1 2013-09-01 Jan Hubicka <jh@suse.cz>
3 * Makefile.in: Add ipa-profile.o
4 (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
5 * cgraph.c (struct cgraph_propagate_frequency_data,
6 cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7 ipa-profile.c; replace cgraph_ by ipa_ prefix.
8 * cgraph.h (cgraph_propagate_frequency): Remove.
9 * ipa-inline-analysis.c: Include ipa-utils.h; drop duplicated cfgloop.h.
10 (inline_update_callee_summaries): Update.
11 * ipa-profile.c: New file.
12 * ipa-utils.h (ipa_propagate_frequency): Declare.
13 * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
14 data-streamer.h, value-prof.h
15 (symtab_remove_unreachable_nodes): Update profile.
16 (struct histogram_entry, histogram, histogram_pool, histogram_hash,
17 account_time_size, cmp_counts, dump_histogram,
18 ipa_profile_generate_summary, ipa_profile_write_summary,
19 ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
20 pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
23 2013-09-01 John David Anglin <danglin@gcc.gnu.org>
25 * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
27 2013-09-01 Jan Hubicka <jh@suse.cz>
29 * common.opt (fdevirtualize-speculatively): New function.
30 * invoke.texi (fdevirtualize-speculatively): Document.
31 * ipa-devirt.c: Include ipa-inline.h
32 (likely_target_p): New function.
33 (ipa_devirt): New function.
34 (gate_ipa_devirt): New function.
35 (pass_data_ipa_devirt): New static var.
36 (pass_ipa_devirt): Likewise.
37 (make_pass_ipa_devirt): New function.
38 * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
39 (common_handle_option): Disable devirtualization when
40 value range profiling is available.
41 * passes.def (pass_ipa_devirt): Add.
42 * timever.def (TV_IPA_DEVIRT): New timevar.
43 * tree-pass.h (make_pass_ipa_devirt):
45 2013-09-01 Iain Sandoe <iain@codesourcery.com>
47 * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
48 include sanitize(undefined).
50 2013-08-31 Jan Hubicka <jh@suse.cz>
52 * bulitins.c (expand_builtin): Do not early exit for gcov
53 instrumented functions.
55 2013-08-31 Marek Polacek <polacek@redhat.com>
57 * ubsan.c: Include tm_p.h.
59 2013-08-31 Jan Hubicka <jh@suse.cz>
61 * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
64 * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
65 * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
66 * tree-cfg.c (verify_gimple_label): ... here.
67 * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
68 (ipa_merge_profiles): New function.
69 * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
70 (lto_input_function_body): Likewise.
71 * ipa-utils.h (ipa_merge_profiles): Declare.
72 * lto-streamer.h (lto_input_function_body): Update prototype.
73 (emit_label_in_global_context_p): Remove.
74 * lto-symtab.c: Include ipa-utils.h
75 (lto_cgraph_replace_node): Use ipa_merge_profiles.
77 2013-08-31 Jan Hubicka <jh@suse.cz>
79 * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
81 2013-08-31 Jan Hubicka <jh@suse.cz>
83 * basic-block.h (apply_scale): Make scale parmeter gcov_type.
85 2013-08-31 Uros Bizjak <ubizjak@gmail.com>
87 * config/alpha/alpha.c (alpha_emit_conditional_move): Update
88 "cmp" RTX before signed_comparison_operator check to account
91 2013-08-30 Jan Hubicka <jh@suse.cz>
93 * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
94 (detect_type_change_1): Rename to ...
95 (detect_type_change): ... this one; early return on non-polymorphic
97 (detect_type_change_ssa): Add comp_type parameter; update
98 use of detect_type_change.
99 (compute_complex_assign_jump_func): Add param_type parameter;
100 update use of detect_type_change_ssa.
101 (compute_complex_ancestor_jump_func): Likewise.
102 (ipa_get_callee_param_type): New function.
103 (ipa_compute_jump_functions_for_edge): Compute parameter type;
104 update calls to the jump function computation functions.
106 2013-08-30 Teresa Johnson <tejohnson@google.com>
107 Steven Bosscher <steven@gcc.gnu.org>
109 * cfgrtl.c (fixup_new_cold_bb): New routine.
110 (commit_edge_insertions): Invoke fixup_partitions.
111 (find_partition_fixes): New routine.
112 (fixup_partitions): Ditto.
113 (verify_hot_cold_block_grouping): Update comments.
114 (rtl_verify_edges): Invoke find_partition_fixes.
115 (rtl_verify_bb_pointers): Update comments.
116 (rtl_verify_bb_layout): Ditto.
117 * basic-block.h (probably_never_executed_edge_p): Declare.
118 (fixup_partitions): Ditto.
119 * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
120 * bb-reorder.c (sanitize_hot_paths): New function.
121 (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
123 * predict.c (probably_never_executed_edge_p): New routine.
124 * cfg.c (check_bb_profile): Add partition insanity warnings.
126 2013-08-30 Meador Inge <meadori@codesourcery.com>
128 * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
130 2013-08-30 Marek Polacek <polacek@redhat.com>
132 * Makefile.in (ubsan.o): Add.
133 (c-family/c-ubsan.o): Add.
134 (builtins.o): Add ubsan.h dependency.
137 * common.opt: Add -fsanitize=undefined option.
138 (flag_sanitize): Add variable.
139 (fsanitize=): Add option. Add Driver.
140 (fsanitize=thread): Remove option.
141 (fsanitize=address): Likewise.
142 (static-libubsan): New option.
143 * doc/invoke.texi: Document the new flag and -static-libubsan.
144 * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
145 (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
146 * builtin-attrs.def (ATTR_COLD): Define.
147 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
148 * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
149 BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
150 * flag-types.h (sanitize_code): New enum.
151 * opts.c (common_handle_option): Parse command line arguments
152 of -fsanitize=. Add -fsanitize=unreachable option.
153 * varasm.c (get_variable_section): Adjust.
154 (assemble_noswitch_variable): Likewise.
155 (assemble_variable): Likewise.
156 (output_constant_def_contents): Likewise.
157 (categorize_decl_for_section): Likewise.
158 (place_block_symbol): Likewise.
159 (output_object_block): Likewise.
160 * builtins.def: Likewise.
161 * toplev.c (compile_file): Likewise.
162 (process_options): Likewise.
163 * cppbuiltin.c: Likewise.
164 * tsan.c (tsan_pass): Likewise.
165 (tsan_gate): Likewise.
166 (tsan_gate_O0): Likewise.
167 * cfgexpand.c (partition_stack_vars): Likewise.
168 (expand_stack_vars): Likewise.
169 (defer_stack_allocation): Likewise.
170 (expand_used_vars): Likewise.
171 * cfgcleanup.c (old_insns_match_p): Likewise.
172 * asan.c (asan_finish_file): Likewise.
173 (asan_instrument): Likewise.
174 (gate_asan): Likewise.
175 (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
176 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
177 (asan_global_struct): Use pointer_sized_int_node instead
178 calling build_nonstandard_integer_type.
179 (initialize_sanitizer_builtins): Likewise.
180 (asan_finish_file): Likewise.
181 * gcc.c: Document %{%:function(args):X}.
182 (static_spec_functions): Add sanitize.
183 (handle_spec_function): Add retval_nonnull argument and if non-NULL,
184 store funcval != NULL there.
185 (do_spec_1): Adjust handle_spec_function caller.
186 (handle_braces): Allow %:function(args) as condition.
187 (sanitize_spec_function): New function.
188 (ADD_STATIC_LIBUBSAN_LIBS): Define.
189 (LIBUBSAN_SPEC): Likewise.
190 (LIBUBSAN_EARLY_SPEC): Likewise.
191 (SANITIZER_SPEC): Handle libubsan.
192 (SANITIZER_EARLY_SPEC): Likewise.
193 * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
194 instead of fsanitize=address.
195 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
196 instead of fsanitize=address*.
197 * builtins.c: Include ubsan.h.
198 (fold_builtin_0): Instrument __builtin_unreachable.
199 * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
200 instead of flag_asan.
201 * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
202 (pointer_sized_int_node): Define.
203 * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
205 2013-08-30 Mike Stump <mikestump@comcast.net>
207 * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
210 2013-08-29 Jan Hubicka <jh@suse.cz>
212 * cgraph.c (cgraph_function_body_availability): Handle weakref
214 * passes.def: Remove pass_fixup_cfg.
215 * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
216 track when we need to remove functions.
217 (gate_ipa_inline): Execute inlining always; add comment why.
218 (pass_data_ipa_inline): Remove TODO_remove_functions.
219 * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
220 do not produce summaries.
221 * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
222 (symtab_nonoverwritable_alias): Assert we are not called on weakref.
223 * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
224 constant pool and vtable.
226 2013-08-30 Tejas Belagod <tejas.belagod@arm.com>
228 * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
229 New arm_neon.h's internal macros to specify 64-bit constants.
230 Avoid using stdint.h's macros.
232 2013-08-30 Joern Rennecke <joern.rennecke@embecosm.com>
234 * recog.c (verify_changes): Verify that changes[i].old is non-zero
235 before applying REG_P.
237 2013-08-30 Jakub Jelinek <jakub@redhat.com>
239 PR tree-optimization/58277
240 * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
241 after seeing too many stmts with vdef in between dombb and current
242 bb, invalidate everything.
244 2013-08-30 Richard Biener <rguenther@suse.de>
246 * fold-const.c (fold_single_bit_test): Fix overflow test.
248 2013-08-30 Eric Botcazou <ebotcazou@adacore.com>
250 * function.c (assign_parm_setup_reg): For a parameter passed by pointer
251 and which can live in a register, always retrieve the value on entry.
252 * var-tracking.c (add_stores): Treat the copy on entry for a parameter
253 passed by invisible reference specially.
254 (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
255 (vt_add_function_parameter): Correctly deal with a parameter passed by
258 2013-08-30 Jan Hubicka <jh@suse.cz>
260 * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
262 2013-08-30 Richard Biener <rguenther@suse.de>
264 PR tree-optimization/58228
265 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
266 allow invariant loads in nested loop vectorization.
268 2013-08-30 Richard Biener <rguenther@suse.de>
270 PR tree-optimization/58223
271 * tree-loop-distribution.c (has_anti_dependence): Rename to ...
272 (has_anti_or_output_dependence): ... this and adjust to also
273 look for output dependences.
274 (mark_nodes_having_upstream_mem_writes): Adjust.
275 (rdg_flag_uses): Likewise.
277 2013-08-30 Richard Biener <rguenther@suse.de>
279 PR tree-optimization/58010
280 * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
281 assert that we have a loop-closed PHI.
283 2013-08-29 Jan Hubicka <jh@suse.cz>
285 * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
286 * cgraph.c (cgraph_release_function_body): Free decl_in_state.
287 * lto-section-in.c (lto_free_function_in_decl_state): New function.
288 (lto_free_function_in_decl_state_for_node): New function.
290 2013-08-29 Xinliang David Li <davidxl@google.com>
292 * loop-unroll.c (report_unroll_peel): Minor message change.
293 * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
294 Emit alignment peeling message with default -fopt-info.
295 (vect_loop_versioning): Emit loop version info message.
296 * tree-vectorizer.c (vectorize_loops): Minor message change.
297 (execute_vect_slp): Ditto.
299 2013-08-29 Eric Botcazou <ebotcazou@adacore.com>
301 * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
302 of the clone from the DECL_NAME of the original function.
304 2013-08-29 Oleg Endo <olegendo@gcc.gnu.org>
306 * passes.c (register_pass): Add overload.
307 * tree-pass.h (register_pass): Forward declare it. Add comment.
309 2013-08-29 Jan Hubicka <jh@suse.cz>
311 * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
312 DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
313 * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
314 DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
315 (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
316 * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
317 Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
318 (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
320 2013-08-29 Teresa Johnson <tejohnson@google.com>
322 * dumpfile.c (dump_loc): Output column number.
323 * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
324 * doc/invoke.texi: Document optall -fopt-info flag.
325 * profile.c (read_profile_edge_counts): Use new dump framework.
326 (compute_branch_probabilities): Ditto.
327 * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
328 when pass not in any opt group.
329 * pass_manager.h (pass_manager::get_pass_profile): New method.
330 * value-prof.c (check_counter): Use new dump framework.
331 (check_ic_target): Ditto.
332 * coverage.c (get_coverage_counts): Ditto.
333 (coverage_init): Setup new dump framework.
335 2013-08-29 Richard Biener <rguenther@suse.de>
337 PR tree-optimization/58246
338 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
339 handle the dominance check inside a basic-block.
341 2013-08-29 Richard Biener <rguenther@suse.de>
344 * tree-ssa-copy.c (may_propagate_copy): Allow propagating
345 of default defs that appear in abnormal PHI nodes.
347 2013-08-29 Richard Biener <rguenther@suse.de>
349 PR tree-optimization/57685
350 * tree-vrp.c (register_edge_assert_for_1): Recurse only for
351 single-use operands to avoid exponential complexity.
353 2013-08-28 Dehao Chen <dehao@google.com>
355 * ipa-inline.c (edge_badness): Fix integer underflow.
357 2013-08-28 Uros Bizjak <ubizjak@gmail.com>
359 * gtm-builtins.def (_ITM_free): Declare leaf.
