext_operand): Rename from
13907 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
13908 targets. Clarify comment.
13910 2013-03-25 Martin Jambor <mjambor@suse.cz>
13912 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
13913 pass-through jump functions differently.
13914 (ipa_read_jump_function): Likewise. Also use setter functions to set
13917 2013-03-25 Martin Jambor <mjambor@suse.cz>
13919 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
13920 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
13922 (ipa_get_indirect_edge_target): New function.
13923 (devirtualization_time_bonus): New parameter known_aggs, pass it to
13924 ipa_get_indirect_edge_target. Update all callers.
13925 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
13926 ipa_get_indirect_edge_target_1 instead of calling
13927 ipa_get_indirect_edge_target.
13928 (create_specialized_node): Pass aggvlas to
13929 ipcp_discover_new_direct_edges.
13931 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13933 * config/arm/arm.md (f_sels, f_seld): New types.
13934 (*cmov<mode>): New pattern.
13935 * config/arm/predicates.md (arm_vsel_comparison_operator): New
13938 2013-03-25 Kai Tietz <ktietz@redhat.com>
13940 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
13941 POSIX-printf for mingw-hosted builds.
13943 2013-03-25 Richard Biener <rguenther@suse.de>
13945 PR middle-end/56694
13946 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
13947 must-not-throw stmt location.
13949 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13951 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
13952 Emit load-acquire versions when acq is true.
13953 (arm_emit_store_exclusive): Add rel parameter.
13954 Emit store-release versions when rel is true.
13955 (arm_split_compare_and_swap): Use acquire-release instructions
13957 of barriers when appropriate.
13958 (arm_split_atomic_op): Likewise.
13959 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
13960 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
13961 (VUNSPEC_SLX): Likewise.
13962 (VUNSPEC_LDA): Likewise.
13963 (VUNSPEC_STL): Likewise.
13964 * config/arm/sync.md (atomic_load<mode>): New pattern.
13965 (atomic_store<mode>): Likewise.
13966 (arm_load_acquire_exclusive<mode>): Likewise.
13967 (arm_load_acquire_exclusivesi): Likewise.
13968 (arm_load_acquire_exclusivedi): Likewise.
13969 (arm_store_release_exclusive<mode>): Likewise.
13971 2013-03-25 Catherine Moore <clm@codesourcery.com>
13973 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
13974 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
13975 * config/mip/predicates.md (lwsp_swsp_operand,
13976 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
13977 sb16_operand, db4_operand, db7_operand, ib3_operand,
13978 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
13979 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
13980 andi16_operand): New predicates.
13981 * config/mips/mips.md (compression): New attribute.
13982 (enabled): New attribute.
13983 (length): Consider compression in computing length.
13984 (shift_compression): New code attribute.
13985 (*add<mode>3): New operands. Record compression.
13986 (sub<mode>3): Likewise.
13987 (one_cmpl<mode>2): Likewise.
13988 (*and<mode>3): Likewise.
13989 (*ior<mode>3): Likewise.
13990 (unnamed pattern for xor): Likewise.
13991 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
13992 (*<optab><mode>3): Likewise.
13993 (*mov<mode>_internal: Likewise.
13994 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
13995 (mips_unsigned_immediate_p): New.
13996 (umips_lwsp_swsp_address_p): New.
13997 (m16_based_address_p): New.
13998 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
13999 (mips_unsigned_immediate_p): New prototype.
14000 (lwsp_swsp_address_p): New prototype.
14001 (m16_based_address_p): New prototype.
14002 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14003 (mips_signed_immediate_p): New function.
14004 (m16_based_address_p): New function.
14005 (lwsp_swsp_address_p): New function.
14006 (mips_print_operand_punctuation): Recognize short delay slot insns
14007 for microMIPS.add<mode>3"
14009 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14012 * config/arm/iterators.md (v_cmp_result): New mode attribute.
14013 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14015 2013-03-25 Richard Biener <rguenther@suse.de>
14017 PR tree-optimization/56689
14018 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14021 2013-03-25 Richard Biener <rguenther@suse.de>
14023 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14025 (memory_references): Likewise.
14026 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14027 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14028 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14029 (gather_mem_refs_in_loops): Fold into ...
14030 (analyze_memory_references): ... this. Move initialization
14031 to tree_ssa_lim_initialize.
14032 (fill_always_executed_in): Rename to ...
14033 (fill_always_executed_in_1): ... this.
14034 (fill_always_executed_in): Move contains_call computation to
14035 this new function from ...
14036 (tree_ssa_lim_initialize): ... here.
14037 (tree_ssa_lim): Call fill_always_executed_in.
14039 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
14041 * postreload.c (reload_combine): Fix code detecting returns.
14043 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
14045 * function.c (emit_use_return_register_into_block): On cc0 targets,
14046 do not emit the sequence between cc0 setter and user.
14048 2013-03-25 Kai Tietz <ktietz@redhat.com>
14050 * config/i386/predicates.md (local_symbolic_operand): Interpret
14051 dll-imported symbols as none-local.
14053 2013-03-25 Richard Biener <rguenther@suse.de>
14055 * tree-ssa-loop-im.c (struct depend): Remove.
14056 (struct lim_aux_data): Make depends a vec of gimples.
14057 (free_lim_aux_data): Adjust.
14058 (add_dependency): Likewise.
14059 (set_level): Likewise.
14061 2013-03-25 Richard Biener <rguenther@suse.de>
14063 PR middle-end/56434
14064 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14065 the pointer returned by calls with ECF_MALLOC set.
14067 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
14069 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14071 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
14073 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14074 using MMXMODE mode iterator.
14075 (*move<mode>_internal): Merge with *movv2sf_internal and
14076 *movv2sf_internal_rex64 using MMXMODE mode iterator.
14078 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
14080 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14081 (record_last_mem_set_info): Likewise.
14083 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14084 of XNEWVEC followed by memset.
14085 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14087 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
14089 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14090 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14091 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14092 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14093 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14094 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14095 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14096 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14097 BARRIER_P instead of GET_CODE.
14099 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14101 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14102 inaccuracy in the probing code.
14104 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14105 (ctrapdi4): Likewise.
14107 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14109 * calls.c (expand_call): Add missing guard to code handling return
14110 of non-BLKmode structures in MSB.
14111 * function.c (expand_function_end): Likewise.
14113 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14115 * combine.c (try_combine): Adjust comment. Do not add the set of
14116 insn #0 if the destination indirectly is set or dies in insn #2.
14117 Tidy up code to distribute a new note.
14119 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
14121 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14122 also for alternatives 16 and 17.
14124 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
14126 * config/i386/sse.md (*mov<mode>_internal): Merge with
14127 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
14128 Emit insn template depending on type attribute. Use
14129 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14130 movd instead of movq mnemonic for interunit moves. Rewrite mode
14131 attribute calculation. Remove unit attribute calculation.
14132 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14133 Set prefix_data16 attribute for DImode ssemov types.
14134 Use Ym instead of y for SSE-MMX conversion alternatives.
14135 Reorder operand constraints.
14137 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
14139 * df.h (df_insn_delete): Adjust prototype.
14140 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14141 and let it decide whether mark the basic block dirty.
14142 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14143 * df-scan.c (df_insn_info_delete): New helper function, split
14144 off from df_insn_delete.
14145 (df_scan_free_bb_info): Use it.
14146 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14148 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
14149 that the insn is actually an insn and it has a non-NULL basic block.
14150 Do not mark basic block dirty if only deleting a DEBUG_INSN.
14152 2013-03-22 Richard Biener <rguenther@suse.de>
14154 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14156 (mem_ref_alloc): Do not allocate them.
14157 (refs_independent_p): Do not query or maintain a cache.
14159 2013-03-22 Richard Biener <rguenther@suse.de>
14161 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14162 (gather_mem_refs_in_loops): Do not compute it.
14163 (analyze_memory_references): Do not allocate it.
14164 (tree_ssa_lim_finalize): Do not free it.
14165 (for_all_locs_in_loop): Do not query all_refs_in_loop.
14167 2013-03-22 Richard Biener <rguenther@suse.de>
14169 * is-a.h (as_a): Use gcc_checking_assert.
14171 2013-03-22 Ian Bolton <ian.bolton@arm.com>
14173 * config/aarch64/aarch64.c (aarch64_print_operand): New
14174 format specifier for printing a constant in hex.
14175 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14176 format specifier for printing second operand.
14178 2013-03-22 Richard Biener <rguenther@suse.de>
14180 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14182 (gather_mem_refs_in_loops): Perform store accumulation here.
14183 (create_vop_ref_mapping_loop): Remove.
14184 (create_vop_ref_mapping): Likewise.
14185 (analyze_memory_references): Initialize refs_stored_in_loop.
14186 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14187 (record_indep_loop): Remove.
14188 (record_dep_loop): New function.
14189 (ref_indep_loop_p_1): Adjust to only walk over references
14190 in the loop, not its subloops.
14191 (ref_indep_loop_p): Rename to ...
14192 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14193 maintaining a more fine-grained cache.
14194 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14195 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14197 2013-03-22 Richard Biener <rguenther@suse.de>
14199 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14200 (struct mem_ref): Make accesses_in_loop a vec of a vec of
14201 aggregate mem_ref_loc.
14202 (free_mem_ref_locs): Inline into ...
14203 (memref_free): ... this and adjust.
14204 (mem_ref_alloc): Adjust.
14205 (mem_ref_locs_alloc): Remove.
14206 (record_mem_ref_loc): Adjust.
14207 (get_all_locs_in_loop): Rewrite into ...
14208 (for_all_locs_in_loop): ... this iterator.
14209 (rewrite_mem_ref_loc): New functor.
14210 (rewrite_mem_refs): Use for_all_locs_in_loop.
14211 (sm_set_flag_if_changed): New functor.
14212 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14213 (ref_always_accessed): New functor.
14214 (ref_always_accessed_p): Use for_all_locs_in_loop.
14216 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14218 * tree-pass.h (PROP_gimple_lvec): New.
14219 * passes.c (dump_properties): Handle PROP_gimple_lvec.
14220 (init_optimization_passes): Move pass_lower_vector.
14221 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14223 (pass_lower_vector): Provide PROP_gimple_lvec.
14224 (pass_lower_vector_ssa): Likewise.
14225 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14227 2013-03-21 Mark Wielaard <mjw@redhat.com>
14229 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14231 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
14233 * config/i386/i386.md (*movdi_internal): Disparage slightly
14234 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
14235 conversion alternatives.
14237 2013-03-21 Jakub Jelinek <jakub@redhat.com>
14239 PR middle-end/48087
14240 * diagnostic.def (DK_WERROR): New kind.
14241 * diagnostic.h (werrorcount): Define.
14242 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14243 promoted to DK_ERROR, increment DK_WERROR counter instead of
14245 * toplev.c (toplev_main): Call print_ignored_options even if
14246 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
14247 even if just werrorcount is non-zero.
14250 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14252 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14253 (string_cst_pool_decl): New function.
14254 (optimize_one_addr_into_implicit_ptr): New function.
14255 (resolve_addr_in_expr): Optimize DWARF location expression
14256 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14257 which doesn't live in memory, but has DW_AT_location or
14258 DW_AT_const_value, or refers to a string literal, into
14259 DW_OP_GNU_implicit_pointer.
14260 (optimize_location_into_implicit_ptr): New function.
14261 (resolve_addr): If removing DW_AT_location of a variable because
14262 it was DW_OP_addr of address of the variable, but the variable doesn't
14263 live in memory, try to emit const value attribute for the initializer.
14265 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14267 * tree.h (VECTOR_TYPE_P): New macro.
14268 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14269 TYPE_MODE): Use it.
14270 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14271 VEC_COND_EXPR cannot be lvalues.
14272 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14274 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14276 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14277 Restrict the transformation to equal modes.
14279 2013-03-21 Richard Biener <rguenther@suse.de>
14281 PR tree-optimization/39326
14282 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14283 (MEM_ANALYZABLE): Adjust.
14284 (record_mem_ref_loc): Move bitmap ops ...
14285 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
14286 unanalyzable refs, do not record locations for it.
14287 (analyze_memory_references): Allocate ref zero as shared
14289 (refs_independent_p): Do not test for unanalyzed mems here.
14290 (ref_indep_loop_p_1): Special-case disambiguation against
14291 the unanalyzed ref.
14292 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14294 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
14296 * config/arm/arm-protos.h (tune_params): Add
14297 prefer_neon_for_64bits field.
14298 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14299 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14300 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14301 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14302 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14303 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14304 (arm_option_override): Handle -mneon-for-64bits new option.
14305 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14306 (prefer_neon_for_64bits): Declare new variable.
14307 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14308 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14309 (arch_enabled): Handle new arch types. Remove support for onlya8
14311 (one_cmpldi2): Use new arch names.
14312 (zero_extend<mode>di2, extend<mode>di2): Ditto.
14313 * config/arm/arm.opt (mneon-for-64bits): Add option.
14314 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14315 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14316 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14318 * doc/invoke.texi (-mneon-for-64bits): Document.
14320 2013-03-21 Richard Biener <rguenther@suse.de>
14322 PR tree-optimization/39326
14323 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14324 (sort_bbs_in_loop_postorder_cmp): New function.
14325 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14327 2013-03-21 Richard Biener <rguenther@suse.de>
14329 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14330 (vect_insert_into_interleaving_chain): Likewise.
14331 (vect_drs_dependent_in_basic_block): Inline ...
14332 (vect_slp_analyze_data_ref_dependence): ... here. New function,
14334 (vect_analyze_data_ref_dependence): ... here. Simplify.
14335 (vect_check_interleaving): Simplify.
14336 (vect_analyze_data_ref_dependences): Likewise. Split out ...
14337 (vect_slp_analyze_data_ref_dependences): ... this new function.
14338 (dr_group_sort_cmp): New function.
14339 (vect_analyze_data_ref_accesses): Compute data-reference groups
14340 here instead of in vect_analyze_data_ref_dependence. Use
14341 a more efficient algorithm.
14342 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14343 vect_slp_analyze_data_ref_dependences. Call
14344 vect_analyze_data_ref_accesses earlier.
14345 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14346 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14347 (vect_slp_analyze_data_ref_dependences): New prototype.
14349 2013-03-21 Richard Biener <rguenther@suse.de>
14351 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14352 ref is stored in the loop.
14353 (find_refs_for_sm): Walk only over all stores.
14354 (store_motion_loop): Allocate from lim_bitmap_obstack.
14355 (store_motion): Likewise.
14357 2013-03-21 Richard Biener <rguenther@suse.de>
14359 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14360 Update virtual SSA form.
14362 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14364 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14365 * configure: Regenerate.
14366 * config.in: Regenerate.
14367 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14368 if !HAVE_LD_EH_FRAME_CIEV3.
14370 2013-03-21 Richard Biener <rguenther@suse.de>
14372 * tree-cfg.c (verify_expr_no_block): New function.
14373 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14374 nor DECL_VALUE_EXPR have locations with associated blocks.
14375 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14376 (clear_unused_block_pointer): Remove code dealing with
14377 blocks in DECL_DEBUG_EXPR locations.
14379 2013-03-21 Richard Biener <rguenther@suse.de>
14381 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14382 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
14383 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14384 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14385 instead of DECL_DEBUG_EXPR_IS_FROM.
14386 * gimplify.c (gimplify_modify_expr): Likewise.
14387 * tree-cfg.c (verify_expr_location_1): Likewise.
14388 * tree-complex.c (create_one_component_var): Likewise.
14389 * tree-sra.c (create_access_replacement): Likewise.
14390 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14391 (clear_unused_block_pointer): Likewise.
14392 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14393 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14394 * var-tracking.c (var_debug_decl): Likewise.
14395 (track_expr_p): Likewise.
14396 * tree-inline.c (add_local_variables): Likewise. Set
14397 DECL_HAS_DEBUG_EXPR_P after copying it.
14398 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14399 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
14401 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
14404 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14405 * configure: Regenerate.
14406 * config.in: Regenerate.
14407 * config/i386/i386.md (*movdf_internal): Use
14408 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14409 movd instead of movq mnemonic for interunit moves.
14410 (*movdi_internal): Ditto.
14412 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14414 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14415 (abd<mode>_3): New pattern.
14416 (aba<mode>_3): New pattern.
14417 (fabd<mode>_3): New pattern.
14419 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14421 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14422 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14423 occurrence of REGISTER_PREFIX as its empty string.
14425 2013-03-20 Jeff Law <law@redhat.com>
14427 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14428 addititional equivalences for equality comparisons between an SSA_NAME
14429 and a constant where the SSA_NAME was set from a widening conversion.
14431 2013-03-20 Walter Lee <walt@tilera.com>
14433 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14435 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14437 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14438 depending on type attribute.
14439 (*movti_internal): Ditto.
14440 (*movtf_internal): Ditto.
14441 (*movxf_internal): Ditto.
14442 (*movdf_internal): Ditto.
14443 (*movsf_internal): Ditto.
14445 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14447 * config/i386/i386.md (*movti_internal): Set prefix attribute to
14448 maybe_vex for sselog1 and ssemov types.
14449 (*movdi_internal): Reorder operand constraints.
14450 (*movsi_internal): Ditto. Set prefix attribute to
14451 maybe_vex for sselog1 and ssemov types.
14452 (*movtf_internal): Set prefix attribute to maybe_vex
14453 for sselog1 and ssemov types.
14454 (*movdf_internal): Ditto. Set prefix_data16 attribute for
14455 DImode ssemov types. Reorder operand constraints.
14456 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
14457 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
14458 attribute for SImode ssemov types. Reorder operand constraints.
14460 2013-03-20 Martin Jambor <mjambor@suse.cz>
14462 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14463 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14465 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
14467 * config/rs6000/predicates.md (indexed_address, update_address_mem
14468 update_indexed_address_mem): New predicates.
14469 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14470 attribute for load/store instructions.
14471 * config/rs6000/dfp.md (movsd_store): Likewise.
14472 (movsd_load): Likewise.
14473 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14474 (unnamed HI->DI extend define_insn): Likewise.
14475 (unnamed SI->DI extend define_insn): Likewise.
14476 (unnamed QI->SI extend define_insn): Likewise.
14477 (unnamed QI->HI extend define_insn): Likewise.
14478 (unnamed HI->SI extend define_insn): Likewise.
14479 (unnamed HI->SI extend define_insn): Likewise.
14480 (extendsfdf2_fpr): Likewise.
14481 (movsi_internal1): Likewise.
14482 (movsi_internal1_single): Likewise.
14483 (movhi_internal): Likewise.
14484 (movqi_internal): Likewise.
14485 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14486 attribute for load/store instructions.
14487 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14489 (mov<mode>_softfloat): Likewise.
14490 (mov<mode>_hardfloat32): Likewise.
14491 (mov<mode>_hardfloat64): Likewise.
14492 (mov<mode>_softfloat64): Likewise.
14493 (movdi_internal32): Likewise.
14494 (movdi_internal64): Likewise.
14495 (probe_stack_<mode>): Likewise.
14497 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14499 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14500 floating point, and decimal floating point to reload iterator.
14502 * config/rs6000/constraints.md (wl constraint): New constraints to
14503 return FLOAT_REGS if certain options are used to reduce the number
14504 of separate patterns that exist in the file.
14505 (wx constraint): Likewise.
14506 (wz constraint): Likewise.
14508 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14509 -mdebug=reg, print wg, wl, wx, and wz constraints.
14510 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14511 Initialize the reload functions for 64-bit binary/decimal floating
14513 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14514 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14515 create the buffer on the stack to overcome not having a 32-bit
14517 (rs6000_emit_move): Likewise.
14518 (rs6000_secondary_memory_needed_rtx): Likewise.
14519 (rs6000_alloc_sdmode_stack_slot): Likewise.
14520 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14521 via xxlxor, just like DFmode 0.0.
14523 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14524 define as 1 if we are running on a power7 or newer.
14525 (enum r6000_reg_class_enum): Add new constraints.
14527 * config/rs6000/dfp.md (movsd): Delete, combine with binary
14528 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
14529 with other moves by using conditional constraits (wg). Use LFIWZX
14530 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
14531 (movsd splitter): Likewise.
14532 (movsd_hardfloat): Likewise.
14533 (movsd_softfloat): Likewise.
14535 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14536 binary and decimal floating point moves.