361 2013-08-28 Jakub Jelinek <jakub@redhat.com>
364 * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
365 (*tls_local_dynamic_base_64_largepic): Likewise.
366 (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
367 Remove predicate from call operand.
368 * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
369 return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
371 2013-08-28 Jeff Law <law@redhat.com>
373 * tree-ssa-threadedge.c (thread_around_empty_block): Remove
374 checks for the number of predecessors and successors allowed.
375 * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
376 which require copying a joiner block if there is a request which
377 is a subpath that requires no joiner block copying.
379 2013-08-28 Jan Hubicka <jh@suse.cz>
381 * lto-streamer-out.c (DFS_write_tree_body): Drop
382 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
383 (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
384 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
385 * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
386 Do not read DECL_ERROR_ISSUED.
387 (unpack_ts_decl_with_vis_value_fields): Do not read
389 (lto_input_ts_binfo_tree_pointers): Do not read
390 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
391 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
392 write DECL_ERROR_ISSUED..
393 (pack_ts_decl_with_vis_value_fields): Do not write
395 (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
396 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
397 * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
398 * tree.h (tree_decl_common): Update comment.
399 (DECL_ERROR_ISSUED): Remove.
401 2013-08-28 Jakub Jelinek <jakub@redhat.com>
404 * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
406 2013-08-28 Jan Hubicka <jh@suse.cz>
408 * builtins.def (free): Declare leaf.
410 2013-08-27 David Malcolm <dmalcolm@redhat.com>
413 * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
414 * configure: Regenerate.
416 2013-08-27 Martin Jambor <mjambor@suse.cz>
418 * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
419 (ipa_ancestor_jf_data): Likewise.
420 (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
421 (ipa_get_jf_pass_through_type_preserved): New function.
422 (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
423 (ipa_get_jf_ancestor_type_preserved): New function.
424 * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
425 (ipa_get_jf_ancestor_result): Likewise.
426 (propagate_vals_accross_pass_through): Use
427 ipa_get_jf_pass_through_result to do all the value mappings.
428 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
430 (ipa_set_jf_cst_copy): New function.
431 (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
432 (ipa_set_jf_arith_pass_through): Likewise.
433 (ipa_set_ancestor_jf): Likewise.
434 (compute_complex_assign_jump_func): Set type_preserved instead of
436 (ipa_compute_jump_functions_for_edge): Likewise.
437 (combine_known_type_and_ancestor_jfs): Honor type_preserved.
438 (update_jump_functions_after_inlining): Update type_preserved.
439 Explicitely create jump functions when combining one with pass_through.
440 (ipa_write_jump_function): Stream the type_preserved flags.
441 (ipa_read_jump_function): Likewise.
443 2013-08-27 Jakub Jelinek <jakub@redhat.com>
444 Aldy Hernandez <aldyh@redhat.com>
446 * Makefile.in (omp-low.o): Depend on $(TARGET_H).
447 * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
448 * function.h (struct function): Add has_force_vect_loops and
450 * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
451 * gimple.c (gimple_build_omp_critical): Add KIND argument and
453 * gimple.def: Update CLAUSES comments.
454 * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
455 (gimple_build_omp_for): Add argument to prototype.
456 (gimple_omp_for_kind): New.
457 (gimple_omp_for_set_kind): New.
458 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
459 GOVD_DATA_SHARE_CLASS.
460 (enum omp_region_type): Add ORT_SIMD.
461 (gimple_add_tmp_var): Handle ORT_SIMD.
462 (gimplify_var_or_parm_decl): Same.
463 (is_gimple_stmt): Same.
464 (omp_firstprivatize_variable): Same.
465 (omp_add_variable): Only use splay_tree_insert if lookup failed.
466 (omp_notice_variable): Handle ORT_SIMD.
467 (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
468 (omp_check_private): Handle ORT_SIMD.
469 (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
471 (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
472 Handle OMP_CLAUSE_LASTPRIVATE.
473 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
475 (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
476 (gimplify_expr): Handle OMP_SIMD.
477 * internal-fn.c (expand_GOMP_SIMD_LANE): New.
478 (expand_GOMP_SIMD_VF): New.
479 (expand_GOMP_SIMD_LAST_LANE): New.
480 * internal-fn.def (GOMP_SIMD_LANE): New.
482 (GOMP_SIMD_LAST_LANE): New.
483 * omp-low.c: Include target.h.
484 (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
486 (check_omp_nesting_restrictions): Same.
488 (lower_rec_simd_input_clauses): New.
489 (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
491 (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
492 GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
493 (expand_omp_build_assign): New.
494 (expand_omp_for_init_counts): New.
495 (expand_omp_for_init_vars): New.
496 (extract_omp_for_update_vars): New.
497 (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
498 and rewrite accordingly.
499 (expand_omp_simd): New.
500 (expand_omp_for): Use expand_omp_simd.
501 (lower_omp_for_lastprivate): Unshare vinit when appropriate.
502 (lower_omp_for): Do not lower the body.
503 * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
505 * tree-flow.h (find_omp_clause): Remove prototype.
506 * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
507 forcing vectorization of the loop, or if flag_tree_vectorize.
508 (gate_tree_if_conversion): Similarly.
509 * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
510 gimple_build_omp_for.
511 (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
512 * tree-parloops (create_parallel_loop): Pass kind argument to
513 gimple_build_omp_for.
514 * tree-pretty-print.c (dump_omp_clause): Add cases for
515 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
516 OMP_CLAUSE__SIMDUID_.
517 (dump_generic_node): Handle OMP_SIMD.
518 * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
519 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
520 unroll OMP_SIMD loops here.
521 * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
522 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
524 (vect_analyze_data_refs): Handle simd loops.
525 * tree-vect-loop.c (vectorizable_live_operation): Handle
527 * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
528 (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
529 (vectorizable_load): Same.
530 * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
531 (struct simduid_to_vf): New.
532 (simduid_to_vf::hash): New.
533 (simduid_to-vf::equal): New.
534 (struct simd_array_to_simduid): New.
535 (simd_array_to_simduid::hash): New.
536 (simd_array_to_simduid::equal): New.
537 (adjust_simduid_builtins): New.
538 (struct note_simd_array_uses_struct): New.
539 (note_simd_array_uses_cb): New.
540 (note_simd_array_uses): New.
541 (vectorize_loops): Handle simd hints and adjust simd builtins
543 * tree-vectorizer.h (struct _stmt_vec_info): Add
544 simd_lane_access_p field.
545 (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
546 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
547 OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
548 (omp_clause_code_name): Same.
549 (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
550 OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
551 * tree.def (OMP_SIMD): New entry.
552 * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
553 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
554 (OMP_CLAUSE_DECL): Adjust range for new clauses.
555 (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
556 (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
557 (OMP_CLAUSE_LINEAR_STEP): New.
558 (OMP_CLAUSE_SAFELEN_EXPR): New.
559 (OMP_CLAUSE__SIMDUID__DECL): New.
560 (find_omp_clause): New prototype.
562 2013-08-27 H.J. Lu <hongjiu.lu@intel.com>
564 * config/i386/driver-i386.c (host_detect_local_cpu): Update
565 Haswell processor detection.
567 2013-08-27 Christian Widmer <shadow@umbrox.de>
570 * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
571 of Ivy Bridge and Haswell processors. Assume core-avx2 for unknown
572 AVX2 capable processors.
574 2013-08-27 Tejas Belagod <tejas.belagod@arm.com>
576 * config/aarch64/arm_neon.h: Replace all inline asm implementations
577 of vget_low_* with implementations in terms of other intrinsics.
579 2013-08-27 Marc Glisse <marc.glisse@inria.fr>
582 * doc/extend.texi (__builtin_isinf_sign): Restrict the return
583 values to -1, 0 and 1.
585 2013-08-27 Vidya Praveen <vidyapraveen@arm.com>
587 * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
588 UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
589 (<optab><mode>3_insn): Remove.
590 (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
591 (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
592 (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
593 (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
594 (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
595 (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
596 (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
597 (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
598 (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
599 (ror<mode>3_insn): Likewise.
600 * config/aarch64/predicates.md (aarch64_simd_register): New.
602 2013-08-27 Richard Biener <rguenther@suse.de>
604 PR tree-optimization/57521
605 * tree-if-conv.c (if_convertible_bb_p): Verify that at least
606 one edge is non-critical.
607 (find_phi_replacement_condition): Make sure to use a non-critical
608 edge. Cleanup and remove old bug workarounds.
609 (bb_postdominates_preds): Remove.
610 (if_convertible_loop_p_1): Do not compute post-dominators.
611 (combine_blocks): Do not free post-dominators.
612 (main_tree_if_conversion): Likewise.
613 (pass_data_if_conversion): Add TODO_verify_ssa.
615 2013-08-27 DJ Delorie <dj@redhat.com>
617 * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
619 2013-08-27 Yufeng Zhang <yufeng.zhang@arm.com>
621 * function.c (assign_parm_find_data_types): Set passed_mode and
622 nominal_mode to the TYPE_MODE of nominal_type for the built
623 pointer type in case of the struct-pass-by-reference.
625 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
627 * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
628 (UINT16_TYPE): Change default to "unsigned int".
630 * config/avr/avr.opt (mfract-convert-truncate): New option.
631 * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
632 is set, round negative fractional integers according to n1169
633 when converting to integer types.
635 2013-08-26 Jan Hubicka <jh@suse.cz>
637 * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
638 methods can not be called indirectly when their address is not taken.
640 2013-08-26 Jan Hubicka <jh@suse.cz>
642 * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
645 2013-08-26 Jan Hubicka <jh@suse.cz>
647 * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
650 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
652 * reload.c (find_valid_class): Allow classes that do not include
653 FIRST_PSEUDO_REGISTER - 1.
655 2013-08-26 Jan Hubicka <jh@suse.cz>
657 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
658 fix edge count/frequency when speculation failed; fix type check
661 2013-08-26 Jan Hubicka <jh@suse.cz>
663 * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
665 2013-08-26 Jan Hubicka <jh@suse.cz>
667 * ipa-inline-transform.c (inline_transform): Be ready for basic block
668 to be changed by edge redirection.
670 2013-08-26 Jan Hubicka <jh@suse.cz>
672 * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
673 formating; add sanity check.
674 (cgraph_resolve_speculation): Add FIXME about scaling profiles.
675 (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
676 * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
677 (resolve_noninline_speculation): Update callee keys, too.
679 2013-08-26 Jan Hubicka <jh@suse.cz>
681 * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
682 (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
684 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
686 * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
689 2013-08-26 Uros Bizjak <ubizjak@gmail.com>
691 * config/i386/i386.c (ix86_debug_options): Remove prototype.
692 (x86_64_elf_select_section): Ditto.
693 (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
695 (ix86_pass_by_reference): Ditto.
696 (output_set_got): Ditto.
697 (ix86_unary_operator_ok): Ditto.
698 (ix86_expand_builtin): Ditto.
700 2013-08-23 Jan Hubicka <jh@suse.cz>
702 * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
704 2013-08-23 Jan Hubicka <jh@suse.cz>
706 * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
707 (tree_decl_with_vis): Add FINAL field.
709 2013-08-23 Jeff Law <law@redhat.com>
711 * tree-ssa-pre.c (do_regular_insertion): Include the expression in
712 the debugging dump when the expression is fully redundant.
714 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
716 * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
717 * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
718 * pretty-print.c (pp_formatted_text_data): Likewise.
719 (pp_write_text_to_stream): Likewise.
720 (pp_write_text_as_dot_label_to_stream): Likewise.
721 (pp_append_r): Likewise.
722 (pp_format): Likewise.
723 (pp_flush): Likewise.
724 (pp_clear_output_area): Likewise.
725 (pp_append_text): Likewise.
726 (pp_formatted_text): Likewise.
727 (pp_remaining_character_count_for_line): Likewise.
728 (pp_newline): Likewise.
729 (pp_character): Likewise.
730 (output_buffer::~output_buffer): Define.
731 (pretty_printer::~pretty_printer): Destruct output buffer.
732 * pretty-print.h (output_buffer::~output_buffer): Declare.
733 (pretty_printer::~pretty_printer): Declare virtual.
735 2013-08-24 Marc Glisse <marc.glisse@inria.fr>
738 * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
739 HOST_WIDE_INT_M1U): New macros.
740 * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
741 fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
742 unsigned -1 for lshift.
743 * cse.c (cse_insn): Likewise.
744 * double-int.c (rshift_double, lshift_double): Likewise.
745 * builtins.c (fold_builtin_bitop): Likewise.
746 * combine.c (force_to_mode): Likewise.
747 * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
748 * simplify-rtx.c (simplify_const_unary_operation,
749 simplify_const_binary_operation): Likewise.
750 * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
751 check_va_list_escapes): Likewise.
752 * rtlanal.c (nonzero_bits1): Likewise.
753 * expmed.c (expand_smod_pow2): Likewise.
754 * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
756 2013-08-23 Jan Hubicka <jh@suse.cz>
758 * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
759 as having address taken.
761 2013-08-23 Jan Hubicka <jh@suse.cz>
763 * ipa-utils.h (method_class_type): Declare.
764 * ipa-devirt.c (method_class_type): Export.