14537 (fmove_ok): New attributes to combine binary and decimal floating
14538 point moves, and to combine power6x (mfpgpr) moves along normal
14540 (real_value_to_target): Likewise.
14541 (f32_lr): Likewise.
14542 (f32_lm): Likewise.
14543 (f32_li): Likewise.
14544 (f32_sr): Likewise.
14545 (f32_sm): Likewise.
14546 (f32_si): Likewise.
14547 (movsf): Combine binary and decimal floating point moves. Combine
14548 power6x (mfpgpr) moves with other moves by using conditional
14549 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
14550 (mov<mode> for SFmode/SDmode); Likewise.
14551 (SFmode/SDmode splitters): Likewise.
14552 (movsf_hardfloat): Likewise.
14553 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14554 (movsf_softfloat): Likewise.
14555 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14557 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14558 wx and wz constraints.
14560 * config/rs6000/constraints.md (wg constraint): New constraint to
14561 return FLOAT_REGS if -mmfpgpr (power6x) was used.
14563 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14566 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14567 -mdebug=reg, print wg, wl, wx, and wz constraints.
14568 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14569 Initialize the reload functions for 64-bit binary/decimal floating
14571 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14572 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14573 create the buffer on the stack to overcome not having a 32-bit
14575 (rs6000_emit_move): Likewise.
14576 (rs6000_secondary_memory_needed_rtx): Likewise.
14577 (rs6000_alloc_sdmode_stack_slot): Likewise.
14578 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14579 via xxlxor, just like DFmode 0.0.
14581 * config/rs6000/dfp.md (movdd): Delete, combine with binary
14582 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
14583 with other moves by using conditional constraits (wg). Use LFIWZX
14584 and STFIWX for loading SDmode on power7.
14585 (movdd splitters): Likewise.
14586 (movdd_hardfloat32): Likewise.
14587 (movdd_softfloat32): Likewise.
14588 (movdd_hardfloat64_mfpgpr): Likewise.
14589 (movdd_hardfloat64): Likewise.
14590 (movdd_softfloat64): Likewise.
14592 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14593 64-bit binary and decimal floating point moves.
14594 (FMOVE64X): Likewise.
14595 (movdf): Combine 64-bit binary and decimal floating point moves.
14596 Combine power6x (mfpgpr) moves with other moves by using
14597 conditional constraits (wg).
14598 (mov<mode> for DFmode/DDmode): Likewise.
14599 (DFmode/DDmode splitters): Likewise.
14600 (movdf_hardfloat32): Likewise.
14601 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
14602 (movdf_softfloat32): Likewise.
14603 (movdf_hardfloat64_mfpgpr): Likewise.
14604 (movdf_hardfloat64): Likewise.
14605 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
14606 (movdf_softfloat64): Likewise.
14607 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
14608 (reload_<mode>_load): Move to later in the file so they aren't in
14609 the middle of the floating point move insns.
14610 (reload_<mode>_store): Likewise.
14612 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
14615 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
14616 constraint if -mdebug=reg.
14617 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
14618 Enable using dd reload support if needed.
14620 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
14621 binary and decimal floating point moves in rs6000.md.
14622 (movtd_internal): Likewise.
14624 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
14625 decimal floating point moves.
14627 (movtf_internal): Likewise.
14628 (mov<mode>_internal, TDmode/TFmode): Likewise.
14629 (movtf_softfloat): Likewise.
14630 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
14632 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
14633 movdi_internal64, using wg constraint for move direct operations.
14634 (movdi_internal64): Likewise.
14636 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
14637 MODES_TIEABLE_P for selected modes. Print the numerical value of
14638 the various virtual registers. Use GPR/FPR first/last values,
14639 instead of hard coding the register numbers. Print which modes
14640 have reload functions registered.
14641 (rs6000_option_override_internal): If -mdebug=reg, trace the options
14642 settings before/after setting cpu, target and subtarget settings.
14643 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
14644 and for secondary reload failures in rs6000_secondary_reload_inner.
14645 (rs6000_secondary_reload_fail): Likewise.
14646 (rs6000_secondary_reload_inner): Likewise.
14648 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
14649 macros for first/last GPR and FPR registers.
14650 (LAST_GPR_REGNO): Likewise.
14651 (FIRST_FPR_REGNO): Likewise.
14652 (LAST_FPR_REGNO): Likewise.
14654 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
14655 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
14656 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
14657 (vcond<mode><mode>): Likewise.
14658 (vcondu<mode><mode>): Likewise.
14659 (vector_gtu<mode>): Likewise.
14660 (vector_gte<mode>): Likewise.
14661 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
14662 to prevent the compiler from converting DImode operations to TImode.
14663 (ior<mode>3): Likewise.
14664 (and<mode>3): Likewise.
14665 (one_cmpl<mode>2): Likewise.
14666 (nor<mode>3): Likewise.
14667 (andc<mode>3): Likewise.
14669 * config/rs6000/constraints.md (wt constraint): New constraint
14670 that returns VSX_REGS if TImode is allowed in VSX registers.
14672 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
14673 constant under VSX.
14675 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
14676 similar to TImode, but it is restricted to being in the GPRs.
14678 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
14679 TImode to occupy a single VSX register.
14681 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
14682 -mvsx-timode for power7/power8.
14683 (power7 cpu): Likewise.
14684 (power8 cpu): Likewise.
14686 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
14687 sure that TFmode/TDmode take up two registers if they are ever
14688 allowed in the upper VSX registers.
14689 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
14691 (rs6000_init_hard_regno_mode_ok): Likewise.
14692 (rs6000_debug_reg_global): Add debugging for PTImode and wt
14693 constraint. Print if LRA is turned on.
14694 (rs6000_option_override_internal): Give an error if -mvsx-timode
14695 and VSX is not enabled.
14696 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
14697 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
14698 to reg+offset addressing. Use PTImode when checking offset
14699 addresses for validity.
14700 (reg_offset_addressing_ok_p): Likewise.
14701 (rs6000_legitimate_offset_address_p): Likewise.
14702 (rs6000_legitimize_address): Likewise.
14703 (rs6000_legitimize_reload_address): Likewise.
14704 (rs6000_legitimate_address_p): Likewise.
14705 (rs6000_eliminate_indexed_memrefs): Likewise.
14706 (rs6000_emit_move): Likewise.
14707 (rs6000_secondary_reload): Likewise.
14708 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
14709 reloads to fpr registers to continue to use reg+offset addressing,
14710 but 64-bit reloads to altivec registers need reg+reg addressing.
14711 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
14712 it. Treat LO_SUM like a PLUS operation.
14713 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
14714 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
14715 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
14716 registers to share a register with a smaller sized type, since VSX
14717 puts scalars in the upper 64-bits.
14718 (print_operand): Add support for PTImode.
14719 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
14720 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
14721 registers, but don't have arithmetic support.
14722 (rs6000_memory_move_cost): Add test for VSX.
14723 (rs6000_opt_masks): Add -mvsx-timode.
14725 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
14728 (VSr): Use wt constraint for TImode.
14729 (VSv): Drop TImode support.
14730 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
14731 (vsx_movti_64bit): Likewise.
14732 (vsx_movti_32bit): Likewise.
14733 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
14734 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
14735 one '?' on the appropriate output constraint. Do not allow TImode
14736 logical operations on 32-bit systems.
14737 (vsx_ior<mode>3): Likewise.
14738 (vsx_xor<mode>3): Likewise.
14739 (vsx_one_cmpl<mode>2): Likewise.
14740 (vsx_nor<mode>3): Likewise.
14741 (vsx_andc<mode>3): Likewise.
14742 (vsx_concat_<mode>): Likewise.
14743 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
14745 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
14746 OPTION_MASK_VSX_TIMODE.
14747 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
14748 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
14750 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
14751 (TI2 iterator): New iterator for TImode, PTImode.
14752 (wd mode attribute): Add values for vector types.
14753 (movti_string): Replace TI move operations with operations for TImode
14754 and PTImode. Add support for TImode being allowed in VSX registers.
14755 (mov<mode>_string, TImode/PTImode): Likewise.
14756 (movti_ppc64): Likewise.
14757 (mov<mode>_ppc64, TImode/PTImode): Likewise.
14758 (TI mode splitters): Likewise.
14760 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
14763 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
14765 PR tree-optimization/56355
14766 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
14767 Also handle integers with undefined overflow.
14769 2013-03-20 Catherine Moore <clm@codesourcery.com>
14770 Maciej W. Rozycki <macro@codesourcery.com>
14771 Tom de Vries <tom@codesourcery.com>
14772 Nathan Sidwell <nathan@codesourcery.com>
14773 Iain Sandoe <iain@codesourcery.com>
14774 Nathan Froyd <froydnj@codesourcery.com>
14775 Chao-ying Fu <fu@mips.com>
14777 * doc/extend.texi: (micromips, nomicromips, nocompression):
14778 Document new function attributes.
14779 * doc/invoke.texi (minterlink-compressed, mmicromips,
14780 m14k, m14ke, m14kec): Document new options.
14781 (minterlink-mips16): Update documentation.
14782 * doc/md.texi (ZC, ZD): Document new constraints.
14783 * configure.ac (gcc_cv_as_micromips): Check if linker
14784 supports the .set micromips directive.
14785 * configure: Regenerate.
14786 * config.in: Regenerate.
14787 * config/mips/mips-tables.opt: Regenerate.
14788 * config/mips/micromips.md: New file.
14789 * constraints.md (ZC, ZD): New constraints.
14790 * config/mips/predicates.md (movep_src_register): New predicate.
14791 (movep_src_operand): New predicate.
14792 (non_volatile_mem_operand): New predicate.
14793 * config/mips/mips.md (multimem): New type.
14794 (length): Differentiate between 17-bit and 18-bit branch offsets.
14795 (MOVEP1, MOVEP2): New mode iterator.
14796 (mov_<load>l): Use ZC constraint.
14797 (mov_<load>r): Likewise.
14798 (mov_<store>l): Likewise.
14799 (mov_<store>r): Likewise.
14800 (*branch_equality<mode>_inverted): Add microMIPS support.
14801 (*branch_equality<mode>): Likewise.
14802 (*jump_absolute): Likewise.
14803 (indirect_jump_<mode>): Likewise.
14804 (tablejump_<mode>): Likewise.
14805 (<optab>_internal): Likewise.
14806 (sibcall_internal): Likewise.
14807 (sibcall_value_internal): Likewise.
14808 (prefetch): Use constraint ZD.
14809 * config/mips/mips.opt (minterlink-compressed): New option.
14810 (minterlink-mips16): Now an alias for minterlink-compressed.
14811 (mmicromips): New option.
14812 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14813 (compare_and_swap_12): Likewise.
14814 (sync_add<mode>): Likewise.
14815 (sync_<optab>_12): Likewise.
14816 (sync_old_<optab>_12): Likewise.
14817 (sync_new_<optab>_12): Likewise.
14818 (sync_nand_12): Likewise.
14819 (sync_old_nand_12): Likewise.
14820 (sync_new_nand_12): Likewise.
14821 (sync_sub<mode>): Likewise.
14822 (sync_old_add<mode>): Likewise.
14823 (sync_old_sub<mode>): Likewise.
14824 (sync_new_add<mode>): Likewise.
14825 (sync_new_sub<mode>): Likewise.
14826 (sync_<optab><mode>): Likewise.
14827 (sync_old_<optab><mode>): Likewise.
14828 (sync_new_<optab><mode>): Likewise.
14829 (sync_nand<mode>): Likewise.
14830 (sync_old_nand<mode>): Likewise.
14831 (sync_new_nand<mode>): Likewise.
14832 (sync_lock_test_and_set<mode>): Likewise.
14833 (test_and_set_12): Likewise.
14834 (atomic_compare_and_swap<mode>): Likewise.
14835 (atomic_exchange<mode>_llsc): Likewise.
14836 (atomic_fetch_add<mode>_llsc): Likewise.
14837 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14838 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14839 (umips_save_restore_pattern_p): Likewise.
14840 (umips_load_store_pair_p): Likewise.
14841 (umips_output_load_store_pair): Likewise.
14842 (umips_movep_target_p): Likewise.
14843 (umips_12bit_offset_address_p): Likewise.
14844 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14845 (mips_base_mips16): Rename this...
14846 (mips_base_compression_flags): ...to this. Update all uses.
14847 (mips_attribute_table): Add micromips, nomicromips and nocompression.
14848 (mips_mips16_decl_p): Delete.
14849 (mips_nomips16_decl_p): Delete.
14850 (mips_get_compress_on_flags): New function.
14851 (mips_get_compress_off_flags): New function.
14852 (mips_get_compress_mode): New function.
14853 (mips_get_compress_on_name): New function.
14854 (mips_get_compress_off_name): New function.
14855 (mips_insert_attributes): Support multiple compression types.
14856 (mips_merge_decl_attributes): Likewise.
14857 (umips_12bit_offset_address_p): New function.
14858 (mips_start_function_definition): Emit .set micromips directive.
14859 (mips_call_may_need_jalx_p): New function.
14860 (mips_function_ok_for_sibcall): Add microMIPS support.
14861 (mips_print_operand_punctuation): Support short delay slots and
14863 (umips_swm_mask, umips_swm_encoding): New.
14864 (umips_build_save_restore): New function.
14865 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
14866 (was_mips16_p): Remove.
14867 (old_compression_mode): New.
14868 (mips_set_compression_mode): New function.
14869 (mips_set_current_function): Add microMIPS support.
14870 (mips_option_override): Likewise.
14871 (umips_save_restore_pattern_p): New function.
14872 (umips_output_save_restore): New function.
14873 (umips_load_store_pair_p_1): New function.
14874 (umips_load_store_pair_p): New function.
14875 (umips_output_load_store_pair_1): New function.
14876 (umips_output_load_store_pair): New function.
14877 (umips_movep_target_p) New function.
14878 (mips_prepare_pch_save): Add microMIPS support.
14879 * config/mips/mips.h (TARGET_COMPRESSION): New.
14880 (TARGET_CPU_CPP_BUILTINS): Update macro
14881 to use new compression flags and to support microMIPS.
14882 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
14883 (MIPS_ARCH_FLOAT_SPEC): Likewise.
14884 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
14885 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
14886 (ASM_SPEC): Support mmicromips and mno-micromips.
14887 (M16STORE_REG_P): New macro.
14888 (MIPS_CALL): Support TARGET_MICROMIPS.
14889 (MICROMIPS_J): New macro.
14890 (mips_base_mips16): Rename this...
14891 (mips_base_compression_flags): ...to this.
14892 (UMIPS_12BIT_OFFSET_P): New macro.
14893 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
14894 (MULTILIB_DIRNAMES): Likewise.
14895 2013-03-20 Richard Biener <rguenther@suse.de>
14897 PR tree-optimization/56661
14898 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
14899 the result does not have to be distinct.
14901 2013-03-20 Richard Biener <rguenther@suse.de>
14903 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
14906 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14907 Steven Bosscher <steven@gcc.gnu.org>
14909 PR rtl-optimization/56605
14910 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
14912 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14915 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
14916 that require movd instead of movq.
14918 2013-03-20 Richard Biener <rguenther@suse.de>
14920 * tree-ssa-structalias.c (struct variable_info): Add pointer
14921 to the first field of an aggregate with sub-vars. Make
14922 this and the pointer to the next subfield its ID.
14923 (vi_next): New function.
14924 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
14925 storedanything_id, integer_id): Increment by one.
14926 (new_var_info, get_call_vi, lookup_call_clobber_vi,
14927 get_call_clobber_vi): Adjust.
14928 (solution_set_expand): Simplify and speedup.
14929 (solution_set_add): Inline into ...
14930 (set_union_with_increment): ... this. Adjust accordingly.
14931 (do_sd_constraint): Likewise.
14932 (do_ds_constraint): Likewise.
14933 (do_complex_constraint): Simplify.
14934 (build_pred_graph): Adjust.
14935 (solve_graph): Likewise. Simplify and speedup.
14936 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
14937 get_constraint_for_component_ref, get_constraint_for_1,
14938 first_vi_for_offset, first_or_preceding_vi_for_offset,
14939 create_function_info_for, create_variable_info_for_1,
14940 create_variable_info_for, intra_create_variable_infos): Adjust.
14941 (init_base_vars): Push NULL for ID zero.
14942 (compute_points_to_sets): Adjust.
14944 2013-03-20 Richard Biener <rguenther@suse.de>
14946 * cfgloop.c (verify_loop_structure): Streamline and avoid
14947 ICEing on corrupt loop tree.
14948 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
14951 2013-03-20 Richard Biener <rguenther@suse.de>
14953 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
14954 check whether an SSA update is needed.
14956 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
14958 * config/mips/constraints.md (T): Rename to...
14962 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
14963 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
14965 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14967 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
14968 (*subsi3_carryin_uxtw): Likewise.
14970 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14972 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
14973 (*rorsi3_insn_uxtw): Likewise.
14975 2013-03-19 Ian Bolton <ian.bolton@arm.com>
14977 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
14978 (*extrsi5_insn_uxtw): Likewise.
14980 2013-03-19 Richard Biener <rguenther@suse.de>
14982 PR tree-optimization/56273
14983 * passes.c (init_optimization_passes): Move second VRP after DOM.
14985 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
14987 * config/i386/i386.md (*movti_internal): Merge from
14988 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
14989 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
14990 nox64 isa attributes.
14992 2013-03-18 Richard Biener <rguenther@suse.de>
14994 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
14996 (merge_node_constraints): Likewise.
14997 (build_succ_graph): Likewise.
14998 (valid_graph_edge): Inline into single caller.
14999 (unify_nodes): Likewise. Use bitmap_set_bit return value
15001 (scc_visit): Fix formatting and variable use.
15002 (do_sd_constraint): Use gcc_checking_assert.
15003 (do_ds_constraint): Likewise.
15004 (do_complex_constraint): Likewise.
15005 (condense_visit): Likewise. Cleanup.
15006 (dump_pred_graph): New function.
15007 (perform_var_substitution): Dump the pred-graph before
15008 variable substitution.
15009 (find_equivalent_node): Use gcc_checking_assert.
15010 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15012 2013-03-18 Richard Biener <rguenther@suse.de>
15014 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15015 Remove cond_expr_stmt_list argument and do not gimplify the
15017 (vect_loop_versioning): Adjust.
15018 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15019 Cleanup to use less temporaries.
15020 (vect_create_data_ref_ptr): Cleanup.
15022 2013-03-18 Jakub Jelinek <jakub@redhat.com>
15024 PR tree-optimization/56635
15025 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15026 require types_compatible_p types.
15028 2013-03-18 Nick Clifton <nickc@redhat.com>
15030 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15031 spurious backslash.
15033 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15034 Add missing line to comment describing function.
15036 2013-03-18 Richard Biener <rguenther@suse.de>
15038 PR tree-optimization/56210
15039 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15040 Handle string / character search functions.
15041 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15043 2013-03-18 Richard Biener <rguenther@suse.de>
15045 PR middle-end/56483
15046 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15047 and implement properly.
15048 * gimple.h (gimple_cond_single_var_p): Remove.
15050 2013-03-18 Richard Biener <rguenther@suse.de>
15052 * tree-data-ref.h (find_data_references_in_loop): Declare.
15053 * tree-data-ref.c (get_references_in_stmt): Use a stack
15054 vector pre-allocated in the callers.
15055 (find_data_references_in_stmt): Adjust.
15056 (graphite_find_data_references_in_stmt): Likewise.
15057 (create_rdg_vertices): Likewise.
15058 (find_data_references_in_loop): Export.
15059 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15060 Compute dependences here...
15061 (vect_analyze_data_refs): ...not here. When we encounter
15062 a non-vectorizable data reference in basic-block vectorization
15063 truncate the data reference vector. Do not bother to
15064 fixup data-dependence information for gather loads.
15065 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15066 of data references, as reported.
15068 2013-03-18 Richard Biener <rguenther@suse.de>
15070 PR tree-optimization/3713
15071 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
15072 has_constants and expr.
15073 (stmt_has_constants): Properly valueize SSA names when deciding
15074 whether the stmt has constants.
15076 2013-03-18 Richard Biener <rguenther@suse.de>
15078 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15079 whole function when there is nothing to do.
15080 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15081 * tree-vectorizer.c (vectorize_loops): Update virtual and
15082 loop-closed SSA once.