766 * cgraphunit.c (analyze_functions): Do basic devirtualization;
767 do not walk base classes of anonymous types.
769 2013-08-23 Kaz Kojima <kkojima@gcc.gnu.org>
771 PR rtl-optimization/58220
773 * final.c (reemit_insn_block_notes): Use NEXT_INSN to
774 handle SEQUENCE insns properly.
776 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
778 * pretty-print.h (pp_newline_and_flush): Declare. Remove macro
780 (pp_newline_and_indent): Likewise.
781 (pp_separate_with): Likewise.
782 * pretty-print.c (pp_newline_and_flush): Define.
783 (pp_newline_and_indent): Likewise.
784 (pp_separate_with): Likewise.
786 2013-08-23 Jakub Jelinek <jakub@redhat.com>
789 * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
790 * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
792 2013-08-23 Kirill Yukhin <kirill.yukhin@intel.com>
794 * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
795 * gcc/config/i386/i386.md (*movti_internal): Use
796 predicate to determine if EVEX is needed.
797 (*movsi_internal): Ditto.
798 (*movdf_internal): Ditto.
799 (*movsf_internal): Ditto.
800 * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
802 2013-08-23 Jakub Jelinek <jakub@redhat.com>
804 PR tree-optimization/58209
805 * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
806 (find_tail_calls): Give up for pointer result types if m is non-NULL.
807 (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
808 emit POINTER_PLUS_EXPR.
809 (create_tailcall_accumulator): For pointer result type accumulate in
812 2013-08-22 Paolo Carlini <paolo.carlini@oracle.com>
814 * configure.ac: Add backslashes missing from the last change.
815 * configure: Regenerate.
817 2013-08-22 Jan Hubicka <jh@susue.cz>
819 * ipa.c (function_and_variable_visibility): First remember function
820 was global and then make it local.
822 2013-08-22 Julian Brown <julian@codesourcery.com>
824 * configure.ac: Add aarch64 to list of arches which use "nop" in
826 * configure: Regenerate.
828 2013-08-22 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
830 * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
831 gnu_libc_has_function.
832 * config/s390/tpf.h: Likewise.
834 2013-08-22 Jan Hubicka <jh@susue.cz>
836 * timevar.c (validate_phases): Add cast.
838 2013-08-22 Jan Hubicka <jh@susue.cz>
840 * timevar.c (validate_phases): Use size_t for memory.
841 * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
843 2013-08-22 Gabriel Dos Reis <gdr@integrable-solutions.net>
845 * pretty-print.h (output_buffer::output_buffer): Declare.
846 (pretty_printer::pretty_printer): Likewise.
847 (pp_construct): Remove.
848 * pretty-print.c (output_buffer::output_buffer): Define.
849 (pretty_printer::pretty_printer): Rename from pp_construct. Simplify.
850 * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
851 (print_gimple_expr): Likewise.
852 (print_gimple_seq): Likewise.
853 (gimple_dump_bb): Likewise.
854 * sched-vis.c (dump_value_slim): Likewise.
855 (dump_insn_slim): Likewise.
856 (dump_rtl_slim): Likewise.
857 (str_pattern_slim): Likewise.
858 * tree-mudflap.c (mf_varname_tree): Likewise.
859 * graph.c (print_graph_cfg): Likewise.
860 (start_graph_dump): Likewise.
861 * tree-pretty-print.c (maybe_init_pretty_print): Likewise. Use
863 * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
864 pretty printer initialization.
865 * coretypes.h (diagnostic_context): Remove superflous type alias
867 (pretty_printer): Likewise. Declare directly as a class.
868 (pretty_print_info): Remove declaration as class.
869 * asan.c (asan_emit_stack_protection): Remove call to pp_construct
870 and pp_clear_output_area.
871 (asan_add_global): Likewise.
873 2013-08-22 Jan Hubicka <jh@suse.cz>
875 * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
876 * ipa-utils.h (update_type_inheritance_graph): Declare.
877 (possible_polymorphic_call_target_p): Declare.
878 (possible_polymorphic_call_target_p): New.
879 * ipa-devirt.c: Update toplevel comments.
880 (cached_polymorphic_call_targets): Move up.
881 (odr_type_d): Move ID down.
882 (polymorphic_type_binfo_p): Update comment.
883 (odr_hasher::remove): Likewise;
884 (get_odr_type): Set anonymous_namespace.
885 (dump_odr_type): Dump it.
886 (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
887 (maybe_record_node): Record node in cached_polymorphic_call_targets.
888 (record_binfo): Add comment.
889 (free_polymorphic_call_targets_hash): Do not ICE when cache is not
891 (devirt_node_removal_hook): Do not iCE when cache is freed.
892 (possible_polymorphic_call_target_p): New predicate.
893 (update_type_inheritance_graph): New function.
895 2013-08-22 Alexander Ivchenko <alexander.ivchenko@intel.com>
896 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
897 Sergey Lega <sergey.s.lega@intel.com>
898 Anna Tikhonova <anna.tikhonova@intel.com>
899 Ilya Tocar <ilya.tocar@intel.com>
900 Andrey Turetskiy <andrey.turetskiy@intel.com>
901 Ilya Verbin <ilya.verbin@intel.com>
902 Kirill Yukhin <kirill.yukhin@intel.com>
903 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
905 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
906 (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
907 (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
908 (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
909 (OPTION_MASK_ISA_AVX2_UNSET): Update.
910 (OPTION_MASK_ISA_AVX512F_UNSET): New.
911 (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
912 (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
913 (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
914 (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
915 OPT_mavx512pf, OPT_mavx512er cases.
916 * config/i386/constraints.md (v): New constraint.
917 (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
918 * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
920 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
921 AVX512F, AVX512ER, AVX512PF, AVX512CD features.
922 * config/i386/i386-c.c (ix86_target_macros_internal):
923 Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
925 * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
926 (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
927 * config/i386/i386.c (regclass_map, dbx_register_map)
928 (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
929 (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
930 (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
932 (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
933 PTA_AVX512PF, PTA_AVX512CD. Handle -mavx512f, -mavx512er, -mavx512cd,
934 -mavx512pf options. Fix formatting.
935 (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
936 targets. Squash EVEX_SSE_REGS if AVX512F is disabled.
937 (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
938 -mavx512cd, -mavx512pf options.
939 (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
940 (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
941 (ix86_preferred_output_reload_class): Replace SSE_REGS with
943 (ix86_hard_regno_mode_ok): Support 512-bit registers.
944 (ix86_set_reg_reg_cost): Ditto.
945 (x86_order_regs_for_local_alloc): Ditto.
946 (MAX_VECT_LEN): Extend to 64-byte.
947 (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
948 * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
949 (TARGET_AVX512ER, TARGET_AVX512CD): New.
950 (BIGGEST_ALIGNMENT): Extend to 512-bits.
951 (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
952 (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
953 (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
954 (SSE_REG_MODE_P): Support new modes.
955 (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
956 (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
957 (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
958 (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
959 (REG_CLASS_CONTENTS): Add new registers.
960 (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
961 (EXT_REX_SSE_REGNO_P): New.
962 (HI_REGISTER_NAMES): Add new registers.
963 * config/i386/i386.md: Define constants for new registers.
964 (mode): Add new 512-bit modes.
965 (prefix): Support evex prefix.
966 (isa): Support avx512f, noavx512f, fma_avx512f.
967 (ssemodesuffix): Add new 512-bit modes.
969 (*movxi_internal_avx512f): Ditto.
970 (*movdi_internal): Replace constraint "x" with the new constraint "v".
972 (*movsi_internal): Likewise.
973 (*movdf_internal): Likewise.
974 (*movsf_internal): Likewise.
975 (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
976 (<code><mode>3): Likewise.
977 * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
979 * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
980 with the new constraint "v".
981 * config/i386/sse.md (*mov<mode>_internal): Support new registers and
983 (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
984 with the new constraint "v".
985 (<sse2>_loaddqu<avxsizesuffix>): Likewise.
986 (<sse2>_storedqu<avxsizesuffix>): Likewise.
987 (*<plusminus_insn><mode>3): Likewise.
988 (<sse>_vm<plusminus_insn><mode>3): Likewise.
989 (*mul<mode>3): Likewise.
990 (<sse>_vmmul<mode>3): Likewise.
991 (<sse>_div<mode>3): Likewise.
992 (<sse>_vmdiv<mode>3): Likewise.
993 (<sse>_sqrt<mode>2): Likewise.
994 (<sse>_vmsqrt<mode>2): Likewise.
995 (*<code><mode>3_finite): Likewise.
996 (*<code><mode>3) <smaxmin>: Likewise.
997 (<sse>_vm<code><mode>3): Likewise.
998 (*<code><mode>3) <any_logic>: Likewise.
999 (*fma_fmadd_<mode>): Likewise.
1000 (*fma_fmsub_<mode>): Likewise.
1001 (*fma_fnmadd_<mode>): Likewise.
1002 (*fma_fnmsub_<mode>): Likewise.
1003 (*fma_fmaddsub_<mode>): Likewise.
1004 (*fma_fmsubadd_<mode>): Likewise.
1005 (*fmai_fmadd_<mode>): Likewise.
1006 (*fmai_fmsub_<mode>): Likewise.
1007 (*fmai_fnmadd_<mode>): Likewise.
1008 (*fmai_fnmsub_<mode>): Likewise.
1009 (sse_cvtsi2ss): Likewise.
1010 (sse_cvtsi2ssq): Likewise.
1011 (sse_cvtss2si): Likewise.
1012 (sse_cvtss2si_2): Likewise.
1013 (sse_cvtss2siq): Likewise.
1014 (sse_cvtss2siq_2): Likewise.
1015 (sse_cvttss2si): Likewise.
1016 (sse_cvtss2siq_2): Likewise.
1017 (float<sseintvecmodelower><mode>2): Likewise.
1018 (sse2_cvtsd2si_2): Likewise.
1019 (sse2_cvtsd2siq_2): Likewise.
1020 (*<plusminus_insn><mode>3): Likewise.
1021 (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
1022 (*<sse4_1_avx2>_mul<mode>3): Likewise.
1023 (ashr<mode>3): Likewise.
1024 (<shift_insn><mode>3): Likewise.
1025 (avx2_<code><mode>3): Likewise.
1026 (*avx2_<code><mode>3): Likewise.
1027 (*andnot<mode>3): Likewise.
1028 (*<code><mode>3) <any_logic>: Likewise.
1029 (abs<mode>2): Likewise.
1030 (avx2_permvar<mode>): Likewise.
1031 (avx2_perm<mode>_1): Likewise.
1032 (*avx_vpermilp<mode>): Likewise.
1033 (avx_vpermilvar<mode>3): Likewise.
1034 (avx2_ashrv<mode>): Likewise.
1035 (avx2_<shift_insn>v<mode>): Likewise.
1036 * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
1038 * doc/rtl.texi: Document XImode.
1040 2013-08-21 Jeff Law <law@redhat.com>
1042 * tree-flow.h (register_jump_thread): Pass vector of edges
1043 instead of each important edge.
1044 * tree-ssa-threadedge.c (thread_across_edge): Build the jump
1045 thread path into a vector and pass that to register_jump_thread.
1046 * tree-ssa-threadupdate.c (register_jump_thread): Conver the
1047 passed in edge vector to the current 3-edge form.
1050 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
1052 * dce.c (fini_dce): Call df_analyze again just in case
1053 delete_unmarked_insns removed anything.
1055 2013-08-21 Joern Rennecke <joern.rennecke@embecosm.com>
1057 * reload.h (struct reg_equivs): Rename to ..
1058 (struct reg_equivs_s): .. this.
1060 2013-08-20 Martin Liska <marxin.liska@gmail.com>
1062 * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
1064 2013-08-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1066 * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
1068 2013-08-21 Jeff Law <law@redhat.com>
1070 * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
1071 simplify assignments too. If the RHS collapses to a singleton
1072 range, then return the value for the range.
1074 2013-08-21 Kirill Yukhin <kirill.yukhin@intel.com>
1076 * config/i386/sse.md (V16): Rename to...
1078 (mov<mode>): Update iterator name.
1079 (*mov<mode>_internal): Ditto.
1080 (push<mode>1): Ditto.
1081 (movmisalign<mode>): Ditto.
1083 2013-08-20 Jan Hubicka <jh@suse.cz>
1086 * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
1087 entry for direct edges.
1088 (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
1090 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1092 Revert my last two changes, r201865 and r201864:
1095 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1097 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1098 instances can own GC refs.
1100 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1101 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1102 (gcc::context::gt_pch_nx): Likewise.
1103 (gcc::context::gt_pch_nx): Likewise.
1104 * ggc.h (gt_ggc_mx <T>): New.
1105 (gt_pch_nx_with_op <T>): New.
1106 (gt_pch_nx <T>): New.
1107 * passes.c (opt_pass::gt_ggc_mx): New.
1108 (opt_pass::gt_pch_nx): New.
1109 (opt_pass::gt_pch_nx_with_op): New.
1110 (pass_manager::gt_ggc_mx): New.
1111 (pass_manager::gt_pch_nx): New.
1112 (pass_manager::gt_pch_nx_with_op): New.
1113 (pass_manager::operator new): Use
1114 ggc_internal_cleared_alloc_stat rather than xcalloc.
1115 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1116 (pass_manager::gt_ggc_mx): New.
1117 (pass_manager::gt_pch_nx): New.
1118 (pass_manager::gt_pch_nx_with_op): New.
1119 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1120 (opt_pass::operator new): New.
1121 (opt_pass::gt_ggc_mx): New.
1122 (opt_pass::gt_pch_nx): New.
1123 (opt_pass::gt_pch_nx_with_op): New.
1126 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1128 * Makefile.in (GTFILES): Add context.h.
1129 * context.c (gcc::context::operator new): New.
1130 (gcc::context::gt_ggc_mx): New.
1131 (gcc::context::gt_pch_nx): New.
1132 (gcc::context::gt_pch_nx): New.
1133 * context.h (gcc::context): Add GTY((user)) marking.
1134 (gcc::context::operator new): New.
1135 (gcc::context::gt_ggc_mx): New.
1136 (gcc::context::gt_pch_nx): New.
1137 (gcc::context::gt_pch_nx): New.
1138 (g): Add GTY marking.
1139 (gt_ggc_mx (gcc::context *)): New.
1140 (gt_pch_nx (gcc::context *)): New.
1141 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1142 void *cookie)): New.
1143 * gengtype.c (open_base_files) <ifiles>: Add context.h.
1145 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
1147 * dce.c (fini_dce): Call df_analyze again just in case
1148 delete_unmarked_insns removed anything.
1150 2013-08-20 Teresa Johnson <tejohnson@google.com>
1152 PR rtl-optimizations/57451
1153 * final.c (reemit_insn_block_notes): Prevent lexical blocks
1154 from crossing split section boundaries.
1156 2013-08-20 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1158 * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
1159 * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
1160 with MULTLIB_DEFAULTS.
1162 2013-08-20 Nick Clifton <nickc@redhat.com>
1164 * target.def (narrow_volatile_bitfield): Note that the default
1165 value is false, not !TARGET_STRICT_ALIGN.
1166 * doc/tm.texi: Regenerate.
1168 2013-08-20 Pavel Chupin <pavel.v.chupin@intel.com>
1170 Fix LIB_SPEC for systems without libpthread.
1172 * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
1173 * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
1175 * config/i386/linux-common.h: Likewise.
1176 * config/mips/linux-common.h: Likewise.
1178 2013-08-20 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
1180 * tree-ssa-ccp.c (get_default_value): Remove redundant condition
1183 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1185 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1186 instances can own GC refs.
1188 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1189 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1190 (gcc::context::gt_pch_nx): Likewise.
1191 (gcc::context::gt_pch_nx): Likewise.
1192 * ggc.h (gt_ggc_mx <T>): New.
1193 (gt_pch_nx_with_op <T>): New.
1194 (gt_pch_nx <T>): New.
1195 * passes.c (opt_pass::gt_ggc_mx): New.
1196 (opt_pass::gt_pch_nx): New.
1197 (opt_pass::gt_pch_nx_with_op): New.
1198 (pass_manager::gt_ggc_mx): New.
1199 (pass_manager::gt_pch_nx): New.
1200 (pass_manager::gt_pch_nx_with_op): New.
1201 (pass_manager::operator new): Use
1202 ggc_internal_cleared_alloc_stat rather than xcalloc.
1203 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1204 (pass_manager::gt_ggc_mx): New.
1205 (pass_manager::gt_pch_nx): New.
1206 (pass_manager::gt_pch_nx_with_op): New.
1207 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1208 (opt_pass::operator new): New.
1209 (opt_pass::gt_ggc_mx): New.
1210 (opt_pass::gt_pch_nx): New.
1211 (opt_pass::gt_pch_nx_with_op): New.
1213 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1215 * Makefile.in (GTFILES): Add context.h.
1216 * context.c (gcc::context::operator new): New.
1217 (gcc::context::gt_ggc_mx): New.
1218 (gcc::context::gt_pch_nx): New.
1219 (gcc::context::gt_pch_nx): New.
1220 * context.h (gcc::context): Add GTY((user)) marking.
1221 (gcc::context::operator new): New.
1222 (gcc::context::gt_ggc_mx): New.
1223 (gcc::context::gt_pch_nx): New.
1224 (gcc::context::gt_pch_nx): New.
1225 (g): Add GTY marking.
1226 (gt_ggc_mx (gcc::context *)): New.
1227 (gt_pch_nx (gcc::context *)): New.
1228 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1229 void *cookie)): New.
1230 * gengtype.c (open_base_files) <ifiles>: Add context.h.
1232 2013-08-20 Alan Modra <amodra@gmail.com>
1235 * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
1236 (rs6000_emit_epilogue): Likewise.
1238 2013-08-19 Dehao Chen <dehao@google.com>
1240 * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
1242 2013-08-19 Peter Bergner <bergner@vnet.ibm.com>
1243 Jakub Jelinek <jakub@redhat.com>
1245 * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
1246 (BUILT_IN_FABSD64): Likewise.
1247 (BUILT_IN_FABSD128): Likewise.
1248 * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
1249 (fold_builtin_1): Likewise.
1250 * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
1251 destination and source operands.
1252 (*abstd2_fpr): Likewise.
1253 (*nabstd2_fpr): Likewise.
1255 2013-08-19 Richard Sandiford <rdsandiford@googlemail.com>
1257 * config/mips/mips.c (mips_adjust_insn_length): Add checks for
1260 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
1262 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
1265 2013-08-19 Alexander Ivchenko <alexander.ivchenko@intel.com>
1267 * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
1268 * builtins.c (default_libc_has_function): New.
1269 (gnu_libc_has_function): Ditto.
1270 (no_c99_libc_has_function): Ditto.
1271 (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
1272 instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
1273 (fold_builtin_sincos): Likewise.
1274 (fold_builtin_cexp): Likewise.
1275 * builtins.def (DEF_C94_BUILTIN): Likewise.
1276 (DEF_C99_BUILTIN): Likewise.
1277 (DEF_C99_C90RES_BUILTIN): Likewise.
1278 (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
1279 definitions to using this define.
1280 * config/darwin-protos.h (darwin_libc_has_function): New.
1281 * config/darwin.c (darwin_libc_has_function): Ditto.
1282 * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
1283 TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
1284 * config/darwin.h: Ditto.
1285 * config/elfos.h: Ditto.
1286 * config/freebsd.h: Ditto.
1287 * config/i386/cygming.h: Ditto.
1288 * config/i386/djgpp.h: Ditto.
1289 * config/i386/i386-interix.h: Ditto.
1290 * config/microblaze/microblaze.h: Ditto.
1291 * config/mmix/mmix.h: Ditto.
1292 * config/gnu-user.h: Ditto.
1293 * config/ia64/hpux.h: Ditto.
1294 * config/pa/pa-hpux.h: Ditto.
1295 * config/pdp11/pdp11.h: Ditto.
1296 * config/picochip/picochip.h: Ditto.
1297 * config/linux.h: Ditto.
1298 * config/netbsd.h: Ditto.
1299 * config/openbsd.h: Ditto.
1300 * config/rs6000/aix43.h: Ditto.
1301 * config/rs6000/aix51.h: Ditto.
1302 * config/rs6000/aix52.h: Ditto.
1303 * config/rs6000/aix53.h: Ditto.
1304 * config/rs6000/aix61.h: Ditto.
1305 * config/rs6000/darwin.h: Ditto.
1306 * config/rs6000/linux.h: Ditto.
1307 * config/rs6000/linux64.h: Ditto.
1308 * config/s390/tpf.h: Ditto.
1309 * config/sol2-10.h: Ditto.
1310 * config/sol2.h: Ditto.
1311 * config/vms/vms.h: Ditto.
1312 * config/vxworks.h: Ditto.
1313 * config/linux-android.c (linux_android_libc_has_function):
1314 New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
1315 * config/linux-protos.h (linux_android_libc_has_function):
1317 * config/i386/i386.c (ix86_libc_has_function): New.
1318 * config/i386/i386-protos.h
1319 (ix86_libc_has_function): New declaration.
1320 * config/i386/i386.md
1321 ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
1322 ("isinf<mode>2): Likewise.
1323 * convert.c (convert_to_integer): Using new target hook
1324 TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
1325 TARGET_C99_FUNCTIONS.
1326 * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
1327 * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
1328 * coretypes.h (function_class): New enum for different
1329 classes of functions.
1330 * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
1331 * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
1332 (TARGET_HAS_SINCOS): Likewise.
1333 (TARGET_LIBC_HAS_FUNCTION): New.
1334 * doc/tm.texi: Regenerated.
1335 * targhooks.h (default_libc_has_function): New declaration.
1336 (no_c99_libc_has_function): Ditto.
1337 (gnu_libc_has_function): Ditto.
1338 * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
1339 and TARGET_HAS_SINCOS.
1341 2013-08-18 Jan Hubicka <jh@suse.cz>
1343 * Makeifle-in (ipa-devirt.o): New.
1344 (GTFILES): Add ipa-utils.h and ipa-devirt.c
1345 * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
1346 (analyze_functions): Look into possible targets of polymorphic call.
1347 * dumpfile.c (dump_files): Add type-inheritance dump.
1348 * dumpfile.h (TDI_inheritance): New.
1349 * ipa-devirt.c: New file.
1350 * ipa-utils.h (odr_type_d): Forward declare.
1351 (odr_type): New type.
1352 (build_type_inheritance_graph): Declare.
1353 (possible_polymorphic_call_targets): Declare and introduce inline
1354 variant when only edge is pased.
1355 (dump_possible_polymorphic_call_targets): Likewise.
1356 * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
1357 * tree.c (type_in_anonymous_namespace_p): Break out from ...
1358 (types_same_for_odr): ... here.
1359 * tree.h (type_in_anonymous_namespace_p): Declare.
1361 2013-08-18 Jakub Jelinek <jakub@redhat.com>
1363 PR tree-optimization/58006
1364 * tree-parloops.c (take_address_of): Don't ICE if get_name
1366 (eliminate_local_variables_stmt): Remove clobber stmts.
1368 2013-08-18 Eric Botcazou <ebotcazou@adacore.com>
1370 * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
1371 error message is issued for an alias to undefined symbol.
1373 2013-08-18 Jan Hubicka <jh@suse.cz>
1375 * cgraph.c (cgraph_create_indirect_edge): Discover
1376 polymorphic calls and record basic info into indirect_info.
1377 * gimple-fold.c (gimple_fold_call): When doing BINFO based
1378 devirtualization, ignore objc function calls.
1379 * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
1380 call with no parm index info.
1381 * ipa-prop.c (ipa_analyze_call_uses): Likewise.
1382 * tree.c (virtual_method_call_p): New function.
1383 * tree.h (virtual_method_call_p): Declare.
1385 2013-08-16 Jan Hubicka <jh@suse.cz>
1388 * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
1390 2013-08-16 David Edelsohn <dje.gcc@gmail.com>
1392 * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
1395 2013-08-16 David Malcolm <dmalcolm@redhat.com>
1397 * gengtype.c (type_for_name): Add special-case support for
1398 locating types within the "gcc::" namespace.
1399 (open_base_files): Emit a "using namespace gcc" directive.
1401 2013-08-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1404 * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1405 memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1407 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1408 array instead of each individual operand as a separate argument.
1409 (emit_fusion_gpr_load): Likewise.
1410 (expand_fusion_gpr_load): Add new function declaration.
1412 * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1413 signature to have the operands passed as an array, instead of as
1414 separate arguments. Allow ZERO_EXTEND to be in the memory
1415 address, and also SIGN_EXTEND if -mpower8-fusion-sign. Do not
1416 depend on the register live/dead flags when peepholes are run.
1417 (expand_fusion_gpr_load): New function to be called from the
1418 peephole2 pass, to change the register that addis sets to be the
1420 (emit_fusion_gpr_load): Change the calling signature to have the
1421 operands passed as an array, instead of as separate arguments.
1422 Allow ZERO_EXTEND to be in the memory address, and also
1423 SIGN_EXTEND if -mpower8-fusion-sign.
1425 * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1427 (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1428 adjust the register addis loads up in the peephole2 pass. Do not
1429 depend on the register live/dead state when the peephole pass is done.
1431 2013-08-16 David Malcolm <dmalcolm@redhat.com>
1433 * gengtype.c (create_user_defined_type): Ensure that the kind
1434 is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1435 declaration is seen before the GTY((user)) marking.
1437 2013-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
1440 * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1442 2013-08-16 Jan Hubicka <jh@suse.cz>
1444 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1445 arugment expected_type.
1446 (gimple_fold_call): Use it.
1447 * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1448 * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1449 * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1450 (try_make_edge_direct_virtual_call): Likewise.
1451 * tree.c (obj_type_ref_class): New.
1452 * tree.h (obj_type_ref_class): Use it.
1454 2013-08-16 Gabriel Dos Reis <gdr@integrable-solutions.net>
1456 * sched-vis.c (rtl_slim_pp_initialized): Remove.
1457 (rtl_slim_pp): Likewise.
1458 (init_rtl_slim_pretty_print): Likewise.
1459 (dump_value_slim): Don't call it. Use local pretty printer.
1460 (dump_insn_slim): Likewise.
1461 (dump_rtl_slim): Likewise.
1462 (str_pattern_slim): Likewise.