15083 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15085 2013-03-18 Richard Biener <rguenther@suse.de>
15087 PR middle-end/56113
15088 * domwalk.c (bb_postorder): New global static.
15089 (cmp_bb_postorder): New function.
15090 (walk_dominator_tree): Replace scheme imposing an order for
15091 visiting dominator sons by one sorting them at the time they
15092 are pushed on the stack.
15094 2013-03-18 Richard Biener <rguenther@suse.de>
15096 PR tree-optimization/39326
15097 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15098 (struct mem_ref): Replace mem member with ao_ref typed member.
15099 (MEM_ANALYZABLE): Adjust.
15100 (memref_eq): Likewise.
15101 (mem_ref_alloc): Likewise.
15102 (gather_mem_refs_stmt): Likewise.
15103 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15104 (execute_sm_if_changed_flag_set): Adjust.
15105 (execute_sm): Likewise.
15106 (ref_always_accessed_p): Likewise.
15107 (refs_independent_p): Likewise.
15108 (can_sm_ref_p): Likewise.
15110 2013-03-18 Jakub Jelinek <jakub@redhat.com>
15113 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15114 return 1 even for !unsignedp.
15116 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15118 * config/i386/i386.md (isa): Add x64 and nox64.
15119 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15120 (*pushtf): Enable *roF alternative for x64 isa only.
15121 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
15122 mode attribute of integer alternatives to DImode for TARGET_64BIT.
15123 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
15124 (*movtf_internal): Merge from *movtf_internal_rex64 and
15125 *movtf_internal_sse. Use x64 and nox64 isa attributes.
15126 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
15127 nox64 isa attributes.
15128 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
15129 nox64 isa attributes.
15130 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15132 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15134 * config/alpha/alpha.c (TARGET_LRA_P): New define.
15136 2013-03-17 Jakub Jelinek <jakub@redhat.com>
15139 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15140 class names. Remove trailing comma after "ALL_REGS".
15142 2013-03-16 Jan Hubicka <jh@suse.cz>
15144 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15145 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15146 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15147 of cgraph_get_create_node.
15148 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15150 2013-03-16 Jason Merrill <jason@redhat.com>
15153 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15154 with DW_AT_default_value.
15156 2013-03-16 Jakub Jelinek <jakub@redhat.com>
15158 * BASE-VER: Set to 4.9.0.
15160 2013-03-14 Andi Kleen <ak@linux.intel.com>
15163 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15164 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15165 Document _x* TSX intrinsics.
15167 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
15168 David Holsgrove <david.holsgrove@xilinx.com>
15170 * configure.ac: Add MicroBlaze TLS support detection.
15171 * configure: Regenerate.
15172 * config/microblaze/microblaze-protos.h
15173 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15174 symbol_mentioned_p, label_mentioned_p): Add prototypes.
15175 * config/microblaze/microblaze.c (microblaze_address_type): Add
15176 ADDRESS_TLS and tls_reloc address types.
15177 (microblaze_address_info): Add tls_reloc.
15178 (TARGET_HAVE_TLS): Define.
15179 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15180 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15181 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15182 load_tls_operand, microblaze_call_tls_get_addr,
15183 microblaze_legitimize_tls_address): New functions.
15184 (microblaze_classify_unspec): Handle UNSPEC_TLS.
15185 (get_base_reg): Use microblaze_tls_symbol_p.
15186 (microblaze_classify_address): Handle TLS.
15187 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15188 label_mentioned_p and microblaze_tls_referenced_p.
15189 (microblaze_legitimize_address): Handle TLS.
15190 (microblaze_address_insns): Handle ADDRESS_TLS.
15191 (pic_address_needs_scratch): Handle TLS.
15192 (print_operand_address): Handle TLS.
15193 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15194 (microblaze_expand_move): Handle TLS.
15195 (microblaze_legitimate_constant_p): Check
15196 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15197 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15198 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15199 (PIC_OFFSET_TABLE_REGNUM): Set.
15200 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15201 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15202 (addsi3, movsi_internal2, movdf_internal): Update constraints
15203 * config/microblaze/predicates.md (arith_plus_operand): Define
15204 (move_operand): Redefine as move_src_operand,
15205 check microblaze_tls_referenced_p.
15207 2013-03-14 Ian Bolton <ian.bolton@arm.com>
15209 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15210 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15212 2013-03-14 Ian Bolton <ian.bolton@arm.com>
15214 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15217 2013-03-14 Jakub Jelinek <jakub@redhat.com>
15219 PR tree-optimization/53265
15220 * common.opt (Waggressive-loop-optimizations): New option.
15221 * tree-ssa-loop-niter.c: Include tree-pass.h.
15222 (do_warn_aggressive_loop_optimizations): New function.
15223 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
15224 if number_of_latch_executions returned constant.
15225 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15226 early. If number_of_latch_executions returned constant, set
15227 nb_iterations_upper_bound back to it.
15228 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15230 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15231 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15233 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15234 (MULTILIB_OSDIRNAMES): Set.
15235 * genmultilib: If defaultosdirname doesn't start with :: , set
15236 defaultosdirname2 instead, clear it and emit two . multilib_raw
15237 entries instead of just one.
15239 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
15241 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15242 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15243 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15244 (SUBTARGET_OVERRIDE_OPTIONS): New.
15246 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
15249 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15250 (musermode): Convert to Var(TARGET_USERMODE).
15251 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15252 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15253 * config/sh/sh.c (sh_option_override): Use
15254 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15255 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15257 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15259 (udivsi3_i4_single, divsi3_i4_single): Use
15260 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15262 2013-03-13 Dave Korn <dave.korn.cygwin@....>
15264 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15267 2013-03-13 Richard Biener <rguenther@suse.de>
15269 PR tree-optimization/56608
15270 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15271 calls when vectorizing basic-blocks.
15273 2013-03-13 Jakub Jelinek <jakub@redhat.com>
15276 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15279 2013-03-12 Jakub Jelinek <jakub@redhat.com>
15281 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15283 2013-03-11 Jan Hubicka <jh@suse.cz>
15286 * lto-streamer-out.c (output_symbol_p): Skip references from
15287 constructors of external variables.
15289 2013-03-11 Jan Hubicka <jh@suse.cz>
15291 PR middle-end/56571
15292 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15294 * emit-rtl.c (verify_rtx_sharing): Likewise.
15295 (copy_insn_1): Likewise.
15296 * rtl.c (copy_rtx): Likewise.
15298 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
15301 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15302 output_operand_lossage message.
15304 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
15307 * arm.c (shift_op): Validate RTL pattern on the fly.
15308 (arm_print_operand, case 'S'): Don't use shift_operator to validate
15311 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
15314 * config/pa/pa.md (call_value): Check for calls to powf and direct to
15315 new call patterns that clobber %fr12.
15316 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15317 split and postreload patterns.
15318 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15319 registers %fr12 and %fr12R as call used.
15321 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
15323 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15324 (canon_address, record_store, replace_read, check_mem_read_rtx,
15325 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15326 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15327 rest_of_handle_dse): Likewise.
15329 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
15331 PR middle-end/56524
15332 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15334 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15335 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15336 (save_optabs_if_changed): Replace with...
15337 (init_tree_optimization_optabs): ...this.
15338 * optabs.c (save_optabs_if_changed): Rename to...
15339 (init_tree_optimization_optabs): ...this. Take the optimization node
15340 as argument. Do nothing if the base optabs are already correct.
15341 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15342 to recompute optabs.
15343 * function.h (function): Remove optabs field.
15344 * function.c (invoke_set_current_function_hook): Call
15345 init_tree_optimization_optabs. Use the result to initialize
15348 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
15350 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15351 if GTMA_HAS_NO_INSTRUMENTATION.
15352 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15353 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15354 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15355 * gimple-pretty-print.c (dump_gimple_transaction): Handle
15356 GTMA_HAS_NO_INSTRUMENTATION.
15358 2013-03-08 Jakub Jelinek <jakub@redhat.com>
15360 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15363 2013-03-08 Marek Polacek <polacek@redhat.com>
15364 Jakub Jelinek <jakub@redhat.com>
15366 PR tree-optimization/56478
15367 * predict.c (is_comparison_with_loop_invariant_p): Change the
15368 type of loop_step to tree.
15369 (predict_loops): Adjust.
15370 (predict_iv_comparison): Perform the computations on double_ints.
15372 2013-03-08 Richard Biener <rguenther@suse.de>
15374 PR tree-optimization/56570
15375 * tree-cfg.c (verify_expr_location_1): Verify locations for
15377 * tree-sra.c (create_access_replacement): Strip locations
15378 from DECL_DEBUG_EXPRs.
15380 2013-03-08 Richard Biener <rguenther@suse.de>
15382 * tree-inline.c (expand_call_inline): Do not associate
15383 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15384 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15386 2013-03-08 Richard Biener <rguenther@suse.de>
15388 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15389 or block changes with -Og. Fix for location / block encoding
15390 changes and PHI arguments with locations.
15392 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
15394 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15396 (struct output_info): Likewise.
15397 (register_overhead): Remove bad gcc_assert.
15398 (bitmap_find_bit): If there is only a single bitmap element, do not
15399 count a miss as a search.
15400 (print_statistics): Update for counter type changes.
15401 (dump_bitmap_statistics): Likewise. Print headers such that they
15402 are properly lined up with the printed counters.
15404 2013-03-07 Jakub Jelinek <jakub@redhat.com>
15406 PR tree-optimization/56559
15407 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15408 check that it has only a single use.
15410 2013-03-07 Richard Biener <rguenther@suse.de>
15412 * doc/invoke.texi (fwhole-program): Discourage use in combination
15415 2013-03-06 Jakub Jelinek <jakub@redhat.com>
15417 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15419 PR tree-optimization/56539
15420 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15421 instead of GSI_CONTINUE_LINKING as last argument to
15422 force_gimple_operand_gsi. Adjust function comment.
15424 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15427 PR middle-end/56548
15428 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15429 promoted mode, convert the result back to the original mode.
15431 2013-03-06 Richard Biener <rguenther@suse.de>
15433 PR middle-end/56294
15434 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15435 (insert_updated_phi_nodes_compare_uids): New function.
15436 (update_ssa): Sort symbols_to_rename after UID before
15437 traversing it to insert PHI nodes.
15439 2013-03-06 Richard Biener <rguenther@suse.de>
15441 PR middle-end/50494
15442 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15443 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15446 2013-02-13 Richard Biener <rguenther@suse.de>
15449 * varasm.c (output_constant_def_1): Get the decl representing
15450 the constant as argument.
15451 (output_constant_def): Wrap output_constant_def_1.
15452 (make_decl_rtl): Use output_constant_def_1 with the decl
15453 representing the constant.
15454 (build_constant_desc): Optionally re-use a decl already
15455 representing the constant.
15456 (tree_output_constant_def): Adjust.
15458 2013-03-06 Joey Ye <joey.ye@arm.com>
15461 * gcc.c (convert_white_space): New function.
15462 (main): Handles white space in function name.
15464 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
15467 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15468 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
15469 to SH_DIV_CALL_TABLE for TARGET_SH2.
15470 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15472 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15473 call-table options.
15475 2013-03-05 Sterling Augustine <saugustine@google.com>
15476 Cary Coutant <ccoutant@google.com>
15479 * dwarf2out.c (resolve_addr): Don't call
15480 remove_loc_list_addr_table_entries a second time for the same
15483 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15486 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15487 (avoid_complex_debug_insns): New function.
15488 (expand_debug_locations): Call it.
15490 PR rtl-optimization/56484
15491 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15492 lifetimes of hard registers on small register class machines.
15494 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
15496 * config/microblaze/microblaze-protos.h: Rename
15497 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15498 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15500 (microblaze_fast_interrupt_function_p): New function.
15501 (microblaze_is_interrupt_handler): Rename to
15502 microblaze_is_interrupt_variant and add fast_interrupt check.
15503 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15504 (save_restore_insns): Likewise.
15505 (compute_frame_size): Likewise.
15506 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15507 (microblaze_globalize_label): Likewise.
15508 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15509 * config/microblaze/microblaze.md: Use wrapper
15510 microblaze_is_interrupt_variant.
15512 2013-03-05 Kai Tietz <ktietz@redhat.com>
15514 * sdbout.c (sdbout_one_type): Switch to current function's section
15515 supporting cold/hot.
15517 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
15519 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15522 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15524 PR middle-end/56461
15525 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15526 if VALGRIND_GET_VBITS is defined, temporarily make object
15527 memory all defined, and restore previous valgrind addressability
15528 and definability afterwards. Free this_object at the end.
15530 PR middle-end/56461
15531 * lra.c (lra): Call lra_clear_live_ranges if live_p,
15532 right before calling lra_create_live_ranges, also call it
15533 when clearing live_p. Only call lra_clear_live_ranges
15534 at the end if live_p.
15536 PR middle-end/56461
15537 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15539 2013-03-05 Richard Biener <rguenther@suse.de>
15541 PR tree-optimization/56521
15542 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15545 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
15548 * except.h (remove_unreachable_eh_regions): New prototype.
15549 * except.c (remove_eh_handler_splicer): New function, split out
15550 of remove_eh_handler.
15551 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
15552 warning about running it on many EH regions one at a time.
15553 (remove_unreachable_eh_regions_worker): New function, walk the
15554 EH tree in depth-first order and remove non-marked regions.
15555 (remove_unreachable_eh_regions): New function.
15556 * tree-eh.c (mark_reachable_handlers): New function, split out
15557 from remove_unreachable_handlers.
15558 (remove_unreachable_handlers): Use mark_reachable_handlers and
15559 remove_unreachable_eh_regions.
15560 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15561 and remove_unreachable_eh_regions.
15563 2013-03-05 Richard Biener <rguenther@suse.de>
15565 PR middle-end/56525
15566 * loop-init.c (fix_loop_structure): Remove loops in two stages,
15567 not freeing them until the end.
15569 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
15571 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15573 2013-03-05 Richard Biener <rguenther@suse.de>
15575 PR tree-optimization/56270
15576 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15577 of loads after scheduling an SLP instance.
15579 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15581 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15583 (check_gcc_parallelize): Run guality.exp as a separate job from
15584 vect.exp with unsorted.exp and $(dg_target_exps) separately from
15585 struct-layout-1.exp with stackalign.exp.
15587 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15589 PR middle-end/56461
15590 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15591 load_index sbitmap even if some bit in it isn't set.
15593 PR middle-end/56461
15594 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
15595 (discover_iteration_bound_by_body_walk): Change queues to
15596 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
15597 spelling in comment. Call safe_push on queues[bound_index] directly.
15598 Release queues[queue_index] in every iteration unconditionally.
15599 Release bounds vector.
15601 PR middle-end/56461
15602 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
15603 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
15604 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
15607 2013-03-05 Richard Biener <rguenther@suse.de>
15610 * tree-inline.c (remap_blocks_to_null): New function.
15611 (expand_call_inline): When expanding a call stmt without
15612 an associated block inline remap all callee blocks to NULL.
15614 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15616 PR rtl-optimization/56494
15617 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
15618 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
15619 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
15621 PR middle-end/56461
15622 * sel-sched-ir.c (free_sched_pools): Release
15623 succs_info_pool.stack[succs_info_pool.max_top] vectors too
15624 if succs_info_pool.max_top isn't -1.
15627 * opts.c (opts_obstack, opts_concat): Moved to...
15628 * opts-common.c (opts_obstack, opts_concat): ... here.
15630 2013-03-04 Jakub Jelinek <jakub@redhat.com>
15632 PR middle-end/56461
15633 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
15635 2013-03-04 Martin Jambor <mjambor@suse.cz>
15637 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
15638 all appropriate places.
15640 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
15642 PR tree-optimization/56424
15643 * ipa-split.c (split_function): Do not set the RSO flag if result is
15644 not by reference and its type is a register type.
15646 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
15648 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
15649 (microblaze_legitimate_pic_operand): Likewise
15650 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
15651 new function microblaze_legitimate_pic_operand
15652 * config/microblaze/microblaze-protos.h
15653 (microblaze_legitimate_pic_operand): Declare.
15655 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15657 * config/microblaze/predicates.md (call_insn_simple_operand):
15658 New predicate for supported rtx code types.
15659 * config/microblaze/microblaze.md (call_internal1): Use
15660 call_insn_simple_operand predicate.
15662 2013-03-04 Jakub Jelinek <jakub@redhat.com>
15664 PR middle-end/56461
15665 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
15666 partitions.ordered_remove.
15668 PR middle-end/56461
15669 * tree-vect-stmts.c (vectorizable_conversion): Don't call
15670 vec_oprnds0.create (1) for modifier == NONE.
15672 PR middle-end/56461
15673 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
15674 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
15675 vec_oprnds1 right before pushing anything to it for
15678 PR middle-end/56461
15679 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
15680 set nbbs to 0 instead of having separate code path.
15681 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
15682 instead of false as last argument if returning NULL.
15684 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
15686 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
15687 the attribute is now called "target" instead of "option".
15688 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
15689 * doc/tm.texi.in (Target Attributes): Likewise document the correct
15690 attribute/pragma name for TARGET_OPTION_VALID_P and
15691 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
15692 * doc/tm.texi: Regenerated.
15694 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
15696 * config/microblaze/microblaze.c:
15697 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
15698 * config/microblaze/microblaze.h: Add -mxl-reorder to
15700 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
15701 instructions emitted if TARGET_REORDER.
15702 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
15703 or 0 for -m/-mno case, but initialises as 2 to detect default use case
15706 2013-03-01 Xinliang David Li <davidxl@google.com>
15708 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
15711 2013-03-01 Jakub Jelinek <jakub@redhat.com>
15713 PR middle-end/56461
15714 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
15715 vector even when returning true. Fix up function comment formatting.
15717 PR middle-end/56461
15718 * ira-build.c (ira_loop_nodes_count): New variable.
15719 (create_loop_tree_nodes): Initialize it.
15720 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
15722 PR middle-end/56461
15723 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
15724 method on dr_chain and result_chain.
15725 * tree-vect-stmts.c (vectorizable_store): Only call
15726 result_chain.create if j == 0.
15728 PR middle-end/56461
15729 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
15730 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
15731 before overwriting it.
15733 2013-03-01 Tobias Burnus <burnus@net-b.de>
15735 * doc/extended.texi (C Extensions): Change order in @menu
15737 (Other MIPS Built-in Functions): Move last MIPS entry before
15738 "picoChip Built-in Functions".
15739 (SH Built-in Functions): Move after RX Built-in Functions.
15740 * doc/gcc.texi (Introduction): Change order in @menu
15742 * doc/md.texi (Constraints): Ditto.
15743 * gty.texi (Type Information): Ditto.
15744 (User-provided marking routines for template types): Make
15746 * doc/invoke.texi (AArch64 Options): Move before
15747 "Adapteva Epiphany Options".
15749 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
15750 Jakub Jelinek <jakub@redhat.com>
15753 * asan.c (gate_asan): Lookup no_sanitize_address instead of
15754 no_address_safety_analysis attribute.
15755 * doc/extend.texi (no_address_safety_attribute): Rename to
15756 no_sanitize_address attribute, mention no_address_safety_analysis
15757 attribute as deprecated alias.
15759 2013-02-28 Jakub Jelinek <jakub@redhat.com>
15761 PR middle-end/56461
15762 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
15763 type to vec<vec<tree> > *.
15764 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
15765 to be vec<tree> instead of vec<tree> *, set vec_defs
15766 to vNULL and call vec_defs.create (number_of_vects), adjust other
15768 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
15769 vectorizable_condition): Adjust vect_get_slp_defs callers.
15771 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
15773 * config/aarch64/aarch64.c
15774 (aarch64_float_const_representable): Remove unused variable.
15776 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
15778 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
15780 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
15782 * config/aarch64/aarch64-builtins.c
15783 (aarch64_init_simd_builtins): Make static.
15785 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
15787 * config/aarch64/aarch64.c
15788 (aarch64_simd_make_constant): Make static.
15790 2013-02-28 Martin Jambor <mjambor@suse.cz>
15792 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
15793 with no initialization to the RHS of debug statements.
15795 2013-02-28 Martin Jambor <mjambor@suse.cz>
15797 PR tree-optimization/56294
15798 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
15800 (get_access_replacement): Do not call create_access_replacement.
15801 Assert a replacement exists.
15802 (get_repl_default_def_ssa_name): Create the replacement declaration
15805 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15807 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
15808 final_end_function.