1463 * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1466 2013-08-16 Jakub Jelinek <jakub@redhat.com>
1468 PR tree-optimization/58164
1469 * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1470 walk gimple_goto_dest of GIMPLE_GOTO.
1472 PR tree-optimization/58165
1473 * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1474 bi_call must be the last stmt in a bb, don't split_block, instead
1475 use fallthru edge from it and give up if there is none.
1476 Release conds vector when returning early.
1478 2013-08-14 Xinliang David Li <davidxl@google.com>
1480 * config/i386/i386.c (ix86_option_override_internal):
1481 Remove unused variable and field.
1483 2013-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1486 * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1487 * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1488 * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1489 and Linux, correct BLKmode alignment when 128-bit alignment is
1490 required and compatibility flag is not set.
1491 (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1492 for zero-size arguments when compatibility flag is not set.
1494 2013-08-14 Jakub Jelinek <jakub@redhat.com>
1496 PR tree-optimization/58145
1497 * tree-sra.c (build_ref_for_offset): If prev_base has
1498 TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1500 2013-08-14 Xinliang David Li <davidxl@google.com>
1502 * config/i386/i386.c (ix86_option_override_internal):
1503 Fix uninitialized variable error.
1505 2013-08-14 Xinliang David Li <davidxl@google.com>
1507 * config/i386/i386.opt: Define two new options.
1508 * config/i386/x86-tune.def: Add arch selector field in macros.
1509 * config/i386/i386.h: Adjust macro definition.
1510 * config/i386/i386.c (ix86_option_override_internal):
1512 (parse_mtune_ctrl_str): New function.
1513 (set_ix86_tune_features): New function.
1514 (ix86_function_specific_restore): Call the new helper function.
1516 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
1518 PR rtl-optimization/57662
1519 * sel-sched.c (code_motion_process_successors): When the current insn
1520 is removed after the recursive traversal, break from the loop.
1521 Add comments and debug printouts.
1523 2013-08-14 Jakub Jelinek <jakub@redhat.com>
1524 Alexandre Oliva <aoliva@redhat.com>
1527 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1528 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1529 there also UNSPEC_PLTOFF.
1531 2013-08-14 Marek Polacek <polacek@redhat.com>
1533 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1534 NUM_CONDITIONS bit positions.
1536 2013-08-13 Cary Coutant <ccoutant@google.com>
1538 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1539 (attr_checksum): Hash vector contents instead of pointer.
1540 (attr_checksum_ordered): Likewise.
1542 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
1544 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1545 when Pmode != word_mode. Add length_address attribute.
1546 (sse3_monitor_<mode>): Merge from sse3_monitor and
1547 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
1548 Pmode != word_mode. Update insn length attribute.
1549 * config/i386/i386.c (ix86_option_override_internal): Update
1550 ix86_gen_monitor selection for merged sse3_monitor insn.
1552 2013-08-13 Julian Brown <julian@codesourcery.com>
1554 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1555 perform invalid legitimization on greater-than-word-size modes for
1558 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
1560 * ira.c (setup_class_translate_array): Use aclass instead of cl
1561 for classes not fully covered by allocno classes.
1563 2013-08-13 Jakub Jelinek <jakub@redhat.com>
1565 PR tree-optimization/57661
1566 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1567 * tree-inline.c (tree_function_versioning): Initialize it.
1568 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1569 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1570 that is not being copied.
1573 * asan.c (instrument_strlen_call): Fix typo in comment.
1574 Use char * type even for the lhs of POINTER_PLUS_EXPR.
1576 2013-08-13 Steve Ellcey <sellcey@mips.com>
1578 * config/mips/mips.md (prefetch): Use lw instead of ld on
1579 loongson in 32bit mode.
1581 2013-08-13 Nick Clifton <nickc@redhat.com>
1583 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1585 2013-08-13 Jan Hubicka <jh@suse.cz>
1587 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1588 introduced edge; fix typo in sanity check.
1589 (cgraph_resolve_speculation): Export; improve diagnostic.
1590 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1591 speculation at type mismatch.
1592 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1593 (cgraph_resolve_speculation): Declare.
1594 (symtab_can_be_discarded): New function.
1595 * value-prof.c (gimple_ic_transform): Remove actual transform code.
1596 * ipa-inline-transform.c (speculation_removed): New global var.
1597 (clone_inlined_nodes): See if speculation can be removed.
1598 (inline_call): If speculations was removed, we growths may not match.
1599 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1600 (speculation_useful_p): New function.
1601 (resolve_noninline_speculation): New function.
1602 (inline_small_functions): Resolve useless speculations.
1603 * ipa-inline.h (speculation_useful_p): Declare
1604 * ipa.c (can_replace_by_local_alias): Simplify.
1605 (ipa_profile): Produce speculative calls in non-lto, too;
1606 add simple cost model; produce local aliases.
1608 2013-08-13 David Malcolm <dmalcolm@redhat.com>
1610 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1613 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
1615 * config/i386/i386.c (ix86_function_versions): Use error + inform.
1617 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
1619 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1620 iterator instead of X87MODEF.
1622 2013-08-12 Perez Read <netfirewall@gmail.com>
1625 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
1626 operand 0 for intel asm alternative.
1627 (*movabs<mode>_2): Ditto for operand 1.
1629 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
1631 * config/aarch64/arm_none.h
1632 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1634 2013-08-12 Nick Clifton <nickc@redhat.com>
1636 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1638 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
1640 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1641 expand for QI/HImode operand to produce more effictive code for
1642 unsigned char(short) --> float(double) conversion.
1644 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
1646 * doc/invoke.texi: Mention that -ftls-model does not force the final
1649 2013-08-12 Marek Polacek <polacek@redhat.com>
1650 Marc Glisse <marc.glisse@inria.fr>
1652 PR tree-optimization/57980
1653 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1654 when creating -1 constant.
1656 2013-08-10 Jan Hubicka <jh@suse.cz>
1658 Workaround binutils PR14342.
1659 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1660 (gimple_init_edge_profiler): Likewise.
1661 (gimple_gen_ic_func_profiler): Likewise.
1663 2013-08-09 Jan Hubicka <jh@suse.cz>
1665 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1667 2013-08-09 Xinliang David Li <davidxl@google.com>
1669 * config/i386/stringop.def: New file.
1670 * config/i386/stringop.opt: New file.
1671 * config/i386/i386-opts.h: Include stringopt.def.
1672 * config/i386/i386.opt: Include stringopt.opt.
1673 * config/i386/i386.c (ix86_option_override_internal):
1674 Override default size based stringop inline strategies with options.
1675 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1678 2013-08-09 Jan Hubicka <jh@suse.cz>
1680 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1682 2013-08-09 Jan Hubicka <jh@suse.cz>
1684 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1686 (dump_cgraph_node): Dump profile-id.
1687 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1688 and common_target_probability.
1689 * lto-cgraph.c (lto_output_edge): Stream common targets.
1690 (lto_output_node): Stream profile ids.
1691 (input_node): Stream profile ids.
1692 (input_edge): Stream common targets.
1693 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1694 * ipa.c: Include value-prof.h
1695 (ipa_profile_generate_summary): Turn indirect call statement histograms
1696 into common targets.
1697 (ipa_profile): Turn common targets into speculative edges.
1699 2013-08-09 Jan Hubicka <jh@suse.cz>
1701 * cgraph.h (cgraph_node): Add profile_id.
1702 * value-prof.c (cgraph_node_map): Turn into pointer_map.
1703 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1704 (del_node_map): Update.
1705 (find_func_by_funcdef_no): Replace by ...
1706 (find_func_by_profile_id): ... this one.
1707 (gimple_ic_transform): Do not remove useful histograms when
1708 speculation is not done; dump info when indirect call removal
1710 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1711 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1712 (__gcov_indirect_call_profiler_v2): .. this one.
1713 * profile.h (init_node_map): Update.
1714 * coverage.c (coverage_compute_profile_id): New function.
1715 * coverage.h (coverage_compute_profile_id): Declare.
1716 * tree-profile.c (init_ic_make_global_vars): Make
1717 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
1718 (gimple_init_edge_profiler): Update prototype of
1719 __gcov_indirect_call_profiler.
1720 (gimple_gen_ic_func_profiler): Simplify.
1721 (tree_profiling): Use init_node_map
1723 2013-08-09 Jan Hubicka <jh@suse.cz>
1725 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1726 non-speculative refs.
1727 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1728 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1729 (cgraph_set_call_stmt): Likewise.
1730 (cgraph_create_edge_1): Fix release checking compilatoin;
1732 (cgraph_free_edge): Free indirect info.
1733 (cgraph_turn_edge_to_speculative): New function.
1734 (cgraph_speculative_call_info): New function.
1735 (cgraph_make_edge_direct): Return direct edge; handle speculation.
1736 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1737 (dump_cgraph_node): Dump speculation.
1738 (verify_edge_count_and_frequency): Accept speculative edges.
1739 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1740 (verify_cgraph_node): Handle speculation.
1741 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1742 (cgraph_set_call_stmt): Update prototype.
1743 (cgraph_make_edge_direct): Update prototype.
1744 (cgraph_speculative_call_info): Declare.
1745 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1746 to change; update call of ipa_find_references.
1747 * ipa-ref.c (ipa_record_reference): Fix return value; clear
1748 lto_stmt_uid and speculative flags.
1749 (ipa_dump_references): Dump speculation.
1750 (ipa_clone_references): Clone speculative flag.
1751 (ipa_clone_referring): Likewise.
1752 (ipa_clone_ref): New function.
1753 (ipa_find_reference): Look into lto_stmt_uids
1754 (ipa_clear_stmts_in_references): Do not clear speculative calls.
1755 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1756 (ipa_find_reference): Update declaration.
1757 (ipa_clone_ref): Declare.
1758 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1759 stream speculative flag.
1760 (lto_output_ref): Stream statements uids and speculation.
1761 (input_ref): Likewise.
1762 (input_edge): Stream speuclation.
1763 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1764 (cgraph_set_call_stmt_including_clones): Handle speculation.
1765 * ipa-inline.c (heap_edge_removal_hook): New function.
1766 (inline_small_functions): Register it.
1767 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1768 also initialize refs.
1769 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1771 (try_make_edge_direct_simple_call): Likewise.
1772 (try_make_edge_direct_simple_call): Likewise.
1773 (update_indirect_edges_after_inlining): Likewise.
1774 (remove_described_reference): Look proper lto_stmt_uid.
1775 (propagate_controlled_uses): Likewise.
1776 (propagate_controlled_uses): Liekwise.
1777 * tree-inline.c (copy_bb): Copy speculative edges.
1778 (redirect_all_calls): New function.
1779 (copy_cfg_body): Do redirection after loop info is updated.
1780 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1782 2013-08-09 Jan Hubicka <jh@suse.cz>
1784 * lto-streamer-out.c (output_function): Renumber PHIs.
1785 * lto-streamer-in.c (input_function): Likewise.
1787 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
1789 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1790 (get_lane_unsigned): Likewise.
1791 (dup_lane_scalar): Likewise.
1792 (get_lane): enable for VALL.
1793 * config/aarch64/aarch64-simd.md
1794 (aarch64_dup_lane_scalar<mode>): Remove.
1795 (aarch64_get_lane_signed<mode>): Likewise.
1796 (aarch64_get_lane_unsigned<mode>): Likewise.
1797 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1798 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1799 (aarch64_get_lane<mode>): Enable for all vector modes.
1800 (aarch64_get_lanedi): Remove misleading constraints.
1801 * config/aarch64/arm_neon.h
1802 (__aarch64_vget_lane_any): Define.
1803 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1804 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1805 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1806 * config/aarch64/iterators.md (VDQQH): New.
1810 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
1812 * configure.ac: Add GAS check for LEON instructions on SPARC.
1813 * configure: Regenerate.
1814 * config.in: Likewise.
1815 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1817 * config/sparc/sparc.opt (LEON, LEON3): New masks.
1818 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1820 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1821 (AS_LEON_FLAG): New macro.
1822 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1823 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1824 Deal with LEON and LEON3 for the memory model.
1825 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1826 (atomic_compare_and_swap<mode>_1): Likewise.
1827 (*atomic_compare_and_swap<mode>_1): Likewise.
1829 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1831 * config/arm/neon.md (vcond): Fix floating-point vector
1832 comparisons against 0.
1834 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
1836 * lra-constraints.c (emit_spill_move): Remove assert.
1837 (process_alt_operands): Add more debugging
1838 output. Increase reject for spilling into memory. Decrease
1839 reject for reloading scratch.
1840 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1842 2013-08-08 Steve Ellcey <sellcey@mips.com>
1844 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1845 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1846 micromips incompatible. Add nan2008.
1847 (MULTILIB_DIRNAMES): Add nan2008.
1848 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1849 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1850 and micromips incompatible. Add nan2008.
1851 (MULTILIB_DIRNAMES): Add nan2008.
1852 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1854 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
1856 PR rtl-optimization/58079
1857 * combine.c (combine_simplify_rtx): Avoid using SUBST if
1858 simplify_comparison has widened a comparison with an integer.
1860 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1862 * config/arm/neon.md (movmisalign<mode>): Disable when we
1863 don't allow unaligned accesses.
1864 (*movmisalign<mode>_neon_store): Likewise.