15810 2013-02-28 Marek Polacek <polacek@redhat.com>
15812 PR rtl-optimization/56466
15813 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
15814 if we're changing a loop.
15815 (peel_loops_completely): Likewise.
15817 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
15820 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15822 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
15825 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
15826 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
15827 INTX_FTYPE_FX, FX_FTYPE_INTX.
15828 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
15830 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
15832 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
15833 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
15834 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15835 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15836 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15837 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15838 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15839 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15840 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15841 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15842 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15843 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15844 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15845 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15846 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15847 (avrxmega6): Increase max flash segments from 5 to 6.
15848 * config/avr/t-multilib: Regenerate.
15849 * config/avr/avr-tables.opt: Regenerate.
15850 * doc/avr-mmcu.texi: Regenerate.
15852 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
15854 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
15855 (avr_device_to_arch): Rename to avr_device_to_ld.
15856 (avr_device_to_as): New prototype.
15857 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
15858 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
15859 * config/avr/driver-avr.c (avr_device_to_as): New.
15860 (avr_device_to_arch): Rename to avr_device_to_ld.
15862 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15864 PR middle-end/56461
15865 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
15866 method on dr_chain and result_chain.
15868 PR middle-end/56461
15869 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
15870 pointer_set_destroy on not_executed_last_iteration.
15872 PR middle-end/56461
15873 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
15875 PR middle-end/56461
15876 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
15877 FOR_EACH_DEFINED_FUNCTION when freeing state.
15879 PR middle-end/56461
15880 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
15882 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
15885 PR middle-end/56461
15886 * ipa-cp.c (decide_whether_version_node): Call vec_free on
15887 known_aggs[i].items and release known_aggs vector.
15889 PR middle-end/56461
15890 * ipa-reference.c (propagate): Free node_info even for alias nodes.
15892 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
15894 * config/microblaze/microblaze.c (microblaze_emit_compare):
15895 Use xor for EQ/NE comparisions.
15896 * config/microblaze/microblaze.md (cstoresf4): Add constraints
15897 (cbranchsf4): Adjust operator to comparison_operator.
15899 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15901 PR middle-end/56461
15902 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
15904 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
15905 vec_safe_push, always update *slot.
15906 (redirect_edge_var_map_clear): Use vec_free.
15907 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
15908 (free_var_map_entry): Use vec_free.
15909 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
15910 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
15912 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
15914 PR middle-end/45472
15915 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
15916 when the may_trap_p bit of the exprs being merged differs.
15917 Reorder tests for speculativeness in the logical and operator.
15919 2013-02-27 Jakub Jelinek <jakub@redhat.com>
15921 * incpath.c (add_standard_paths): Use reconcat instead of concat
15922 where appropriate and avoid leaking memory.
15924 * opts.h: Include obstack.h.
15925 (opts_concat): New prototype.
15926 (opts_obstack): New declaration.
15927 * opts.c (opts_concat): New function.
15928 (opts_obstack): New variable.
15929 (init_options_struct): Call gcc_init_obstack on opts_obstack.
15930 (finish_options): Use opts_concat instead of concat
15931 and XOBNEWVEC instead of XNEWVEC.
15932 * opts-common.c (generate_canonical_option, decode_cmdline_option,
15933 generate_option): Likewise.
15934 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
15935 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
15938 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
15939 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
15941 2013-02-26 Jakub Jelinek <jakub@redhat.com>
15943 PR middle-end/56461
15944 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
15946 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
15948 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
15949 (arm_block_move_unaligned_straight): Likewise.
15950 (arm_adjust_block_mem): Likewise.
15952 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
15955 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
15956 temp, cond and label.
15957 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
15960 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
15961 * config/c6x/c6x.h (dbx_register_map): Update declaration.
15964 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
15965 of prologue/epilogue functions.
15968 * config/tilegx/tilegx.c (tilegx_expand_prologue):
15969 Remove unused variable cfa_offset.
15970 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
15973 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
15974 type promotion to unsigned.
15977 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
15978 for HOST_WIDE_INT of 32 bit / same size as int.
15979 (arm_block_move_unaligned_straight): Likewise.
15980 (arm_adjust_block_mem): Likewise.
15983 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
15986 2013-02-26 Marek Polacek <polacek@redhat.com>
15988 PR tree-optimization/56426
15989 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
15991 2013-02-26 Richard Biener <rguenther@suse.de>
15994 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
15995 unused variable loops.
15997 2013-02-26 Jakub Jelinek <jakub@redhat.com>
15999 PR tree-optimization/56448
16000 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16001 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16002 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16003 later operands of the references, or even first operand for
16004 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16006 PR tree-optimization/56443
16007 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16008 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16009 to type_for_mode langhook.
16011 2013-02-25 Matt Turner <mattst88@gmail.com>
16013 * doc/invoke.texi: Document r4700.
16015 2013-02-25 Richard Biener <rguenther@suse.de>
16017 PR tree-optimization/56175
16018 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16020 (simplify_bitwise_binary): ... here. Also guard the conversion
16021 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16023 2013-02-25 Catherine Moore <clm@codesourcery.com>
16026 2013-02-24 Catherine Moore <clm@codesourcery.com>
16027 Maciej W. Rozycki <macro@codesourcery.com>
16028 Tom de Vries <tom@codesourcery.com>
16029 Nathan Sidwell <nathan@codesourcery.com>
16030 Iain Sandoe <iain@codesourcery.com>
16031 Nathan Froyd <froydnj@codesourcery.com>
16032 Chao-ying Fu <fu@mips.com>
16034 * doc/extend.texi: (micromips, nomicromips, nocompression):
16035 Document new function attributes.
16036 * doc/invoke.texi (minterlink-compressed, mmicromips,
16037 m14k, m14ke, m14kec): Document new options.
16038 (minterlink-mips16): Update documentation.
16039 * doc/md.texi (ZC, ZD): Document new constraints.
16040 * configure.ac (gcc_cv_as_micromips): Check if linker
16041 supports the .set micromips directive.
16042 * configure: Regenerate.
16043 * config.in: Regenerate.
16044 * config/mips/mips-tables.opt: Regenerate.
16045 * config/mips/micromips.md: New file.
16046 * constraints.md (ZC, AD): New constraints.
16047 * config/mips/predicates.md (movep_src_register): New predicate.
16048 (movep_src_operand): New predicate.
16049 (non_volatile_mem_operand): New predicate.
16050 * config/mips/mips.md (multimem): New type.
16051 (length): Differentiate between 17-bit and 18-bit branch offsets.
16052 (MOVEP1, MOVEP2): New mode iterator.
16053 (mov_<load>l): Use ZC constraint.
16054 (mov_<load>r): Likewise.
16055 (mov_<store>l): Likewise.
16056 (mov_<store>r): Likewise.
16057 (*branch_equality<mode>_inverted): Add microMIPS support.
16058 (*branch_equality<mode>): Likewise.
16059 (*jump_absolute): Likewise.
16060 (indirect_jump_<mode>): Likewise.
16061 (tablejump_<mode>): Likewise.
16062 (<optab>_internal): Likewise.
16063 (sibcall_internal): Likewise.
16064 (sibcall_value_internal): Likewise.
16065 (prefetch): Use constraint ZD.
16066 * config/mips/mips.opt (minterlink-compressed): New option.
16067 (minterlink-mips16): Now an alias for minterlink-compressed.
16068 (mmicromips): New option.
16069 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16070 (compare_and_swap_12): Likewise.
16071 (sync_add<mode>): Likewise.
16072 (sync_<optab>_12): Likewise.
16073 (sync_old_<optab>_12): Likewise.
16074 (sync_new_<optab>_12): Likewise.
16075 (sync_nand_12): Likewise.
16076 (sync_old_nand_12): Likewise.
16077 (sync_new_nand_12): Likewise.
16078 (sync_sub<mode>): Likewise.
16079 (sync_old_add<mode>): Likewise.
16080 (sync_old_sub<mode>): Likewise.
16081 (sync_new_add<mode>): Likewise.
16082 (sync_new_sub<mode>): Likewise.
16083 (sync_<optab><mode>): Likewise.
16084 (sync_old_<optab><mode>): Likewise.
16085 (sync_new_<optab><mode>): Likewise.
16086 (sync_nand<mode>): Likewise.
16087 (sync_old_nand<mode>): Likewise.
16088 (sync_new_nand<mode>): Likewise.
16089 (sync_lock_test_and_set<mode>): Likewise.
16090 (test_and_set_12): Likewise.
16091 (atomic_compare_and_swap<mode>): Likewise.
16092 (atomic_exchange<mode>_llsc): Likewise.
16093 (atomic_fetch_add<mode>_llsc): Likewise.
16094 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16095 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16096 (umips_save_restore_pattern_p): Likewise.
16097 (umips_load_store_pair_p): Likewise.
16098 (umips_output_load_store_pair): Likewise.
16099 (umips_movep_target_p): Likewise.
16100 (umips_12bit_offset_address_p): Likewise.
16101 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16102 (mips_base_mips16): Rename this...
16103 (mips_base_compression_flags): ...to this. Update all uses.
16104 (mips_attribute_table): Add micromips, nomicromips and nocompression.
16105 (mips_mips16_decl_p): Delete.
16106 (mips_nomips16_decl_p): Delete.
16107 (mips_get_compress_on_flags): New function.
16108 (mips_get_compress_off_flags): New function.
16109 (mips_get_compress_mode): New function.
16110 (mips_get_compress_on_name): New function.
16111 (mips_get_compress_off_name): New function.
16112 (mips_insert_attributes): Support multiple compression types.
16113 (mips_merge_decl_attributes): Likewise.
16114 (umips_12bit_offset_address_p): New function.
16115 (mips_start_function_definition): Emit .set micromips directive.
16116 (mips_call_may_need_jalx_p): New function.
16117 (mips_function_ok_for_sibcall): Add microMIPS support.
16118 (mips_print_operand_punctuation): Support short delay slots and
16120 (umips_swm_mask, umips_swm_encoding): New.
16121 (umips_build_save_restore): New function.
16122 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16123 (was_mips16_p): Remove.
16124 (old_compression_mode): New.
16125 (mips_set_compression_mode): New function.
16126 (mips_set_current_function): Add microMIPS support.
16127 (mips_option_override): Likewise.
16128 (umips_save_restore_pattern_p): New function.
16129 (umips_output_save_restore): New function.
16130 (umips_load_store_pair_p_1): New function.
16131 (umips_load_store_pair_p): New function.
16132 (umips_output_load_store_pair_1): New function.
16133 (umips_output_load_store_pair): New function.
16134 (umips_movep_target_p) New function.
16135 (mips_prepare_pch_save): Add microMIPS support.
16136 * config/mips/mips.h (TARGET_COMPRESSION): New.
16137 (TARGET_CPU_CPP_BUILTINS): Update macro
16138 to use new compression flags and to support microMIPS.
16139 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16140 (MIPS_ARCH_FLOAT_SPEC): Likewise.
16141 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16142 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16143 (ASM_SPEC): Support mmicromips and mno-micromips.
16144 (M16STORE_REG_P): New macro.
16145 (MIPS_CALL): Support TARGET_MICROMIPS.
16146 (MICROMIPS_J): New macro.
16147 (mips_base_mips16): Rename this...
16148 (mips_base_compression_flags): ...to this.
16149 (UMIPS_12BIT_OFFSET_P): New macro.
16150 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16151 (MULTILIB_DIRNAMES): Likewise.
16153 2013-02-25 Tom de Vries <tom@codesourcery.com>
16155 PR rtl-optimization/56131
16156 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16157 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16158 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16160 2013-02-25 Tobias Burnus <burnus@net-b.de>
16162 * doc/invoke.texi (-fsanitize=): Move from optimization
16163 to debugging options.
16165 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
16167 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16169 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
16170 Alexander Monakov <amonakov@ispras.ru>
16172 PR middle-end/56077
16173 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16174 flush pending lists also on non-jumps. Adjust comment.
16176 2013-02-24 Catherine Moore <clm@codesourcery.com>
16177 Maciej W. Rozycki <macro@codesourcery.com>
16178 Tom de Vries <tom@codesourcery.com>
16179 Nathan Sidwell <nathan@codesourcery.com>
16180 Iain Sandoe <iain@codesourcery.com>
16181 Nathan Froyd <froydnj@codesourcery.com>
16182 Chao-ying Fu <fu@mips.com>
16184 * doc/extend.texi: (micromips, nomicromips, nocompression):
16185 Document new function attributes.
16186 * doc/invoke.texi (minterlink-compressed, mmicromips,
16187 m14k, m14ke, m14kec): Document new options.
16188 (minterlink-mips16): Update documentation.
16189 * doc/md.texi (ZC, ZD): Document new constraints.
16190 * configure.ac (gcc_cv_as_micromips): Check if linker
16191 supports the .set micromips directive.
16192 * configure: Regenerate.
16193 * config.in: Regenerate.
16194 * config/mips/mips-tables.opt: Regenerate.
16195 * config/mips/micromips.md: New file.
16196 * constraints.md (ZC, AD): New constraints.
16197 * config/mips/predicates.md (movep_src_register): New predicate.
16198 (movep_src_operand): New predicate.
16199 (non_volatile_mem_operand): New predicate.
16200 * config/mips/mips.md (multimem): New type.
16201 (length): Differentiate between 17-bit and 18-bit branch offsets.
16202 (MOVEP1, MOVEP2): New mode iterator.
16203 (mov_<load>l): Use ZC constraint.
16204 (mov_<load>r): Likewise.
16205 (mov_<store>l): Likewise.
16206 (mov_<store>r): Likewise.
16207 (*branch_equality<mode>_inverted): Add microMIPS support.
16208 (*branch_equality<mode>): Likewise.
16209 (*jump_absolute): Likewise.
16210 (indirect_jump_<mode>): Likewise.
16211 (tablejump_<mode>): Likewise.
16212 (<optab>_internal): Likewise.
16213 (sibcall_internal): Likewise.
16214 (sibcall_value_internal): Likewise.
16215 (prefetch): Use constraint ZD.
16216 * config/mips/mips.opt (minterlink-compressed): New option.
16217 (minterlink-mips16): Now an alias for minterlink-compressed.
16218 (mmicromips): New option.
16219 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16220 (compare_and_swap_12): Likewise.
16221 (sync_add<mode>): Likewise.
16222 (sync_<optab>_12): Likewise.
16223 (sync_old_<optab>_12): Likewise.
16224 (sync_new_<optab>_12): Likewise.
16225 (sync_nand_12): Likewise.
16226 (sync_old_nand_12): Likewise.
16227 (sync_new_nand_12): Likewise.
16228 (sync_sub<mode>): Likewise.
16229 (sync_old_add<mode>): Likewise.
16230 (sync_old_sub<mode>): Likewise.
16231 (sync_new_add<mode>): Likewise.
16232 (sync_new_sub<mode>): Likewise.
16233 (sync_<optab><mode>): Likewise.
16234 (sync_old_<optab><mode>): Likewise.
16235 (sync_new_<optab><mode>): Likewise.
16236 (sync_nand<mode>): Likewise.
16237 (sync_old_nand<mode>): Likewise.
16238 (sync_new_nand<mode>): Likewise.
16239 (sync_lock_test_and_set<mode>): Likewise.
16240 (test_and_set_12): Likewise.
16241 (atomic_compare_and_swap<mode>): Likewise.
16242 (atomic_exchange<mode>_llsc): Likewise.
16243 (atomic_fetch_add<mode>_llsc): Likewise.
16244 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16245 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16246 (umips_save_restore_pattern_p): Likewise.
16247 (umips_load_store_pair_p): Likewise.
16248 (umips_output_load_store_pair): Likewise.
16249 (umips_movep_target_p): Likewise.
16250 (umips_12bit_offset_address_p): Likewise.
16251 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16252 (mips_base_mips16): Rename this...
16253 (mips_base_compression_flags): ...to this. Update all uses.
16254 (mips_attribute_table): Add micromips, nomicromips and nocompression.
16255 (mips_mips16_decl_p): Delete.
16256 (mips_nomips16_decl_p): Delete.
16257 (mips_get_compress_on_flags): New function.
16258 (mips_get_compress_off_flags): New function.
16259 (mips_get_compress_mode): New function.
16260 (mips_get_compress_on_name): New function.
16261 (mips_get_compress_off_name): New function.
16262 (mips_insert_attributes): Support multiple compression types.
16263 (mips_merge_decl_attributes): Likewise.
16264 (umips_12bit_offset_address_p): New function.
16265 (mips_start_function_definition): Emit .set micromips directive.
16266 (mips_call_may_need_jalx_p): New function.
16267 (mips_function_ok_for_sibcall): Add microMIPS support.
16268 (mips_print_operand_punctuation): Support short delay slots and
16270 (umips_swm_mask, umips_swm_encoding): New.
16271 (umips_build_save_restore): New function.
16272 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16273 (was_mips16_p): Remove.
16274 (old_compression_mode): New.
16275 (mips_set_compression_mode): New function.
16276 (mips_set_current_function): Add microMIPS support.
16277 (mips_option_override): Likewise.
16278 (umips_save_restore_pattern_p): New function.
16279 (umips_output_save_restore): New function.
16280 (umips_load_store_pair_p_1): New function.
16281 (umips_load_store_pair_p): New function.
16282 (umips_output_load_store_pair_1): New function.
16283 (umips_output_load_store_pair): New function.
16284 (umips_movep_target_p) New function.
16285 (mips_prepare_pch_save): Add microMIPS support.
16286 * config/mips/mips.h (TARGET_COMPRESSION): New.
16287 (TARGET_CPU_CPP_BUILTINS): Update macro
16288 to use new compression flags and to support microMIPS.
16289 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16290 (MIPS_ARCH_FLOAT_SPEC): Likewise.
16291 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16292 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16293 (ASM_SPEC): Support mmicromips and mno-micromips.
16294 (M16STORE_REG_P): New macro.
16295 (MIPS_CALL): Support TARGET_MICROMIPS.
16296 (MICROMIPS_J): New macro.
16297 (mips_base_mips16): Rename this...
16298 (mips_base_compression_flags): ...to this.
16299 (UMIPS_12BIT_OFFSET_P): New macro.
16300 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16301 (MULTILIB_DIRNAMES): Likewise.
16303 2013-02-24 Jakub Jelinek <jakub@redhat.com>
16306 * target-globals.c (save_target_globals): For init_reg_sets and
16307 target_reinit remporarily set this_fn_optabs to this_target_optabs.
16309 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
16311 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16312 * config/aarch64/t-aarch64
16313 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16315 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
16317 PR inline-asm/56148
16318 * lra-constraints.c (process_alt_operands): Reload operand
16319 conflicting with earlier clobber only if no more other conflicting
16322 2013-02-22 Jakub Jelinek <jakub@redhat.com>
16325 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16326 if not linking a shared library.
16328 2013-02-22 Seth LaForge <sethml@google.com>
16330 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16332 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
16334 * config/arm/arm.md (split for extendsidi): Update condition.
16335 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16336 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16337 (qhs_zextenddi_cstr): Likewise.
16339 2013-02-21 Jakub Jelinek <jakub@redhat.com>
16341 PR middle-end/56420
16342 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16343 avoid signed wrapping.
16344 (expand_mult): Handle properly multiplication by
16345 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
16346 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
16347 in the compiler if coeff is HOST_WIDE_INT_MIN.
16348 (expand_divmod): Don't make ext_op1 static, change it's type to
16349 uhwi. Avoid undefined behavior in -INTVAL (op1).
16351 PR rtl-optimization/50339
16352 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16354 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16355 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16356 into splitting_ashiftrt field.
16357 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16359 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16362 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
16364 PR middle-end/56108
16365 * trans-mem.c (execute_tm_mark): Do not expand transactions that
16366 are sure to go irrevocable.
16368 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
16370 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16371 scalars are valid operands.
16373 2013-02-21 Martin Jambor <mjambor@suse.cz>
16375 PR tree-optimization/56310
16376 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16377 only matching indices and non-negative final offsets.
16378 (intersect_aggregates_with_edge): Pass src_idx to
16379 agg_replacements_to_vector. Pass src_idx insstead of index to
16380 intersect_with_agg_replacements.
16382 2013-02-21 Martin Jambor <mjambor@suse.cz>
16384 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16385 instead of hard-wired defaults.
16387 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
16389 * doc/invoke.texi (MIPS Options): Update documentation of the
16390 floating-point multiply-accumulate instruction restrictions.
16392 2013-02-21 Kostya Serebryany <kcc@google.com>
16394 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16395 asan_shadow_offset on x86_64 linux.
16397 2013-02-21 Richard Biener <rguenther@suse.de>
16399 PR tree-optimization/56415
16401 2013-02-11 Richard Biener <rguenther@suse.de>
16403 PR tree-optimization/56273
16404 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16407 2013-02-21 Jakub Jelinek <jakub@redhat.com>
16410 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16413 PR inline-asm/56405
16414 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16415 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16417 2013-02-20 Jan Hubicka <jh@suse.cz>
16419 PR tree-optimization/56265
16420 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16421 when target is referenced for first time.
16423 2013-02-20 Richard Biener <rguenther@suse.de>
16425 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16426 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16427 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16428 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16429 not return anything.
16430 (rename_ssa_copies): Do not remove unused locals.
16431 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16432 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16433 * passes.c (execute_function_todo): Do not schedule unused locals
16434 removal if cleanup_tree_cfg did something.
16435 * tree-ssa-live.c (remove_unused_locals): Dump statistics
16436 about the number of removed locals.
16438 2013-02-20 Richard Biener <rguenther@suse.de>
16440 PR tree-optimization/56398
16441 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16443 2013-02-20 Martin Jambor <mjambor@suse.cz>
16445 PR tree-optimization/55334
16446 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16447 restricted pointers to arrays.
16449 2013-02-20 Richard Biener <rguenther@suse.de>
16450 Jakub Jelinek <jakub@redhat.com>
16452 PR tree-optimization/56396
16453 * tree-ssa-ccp.c (n_const_val): New static variable.
16454 (get_value): Return NULL for SSA names we don't have a lattice
16456 (ccp_initialize): Initialize n_const_val.
16457 * tree-ssa-copy.c (n_copy_of): New static variable.
16458 (init_copy_prop): Initialize n_copy_of.
16459 (get_value): Return NULL_TREE for SSA names we don't have a
16462 2013-02-20 Martin Jambor <mjambor@suse.cz>
16464 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16466 2013-02-20 Richard Biener <rguenther@suse.de>
16468 * genpreds.c (write_lookup_constraint): Do not compare first
16469 letter of the constraint again.
16471 2013-02-20 Richard Biener <rguenther@suse.de>
16473 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16475 (get_use_iv_cost): Terminate hashtable walk when coming across
16478 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
16480 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16481 reassociation for avx2 targets.
16483 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16485 * config/microblaze/microblaze.c: microblaze_has_clz = 0
16486 Add version check for v8.10.a to enable microblaze_has_clz
16487 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16488 version and TARGET_PATTERN_COMPARE check
16489 * config/microblaze/microblaze.md: New clzsi2 instruction
16491 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16493 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16494 function before branching.
16496 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
16498 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16500 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16502 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
16504 PR middle-end/55889
16505 * sel-sched.c: Include ira.h.
16506 (implicit_clobber_conflict_p): New function.
16507 (moveup_expr): Use it.
16508 * Makefile.in (sel-sched.o): Depend on ira.h.
16510 2013-02-19 Richard Biener <rguenther@suse.de>
16512 PR tree-optimization/56384
16513 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16514 (vn_hash_type): Split out from ...
16515 (vn_hash_constant_with_type): ... here.
16516 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16517 (vn_phi_eq): Compare types from vn_phi_s structure.
16518 (vn_phi_lookup): Populate vn_phi_s type.
16519 (vn_phi_insert): Likewise.
16521 2013-02-19 Jakub Jelinek <jakub@redhat.com>
16523 PR tree-optimization/56350
16524 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16525 if haven't found reduction or nested cycle operand, rather than
16526 asserting we must find it.
16528 PR tree-optimization/56381
16529 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16532 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
16533 Jakub Jelinek <jakub@redhat.com>
16536 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16537 (swap_optab_enable): Same.
16538 (init_all_optabs): Use argument instead of global.
16539 * tree.h (struct tree_optimization_option): New field target_optabs.
16540 * expr.h (init_all_optabs): Add argument to prototype.
16541 (TREE_OPTIMIZATION_OPTABS): New.
16542 (save_optabs_if_changed): Protoize.
16543 * optabs.h: Declare this_fn_optabs.
16544 * optabs.c (save_optabs_if_changed): New.
16545 Declare this_fn_optabs.
16546 (init_optabs): Add argument to init_all_optabs() call.
16547 * function.c (invoke_set_current_function_hook): Handle per
16549 * function.h (struct function): New field optabs.
16550 * config/mips/mips.c (mips_set_mips16_mode): Handle when
16551 optimization_current_node has changed.
16552 * target-globals.h (save_target_globals_default_opts): Protoize.
16553 * target-globals.c (save_target_globals_default_opts): New.
16555 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16558 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16559 registers %fr12 and %fr12R as call used.
16562 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16563 and HImode, require all displacements to be an integer multiple of
16565 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16566 only allow QImode and HImode when reload is in progress and strict is
16567 true. Likewise for symbolic addresses. Use base14_operand to check
16568 displacements in REG+BASE addresses.
16570 2013-02-18 Richard Biener <rguenther@suse.de>
16572 PR tree-optimization/56366
16573 * tree-vect-loop.c (get_initial_def_for_induction): Properly
16574 handle sign-conversion of outer-loop initial induction value.
16576 2013-02-18 Richard Biener <rguenther@suse.de>
16578 PR middle-end/56349
16579 * cfghooks.c (merge_blocks): If we merge a latch into another
16580 block adjust references to it.
16581 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16582 (verify_loop_structure): Verify that a recorded latch is in fact
16585 2013-02-18 Richard Biener <rguenther@suse.de>
16587 PR tree-optimization/56321
16588 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16589 order SSA name release and virtual operand unlinking.
16591 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16593 * config/microblaze/microblaze.md (save_stack_block): Define.
16594 (restore_stack_block): Likewise.
16596 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16598 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
16599 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
16600 * config/microblaze/microblaze.c (microblaze_option_override):
16601 Bail out early for PIC modes when target does not support PIC.
16603 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16605 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
16606 Replace with a microblaze version.
16607 (microblaze_trampoline_init): Adapt for microblaze.
16608 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
16611 2013-02-16 Jakub Jelinek <jakub@redhat.com>
16612 Dodji Seketeli <dodji@redhat.com>
16615 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
16616 (instrument_mem_region_access): Do not forget to always put
16617 instrumentation of the of 'base' and 'base + len' in a "if (len !=
16618 0) statement, even for cases where either 'base' or 'base + len'
16619 are not instrumented -- because they have been previously
16620 instrumented. Simplify the logic by putting all the statements
16621 instrument 'base + len' inside a sequence, and then insert that
16622 sequence right before the current insertion point. Then, to
16623 instrument 'base + len', just get an iterator on that statement.
16624 And do not forget to update the pointer to iterator the function
16625 received as argument.
16627 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
16629 PR rtl-optimization/56348
16630 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
16632 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
16634 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
16635 (clean_graph_dump_file): Pass base to start_graph_dump.
16637 2013-02-14 Richard Henderson <rth@redhat.com>
16640 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
16642 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
16644 * collect2-aix.h: Define F_LOADONLY.
16646 2013-02-14 Richard Biener <rguenther@suse.de>
16649 * varasm.c (output_constant_def_1): Get the decl representing
16650 the constant as argument.
16651 (output_constant_def): Wrap output_constant_def_1.
16652 (make_decl_rtl): Use output_constant_def_1 with the decl
16653 representing the constant.
16654 (build_constant_desc): Optionally re-use a decl already
16655 representing the constant.
16656 (tree_output_constant_def): Adjust.
16658 2013-02-14 Dodji Seketeli <dodji@redhat.com>
16661 * asan.c (instrument_builtin_call): Really put the length of the
16662 second source argument into src1_len.
16664 2013-02-13 Jakub Jelinek <jakub@redhat.com>
16666 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
16667 argument. If it is false, don't create edge from then_bb to
16669 (insert_if_then_before_iter): Pass true to it.
16670 (build_check_stmt): Pass false to it.
16671 (transform_statements): Flush hash table only on extended basic
16672 block boundaries, rather than at the beginning of every bb.
16673 Don't flush hash table on nonfreeing_call_p calls.
16674 * tree-flow.h (nonfreeing_call_p): New prototype.
16675 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
16677 2013-02-13 David S. Miller <davem@davemloft.net>
16679 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
16681 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
16684 * ira.c (max_regno_before_ira): Move from ...
16686 (fix_reg_equiv_init): Use max_regno_before_ira instead of
16689 2013-02-13 Jakub Jelinek <jakub@redhat.com>
16691 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
16693 2013-02-13 Richard Biener <rguenther@suse.de>
16696 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
16697 globals in MEM_REFs.
16699 2013-02-13 Richard Biener <rguenther@suse.de>
16701 * loop-init.c (loop_optimizer_init): Clear loop state when
16702 re-initializing preserved loops.
16703 * loop-unswitch.c (unswitch_single_loop): Return whether
16704 we unswitched the loop. Do not verify loop state here.
16705 (unswitch_loops): When we unswitched a loop discover new loops.
16707 2013-02-13 Kostya Serebryany <kcc@google.com>
16709 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
16711 * sanitizer.def: Rename __asan_init to __asan_init_v1.
16713 2013-02-12 Dodji Seketeli <dodji@redhat.com>
16715 Avoid instrumenting duplicated memory access in the same basic block
16716 * Makefile.in (asan.o): Add new dependency on hash-table.h
16717 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
16718 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
16719 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
16720 (free_mem_ref_resources, has_mem_ref_been_instrumented)
16721 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
16722 (get_mem_ref_of_assignment): New functions.
16723 (get_mem_refs_of_builtin_call): Extract from
16724 instrument_builtin_call and tweak a little bit to make it fit with
16726 (instrument_builtin_call): Use the new
16727 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
16728 of is_gimple_builtin_call.
16729 (instrument_derefs, instrument_mem_region_access): Insert the
16730 instrumented memory reference into the hash table.
16731 (maybe_instrument_assignment): Renamed instrument_assignment into
16732 this, and change it to advance the iterator when instrumentation
16733 actually happened and return true in that case. This makes it
16734 homogeneous with maybe_instrument_assignment, and thus give a
16735 chance to callers to be more 'regular'.
16736 (transform_statements): Clear the memory reference hash table
16737 whenever we enter a new BB, when we cross a function call, or when
16738 we are done transforming statements. Use
16739 maybe_instrument_assignment instead of instrumentation. No more
16740 need to special case maybe_instrument_assignment and advance the
16741 iterator after calling it; it's now handled just like
16742 maybe_instrument_call. Update comment.
16744 2013-02-13 Richard Biener <rguenther@suse.de>
16746 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
16747 Fix loop discovery code.
16749 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
16751 PR inline-asm/56148
16752 * lra-constraints.c (process_alt_operands): Match early clobber
16753 operand with itself. Check conflicts with earlyclobber only if
16754 the operand is not reloaded. Prefer to reload conflicting operand
16755 if earlyclobber and matching operands are the same.
16757 2013-02-12 Richard Biener <rguenther@suse.de>
16760 * lto-streamer-out.c (write_symbol): Do not output symbols
16761 for hard register variables.
16763 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
16766 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
16767 (umulsidi3_insn, mulsidi3_insn): New insns.
16769 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
16771 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
16772 (struct tune_params): Add vec_costs field.
16773 * config/arm/arm.c (arm_builtin_vectorization_cost)
16774 (arm_add_stmt_cost): New functions.
16775 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
16776 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
16777 (arm_default_vec_cost): New struct of type cpu_vec_costs.
16778 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
16779 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
16780 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
16781 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
16783 2013-02-12 Richard Biener <rguenther@suse.de>
16786 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
16787 decls again if possible.
16789 2013-02-12 Richard Biener <rguenther@suse.de>
16791 PR middle-end/56288
16792 * tree-ssa.c (verify_ssa_name): Fix check, move
16793 SSA_NAME_IN_FREE_LIST check up.
16795 2013-02-12 Jakub Jelinek <jakub@redhat.com>
16796 Steven Bosscher <steven@gcc.gnu.org>
16798 PR rtl-optimization/56151
16799 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
16800 equal to op0 or op1, and last_insn pattern is CODE operation
16801 with MEM dest and one of the operands matches that MEM.
16803 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
16805 * doc/extend.texi: Document Function Multiversioning and "default"
16806 parameter string to target attribute.
16807 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
16808 target attribute parameter is "default".
16809 (ix86_compare_version_priority): Remove checks for target attribute.
16810 (ix86_mangle_function_version_assembler_name): Change error to sorry.
16811 Remove check for target attribute equal to NULL. Add assert.
16812 (ix86_generate_version_dispatcher_body): Change error to sorry.
16814 2013-02-11 Iain Sandoe <iain@codesourcery.com>
16815 Jack Howarth <howarth@bromo.med.uc.edu>
16816 Patrick Marlier <patrick.marlier@gmail.com>
16819 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
16820 define ENDFILE_SPEC as TM_DESTRUCTOR.
16821 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
16823 2013-02-11 Alexander Potapenko <glider@google.com>
16824 Jack Howarth <howarth@bromo.med.uc.edu>
16825 Jakub Jelinek <jakub@redhat.com>
16828 * config/darwin.c (cdtor_record): Rename ctor_record.
16829 (sort_cdtor_records): Rename sort_ctor_records.
16830 (finalize_dtors): New routine to sort destructors by
16831 priority before use in assemble_integer.
16832 (machopic_asm_out_destructor): Use finalize_dtors if needed.
16834 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
16836 PR rtl-optimization/56275
16837 * simplify-rtx.c (avoid_constant_pool_reference): Check that
16838 offset is non-negative and less than cmode size before
16839 calling simplify_subreg.
16841 2013-02-11 Richard Biener <rguenther@suse.de>
16843 PR tree-optimization/56264
16844 * cfgloop.h (fix_loop_structure): Adjust prototype.
16845 * loop-init.c (fix_loop_structure): Return the number of
16846 newly discovered loops.
16847 * tree-cfgcleanup.c (repair_loop_structures): When new loops
16848 are discovered, do a full loop-closed SSA rewrite.
16850 2013-02-11 Richard Biener <rguenther@suse.de>
16852 PR tree-optimization/56273
16853 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16855 (check_array_ref): Fix missing newline in dumps.
16856 (search_for_addr_array): Likewise.
16858 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
16860 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
16862 2013-02-09 Jakub Jelinek <jakub@redhat.com>
16865 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
16867 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
16869 PR rtl-optimization/56246
16870 * lra-constraints.c (simplify_operand_subreg): Try to reuse
16872 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
16873 constraints are satisfied.
16875 2013-02-08 Jeff Law <law@redhat.com>
16878 * emit-rtl.c (reg_is_parm_p): New function.
16879 * regs.h (reg_is_parm_p): New prototype.
16880 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
16881 callee-clobbered registers.
16883 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
16886 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
16887 If there is no implicit builtin declaration, just return NULL.
16889 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16891 * config/i386/sse.md (FMAMODEM): New mode iterator.
16892 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
16893 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
16895 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16897 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
16898 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
16899 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
16901 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16903 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
16904 (microblaze*-*-elf): Likewise.
16905 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
16907 * config/microblaze/microblaze-c.c: Add builtin defines for
16908 _LITTLE_ENDIAN and _BIG_ENDIAN.
16909 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
16910 add to TARGET_DEFAULT flags.
16911 Expand ASM_SPEC and LINK_SPEC.
16912 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
16913 * config/microblaze/microblaze.md: Update extendsidi2 and
16914 movdi_internal instructions to use low-order / high-order reg
16916 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
16917 options and inversemask / mask of LITTLE_ENDIAN.
16918 * config/microblaze/t-microblaze: Expand multilib options to
16919 include mlittle-endian (le) and update exceptions patterns.
16921 2013-02-08 Jakub Jelinek <jakub@redhat.com>
16923 PR rtl-optimization/56195
16924 * lra-constraints.c (get_reload_reg): Don't reuse regs
16925 if they have smaller mode than requested, if they have
16926 wider mode than requested, try to return a SUBREG.
16928 PR tree-optimization/56250
16929 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
16930 if type is unsigned and code isn't MULT_EXPR.
16932 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
16934 PR tree-optimization/56064
16935 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
16936 bits according to mode.
16937 * fixed-value.h (fixed_from_double_int)
16938 (const_fixed_from_double_int): Adjust comments.
16940 2013-02-08 Richard Biener <rguenther@suse.de>
16943 * lto-streamer.h (struct data_in): Remove current_file, current_line
16944 and current_col members.
16945 * lto-streamer-out.c (lto_output_location): Stream changed bits
16946 en-block for efficiency.
16947 * lto-streamer-in.c (clear_line_info): Remove.
16948 (lto_input_location): Cache current file, line and column
16949 globally via local statics. Read changed bits en-block.
16950 (input_function): Do not call clear_line_info.
16951 (lto_read_body): Likewise.
16952 (lto_input_toplevel_asms): Likewise.
16954 2013-02-08 Michael Matz <matz@suse.de>
16956 PR tree-optimization/52448
16957 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
16958 (nt_call_phase): New static.
16959 (add_or_mark_expr): Only mark accesses with newer phase than any
16961 (nonfreeing_call_p): New.
16962 (nt_init_block): Update nt_call_phase, mark blocks as visited.
16963 (nt_fini_block): Keep blocks marked as visited.
16964 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
16966 2013-02-08 Richard Biener <rguenther@suse.de>
16968 * ira.c (ira): Free broken dominator information.
16970 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
16972 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
16974 2013-02-08 Marek Polacek <polacek@redhat.com>
16976 * cfgloop.c (verify_loop_structure): Add more checking of headers.
16978 2013-02-08 Richard Biener <rguenther@suse.de>
16980 PR middle-end/56181
16981 * cfgloop.h (flow_loops_find): Adjust.
16982 (bb_loop_header_p): Declare.
16983 * cfgloop.c (bb_loop_header_p): New function split out from ...
16984 (flow_loops_find): ... here. Adjust function signature,
16985 support incremental loop structure update.
16986 (verify_loop_structure): Cleanup. Verify a loop is a loop.
16987 * cfgloopmanip.c (fix_loop_structure): Move ...
16988 * loop-init.c (fix_loop_structure): ... here.
16989 (apply_loop_flags): Split out from ...
16990 (loop_optimizer_init): ... here.
16991 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
16992 in incremental mode, only remove dead loops here.
16994 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
16997 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
16998 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
16999 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17000 (*round<mode>3.libgcc): New insns for fixed-modes.
17001 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17002 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17003 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17004 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17005 implementations. Define to __builtin_avr_absFX,
17006 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17007 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17008 __builtin_avr_countlsFX, respectively.
17009 * config/avr/avr-c.c (target.h): Include it.
17010 (enum avr_builtin_id): New enum.
17011 (avr_resolve_overloaded_builtin): New static function.
17012 (avr_register_target_pragmas): Use it to set
17013 targetm.resolve_overloaded_builtin.
17014 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17015 tree nodes used by DEF_BUILTIN.
17016 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17017 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17018 <AVR_BUILTIN_xxBITS>: Same.
17020 2013-02-08 Richard Biener <rguenther@suse.de>
17022 * cfgloop.c (verify_loop_structure): Properly handle
17023 a loop exiting to another loop header.
17024 * ira-int.h (ira_loops): Remove.
17025 * ira.c (ira_loops): Remove.
17026 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17027 (do_reload): Use loop_optimizer_finalize.
17028 * ira-build.c (create_loop_tree_nodes): Use get_loops and
17029 number_of_loops to access the loop tree.
17030 (more_one_region_p): Likewise.
17031 (finish_loop_tree_nodes): Likewise.
17032 (rebuild_regno_allocno_maps): Likewise.
17033 (mark_loops_for_removal): Likewise.
17034 (mark_all_loops_for_removal): Likewise.
17035 (remove_unnecessary_regions): Likewise.
17036 (ira_build): Likewise.
17037 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17039 2013-02-08 Richard Biener <rguenther@suse.de>
17041 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17042 * ipa-pure-const.c (analyze_function): Avoid calling
17043 mark_irreducible_loops twice.