1865 (*movmisalign<mode>_neon_load): Likewise.
1866 (*movmisalign<mode>_neon_store): Likewise.
1867 (*movmisalign<mode>_neon_load): Likewise.
1869 2013-08-08 Jan Hubicka <jh@suse.cz>
1871 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1872 (make_pass_rebuild_cgraph_edges): Also clear references.
1873 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1874 * ipa-inline-transform.c (inline_transform): Remove all references
1876 * cgraphunit.c (expand_function): Remove all references after
1878 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1879 (ipa_find_reference): Rewrite to iterator.
1880 (remove_stmt_references): Likewise.
1881 (ipa_clear_stmts_in_references): New function.
1882 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1883 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1885 * ipa-split.c (split_function): Remove references in split function.
1887 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
1890 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1891 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1893 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
1896 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1897 suggested mode for the assignment isn't compatible with the
1900 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
1903 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1905 2013-08-07 Xinliang David Li <davidxl@google.com>
1907 * config/i386/i386.opt: New option -mtune-ctrl=.
1908 * config/i386/x86-tune.def: New file.
1909 * config/i386/i386.h: include x86-tune.def.
1910 * config/i386/i386.c (ix86_option_override_internal):
1911 Parsing -mtune-ctrl= option and set tune features.
1913 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
1916 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1917 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1920 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
1922 * rtl.h (update_alignments): Declare.
1923 * final.c (grow_label_align): New function extracted from...
1924 (shorten_branches): ...here. Call it.
1925 (update_alignments): New function.
1926 * reorg.c (sibling_labels): New variable.
1927 (get_label_before): Add SIBLING parameter. If it is non-zero, push
1928 the new label along with it onto the sibling_labels vector.
1929 (fill_simple_delay_slots): Adjust call to get_label_before.
1930 (fill_slots_from_thread): Likewise.
1931 (relax_delay_slots): Likewise.
1932 (make_return_insns): Likewise.
1933 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1935 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
1937 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1938 document its semantics.
1939 (diagnostic_report_diagnostic): Adjust accordingly.
1941 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1943 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1944 (sparc_option_override): ...and port to new C++ pass API.
1945 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1947 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
1949 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1951 2013-08-06 Caroline Tice <cmtice@google.com>
1953 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1954 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1955 * tree-pass.h: Add pass_vtable_verify.
1956 * varasm.c (assemble_variable): Add code to properly set the comdat
1957 section and name for the .vtable_map_vars section.
1958 (assemble_vtyv_preinit_initializer): New function.
1959 (default_sectin_type_flags): Make sure .vtable_map_vars section has
1961 * output.h: Add function decl for assemble_vtv_preinit_initializer.
1962 * vtable-verify.c: New file.
1963 * vtable-verify.h: New file.
1964 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1965 initialiation levels.
1966 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1967 * passes.def: Insert pass_vtable_verify.
1968 * aclocal.m4: Reorder includes.
1969 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
1970 -fvtv-counts options.
1971 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1972 as appropriate, if -fvtable-verify=... is used.
1973 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1974 -fvtable-verify=... is used.
1975 * Makefile.in (OBJS): Add vtable-verify.o to list.
1976 (vtable-verify.o): Add new build rule.
1977 (GTFILES): Add vtable-verify.c to list.
1978 * common.opt (fvtable-verify=): New flag.
1979 (vtv_priority): Values for fvtable-verify= flag.
1980 (fvtv-counts): New flag.
1981 (fvtv-debug): New flag.
1982 * tree.h (save_vtable_map_decl): New extern function decl.
1984 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1986 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1987 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1988 (pass_data_rl78_devirt): ...new pass_data instance and...
1989 (make_pass_rl78_devirt): ...new function.
1990 (rl78_asm_file_start): Port pass registration to new C++ API.
1992 2013-08-07 David Malcolm <dmalcolm@redhat.com>
1994 * coretypes.h (rtl_opt_pass): Add.
1995 (gcc::context): Add.
1996 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1997 (epiphany_init): Port to new C++ pass API.
1998 (epiphany_optimize_mode_switching): Likewise.
1999 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
2000 (pass_manager::get_pass_mode_switching): New.
2001 (pass_manager::get_pass_peephole2): New.
2002 * mode-switching.c (pass_mode_switching): Add clone method.
2003 * recog.c (pass_peephole2): Add clone method.
2004 (pass_split_all_insns): Add clone method.
2006 2013-08-06 David Malcolm <dmalcolm@redhat.com>
2008 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
2009 (mips_option_override): ...here, porting to new C++ API for passes.
2011 2013-08-06 Jan Hubicka <jh@suse.cz>
2013 * cgraph.c (cgraph_get_body): New function based on lto.c
2015 * cgraph.h (cgraph_get_body): Declare.
2016 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
2018 * cgraphunit.c (expand_function): Get body prior expanding.
2019 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
2020 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
2022 * passes.c (do_per_function_toporder): Get body.
2023 * tree-inline.c (expand_call_inline): Get body prior inlining it.
2024 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
2026 2013-08-06 Martin Jambor <mjambor@suse.cz>
2029 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
2030 is not re-finalized. Rename second parameter to no_collect.
2032 2013-08-06 Martin Jambor <mjambor@suse.cz>
2035 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
2036 MEM_REF has proper alignment information.
2038 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
2041 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
2043 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
2044 rtx (*) (rtx, ...) with insn_gen_fn.
2045 * genoutput.c (output_insn_data): Cast gen_? function pointers to
2046 insn_gen_fn::stored_funcptr. Add initializer braces.
2048 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2050 Rewrite how instances of passes are cloned to remove assumptions
2051 about their sizes (thus allowing pass subclasses to have
2052 additional data fields, albeit non-GC-managed ones at this point).
2054 * passes.c (make_pass_instance): Now that passes have clone
2055 methods, rewrite this function to eliminate XNEW and memcpy
2056 calls that used hardcoded sizes. Since this function no longer
2057 creates pass instances, rename it to...
2058 (add_pass_instance): ...this. Document the old way that passes were
2059 numbered and flagged, and rework this function to continue using it.
2060 (next_pass_1): Add an initial_pass argument for use by
2062 (position_pass): When adding multiple instances of a pass, use
2063 the pass's clone method, rather than relying on the XNEW/memcpy
2064 within the former make_pass_instance (now add_pass_instance).
2065 (pass_manager::pass_manager): When invoking next_pass_1, also supply
2066 the initial instance of the current pass within the pass manager.
2068 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2070 This is the automated part of the conversion of passes from C
2071 structs to C++ classes.
2073 Patch autogenerated by refactor_passes.py from
2074 https://github.com/davidmalcolm/gcc-refactoring-scripts
2075 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
2077 * asan.c (pass_asan): Convert from a global struct to a subclass of
2078 gimple_opt_pass along with...
2079 (pass_data_asan): ...new pass_data instance and...
2080 (make_pass_asan): ...new function.
2081 (pass_asan_O0): Convert from a global struct to a subclass of
2082 gimple_opt_pass along with...
2083 (pass_data_asan_O0): ...new pass_data instance and...
2084 (make_pass_asan_O0): ...new function.
2085 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
2086 subclass of rtl_opt_pass along with...
2087 (pass_data_inc_dec): ...new pass_data instance and...
2088 (make_pass_inc_dec): ...new function.
2089 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
2090 a subclass of rtl_opt_pass along with...
2091 (pass_data_reorder_blocks): ...new pass_data instance and...
2092 (make_pass_reorder_blocks): ...new function.
2093 (pass_duplicate_computed_gotos): Convert from a global struct to a
2094 subclass of rtl_opt_pass along with...
2095 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
2096 (make_pass_duplicate_computed_gotos): ...new function.
2097 (pass_partition_blocks): Convert from a global struct to a subclass of
2098 rtl_opt_pass along with...
2099 (pass_data_partition_blocks): ...new pass_data instance and...
2100 (make_pass_partition_blocks): ...new function.
2101 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
2102 struct to a subclass of rtl_opt_pass along with...
2103 (pass_data_branch_target_load_optimize1): ...new pass_data instance
2105 (make_pass_branch_target_load_optimize1): ...new function.
2106 (pass_branch_target_load_optimize2): Convert from a global struct to a
2107 subclass of rtl_opt_pass along with...
2108 (pass_data_branch_target_load_optimize2): ...new pass_data instance
2110 (make_pass_branch_target_load_optimize2): ...new function.
2111 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
2112 of rtl_opt_pass along with...
2113 (pass_data_jump): ...new pass_data instance and...
2114 (make_pass_jump): ...new function.
2115 (pass_jump2): Convert from a global struct to a subclass of
2116 rtl_opt_pass along with...
2117 (pass_data_jump2): ...new pass_data instance and...
2118 (make_pass_jump2): ...new function.
2119 * cfgexpand.c (pass_expand): Convert from a global struct to a
2120 subclass of rtl_opt_pass along with...
2121 (pass_data_expand): ...new pass_data instance and...
2122 (make_pass_expand): ...new function.
2123 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
2124 of rtl_opt_pass along with...
2125 (pass_data_free_cfg): ...new pass_data instance and...
2126 (make_pass_free_cfg): ...new function.
2127 (pass_into_cfg_layout_mode): Convert from a global struct to a
2128 subclass of rtl_opt_pass along with...
2129 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
2130 (make_pass_into_cfg_layout_mode): ...new function.
2131 (pass_outof_cfg_layout_mode): Convert from a global struct to a
2132 subclass of rtl_opt_pass along with...
2133 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
2134 (make_pass_outof_cfg_layout_mode): ...new function.
2135 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
2136 struct to a subclass of gimple_opt_pass along with...
2137 (pass_data_build_cgraph_edges): ...new pass_data instance and...
2138 (make_pass_build_cgraph_edges): ...new function.
2139 (pass_rebuild_cgraph_edges): Convert from a global struct to a
2140 subclass of gimple_opt_pass along with...
2141 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
2142 (make_pass_rebuild_cgraph_edges): ...new function.
2143 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
2144 subclass of gimple_opt_pass along with...
2145 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
2147 (make_pass_remove_cgraph_callee_edges): ...new function.
2148 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
2149 struct to a subclass of rtl_opt_pass along with...
2150 (pass_data_stack_adjustments): ...new pass_data instance and...
2151 (make_pass_stack_adjustments): ...new function.
2152 * combine.c (pass_combine): Convert from a global struct to a subclass
2153 of rtl_opt_pass along with...
2154 (pass_data_combine): ...new pass_data instance and...
2155 (make_pass_combine): ...new function.
2156 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
2157 global struct to a subclass of rtl_opt_pass along with...
2158 (pass_data_compare_elim_after_reload): ...new pass_data instance
2160 (make_pass_compare_elim_after_reload): ...new function.
2161 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
2162 of rtl_opt_pass along with...
2163 (pass_data_rtl_cprop): ...new pass_data instance and...
2164 (make_pass_rtl_cprop): ...new function.
2165 * cse.c (pass_cse): Convert from a global struct to a subclass of
2166 rtl_opt_pass along with...
2167 (pass_data_cse): ...new pass_data instance and...
2168 (make_pass_cse): ...new function.
2169 (pass_cse2): Convert from a global struct to a subclass of
2170 rtl_opt_pass along with...
2171 (pass_data_cse2): ...new pass_data instance and...
2172 (make_pass_cse2): ...new function.
2173 (pass_cse_after_global_opts): Convert from a global struct to a
2174 subclass of rtl_opt_pass along with...
2175 (pass_data_cse_after_global_opts): ...new pass_data instance and...
2176 (make_pass_cse_after_global_opts): ...new function.
2177 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
2178 of rtl_opt_pass along with...
2179 (pass_data_ud_rtl_dce): ...new pass_data instance and...
2180 (make_pass_ud_rtl_dce): ...new function.
2181 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
2182 rtl_opt_pass along with...
2183 (pass_data_fast_rtl_dce): ...new pass_data instance and...
2184 (make_pass_fast_rtl_dce): ...new function.
2185 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
2186 a subclass of rtl_opt_pass along with...
2187 (pass_data_df_initialize_opt): ...new pass_data instance and...
2188 (make_pass_df_initialize_opt): ...new function.
2189 (pass_df_initialize_no_opt): Convert from a global struct to a
2190 subclass of rtl_opt_pass along with...
2191 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
2192 (make_pass_df_initialize_no_opt): ...new function.
2193 (pass_df_finish): Convert from a global struct to a subclass of
2194 rtl_opt_pass along with...
2195 (pass_data_df_finish): ...new pass_data instance and...
2196 (make_pass_df_finish): ...new function.
2197 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
2198 rtl_opt_pass along with...
2199 (pass_data_rtl_dse1): ...new pass_data instance and...
2200 (make_pass_rtl_dse1): ...new function.
2201 (pass_rtl_dse2): Convert from a global struct to a subclass of
2202 rtl_opt_pass along with...
2203 (pass_data_rtl_dse2): ...new pass_data instance and...
2204 (make_pass_rtl_dse2): ...new function.
2205 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
2206 subclass of rtl_opt_pass along with...
2207 (pass_data_dwarf2_frame): ...new pass_data instance and...
2208 (make_pass_dwarf2_frame): ...new function.