17044 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17046 2013-02-07 David S. Miller <davem@davemloft.net>
17048 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17050 * var-tracking.c (vt_add_function_parameter): Test the presence of
17051 HAVE_window_save properly and do not remap argument registers when
17052 we have a leaf function.
17054 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
17057 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17059 * config/i386/i386.c (ix86_print_operand): Ditto.
17061 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
17063 * lra-constraints.c (process_alt_operands): Fix recently added comment.
17065 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
17067 PR rtl-optimization/56225
17068 * lra-constraints.c (process_alt_operands): Check that reload hard
17069 reg can hold value for strict_low_part.
17071 2013-02-07 Jakub Jelinek <jakub@redhat.com>
17074 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17075 dwarf2out_end_function.
17076 (in_first_function_p, maybe_at_text_label_p,
17077 first_loclabel_num_not_at_text_label): New variables.
17078 (dwarf2out_var_location): In the first function find out
17079 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17080 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17083 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
17085 PR rtl-optimization/56178
17086 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17087 SUBREG of a register. Tidy up related block of code.
17088 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17089 note if the source is a register or a SUBREG of a register.
17091 2013-02-07 Jakub Jelinek <jakub@redhat.com>
17094 * config/rs6000/rs6000.md (ptrm): New mode attr.
17095 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17096 call_value_indirect_aix<pttrsize>,
17097 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17100 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
17102 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17103 if -bnortl. Convert to strcmp and strncmp.
17105 2013-02-07 Alan Modra <amodra@gmail.com>
17108 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17109 addresses won't wrap when offsetting.
17110 (rs6000_secondary_reload): Provide secondary reloads needed for
17111 wrapping LO_SUM addresses.
17113 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
17115 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17116 MACH, just __MACH__.
17118 2013-02-06 Richard Biener <rguenther@suse.de>
17120 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17121 instead of calling fix_loop_structure.
17123 2013-02-06 Jakub Jelinek <jakub@redhat.com>
17125 PR middle-end/56217
17126 * omp-low.c (use_pointer_for_field): Return false if
17127 lower_send_shared_vars doesn't generate any copy-out code.
17129 2013-02-06 Tom de Vries <tom@codesourcery.com>
17131 PR rtl-optimization/56131
17132 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17133 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
17134 of the label is NULL. Add comment.
17136 2013-02-05 Jakub Jelinek <jakub@redhat.com>
17138 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17141 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17142 (STATIC_LIBTSAN_LIBS): Likewise.
17143 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17144 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17145 is defined, don't add anything else beyond that.
17146 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17147 (LINK_COMMAND_SPEC): Use them.
17149 PR tree-optimization/56205
17150 * tree-stdarg.c (check_all_va_list_escapes): Return true if
17151 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17152 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17154 2013-02-05 Richard Biener <rguenther@suse.de>
17156 PR tree-optimization/53342
17157 PR tree-optimization/53185
17158 * tree-vectorizer.h (vect_check_strided_load): Remove.
17159 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17160 not disallow peeling for vectorized strided loads.
17161 (vect_check_strided_load): Make static and simplify.
17162 (vect_analyze_data_refs): Adjust.
17163 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17164 correctly when vectorizing strided loads.
17166 2013-02-05 Richard Biener <rguenther@suse.de>
17168 * doc/install.texi: Refer to ISL, not PPL.
17170 2013-02-05 Jan Hubicka <jh@suse.cz>
17172 PR tree-optimization/55789
17173 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17175 2013-02-05 Jan Hubicka <jh@suse.cz>
17177 PR tree-optimization/55789
17178 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17179 the dead call anyway.
17181 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
17184 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17186 2013-02-04 Alexander Potapenko <glider@google.com>
17187 Jack Howarth <howarth@bromo.med.uc.edu>
17188 Jakub Jelinek <jakub@redhat.com>
17191 * config/darwin.c (sort_ctor_records): Stabilized qsort
17192 on constructor priority by using original position.
17193 (finalize_ctors): New routine to sort constructors by
17194 priority before use in assemble_integer.
17195 (machopic_asm_out_constructor): Use finalize_ctors if needed.
17197 2013-02-04 Jakub Jelinek <jakub@redhat.com>
17200 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17201 about visibility on artificial decls.
17202 * config/sol2.c (solaris_assemble_visibility): Likewise.
17204 2013-02-04 Kai Tietz <ktietz@redhat.com>
17207 * config/i386/i386.c (function_value_ms_64): Add additional valtype
17208 argument and improve checking of return-argument types for 16-byte
17210 (ix86_function_value_1): Add additional valtype argument on call
17211 of function_value_64.
17212 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17213 handling infunction_value_64 function.
17215 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
17217 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17219 2013-02-04 Richard Biener <rguenther@suse.de>
17221 PR tree-optimization/56188
17222 * tree-ssa-structalias.c (label_visit): Consider case with
17223 initially non-empty points-to set.
17224 (perform_var_substitution): Dump node mapping and clean up.
17226 2013-02-04 Richard Guenther <rguenther@suse.de>
17229 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17230 node prevail as last resort.
17231 (lto_symtab_merge_decls): Remove guard on LTRANS here.
17232 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17234 2013-02-04 Richard Biener <rguenther@suse.de>
17236 PR tree-optimization/56113
17237 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17239 (equiv_class_lookup_or_add): ... this.
17240 (label_visit): Adjust and fix error in previous patch.
17241 (perform_var_substitution): Adjust.
17243 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
17245 * config/sh/divtab.c: Fix formatting and comments throughout the file.
17246 * config/sh/sh4-300.md: Likewise.
17247 * config/sh/sh4a.md: Likewise.
17248 * config/sh/constraints.md: Likewise.
17249 * config/sh/sh.md: Likewise.
17250 * config/sh/netbsd-elf.h: Likewise.
17251 * config/sh/predicates.md: Likewise.
17252 * config/sh/sh-protos.h: Likewise.
17253 * config/sh/ushmedia.h: Likewise.
17254 * config/sh/linux.h: Likewise.
17255 * config/sh/sh.c: Likewise.
17256 * config/sh/superh.h: Likewise.
17257 * config/sh/elf.h: Likewise.
17258 * config/sh/sh4.md: Likewise.
17259 * config/sh/sh.h: Likewise.
17261 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17263 * config/pa/constraints.md: Adjust unused letters. Change "T"
17264 constraint to match_test floating_point_store_memory_operand().
17265 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17266 (base14_operand): New.
17267 (floating_point_store_memory_operand): New.
17268 (integer_store_memory_operand): Revise to use base14_operand and
17269 reg_plus_base_memory_operand.
17270 (move_dest_operand): Allow symbolic_memory_operands.
17271 (symbolic_memory_operand): Check for LO_SOM.
17272 (symbolic_operand): Change default case to break.
17273 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17274 CONST_DOUBLE values to be reloaded by putting them into memory when
17275 the destination is a floating point register.
17276 (movdf): Remove code to handle CONST_DOUBLE.
17278 (reload_indf_r1): New.
17279 (reload_insf_r1): New.
17280 Consistently use "Q" and "T" constraints with integer and floating
17281 point move instructions, respectively.
17282 (movdi): Remove FAIL.
17283 Change predicate for source operand unamed DImode move from
17284 general_operand to move_src_operand.
17285 (umulsidi3): Change predicate for destination operand to
17287 Likewise for similar unamed patterns.
17288 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17289 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17290 (hppa_legitimize_address): Simplify mask calculation.
17291 (pa_emit_move_sequence): Revised handling of secondary reloads from
17292 REG+D addresses for floating point loads and stores. Directly handle
17293 loading CONST0_RTX (mode) to a floating point register.
17294 (pa_secondary_reload): Handle reloading DF and SFmode constant values
17295 to floating point registers. Don't restrict secondary reloads to
17296 floating point registers to integer modes. Revise some comments and
17298 (TARGET_LEGITIMATE_ADDRESS_P): Define.
17299 (pa_legitimate_address_p): New.
17300 (pa_legitimize_reload_address): New.
17301 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17302 (STRICT_REG_OK_FOR_BASE_P): New.
17303 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
17304 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17306 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
17307 Andrew Dixie <andrewd@gentrack.com>
17309 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17312 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
17314 * expmed.c (extract_bit_field_1): Pass the full width of the
17315 structure to get_best_reg_extraction_insn.
17317 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
17320 * configure.ac (use_cxa_atexit): Add AIX.
17321 * configure: Regenerate.
17323 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17325 2013-02-01 Jakub Jelinek <jakub@redhat.com>
17328 * final.c (need_profile_function): New variable.
17329 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17330 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17331 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17332 notes, targetm.asm_out.function_prologue doesn't emit anything,
17333 HAVE_prologue and profiler should be emitted before prologue,
17334 set need_profile_function instead of emitting it.
17335 (final_scan_insn): If need_profile_function, emit
17336 profile_function on the first NOTE_INSN_BASIC_BLOCK or
17337 NOTE_INSN_FUNCTION_BEG note.
17339 2013-02-01 Richard Henderson <rth@redhat.com>
17341 * config/rs6000/rs6000.md (smulditi3): New.
17344 * config/alpha/alpha.md (umulditi3): New.
17346 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
17348 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17349 (ASM_OUTPUT_ALIGNED_LOCAL): New.
17351 2013-02-01 Richard Biener <rguenther@suse.de>
17353 PR tree-optimization/56113
17354 * tree-ssa-structalias.c (label_visit): Reduce work for
17355 single-predecessor nodes.
17357 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
17359 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17360 range isn't testing for zero.
17362 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
17364 PR middle-end/56113
17365 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17367 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
17368 Nick Clifton <nickc@redhat.com>
17370 * config/v850/constraints.md (Q): Define as a memory constraint.
17371 * config/v850/predicates.md (label_ref_operand): New predicate.
17372 (e3v5_shift_operand): New predicate.
17373 (ior_operator): New predicate.
17374 * config/v850/t-v850: Add e3v5 multilib.
17375 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17376 (v850_gen_movdi): Prototype.
17377 * config/v850/v850.c: Add support for e3v5 architecture.
17378 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17380 (construct_save_jarl): Add e3v5 long JARL support.
17381 (v850_adjust_insn_length): New function. Adjust length of call
17382 insns when using e3v5 instructions.
17383 (v850_gen_movdi): New function: Generate instructions to move a
17385 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17386 (CPP_SPEC): Define __v850e3v5__ as appropriate.
17387 (TARGET_USE_FPU): Enable for e3v5.
17388 (CONST_OK_FOR_W): New macro.
17389 (ADJUST_INSN_LENGTH): Define.
17390 * config/v850/v850.md (UNSPEC_LOOP): Define.
17391 (attr cpu): Add v850e3v5.
17392 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17393 (movdi): New pattern.
17394 (movdi_internal): New pattern.
17395 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17396 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17397 (cstoresf4): Likewise.
17398 (cstoredf4): Likewise.
17399 (insv): New pattern.
17400 (rotlso3_a): New pattern.
17401 (rotlsi3_b): New pattern
17402 (rotlsi3_v850e3v5): New pattern.
17403 (doloop_begin): New pattern.
17404 (fix_loop_counter): New pattern.
17405 (doloop_end): New pattern.
17406 (branch_normal): Add e3v5 long branch support.
17407 (branch_invert): Likewise.
17408 (branch_z_normal): Likewise.
17409 (branch_z_invert): Likewise.
17410 (branch_nz_normal): Likewise.
17411 (branch_nz_invert): Likewise.
17412 (call_internal_short): Add e3v5 register-indirect JARL support.
17413 (call_internal_long): Likewise.
17414 (call_value_internal_short): Likewise.
17415 (call_value_internal_long): Likewise.
17416 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17417 (mloop): New option.
17418 * config.gcc: Add support for configuring v840e3v5 target.
17419 * doc/invoke.texi: Document new v850 specific command line options.
17421 2013-01-31 Paul Koning <ni1d@arrl.net>
17425 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17426 children if parent is a class.
17427 (prune_unused_types_prune): Don't add DW_AT_declaration.
17429 2013-01-31 Richard Biener <rguenther@suse.de>
17431 PR tree-optimization/56157
17432 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17433 match up operand with SLP child.
17435 2013-01-31 Jason Merrill <jason@redhat.com>
17438 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17439 parameters the first time.
17440 (gen_scheduled_generic_parms_dies): Check completeness here.
17442 2013-01-31 Richard Biener <rguenther@suse.de>
17444 PR middle-end/53073
17445 * common.opt (faggressive-loop-optimizations): New flag,
17446 enabled by default.
17447 * doc/invoke.texi (faggressive-loop-optimizations): Document.
17448 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17449 infer_loop_bounds_from_undefined by it.
17451 2013-01-31 Richard Biener <rguenther@suse.de>
17453 PR tree-optimization/56150
17454 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17455 visit virtual operands.
17456 (find_uses_to_rename_bb): Likewise.
17458 2013-01-31 Richard Biener <rguenther@suse.de>
17460 PR tree-optimization/56150
17461 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17462 mixed store non-store stmts.
17464 2013-01-30 Jakub Jelinek <jakub@redhat.com>
17467 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17468 LIBASAN_EARLY_SPEC is defined.
17469 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17470 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17472 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17475 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17476 invalid args instead of ICEing on it.
17477 (ix86_valid_target_attribute_tree): Return error_mark_node if
17478 ix86_valid_target_attribute_inner_p failed.
17479 (ix86_valid_target_attribute_p): Return false only if
17480 ix86_valid_target_attribute_tree returned error_mark_node. Allow
17481 target("default") attribute.
17482 (sorted_attr_string): Change argument from const char * to tree,
17483 merge in all target attribute arguments rather than just one.
17484 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
17485 instead of free. Avoid using strcat.
17486 (ix86_mangle_function_version_assembler_name): Mangle
17487 target("default") as if no target attribute is present. Adjust
17488 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
17489 instead of xmalloc and XDELETEVEC instead of free.
17490 (ix86_function_versions): Don't return true if one of the decls
17491 doesn't have target attribute. If they don't and one of the decls
17492 is DECL_FUNCTION_VERSIONED, report an error. Adjust
17493 sorted_attr_string caller. Use XDELETEVEC instead of free.
17494 (ix86_supports_function_versions): Remove.
17495 (make_name): Fix up formatting.
17496 (make_dispatcher_decl): Remove resolver_name and its initialization.
17497 Avoid leaking memory.
17498 (is_function_default_version): Return true if there is
17499 target("default") attribute rather than no target attribute at all.
17500 (make_resolver_func): Avoid leaking memory.
17501 (ix86_generate_version_dispatcher_body): Likewise.
17502 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17503 * target.def (supports_function_versions): Remove.
17504 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17505 * doc/tm.texi: Regenerated.
17507 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
17509 PR rtl-optimization/56144
17510 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17511 for values with side effects.
17513 2013-01-30 Richard Biener <rguenther@suse.de>
17515 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17516 (sparseset_pop): Likewise.
17517 * cfganal.c (compute_idf): Likewise. Increase work-stack size
17518 to be able to use quick_push in the worker loop.
17520 2013-01-30 Marek Polacek <polacek@redhat.com>
17522 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17524 2013-01-30 Richard Biener <rguenther@suse.de>
17527 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17529 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
17531 PR tree-optimization/56064
17532 * fixed-value.c (fixed_from_double_int): New function.
17533 * fixed-value.h (fixed_from_double_int): New prototype.
17534 (const_fixed_from_double_int): New static inline function.
17535 * fold-const.c (native_interpret_fixed): New static function.
17536 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17537 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17538 (native_encode_fixed): New static function.
17539 (native_encode_expr) <FIXED_CST>: Use it.
17540 (native_interpret_int): Move double_int worker code to...
17541 * double-int.c (double_int::from_buffer): ...this new static method.
17542 * double-int.h (double_int::from_buffer): Prototype it.
17544 2013-01-30 Richard Biener <rguenther@suse.de>
17546 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17547 New pointer-map and obstack.
17548 (init_alias_vars): Allocate pointer-map and obstack.
17549 (delete_points_to_sets): Free them.
17550 (find_what_var_points_to): Cache result.
17551 (find_what_p_points_to): Adjust for changed interface of
17552 find_what_var_points_to.
17553 (compute_points_to_sets): Likewise.
17554 (ipa_pta_execute): Likewise.
17556 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17558 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17559 * configure: Regenerate.
17560 * config.in: Regenerate.
17561 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17562 #nobits/#progbits if supported.
17564 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
17567 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17568 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17569 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17571 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17573 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17574 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17576 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17578 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17579 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17581 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17583 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17585 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17586 * config/arm/cortex-a7.md: New bypasses using
17587 arm_mac_accumulator_is_result.
17589 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17591 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
17592 (cortex_a7_neon_mla): Likewise.
17593 (cortex_a7_fpfmad): New reservation.
17594 (cortex_a7_fpmacs): Use ffmas and update required units.
17595 (cortex_a7_fpmuld): Update required units and latency.
17596 (cortex_a7_fpmacd): Likewise.
17597 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
17598 (cortex_a7_neon). Likewise.
17599 (bypass) Update participating units.
17601 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17603 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
17604 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
17606 * config/arm/vfp11.md (vfp_farith): Use ffmas.
17607 (vfp_fmul): Use ffmad.
17608 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
17609 (cortex_r4_fmacd): Use ffmad.
17610 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
17611 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
17612 (cortex_a9_fmacd): Use ffmad.
17613 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
17614 (cortex_a8_vfp_macd): Use ffmad.
17615 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
17616 (cortex_a5_fpmacd): Use ffmad.
17617 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
17618 (cortex_a15_vfp_macd): Use ffmad.
17619 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
17621 2013-01-29 Jason Merrill <jason@redhat.com>
17624 * varasm.c (default_assemble_visibility): Don't warn about
17625 visibility on artificial decls.
17627 2013-01-29 Richard Biener <rguenther@suse.de>
17629 PR tree-optimization/56113
17630 * tree-ssa-structalias.c (equiv_class_lookup): Also return
17632 (label_visit): Free duplicate bitmaps and record the leader instead.
17633 (perform_var_substitution): Adjust.
17635 2013-01-29 Richard Biener <rguenther@suse.de>
17637 PR tree-optimization/55270
17638 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
17639 the CFG, schedule loops for fixup.
17641 2013-01-29 Nick Clifton <nickc@redhat.com>
17643 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
17646 2013-01-28 Leif Ekblad <leif@rdos.net>
17648 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
17649 * config/i386/i386.h (TARGET_RDOS): New macro.
17650 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
17651 * config/i386/i386.c (ix86_option_override_internal): For 64bit
17652 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
17653 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
17654 DEFAULT_LARGE_SECTION_THRESHOLD.
17655 * config/i386/i386.md (R14_REG, R15_REG): New constants.
17656 * config/i386/rdos.h: New file.
17657 * config/i386/rdos64.h: New file.
17659 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
17662 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
17665 2013-01-28 Jakub Jelinek <jakub@redhat.com>
17667 PR rtl-optimization/56117
17668 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
17669 call cselib_lookup_from_insn on the MEM before calling
17670 add_insn_mem_dependence.
17672 2013-01-28 Richard Biener <rguenther@suse.de>
17674 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
17675 to a stmt that didn't have one.
17676 (copy_phis_for_bb): Likewise for PHI arguments.
17677 (copy_debug_stmt): Likewise for debug stmts.
17679 2013-01-28 Richard Biener <rguenther@suse.de>
17681 PR tree-optimization/56034
17682 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
17683 (partition_builtin_p): Adjust.
17684 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
17685 it is the last partition.
17686 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
17687 up the vertex for the definition.
17688 (classify_partition): Classify whether a partition is a
17689 PKIND_REDUCTION, thus has uses outside of the loop.
17690 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
17691 Merge all PKIND_REDUCTION partitions into the last partition.
17692 (tree_loop_distribution): Seed partitions from reductions as well.
17694 2013-01-28 Jakub Jelinek <jakub@redhat.com>
17696 PR tree-optimization/56125
17697 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
17698 pow(x,c) into sqrt(x) * powi(x, n/2) or
17699 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
17700 optimizing for size.
17701 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
17702 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
17705 PR tree-optimization/56094
17706 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
17707 to UNKNOWN_LOCATION while gimplifying expr.
17709 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
17712 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
17713 operand 0 in movabs insn template for -masm=intel asm alternative.
17714 (*movabs<mode>_2): Ditto for operand 1.
17716 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
17719 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
17722 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17724 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
17727 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
17729 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
17730 Undef to avoid warning.
17732 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
17734 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
17735 * configure: Regenerate.
17737 2013-01-25 Jakub Jelinek <jakub@redhat.com>
17739 PR tree-optimization/56098
17740 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
17741 for stmts with volatile ops.
17742 (cond_store_replacement): Don't optimize if assign has volatile ops.
17743 (cond_if_else_store_replacement_1): Don't optimize if either
17744 then_assign or else_assign have volatile ops.
17745 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
17748 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
17750 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
17752 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
17754 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
17755 missing ':' in asm example.
17757 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
17759 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
17760 entries into lane and laneq entries.
17761 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
17762 Remove AdvSIMD scalar modes.
17763 (aarch64_sq<r>dmulh_laneq<mode>): New.
17764 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
17766 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
17767 builtin implementations to relfect changes in RTL in aarch64-simd.md.
17768 * config/aarch64/iterators.md (VCOND): New.
17771 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
17774 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
17775 Add NULL LIBNAME argument to existing definitions.
17776 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
17777 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
17778 * config/avr/avr.c (DEF_BUILTIN): Same.
17779 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
17780 (avr_expand_builtin): Expand to a vanilla call if a libgcc
17781 implementation is available (DECL_ASSEMBLER_NAME is set).
17782 (avr_fold_absfx): New static function.
17783 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
17784 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
17785 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
17786 AVR_BUILTIN_ABSLLK.
17787 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
17788 (abshk, absk, abslk, absllk): Provide as static inline functions.
17790 2013-01-25 Marek Polacek <polacek@redhat.com>
17792 PR tree-optimization/56035
17793 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
17795 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
17797 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
17798 (*movtf_internal_rex64): Add (!o,C) alternative
17799 (*movxf_internal_rex64): Ditto.
17800 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
17802 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
17804 * doc/invoke.texi: fix typo.
17805 * doc/objc.texi: fix typo.
17807 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
17809 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
17810 for the first two alternatives.
17812 2013-01-24 Diego Novillo <dnovillo@google.com>
17814 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
17815 (ggc-zone.o): Remove.
17816 * configure.ac: Remove option --with-gc.
17817 * configure: Re-generate.
17818 * doc/install.texi: Remove documentation for --with-gc.
17819 * gengtype.c (write_enum_defn): Remove. Update all users.
17820 (write_Types_process_field): Remove generation of gt_e_* argument.
17821 (output_type_enum): Remove. Update all users.
17822 (write_enum_defn): Remove. Update all users.
17823 (enum alloc_zone): Remove. Update all users.
17824 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
17825 * ggc-common.c (ggc_splay_alloc): Remove first argument.
17826 Update all callers.
17827 (struct ptr_data): Remove field TYPE. Update all users.
17828 (gt_pch_note_object): Remove argument TYPE. Update all users.
17829 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
17831 * ggc-none.c (ggc_alloc_typed_stat): Remove.
17832 (struct alloc_zone): Remove.
17833 (ggc_internal_alloc_zone_stat): Remove.
17834 (ggc_internal_cleared_alloc_zone_stat): Remove.
17835 * ggc-page.c (ggc_alloc_typed_stat): Remove.
17836 (ggc_pch_count_object): Remove last argument. Update all users.
17837 (ggc_pch_alloc_object): Remove last argument. Update all users.
17838 (struct alloc_zone): Remove.
17839 * ggc-zone.c: Remove.
17840 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
17841 (struct alloc_zone): Remove.
17842 (ggc_alloc_typed_stat): Remove.
17843 (ggc_alloc_typed): Remove.
17844 (ggc_splay_alloc): Remove first argument.
17845 (rtl_zone): Remove. Update all users.
17846 (tree_zone): Remove. Update all users.
17847 (tree_id_zone): Remove. Update all users.
17848 (ggc_internal_zone_alloc_stat): Remove. Update all users.
17849 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
17850 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
17851 * tree-ssanames.c: Remove references to zone allocator in comments.
17853 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
17855 * config/avr/avr.c (avr_out_fract): Make register numbers that
17856 might be outside of source operand signed.
17858 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
17860 * config/i386/constraints.md (Yf): New constraint.
17861 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
17862 of f constraint to conditionaly disable x87 register preferences.
17863 (*movdf_internal): Ditto.
17864 (*movsf_internal): Ditto.
17866 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
17868 PR inline-asm/55934
17869 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
17870 that have operands with impossible constraints.
17871 Add a FIXME for a speed-up opportunity.
17872 * lra-constraints.c (process_alt_operands): Verify that a class
17873 selected from constraints on asms is valid for the operand mode.
17874 (curr_insn_transform): Remove incorrect comment.
17876 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
17878 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
17879 TOC operand is a valid symbol ref in the constant pool.
17881 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17883 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
17885 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
17888 * config/avr/stdfix.h: New file.
17889 * t-avr (stdfix-gcc.h): New rule to build it.
17890 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
17892 2013-01-23 Kostya Serebryany <kcc@google.com>
17894 * config/darwin.h: remove dependency on
17895 CoreFoundation (asan on Mac OS).
17897 2013-01-23 Jakub Jelinek <jakub@redhat.com>
17900 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
17901 instead of cmpdi_operand for first comparison operand.
17902 Don't assert that comparison operands aren't both constants.
17904 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
17906 * doc/install.texi (Downloading the Source): Update references to
17907 downloading separate components.
17909 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
17911 * doc/extend.texi (__int128): Improve grammar.
17913 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
17916 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
17917 alternative to (o,r).
17918 (*movdi_internal_rex64): Remove (!o,n) alternative.
17919 (DImode immediate->memory splitter): Remove.
17920 (DImode immediate->memory peephole2): Remove.
17921 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
17922 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
17923 alternative to (!o,*r).
17924 (*movtf_internal_sse): New pattern.
17925 (*movxf_internal_rex64): New pattern.
17926 (*movxf_internal): Disable for TARGET_64BIT.
17927 (*movdf_internal_rex64): Remove (!o,F) alternative.
17929 2013-01-22 Jakub Jelinek <jakub@redhat.com>
17931 PR middle-end/56074
17932 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
17933 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
17934 * tree-vect-loop-manip.c (find_loop_location): Also ignore
17935 stmt locations where LOCATION_LOCUS of the stmt location is
17936 UNKNOWN_LOCATION or BUILTINS_LOCATION.
17939 * config/i386/i386.md (UNSPEC_STOS): New.
17940 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
17941 *strsetqi_1): Add UNSPEC_STOS.
17943 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
17946 * doc/invoke.texi: Remove left over -Wsynth example.
17948 2013-01-21 Jakub Jelinek <jakub@redhat.com>
17950 PR tree-optimization/56051
17951 * fold-const.c (fold_binary_loc): Don't fold
17952 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
17953 a narrowing conversion, or widening conversion from signed
17956 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
17958 PR rtl-optimization/56023
17959 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
17960 dependent on debug instruction.
17962 2013-01-21 Martin Jambor <mjambor@suse.cz>
17964 PR middle-end/56022
17965 * function.c (allocate_struct_function): Call
17966 invoke_set_current_function_hook earlier.
17968 2013-01-21 Jakub Jelinek <jakub@redhat.com>
17970 * reload1.c (init_reload): Only initialize reload_obstack
17971 during the first call.
17973 2013-01-21 Marek Polacek <polacek@redhat.com>
17975 * cfgloop.c (verify_loop_structure): Fix up grammar.
17977 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
17979 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
17980 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
17982 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17985 * config/arm/marvell-pj4.md: Update copyright year.
17986 Fix up use of alu to alu_reg and simple_alu_imm.
17988 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
17990 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
17992 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
17995 * lra-constraints.c (curr_insn_transform): Don't reuse original
17996 insn for secondary memory move when memory mode should be different.
17998 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18000 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18001 atomic_storedi_1): New patterns.
18003 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
18005 btver2 pipeline descriptions.
18006 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18008 * config/i386/i386.md (btver2_decode): New type attributes.
18009 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18011 * config/i386/btver2.md: New file describing btver2 pipelines.
18013 2013-01-19 Andrew Pinski <apinski@cavium.com>
18015 PR tree-optimization/52631
18016 * tree-ssa-sccvn (visit_use): Before looking up the original
18017 statement, try looking up the simplified expression.
18019 2013-01-19 Anthony Green <green@moxielogic.com>
18021 * config/moxie/moxie.c (moxie_expand_prologue): Set
18022 current_function_static_stack_size.
18024 2013-01-18 Jakub Jelinek <jakub@redhat.com>
18026 PR tree-optimization/56029
18027 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18028 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18030 2013-01-18 Sharad Singhai <singhai@google.com>
18032 PR tree-optimization/55995
18033 * dumpfile.c (dump_loc): Print location only if available.
18034 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18036 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
18039 * lra-constraints.c (curr_insn_transform): Reuse original insn for
18040 secondary memory move.
18041 (inherit_reload_reg): Use rclass instead of cl for
18042 check_secondary_memory_needed_p.
18044 2013-01-18 Jakub Jelinek <jakub@redhat.com>
18046 PR middle-end/56015
18047 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18048 the case where writing real complex part of target modifies op1.
18050 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
18052 * config/aarch64/aarch64-simd.md
18053 (aarch64_vcond_internal<mode>): Handle unordered cases.
18054 * config/aarch64/iterators.md (v_cmp_result): New.
18056 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
18057 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18059 * config/arm/marvell-pj4.md: New file.
18060 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18061 * config/arm/arm.md (generic_sched): Add marvell_pj4.
18062 (generic_vfp): Likewise.
18063 * config/arm/arm-cores.def: Add marvell-pj4.
18064 * config/arm/arm-tune.md: Regenerate.
18065 * config/arm/arm-tables.opt: Regenerate.
18066 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18067 * doc/invoke.texi: Document marvell-pj4.
18069 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
18071 * config/aarch64/arm_neon.h: Map scalar types to standard types.
18073 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18078 * var-tracking.c (negative_power_of_two_p): New.
18079 (global_get_addr_cache, local_get_addr_cache): New.
18080 (get_addr_from_global_cache, get_addr_from_local_cache): New.
18081 (vt_canonicalize_addr): Rewrite using the above. Adjust the
18083 (vt_stack_offset_p): Remove.
18084 (vt_canon_true_dep): Always canonicalize loc's address.
18085 (clobber_overlapping_mems): Make sure we have a MEM.
18086 (local_get_addr_clear_given_value): New.
18087 (val_reset): Clear local cached entries.
18088 (compute_bb_dataflow): Create and release the local cache.
18089 Disable duplicate MEMs clobbering.
18090 (emit_notes_in_bb): Clobber MEMs likewise.
18091 (vt_emit_notes): Create and release the local cache.
18092 (vt_initialize, vt_finalize): Create and release the global
18093 cache, respectively.
18094 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18096 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18098 PR libmudflap/53359
18099 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18100 not found in the symtab.
18102 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18105 PR rtl-optimization/55547
18106 PR rtl-optimization/53827
18109 * alias.c (offset_overlap_p): New, factored out of...
18110 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
18111 the conservative special case for symbolic constants. Don't
18112 adjust zero sizes on alignment.
18114 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
18116 PR rtl-optimization/52573
18117 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18118 REG_UNUSED for the same register.
18120 2013-01-17 Richard Biener <rguenther@suse.de>
18121 Marek Polacek <polacek@redhat.com>
18123 PR rtl-optimization/55833
18124 * loop-unswitch.c (unswitch_loops): Move loop verification...
18125 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
18126 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18127 Set it to true when we're removing a loop from hierarchy tree in
18128 an irreducible region.
18129 (fix_bb_placements): Adjust caller.
18130 (fix_loop_placements): Likewise.
18132 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
18134 * config/avr/builtins.def (DEF_BUILTIN): Factor out
18135 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18136 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18137 Remove ID. Adjust comments.
18138 * config/avr/avr-c.c (avr_builtin_name): Remove.
18139 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18140 * config/avr/avr.c (avr_tolower): New static function.
18141 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
18142 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18143 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18146 2013-01-17 Jan Hubicka <jh@suse.cz>
18148 PR tree-optimization/55273
18149 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18151 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
18154 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18155 store through atomic_store<mode>_1.
18156 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18158 2013-01-17 Martin Jambor <mjambor@suse.cz>
18160 PR tree-optimizations/55264
18161 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18162 for virtual methods.
18163 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18164 virtual methods before inlining is over.
18165 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18167 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18170 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
18172 PR rtl-optimization/56005
18173 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18174 pending reads for prefetch.
18176 2013-01-16 Ian Bolton <ian.bolton@arm.com>
18178 * config/aarch64/aarch64.md
18179 (*cstoresi_neg_uxtw): New pattern.
18180 (*cmovsi_insn_uxtw): New pattern.
18181 (*<optab>si3_uxtw): New pattern.
18182 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18183 (*<optab>si3_insn_uxtw): New pattern.
18184 (*bswapsi2_uxtw): New pattern.
18186 2013-01-16 Richard Biener <rguenther@suse.de>
18188 * tree-inline.c (tree_function_versioning): Remove set but
18189 never used variable.
18191 2013-01-16 Richard Biener <rguenther@suse.de>
18193 PR tree-optimization/55964
18194 * tree-flow.h (rename_variables_in_loop): Remove.
18195 (rename_variables_in_bb): Likewise.
18196 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18197 (copy_loop_before): Adjust and delete update-ssa status.
18198 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18199 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
18200 (rename_variables_in_loop): Remove.
18201 (slpeel_update_phis_for_duplicate_loop): Likewise.
18202 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18203 use available cfg machinery instead of duplicating it.
18204 Update PHI nodes and perform poor-mans SSA update here.
18205 (slpeel_tree_peel_loop_to_edge): Adjust.
18207 2013-01-16 Richard Biener <rguenther@suse.de>
18209 PR tree-optimization/54767
18210 PR tree-optimization/53465
18211 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18212 (vrp_visit_phi_node): For PHI arguments coming via backedges
18213 drop all symbolical range information.
18214 (execute_vrp): Compute backedges.
18216 2013-01-16 Richard Biener <rguenther@suse.de>
18218 * doc/install.texi: Update CLooG and ISL requirements to
18221 2013-01-16 Christian Bruel <christian.bruel@st.com>
18224 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18225 (broken_move): Handle UNSPECV_SP_SWITCH_B.
18226 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18228 2013-01-16 DJ Delorie <dj@redhat.com>
18230 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18231 (UNSPECV_SP_SWITCH_E): New.
18232 (sp_switch_1): Change to an unspec.
18233 (sp_switch_2): Change to an unspec. Don't use post-inc when we
18236 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
18238 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18239 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18240 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18241 (expand_mem_thread_fence): Ditto.
18242 (expand_mem_signal_fence): Ditto.
18243 (expand_atomic_load): Ditto.
18244 (expand_atomic_store): Ditto.
18246 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
18248 PR rtl-optimization/55547
18249 PR rtl-optimization/53827
18252 * alias.c (memrefs_conflict_p): Set sizes to negative after
18255 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18258 * function.c (thread_prologue_and_epilogue_insns): Always
18259 add crtl->drap_reg to set_up_by_prologue.set, even if
18260 stack_realign_drap is false.
18262 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
18264 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18265 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18266 *call): Fix indention.
18268 2013-01-15 Tom de Vries <tom@codesourcery.com>
18271 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18274 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
18276 PR rtl-optimization/55153
18277 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18279 2013-01-15 Martin Jambor <mjambor@suse.cz>
18281 PR tree-optimization/55920
18282 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18283 accesses as grp_to_be_debug_replaced.
18285 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18287 PR tree-optimization/55920
18288 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18289 there is non-useless type conversion needed from debug rhs to lhs,
18290 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18292 2013-01-15 Joseph Myers <joseph@codesourcery.com>
18293 Mikael Pettersson <mikpe@it.uu.se>
18296 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18298 (ASM_OUTPUT_CASE_LABEL): Remove.
18299 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18300 * final.c (shorten_branches): Update alignment of labels before
18301 jump tables if CASE_VECTOR_SHORTEN_MODE.
18303 2013-01-15 Richard Biener <rguenther@suse.de>
18306 * system.h: Do not include gmp.h for building host tools.
18308 2013-01-15 Richard Biener <rguenther@suse.de>
18310 PR middle-end/55882
18311 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18312 account for bitpos when computing alignment.
18314 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
18316 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18317 (ix86_target_macros_internal): Likewise.
18319 * config/i386/i386.c (m_CORE2I7): Removed.
18320 (m_CORE_HASWELL): New macro.
18321 (m_CORE_ALL): Likewise.
18322 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18323 (initial_ix86_arch_features): Likewise.
18324 (processor_target_table): Initializations for Core avx2.
18325 (cpu_names): New names "core-avx2".
18326 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18327 PROCESSOR_CORE_HASWELL.
18328 (ix86_issue_rate): New case.
18329 (ia32_multipass_dfa_lookahead): Likewise.
18330 (ix86_sched_init_global): Likewise.
18332 * config/i386/i386.h (TARGET_HASWELL): New macro.
18333 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18334 (processor_type): New PROCESSOR_HASWELL.
18336 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18338 PR tree-optimization/55955
18339 * tree-vect-loop.c (vectorizable_reduction): Give up early on
18340 *SHIFT_EXPR and *ROTATE_EXPR codes.
18342 PR tree-optimization/48766
18343 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18344 -ftrapv disable -fwrapv.
18346 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18349 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18350 etc. to 1 and not to __flash.
18351 Use LL suffix for __INT24_MAX__ with -mint8.
18352 Use ULL suffix for __UINT24_MAX__ with -mint8.
18354 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18356 * config/avr/avr-arch.h
18357 (struct base_arch_s): Use typedef avr_arch_t instead.
18358 (struct arch_info_s): Use typedef avr_arch_info_t instead.
18359 (struct mcu_type_s): Use typedef avr_mcu_t instead.
18360 * config/avr/avr.c: Same.
18361 * config/avr/avr-devices.c: Same.
18362 * config/avr/driver-avr.c: Same.
18363 * config/avr/gen-avr-mmcu-texi.c: Same.
18364 * config/avr/avr-mcus.def: Adjust comment.
18366 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
18368 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18369 * config/aarch64/iterators.md (VALLDI): New.
18371 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
18372 Andi Kleen <ak@linux.intel.com>
18375 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18376 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18379 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18381 * config/avr/avr-stdint.h: Remove trailing blanks.
18382 * config/avr/avr-log.h: Same.
18383 * config/avr/avr-arch.h: Same.
18384 * config/avr/avr-devices.c: Same.
18385 * config/avr/avr-dimode.md: Same.
18386 * config/avr/predicates.md: Same.
18387 * config/avr/avr-c.c: Same. And fix typo.
18389 * config/avr/avr-protos.h: Same. And:
18390 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18391 (init_cumulative_args): Rename to avr_init_cumulative_args.
18392 (expand_prologue): Rename to avr_expand_prologue.
18393 (expand_epilogue): Rename to avr_expand_epilogue.
18394 (adjust_insn_length): Rename to avr_adjust_insn_length.
18395 (notice_update_cc): Rename to avr_notice_update_cc.
18396 (final_prescan_insn): Rename to avr_final_prescan_insn.
18397 * config/avr/avr.c: Same.
18398 * config/avr/avr.h: Same.
18399 * config/avr/avr.md: Remove trailing blanks.
18400 (prologue): Use avr_expand_prologue.
18401 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18403 2013-01-14 Richard Biener <rguenther@suse.de>
18405 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18406 verify_location, collect_subblocks): New functions.
18407 (verify_gimple_in_cfg): Verify that locations only reference
18408 BLOCKs in the functions BLOCK tree.
18410 2013-01-14 Richard Biener <rguenther@suse.de>
18412 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18414 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18416 (insert_out_of_ssa_copy_on_edge): Likewise.
18417 (rewrite_close_phi_out_of_ssa): Likewise.
18418 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18420 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18421 propagated constants.
18422 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18425 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18427 * config/avr/avr-modes.def: Add GPL copyright notice.
18429 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
18431 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18432 MEMMODEL_MASK to determine memory model.
18433 (atomic_store<mode>): Ditto from operands[2].
18434 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18436 2013-01-13 Jakub Jelinek <jakub@redhat.com>
18439 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18440 (fold_gimple_assign): Don't call unshare_expr here.
18441 (fold_ctor_reference): Call unshare_expr.