2209 * except.c (pass_set_nothrow_function_flags): Convert from a global
2210 struct to a subclass of rtl_opt_pass along with...
2211 (pass_data_set_nothrow_function_flags): ...new pass_data instance
2213 (make_pass_set_nothrow_function_flags): ...new function.
2214 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
2215 subclass of rtl_opt_pass along with...
2216 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
2218 (make_pass_convert_to_eh_region_ranges): ...new function.
2219 * final.c (pass_compute_alignments): Convert from a global struct to a
2220 subclass of rtl_opt_pass along with...
2221 (pass_data_compute_alignments): ...new pass_data instance and...
2222 (make_pass_compute_alignments): ...new function.
2223 (pass_final): Convert from a global struct to a subclass of
2224 rtl_opt_pass along with...
2225 (pass_data_final): ...new pass_data instance and...
2226 (make_pass_final): ...new function.
2227 (pass_shorten_branches): Convert from a global struct to a subclass of
2228 rtl_opt_pass along with...
2229 (pass_data_shorten_branches): ...new pass_data instance and...
2230 (make_pass_shorten_branches): ...new function.
2231 (pass_clean_state): Convert from a global struct to a subclass of
2232 rtl_opt_pass along with...
2233 (pass_data_clean_state): ...new pass_data instance and...
2234 (make_pass_clean_state): ...new function.
2235 * function.c (pass_instantiate_virtual_regs): Convert from a global
2236 struct to a subclass of rtl_opt_pass along with...
2237 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
2238 (make_pass_instantiate_virtual_regs): ...new function.
2239 (pass_leaf_regs): Convert from a global struct to a subclass of
2240 rtl_opt_pass along with...
2241 (pass_data_leaf_regs): ...new pass_data instance and...
2242 (make_pass_leaf_regs): ...new function.
2243 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
2244 subclass of rtl_opt_pass along with...
2245 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
2247 (make_pass_thread_prologue_and_epilogue): ...new function.
2248 (pass_match_asm_constraints): Convert from a global struct to a
2249 subclass of rtl_opt_pass along with...
2250 (pass_data_match_asm_constraints): ...new pass_data instance and...
2251 (make_pass_match_asm_constraints): ...new function.
2252 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
2253 subclass of rtl_opt_pass along with...
2254 (pass_data_rtl_fwprop): ...new pass_data instance and...
2255 (make_pass_rtl_fwprop): ...new function.
2256 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
2257 rtl_opt_pass along with...
2258 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
2259 (make_pass_rtl_fwprop_addr): ...new function.
2260 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
2261 rtl_opt_pass along with...
2262 (pass_data_rtl_pre): ...new pass_data instance and...
2263 (make_pass_rtl_pre): ...new function.
2264 (pass_rtl_hoist): Convert from a global struct to a subclass of
2265 rtl_opt_pass along with...
2266 (pass_data_rtl_hoist): ...new pass_data instance and...
2267 (make_pass_rtl_hoist): ...new function.
2268 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
2269 subclass of gimple_opt_pass along with...
2270 (pass_data_lower_cf): ...new pass_data instance and...
2271 (make_pass_lower_cf): ...new function.
2272 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
2273 from a global struct to a subclass of gimple_opt_pass along with...
2274 (pass_data_strength_reduction): ...new pass_data instance and...
2275 (make_pass_strength_reduction): ...new function.
2276 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
2277 of rtl_opt_pass along with...
2278 (pass_data_rtl_ifcvt): ...new pass_data instance and...
2279 (make_pass_rtl_ifcvt): ...new function.
2280 (pass_if_after_combine): Convert from a global struct to a subclass of
2281 rtl_opt_pass along with...
2282 (pass_data_if_after_combine): ...new pass_data instance and...
2283 (make_pass_if_after_combine): ...new function.
2284 (pass_if_after_reload): Convert from a global struct to a subclass of
2285 rtl_opt_pass along with...
2286 (pass_data_if_after_reload): ...new pass_data instance and...
2287 (make_pass_if_after_reload): ...new function.
2288 * init-regs.c (pass_initialize_regs): Convert from a global struct to
2289 a subclass of rtl_opt_pass along with...
2290 (pass_data_initialize_regs): ...new pass_data instance and...
2291 (make_pass_initialize_regs): ...new function.
2292 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
2293 of ipa_opt_pass_d along with...
2294 (pass_data_ipa_cp): ...new pass_data instance and...
2295 (make_pass_ipa_cp): ...new function.
2296 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
2297 global struct to a subclass of gimple_opt_pass along with...
2298 (pass_data_inline_parameters): ...new pass_data instance and...
2299 (make_pass_inline_parameters): ...new function.
2300 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
2301 subclass of gimple_opt_pass along with...
2302 (pass_data_early_inline): ...new pass_data instance and...
2303 (make_pass_early_inline): ...new function.
2304 (pass_ipa_inline): Convert from a global struct to a subclass of
2305 ipa_opt_pass_d along with...
2306 (pass_data_ipa_inline): ...new pass_data instance and...
2307 (make_pass_ipa_inline): ...new function.
2308 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
2309 struct to a subclass of gimple_opt_pass along with...
2310 (pass_data_local_pure_const): ...new pass_data instance and...
2311 (make_pass_local_pure_const): ...new function.
2312 (pass_ipa_pure_const): Convert from a global struct to a subclass of
2313 ipa_opt_pass_d along with...
2314 (pass_data_ipa_pure_const): ...new pass_data instance and...
2315 (make_pass_ipa_pure_const): ...new function.
2316 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
2317 to a subclass of ipa_opt_pass_d along with...
2318 (pass_data_ipa_reference): ...new pass_data instance and...
2319 (make_pass_ipa_reference): ...new function.
2320 * ipa-split.c (pass_split_functions): Convert from a global struct to
2321 a subclass of gimple_opt_pass along with...
2322 (pass_data_split_functions): ...new pass_data instance and...
2323 (make_pass_split_functions): ...new function.
2324 (pass_feedback_split_functions): Convert from a global struct to a
2325 subclass of gimple_opt_pass along with...
2326 (pass_data_feedback_split_functions): ...new pass_data instance and...
2327 (make_pass_feedback_split_functions): ...new function.
2328 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
2329 global struct to a subclass of simple_ipa_opt_pass along with...
2330 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
2332 (make_pass_ipa_function_and_variable_visibility): ...new function.
2333 (pass_ipa_free_inline_summary): Convert from a global struct to a
2334 subclass of simple_ipa_opt_pass along with...
2335 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
2336 (make_pass_ipa_free_inline_summary): ...new function.
2337 (pass_ipa_whole_program_visibility): Convert from a global struct to a
2338 subclass of ipa_opt_pass_d along with...
2339 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
2341 (make_pass_ipa_whole_program_visibility): ...new function.
2342 (pass_ipa_profile): Convert from a global struct to a subclass of
2343 ipa_opt_pass_d along with...
2344 (pass_data_ipa_profile): ...new pass_data instance and...
2345 (make_pass_ipa_profile): ...new function.
2346 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
2347 ipa_opt_pass_d along with...
2348 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
2349 (make_pass_ipa_cdtor_merge): ...new function.
2350 * ira.c (pass_ira): Convert from a global struct to a subclass of
2351 rtl_opt_pass along with...
2352 (pass_data_ira): ...new pass_data instance and...
2353 (make_pass_ira): ...new function.
2354 (pass_reload): Convert from a global struct to a subclass of
2355 rtl_opt_pass along with...
2356 (pass_data_reload): ...new pass_data instance and...
2357 (make_pass_reload): ...new function.
2358 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
2359 subclass of rtl_opt_pass along with...
2360 (pass_data_cleanup_barriers): ...new pass_data instance and...
2361 (make_pass_cleanup_barriers): ...new function.
2362 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
2363 of rtl_opt_pass along with...
2364 (pass_data_loop2): ...new pass_data instance and...
2365 (make_pass_loop2): ...new function.
2366 (pass_rtl_loop_init): Convert from a global struct to a subclass of
2367 rtl_opt_pass along with...
2368 (pass_data_rtl_loop_init): ...new pass_data instance and...
2369 (make_pass_rtl_loop_init): ...new function.
2370 (pass_rtl_loop_done): Convert from a global struct to a subclass of
2371 rtl_opt_pass along with...
2372 (pass_data_rtl_loop_done): ...new pass_data instance and...
2373 (make_pass_rtl_loop_done): ...new function.
2374 (pass_rtl_move_loop_invariants): Convert from a global struct to a
2375 subclass of rtl_opt_pass along with...
2376 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
2377 (make_pass_rtl_move_loop_invariants): ...new function.
2378 (pass_rtl_unswitch): Convert from a global struct to a subclass of
2379 rtl_opt_pass along with...
2380 (pass_data_rtl_unswitch): ...new pass_data instance and...
2381 (make_pass_rtl_unswitch): ...new function.
2382 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
2383 subclass of rtl_opt_pass along with...
2384 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
2386 (make_pass_rtl_unroll_and_peel_loops): ...new function.
2387 (pass_rtl_doloop): Convert from a global struct to a subclass of
2388 rtl_opt_pass along with...
2389 (pass_data_rtl_doloop): ...new pass_data instance and...
2390 (make_pass_rtl_doloop): ...new function.
2391 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
2392 a subclass of rtl_opt_pass along with...
2393 (pass_data_lower_subreg): ...new pass_data instance and...
2394 (make_pass_lower_subreg): ...new function.
2395 (pass_lower_subreg2): Convert from a global struct to a subclass of
2396 rtl_opt_pass along with...
2397 (pass_data_lower_subreg2): ...new pass_data instance and...
2398 (make_pass_lower_subreg2): ...new function.
2399 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
2400 struct to a subclass of ipa_opt_pass_d along with...
2401 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
2402 (make_pass_ipa_lto_gimple_out): ...new function.
2403 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
2404 of ipa_opt_pass_d along with...
2405 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2406 (make_pass_ipa_lto_finish_out): ...new function.
2407 * mode-switching.c (pass_mode_switching): Convert from a global struct
2408 to a subclass of rtl_opt_pass along with...
2409 (pass_data_mode_switching): ...new pass_data instance and...
2410 (make_pass_mode_switching): ...new function.
2411 * modulo-sched.c (pass_sms): Convert from a global struct to a
2412 subclass of rtl_opt_pass along with...
2413 (pass_data_sms): ...new pass_data instance and...
2414 (make_pass_sms): ...new function.
2415 * omp-low.c (pass_expand_omp): Convert from a global struct to a
2416 subclass of gimple_opt_pass along with...
2417 (pass_data_expand_omp): ...new pass_data instance and...
2418 (make_pass_expand_omp): ...new function.
2419 (pass_lower_omp): Convert from a global struct to a subclass of
2420 gimple_opt_pass along with...
2421 (pass_data_lower_omp): ...new pass_data instance and...
2422 (make_pass_lower_omp): ...new function.
2423 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2424 of gimple_opt_pass along with...
2425 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2426 (make_pass_diagnose_omp_blocks): ...new function.
2427 * passes.c (pass_early_local_passes): Convert from a global struct to
2428 a subclass of simple_ipa_opt_pass along with...
2429 (pass_data_early_local_passes): ...new pass_data instance and...
2430 (make_pass_early_local_passes): ...new function.
2431 (pass_all_early_optimizations): Convert from a global struct to a
2432 subclass of gimple_opt_pass along with...
2433 (pass_data_all_early_optimizations): ...new pass_data instance and...
2434 (make_pass_all_early_optimizations): ...new function.
2435 (pass_all_optimizations): Convert from a global struct to a subclass
2436 of gimple_opt_pass along with...
2437 (pass_data_all_optimizations): ...new pass_data instance and...
2438 (make_pass_all_optimizations): ...new function.
2439 (pass_all_optimizations_g): Convert from a global struct to a subclass
2440 of gimple_opt_pass along with...
2441 (pass_data_all_optimizations_g): ...new pass_data instance and...
2442 (make_pass_all_optimizations_g): ...new function.
2443 (pass_rest_of_compilation): Convert from a global struct to a subclass
2444 of rtl_opt_pass along with...
2445 (pass_data_rest_of_compilation): ...new pass_data instance and...
2446 (make_pass_rest_of_compilation): ...new function.
2447 (pass_postreload): Convert from a global struct to a subclass of
2448 rtl_opt_pass along with...
2449 (pass_data_postreload): ...new pass_data instance and...
2450 (make_pass_postreload): ...new function.
2451 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2452 subclass of rtl_opt_pass along with...
2453 (pass_data_gcse2): ...new pass_data instance and...
2454 (make_pass_gcse2): ...new function.
2455 * postreload.c (pass_postreload_cse): Convert from a global struct to
2456 a subclass of rtl_opt_pass along with...
2457 (pass_data_postreload_cse): ...new pass_data instance and...
2458 (make_pass_postreload_cse): ...new function.
2459 * predict.c (pass_profile): Convert from a global struct to a subclass
2460 of gimple_opt_pass along with...
2461 (pass_data_profile): ...new pass_data instance and...
2462 (make_pass_profile): ...new function.
2463 (pass_strip_predict_hints): Convert from a global struct to a subclass
2464 of gimple_opt_pass along with...
2465 (pass_data_strip_predict_hints): ...new pass_data instance and...
2466 (make_pass_strip_predict_hints): ...new function.