18443 2013-01-13 Terry Guo <terry.guo@arm.com>
18445 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18446 * doc/fragments.texi: Document MULTILIB_REUSE.
18447 * gcc.c (multilib_reuse): New internal spec.
18448 (set_multilib_dir): Also search multilib from multilib_reuse.
18449 * genmultilib (tmpmultilib3): Refactor code.
18450 (tmpmultilib4): Ditto.
18451 (multilib_reuse): New multilib argument.
18453 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
18455 * Makefile.in: Update copyright.
18457 2013-01-12 Tom de Vries <tom@codesourcery.com>
18459 PR middle-end/55890
18460 * calls.c (expand_call): Check if arg_nr is valid.
18462 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18464 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18465 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18466 documentation. Add missing '__' in front of
18467 __builtin_ia32_packssdw256.
18469 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
18472 * config/s390/s390.c (s390_preferred_reload_class): Do not return
18473 NO_REGS for larl operands.
18474 (s390_reload_larl_operand): Use s390_load_address instead of
18477 2013-01-11 Richard Biener <rguenther@suse.de>
18479 * tree-cfg.c (verify_node_sharing_1): Split out from ...
18480 (verify_node_sharing): ... here.
18481 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18483 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
18485 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18486 Substitute TREECHECKING.
18487 * configure: Regenerate.
18488 * Makefile.in (TREECHECKING): New.
18490 2013-01-11 Richard Guenther <rguenther@suse.de>
18492 PR tree-optimization/44061
18493 * tree-vrp.c (extract_range_basic): Compute zero as
18494 value-range for __builtin_constant_p of function parameters.
18496 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
18498 Update copyright years.
18500 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
18502 PR rtl-optimization/55672
18503 * lra-eliminations.c (mark_not_eliminable): Permit addition with
18504 const to be eliminable.
18506 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
18508 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18509 * configure: Regenerate.
18511 2013-01-10 Richard Biener <rguenther@suse.de>
18513 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18515 2013-01-10 Richard Biener <rguenther@suse.de>
18518 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18519 locations for virtual PHI arguments.
18520 (rewrite_update_phi_arguments): Likewise.
18522 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
18524 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18527 2013-01-10 Jakub Jelinek <jakub@redhat.com>
18529 PR tree-optimization/55921
18530 * tree-complex.c (expand_complex_asm): New function.
18531 (expand_complex_operations_1): Call it for GIMPLE_ASM.
18533 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
18536 * config/s390/s390.c (s390_symref_operand_p)
18537 (s390_loadrelative_operand_p): Merge the two functions.
18538 (s390_check_qrst_address, print_operand_address): Add parameters
18539 to s390_loadrelative_operand_p invokation.
18540 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18541 (s390_reload_larl_operand, s390_secondary_reload): Use
18542 s390_loadrelative_operand_p instead of s390_symref_operand_p.
18543 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18545 2013-01-09 Mike Stump <mikestump@comcast.net>
18547 * dse.c (record_store): Remove unnecessary assert.
18549 2013-01-09 Jan Hubicka <jh@suse.cz>
18551 PR tree-optimization/55569
18552 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18553 * cfgloop.h (scale_loop_profile): Likewise.
18555 2013-01-09 Jan Hubicka <jh@suse.cz>
18558 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18560 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18562 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
18564 PR middle-end/55114
18565 * expr.h (maybe_emit_group_store): Declare.
18566 * expr.c (maybe_emit_group_store): New function.
18567 * builtins.c (expand_builtin_int_roundingfn): Call it.
18568 (expand_builtin_int_roundingfn_2): Likewise.
18570 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
18572 PR rtl-optimization/55829
18573 * lra-constraints.c (match_reload): Add code for absent output.
18574 (curr_insn_transform): Add code for reloads of matched inputs
18577 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
18579 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18580 attribute of movddup insn to DF.
18581 (*vec_interleave_lowv2df): Ditto.
18582 (vec_dupv2df): Ditto.
18584 2013-01-09 Jan Hubicka <jh@suse.cz>
18586 PR tree-optimiation/55875
18587 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18588 EVERY_ITERATION parameter.
18589 (number_of_iterations_exit): Check if exit is executed every iteration.
18590 (idx_infer_loop_bounds): Similarly here.
18591 (n_of_executions_at_most): Simplify
18592 to only test for cases where statement is dominated by the
18593 particular bound; handle correctly the "postdominance" test.
18594 (scev_probably_wraps_p): Use max loop iterations info
18595 as a global bound first.
18597 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
18598 Nick Clifton <nickc@redhat.com>
18600 * config/v850/v850.md (cbranchsf4): New pattern.
18601 (cstoresf4): New pattern.
18602 (cbranchdf4): New pattern.
18603 (cstoredf4): New pattern.
18604 (movsicc): Disallow floating point comparisons.
18605 (cmpsf_le_insn): Fix order of operators.
18606 (cmpsf_lt_insn): Likewise.
18607 (cmpsf_eq_insn): Likewise.
18608 (cmpdf_le_insn): Likewise.
18609 (cmpdf_lt_insn): Likewise.
18610 (cmpdf_eq_insn): Likewise.
18611 (cmpsf_ge_insn): Use LE comparison.
18612 (cmpdf_ge_insn): Likewise.
18613 (cmpsf_gt_insn): Use LT comparison.
18614 (cmpdf_gt_insn): Likewise.
18615 (cmpsf_ne_insn): Delete pattern.
18616 (cmpdf_ne_insn): Delete pattern.
18617 * config/v850/v850.c (v850_gen_float_compare): Use
18618 gen_cmpdf_eq_insn for NE comparison.
18619 (v850_float_z_comparison_operator)
18620 (v850_float_nz_comparison_operator): Move from here ...
18621 * config/v850/predicates.md: ... to here. Move GT and GE
18622 comparisons into v850_float_z_comparison_operator.
18623 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
18625 (v850_float_nz_comparison_operator): Likewise.
18627 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18629 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
18630 with calls to gen_insvsi/gen_insvdi.
18632 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
18634 * config/i386/i386.c (initial_ix86_tune_features): Set up
18635 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
18637 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
18638 Jakub Jelinek <jakub@redhat.com>
18640 PR tree-optimization/48189
18641 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
18642 If nitercst is 0, don't predict the exit edge.
18644 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
18646 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
18647 in asm_fprintf with reg_names.
18648 (aarch64_print_operand_address): Likewise.
18649 (aarch64_return_addr): Likewise.
18650 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
18652 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18654 * config/pa/pa.h (VAL_U6_BITS_P): Define.
18655 (INT_U6_BITS): Likewise.
18656 * config/pa/predicates.md (uint6_operand): New predicate.
18657 (shift5_operand, shift6_operand): Likewise.
18658 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
18660 (lshrdi3): Use shift6_operand.
18661 (shrpsi4, shrpdi4): New insn patterns.
18662 (extzv): Delete expander.
18663 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
18664 predicates in unamed zero extract patterns. Tighten common constraint.
18665 (extv): Delete expander.
18666 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
18667 predicates in unamed sign extract patterns. Tighten common constraint.
18668 (insv): Delete expander.
18669 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
18670 predicates in unamed insert patterns. Tighten common constraint.
18671 Change uint32_operand predicate to uint6_operand predicate in unamed
18672 DImode pattern to insert constant values of type 1...1xxxx.
18674 2013-01-04 Jan Hubicka <jh@suse.cz>
18676 PR tree-optimization/55823
18677 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
18680 2013-01-08 Jakub Jelinek <jakub@redhat.com>
18681 Uros Bizjak <ubizjak@gmail.com>
18683 PR rtl-optimization/55845
18684 * df-problems.c (can_move_insns_across): Stop scanning at
18685 volatile_insn_p source instruction or give up if
18686 across_from .. across_to range contains any volatile_insn_p
18689 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
18691 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
18692 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
18694 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
18695 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
18697 2013-01-08 Jakub Jelinek <jakub@redhat.com>
18700 * asan.c (asan_clear_shadow): New function.
18701 (asan_emit_stack_protection): Use it.
18703 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
18705 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
18706 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
18707 with tab instead of space.
18709 2013-01-08 Nick Clifton <nickc@redhat.com>
18711 * config/rl78/rl78.c (rl78_expand_prologue): Always select
18712 register bank 0 at the start of an interrupt handler.
18713 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
18716 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
18718 * config/aarch64/aarch64-simd.md
18719 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
18720 (aarch64_simd_bsl): Likewise.
18721 (aarch64_vcond_internal<mode>): Likewise.
18722 (vcond<mode><mode>): Likewise.
18723 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
18724 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
18726 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
18728 * config/aarch64/aarch64-builtins.c
18729 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
18731 2013-01-08 Martin Jambor <mjambor@suse.cz>
18734 * tree-sra.c (analyze_access_subtree): Return true also after
18735 potentially creating a debug-only replacement.
18737 2013-01-08 Jakub Jelinek <jakub@redhat.com>
18739 PR middle-end/55890
18740 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
18742 PR tree-optimization/54120
18743 * tree-vrp.c (range_fits_type_p): Don't allow
18744 src_precision < precision from signed vr to unsigned_p
18745 if vr->min or vr->max is negative.
18746 (simplify_float_conversion_using_ranges): Test can_float_p
18747 against CODE_FOR_nothing.
18749 2013-01-08 Jakub Jelinek <jakub@redhat.com>
18750 Richard Biener <rguenther@suse.de>
18752 PR middle-end/55851
18753 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
18754 types instead of just INTEGER_TYPE types.
18756 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
18758 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
18761 2013-01-07 Steve Ellcey <sellcey@mips.com>
18764 * config/mips/mips.opt: Change mad to mmad to match documentation.
18766 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18769 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
18770 .progmemx.data now.
18772 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18775 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
18776 (avr_addrspace_t): Add .section_name field.
18777 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
18779 (avr_addrspace): Same. Initialize .section_name. Remove last
18780 NULL entry. Put __memx into .progmemx.data.
18781 (progmem_section_prefix): Remove.
18782 (avr_asm_init_sections): No need to initialize progmem_section.
18783 (avr_asm_named_section): Use avr_addrspace[].section_name to get
18784 section name prefix.
18785 (avr_asm_select_section): Ditto. And use get_unnamed_section to
18786 retrieve the progmem section.
18787 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
18788 boundary to run over avr_addrspace[].
18789 (avr_register_target_pragmas): Ditto.
18791 2013-01-06 Jakub Jelinek <jakub@redhat.com>
18793 * varasm.c (output_constant_def_contents): For asan_protect_global
18794 protected strings, adjust DECL_ALIGN if needed, before testing for
18796 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18797 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
18799 (output_object_block): For asan protected decls, emit asan padding
18800 after their contents.
18801 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
18802 (asan_finish_file): Test it here instead.
18804 2013-01-07 Nick Clifton <nickc@redhat.com>
18805 Matthias Klose <doko@debian.org>
18806 Doug Kwan <dougkwan@google.com>
18807 H.J. Lu <hongjiu.lu@intel.com>
18810 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
18812 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
18814 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
18816 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
18818 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
18820 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18823 * doc/install.texi (Cross-Compiler-Specific Options): Document
18826 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
18828 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
18829 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
18830 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
18831 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
18832 vqmovun_high_s64): Fix source operand number and update copyright.
18834 2013-01-07 Richard Biener <rguenther@suse.de>
18836 PR middle-end/55890
18837 * gimple.h (gimple_call_builtin_p): New overload.
18838 * gimple.c (validate_call): New function.
18839 (gimple_call_builtin_p): Likewise.
18840 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18841 Use gimple_call_builtin_p.
18842 (find_func_clobbers): Likewise.
18843 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18844 (strlen_optimize_stmt): Likewise.
18846 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
18848 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18849 (vld1q_dup_*): Likewise.
18850 (vld1_*): Likewise.
18851 (vld1q_*): Likewise.
18852 (vld1_lane_*): Likewise.
18853 (vld1q_lane_*): Likewise.
18855 2013-01-07 Richard Biener <rguenther@suse.de>
18857 * lto-streamer.h (LTO_minor_version): Bump to 2.
18859 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
18861 * config/aarch64/aarch64-protos.h
18862 (aarch64_const_double_zero_rtx_p): Rename to...
18863 (aarch64_float_const_zero_rtx_p): ...this.
18864 (aarch64_float_const_representable_p): New.
18865 (aarch64_output_simd_mov_immediate): Likewise.
18866 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
18867 move immediate case.
18868 * config/aarch64/aarch64.c
18869 (aarch64_const_double_zero_rtx_p): Rename to...
18870 (aarch64_float_const_zero_rtx_p): ...this.
18871 (aarch64_print_operand): Allow printing of new constants.
18872 (aarch64_valid_floating_const): New.
18873 (aarch64_legitimate_constant_p): Check for valid floating-point
18875 (aarch64_simd_valid_immediate): Likewise.
18876 (aarch64_vect_float_const_representable_p): New.
18877 (aarch64_float_const_representable_p): Likewise.
18878 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
18879 (aarch64_output_simd_mov_immediate): New.
18880 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
18881 (*movdf_aarch64): Likewise.
18882 * config/aarch64/constraints.md (Ufc): New.
18883 (Y): call aarch64_float_const_zero_rtx.
18884 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
18886 2013-01-07 Richard Biener <rguenther@suse.de>
18888 PR tree-optimization/55888
18889 PR tree-optimization/55862
18890 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
18891 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
18892 not if it is contained therein.
18894 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18896 * config/avr/t-avr: Typo.
18898 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
18901 * config/avr/t-avr: Don't automatically rebuild
18902 $(srcdir)/config/avr/t-multilib
18903 $(srcdir)/config/avr/avr-tables.opt
18904 $(srcdir)/doc/avr-mmcu.texi
18905 (avr-mcus): New phony target to build them on request.
18906 (s-avr-mlib, s-avr-mmcu-texi): Remove.
18907 * avr/avr-mcus.def: Adjust comments.
18909 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
18911 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
18913 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
18915 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
18917 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
18919 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
18921 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
18923 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
18924 to generate profiling.
18925 * config/rs6000/aix64.h (LIB_SPEC): Same.
18927 2013-01-04 Andrew Pinski <apinski@cavium.com>
18929 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
18931 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
18933 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
18935 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
18937 (ix86_expand_move): Ditto.
18938 (ix86_zero_extend_to_Pmode): Ditto.
18939 (ix86_expand_call): Ditto.
18940 (ix86_expand_special_args_builtin): Ditto.
18941 (ix86_expand_builtin): Ditto.
18943 2013-01-04 Richard Biener <rguenther@suse.de>
18945 PR tree-optimization/55862
18946 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
18947 translating them through PHI nodes.
18949 2013-01-04 Martin Jambor <mjambor@suse.cz>
18951 PR tree-optimization/55755
18952 * tree-sra.c (sra_modify_assign): Do not check that an access has no
18953 children when trying to avoid producing a VIEW_CONVERT_EXPR.
18955 2013-01-04 Marek Polacek <polacek@redhat.com>
18957 PR middle-end/55859
18958 * opts.c (default_options_optimization): Clarify error message.
18960 2013-01-04 Richard Biener <rguenther@suse.de>
18962 PR middle-end/55863
18963 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
18966 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18969 * config/pa/pa.md (movsi): Revert previous change.
18970 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
18973 2013-01-03 Richard Henderson <rth@redhat.com>
18975 * config/i386/i386.c (ix86_expand_move): Always assign to op1
18976 after eliminating TLS symbols.
18978 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
18981 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
18982 * graphite-poly.c (debug_gmp_value): Likewise.
18984 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
18987 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
18988 selected code model, define __code_mode_small__, __code_model_medium__,
18989 __code_model_large__, __code_model_32__ or __code_model_kernel__.
18990 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
18991 xchg temporary register with %k. Declare temporary register as
18993 [__x86_64__]: For medium and large code models, preserve %rbx register.
18995 2013-01-03 Richard Biener <rguenther@suse.de>
18997 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
18998 (dump_subscript): Adjust.
18999 (finalize_ddr_dependent): Do not dump redundant info.
19000 (analyze_siv_subscript): Adjust.
19001 (subscript_dependence_tester): Likewise.
19002 (compute_affine_dependence): Likewise.
19004 2013-01-03 Richard Biener <rguenther@suse.de>
19007 2013-01-03 Richard Biener <rguenther@suse.de>
19009 PR tree-optimization/55857
19010 * tree-vect-stmts.c (vectorizable_load): Do not setup
19011 re-alignment for invariant loads.
19013 2013-01-02 Richard Biener <rguenther@suse.de>
19015 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19016 invariant load do not generate a vector load from the scalar location.
19018 2013-01-03 Richard Biener <rguenther@suse.de>
19020 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19021 for not vectorizing.
19022 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19023 not build INDIRECT_REFs, call get_name once only.
19024 (vect_create_data_ref_ptr): Likewise. Dump base object kind
19025 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19027 2013-01-03 Richard Biener <rguenther@suse.de>
19029 PR tree-optimization/55857
19030 * tree-vect-stmts.c (vectorizable_load): Do not setup
19031 re-alignment for invariant loads.
19033 2013-01-03 Richard Biener <rguenther@suse.de>
19036 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19037 prefer a built-in decl.
19039 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19041 * gcc.c (process_command): Update copyright notice dates.
19042 * gcov.c (print_version): Likewise.
19043 * gcov-dump.c (print_version): Likewise.
19045 PR rtl-optimization/55838
19046 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19047 iv0.step, iv1.step and step.
19049 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19050 Marc Glisse <marc.glisse@inria.fr>
19052 PR tree-optimization/55832
19053 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19054 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19055 integer_{one,zero}_node.
19057 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19060 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19061 * var-tracking.c (reverse_op): Don't add reverse ops to
19062 VALUEs that have already
19063 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19065 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19067 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19069 2013-01-02 Teresa Johnson <tejohnson@google.com>
19071 * dumpfile.c (dump_loc): Print filename with location.
19072 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19073 new location_t parameter to emit complete unroll message with
19074 new dump framework.
19075 (canonicalize_loop_induction_variables): Compute loops location
19076 and pass to try_unroll_loop_completely.
19077 * loop-unroll.c (report_unroll_peel): New function.
19078 (peel_loops_completely): Use new dump format with location
19079 for main dumpfile message, and invoke report_unroll_peel on success.
19080 (decide_unrolling_and_peeling): Ditto.
19081 (decide_peel_once_rolling): Remove old dumpfile message subsumed
19082 by report_unroll_peel.
19083 (decide_peel_completely): Ditto.
19084 (decide_unroll_constant_iterations): Ditto.
19085 (decide_unroll_runtime_iterations): Ditto.
19086 (decide_peel_simple): Ditto.
19087 (decide_unroll_stupid): Ditto.
19088 * cfgloop.c (get_loop_location): New function.
19089 * cfgloop.h (get_loop_location): Declare.
19091 2013-01-02 Sriraman Tallam <tmsriram@google.com>
19093 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19096 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19098 PR middle-end/55198
19099 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19100 BLKmode objects when EXPAND_MEMORY is specified.
19102 2013-01-02 Sriraman Tallam <tmsriram@google.com>
19104 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19106 (fold_builtin_cpu): Do not share cpu model decls across statements.
19108 2013-01-02 Jason Merrill <jason@redhat.com>
19111 * tree.c (build_array_type_1): Revert earlier change.
19113 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
19115 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19117 * config/aarch64/aarch64-tune.md: Re-generate.
19119 2013-01-02 Richard Biener <rguenther@suse.de>
19121 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19122 invariant load do not generate a vector load from the scalar location.
19124 2013-01-02 Richard Biener <rguenther@suse.de>
19127 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19128 * configure: Regenerate.
19130 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
19132 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19133 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19134 (expand_builtin_int_roundingfn_2): Keep the original target around
19135 for the fallback case.
19137 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
19139 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19140 to be clear for sign changes.
19142 2013-01-01 Jan Hubicka <jh@suse.cz>
19144 * ipa-inline-analysis.c: Fix formatting.
19146 2013-01-01 Jakub Jelinek <jakub@redhat.com>
19148 PR tree-optimization/55831
19149 * tree-vect-loop.c (get_initial_def_for_induction): Use
19150 gsi_after_labels instead of gsi_start_bb.
19152 Copyright (C) 2013 Free Software Foundation, Inc.
19154 Copying and distribution of this file, with or without modification,
19155 are permitted in any medium without royalty provided the copyright
19156 notice and this notice are preserved.