2467 * recog.c (pass_peephole2): Convert from a global struct to a subclass
2468 of rtl_opt_pass along with...
2469 (pass_data_peephole2): ...new pass_data instance and...
2470 (make_pass_peephole2): ...new function.
2471 (pass_split_all_insns): Convert from a global struct to a subclass of
2472 rtl_opt_pass along with...
2473 (pass_data_split_all_insns): ...new pass_data instance and...
2474 (make_pass_split_all_insns): ...new function.
2475 (pass_split_after_reload): Convert from a global struct to a subclass
2476 of rtl_opt_pass along with...
2477 (pass_data_split_after_reload): ...new pass_data instance and...
2478 (make_pass_split_after_reload): ...new function.
2479 (pass_split_before_regstack): Convert from a global struct to a
2480 subclass of rtl_opt_pass along with...
2481 (pass_data_split_before_regstack): ...new pass_data instance and...
2482 (make_pass_split_before_regstack): ...new function.
2483 (pass_split_before_sched2): Convert from a global struct to a subclass
2484 of rtl_opt_pass along with...
2485 (pass_data_split_before_sched2): ...new pass_data instance and...
2486 (make_pass_split_before_sched2): ...new function.
2487 (pass_split_for_shorten_branches): Convert from a global struct to a
2488 subclass of rtl_opt_pass along with...
2489 (pass_data_split_for_shorten_branches): ...new pass_data instance
2491 (make_pass_split_for_shorten_branches): ...new function.
2492 * ree.c (pass_ree): Convert from a global struct to a subclass of
2493 rtl_opt_pass along with...
2494 (pass_data_ree): ...new pass_data instance and...
2495 (make_pass_ree): ...new function.
2496 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2497 subclass of rtl_opt_pass along with...
2498 (pass_data_stack_regs): ...new pass_data instance and...
2499 (make_pass_stack_regs): ...new function.
2500 (pass_stack_regs_run): Convert from a global struct to a subclass of
2501 rtl_opt_pass along with...
2502 (pass_data_stack_regs_run): ...new pass_data instance and...
2503 (make_pass_stack_regs_run): ...new function.
2504 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2505 subclass of rtl_opt_pass along with...
2506 (pass_data_cprop_hardreg): ...new pass_data instance and...
2507 (make_pass_cprop_hardreg): ...new function.
2508 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2509 subclass of rtl_opt_pass along with...
2510 (pass_data_reginfo_init): ...new pass_data instance and...
2511 (make_pass_reginfo_init): ...new function.
2512 * regmove.c (pass_regmove): Convert from a global struct to a subclass
2513 of rtl_opt_pass along with...
2514 (pass_data_regmove): ...new pass_data instance and...
2515 (make_pass_regmove): ...new function.
2516 * regrename.c (pass_regrename): Convert from a global struct to a
2517 subclass of rtl_opt_pass along with...
2518 (pass_data_regrename): ...new pass_data instance and...
2519 (make_pass_regrename): ...new function.
2520 * reorg.c (pass_delay_slots): Convert from a global struct to a
2521 subclass of rtl_opt_pass along with...
2522 (pass_data_delay_slots): ...new pass_data instance and...
2523 (make_pass_delay_slots): ...new function.
2524 (pass_machine_reorg): Convert from a global struct to a subclass of
2525 rtl_opt_pass along with...
2526 (pass_data_machine_reorg): ...new pass_data instance and...
2527 (make_pass_machine_reorg): ...new function.
2528 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2529 of rtl_opt_pass along with...
2530 (pass_data_sched): ...new pass_data instance and...
2531 (make_pass_sched): ...new function.
2532 (pass_sched2): Convert from a global struct to a subclass of
2533 rtl_opt_pass along with...
2534 (pass_data_sched2): ...new pass_data instance and...
2535 (make_pass_sched2): ...new function.
2536 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2537 to a subclass of rtl_opt_pass along with...
2538 (pass_data_stack_ptr_mod): ...new pass_data instance and...
2539 (make_pass_stack_ptr_mod): ...new function.
2540 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2541 to a subclass of rtl_opt_pass along with...
2542 (pass_data_rtl_store_motion): ...new pass_data instance and...
2543 (make_pass_rtl_store_motion): ...new function.
2544 * tracer.c (pass_tracer): Convert from a global struct to a subclass
2545 of gimple_opt_pass along with...
2546 (pass_data_tracer): ...new pass_data instance and...
2547 (make_pass_tracer): ...new function.
2548 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2549 to a subclass of gimple_opt_pass along with...
2550 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2551 (make_pass_diagnose_tm_blocks): ...new function.
2552 (pass_lower_tm): Convert from a global struct to a subclass of
2553 gimple_opt_pass along with...
2554 (pass_data_lower_tm): ...new pass_data instance and...
2555 (make_pass_lower_tm): ...new function.
2556 (pass_tm_init): Convert from a global struct to a subclass of
2557 gimple_opt_pass along with...
2558 (pass_data_tm_init): ...new pass_data instance and...
2559 (make_pass_tm_init): ...new function.
2560 (pass_tm_mark): Convert from a global struct to a subclass of
2561 gimple_opt_pass along with...
2562 (pass_data_tm_mark): ...new pass_data instance and...
2563 (make_pass_tm_mark): ...new function.
2564 (pass_tm_edges): Convert from a global struct to a subclass of
2565 gimple_opt_pass along with...
2566 (pass_data_tm_edges): ...new pass_data instance and...
2567 (make_pass_tm_edges): ...new function.
2568 (pass_tm_memopt): Convert from a global struct to a subclass of
2569 gimple_opt_pass along with...
2570 (pass_data_tm_memopt): ...new pass_data instance and...
2571 (make_pass_tm_memopt): ...new function.
2572 (pass_ipa_tm): Convert from a global struct to a subclass of
2573 simple_ipa_opt_pass along with...
2574 (pass_data_ipa_tm): ...new pass_data instance and...
2575 (make_pass_ipa_tm): ...new function.
2576 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2577 subclass of gimple_opt_pass along with...
2578 (pass_data_call_cdce): ...new pass_data instance and...
2579 (make_pass_call_cdce): ...new function.
2580 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2581 subclass of gimple_opt_pass along with...
2582 (pass_data_build_cfg): ...new pass_data instance and...
2583 (make_pass_build_cfg): ...new function.
2584 (pass_split_crit_edges): Convert from a global struct to a subclass of
2585 gimple_opt_pass along with...
2586 (pass_data_split_crit_edges): ...new pass_data instance and...
2587 (make_pass_split_crit_edges): ...new function.
2588 (pass_warn_function_return): Convert from a global struct to a
2589 subclass of gimple_opt_pass along with...
2590 (pass_data_warn_function_return): ...new pass_data instance and...
2591 (make_pass_warn_function_return): ...new function.
2592 (pass_warn_function_noreturn): Convert from a global struct to a
2593 subclass of gimple_opt_pass along with...
2594 (pass_data_warn_function_noreturn): ...new pass_data instance and...
2595 (make_pass_warn_function_noreturn): ...new function.
2596 (pass_warn_unused_result): Convert from a global struct to a subclass
2597 of gimple_opt_pass along with...
2598 (pass_data_warn_unused_result): ...new pass_data instance and...
2599 (make_pass_warn_unused_result): ...new function.
2600 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2601 a subclass of gimple_opt_pass along with...
2602 (pass_data_merge_phi): ...new pass_data instance and...
2603 (make_pass_merge_phi): ...new function.
2604 * tree-complex.c (pass_lower_complex): Convert from a global struct to
2605 a subclass of gimple_opt_pass along with...
2606 (pass_data_lower_complex): ...new pass_data instance and...
2607 (make_pass_lower_complex): ...new function.
2608 (pass_lower_complex_O0): Convert from a global struct to a subclass of
2609 gimple_opt_pass along with...
2610 (pass_data_lower_complex_O0): ...new pass_data instance and...
2611 (make_pass_lower_complex_O0): ...new function.
2612 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2613 subclass of gimple_opt_pass along with...
2614 (pass_data_lower_eh): ...new pass_data instance and...
2615 (make_pass_lower_eh): ...new function.
2616 (pass_refactor_eh): Convert from a global struct to a subclass of
2617 gimple_opt_pass along with...
2618 (pass_data_refactor_eh): ...new pass_data instance and...
2619 (make_pass_refactor_eh): ...new function.
2620 (pass_lower_resx): Convert from a global struct to a subclass of
2621 gimple_opt_pass along with...
2622 (pass_data_lower_resx): ...new pass_data instance and...
2623 (make_pass_lower_resx): ...new function.
2624 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2625 of gimple_opt_pass along with...
2626 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2627 (make_pass_lower_eh_dispatch): ...new function.
2628 (pass_cleanup_eh): Convert from a global struct to a subclass of
2629 gimple_opt_pass along with...
2630 (pass_data_cleanup_eh): ...new pass_data instance and...
2631 (make_pass_cleanup_eh): ...new function.
2632 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2633 to a subclass of simple_ipa_opt_pass along with...
2634 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2635 (make_pass_ipa_lower_emutls): ...new function.
2636 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2637 a subclass of gimple_opt_pass along with...
2638 (pass_data_if_conversion): ...new pass_data instance and...
2639 (make_pass_if_conversion): ...new function.
2640 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2641 subclass of gimple_opt_pass along with...
2642 (pass_data_build_ssa): ...new pass_data instance and...
2643 (make_pass_build_ssa): ...new function.
2644 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2645 global struct to a subclass of gimple_opt_pass along with...
2646 (pass_data_loop_distribution): ...new pass_data instance and...
2647 (make_pass_loop_distribution): ...new function.
2648 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2649 subclass of gimple_opt_pass along with...
2650 (pass_data_mudflap_1): ...new pass_data instance and...
2651 (make_pass_mudflap_1): ...new function.
2652 (pass_mudflap_2): Convert from a global struct to a subclass of
2653 gimple_opt_pass along with...
2654 (pass_data_mudflap_2): ...new pass_data instance and...
2655 (make_pass_mudflap_2): ...new function.
2656 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2657 subclass of gimple_opt_pass along with...
2658 (pass_data_mudflap_1): ...new pass_data instance and...
2659 (make_pass_mudflap_1): ...new function.
2660 (pass_mudflap_2): Convert from a global struct to a subclass of
2661 gimple_opt_pass along with...
2662 (pass_data_mudflap_2): ...new pass_data instance and...
2663 (make_pass_mudflap_2): ...new function.
2664 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2665 gimple_opt_pass along with...
2666 (pass_data_nrv): ...new pass_data instance and...
2667 (make_pass_nrv): ...new function.
2668 (pass_return_slot): Convert from a global struct to a subclass of
2669 gimple_opt_pass along with...
2670 (pass_data_return_slot): ...new pass_data instance and...
2671 (make_pass_return_slot): ...new function.
2672 * tree-object-size.c (pass_object_sizes): Convert from a global struct
2673 to a subclass of gimple_opt_pass along with...
2674 (pass_data_object_sizes): ...new pass_data instance and...
2675 (make_pass_object_sizes): ...new function.
2676 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2677 global struct to a subclass of gimple_opt_pass along with...
2678 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2680 (make_pass_cleanup_cfg_post_optimizing): ...new function.
2681 (pass_fixup_cfg): Convert from a global struct to a subclass of
2682 gimple_opt_pass along with...
2683 (pass_data_fixup_cfg): ...new pass_data instance and...
2684 (make_pass_fixup_cfg): ...new function.
2685 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2686 (make_pass_mudflap_1): ...new function.
2687 (pass_mudflap_2): Replace declaration with that of...
2688 (make_pass_mudflap_2): ...new function.
2689 (pass_asan): Replace declaration with that of...
2690 (make_pass_asan): ...new function.
2691 (pass_asan_O0): Replace declaration with that of...
2692 (make_pass_asan_O0): ...new function.
2693 (pass_tsan): Replace declaration with that of...
2694 (make_pass_tsan): ...new function.
2695 (pass_tsan_O0): Replace declaration with that of...
2696 (make_pass_tsan_O0): ...new function.
2697 (pass_lower_cf): Replace declaration with that of...
2698 (make_pass_lower_cf): ...new function.
2699 (pass_refactor_eh): Replace declaration with that of...
2700 (make_pass_refactor_eh): ...new function.
2701 (pass_lower_eh): Replace declaration with that of...
2702 (make_pass_lower_eh): ...new function.
2703 (pass_lower_eh_dispatch): Replace declaration with that of...
2704 (make_pass_lower_eh_dispatch): ...new function.
2705 (pass_lower_resx): Replace declaration with that of...
2706 (make_pass_lower_resx): ...new function.
2707 (pass_build_cfg): Replace declaration with that of...
2708 (make_pass_build_cfg): ...new function.
2709 (pass_early_tree_profile): Replace declaration with that of...
2710 (make_pass_early_tree_profile): ...new function.
2711 (pass_cleanup_eh): Replace declaration with that of...
2712 (make_pass_cleanup_eh): ...new function.
2713 (pass_sra): Replace declaration with that of...
2714 (make_pass_sra): ...new function.
2715 (pass_sra_early): Replace declaration with that of...
2716 (make_pass_sra_early): ...new function.
2717 (pass_early_ipa_sra): Replace declaration with that of...
2718 (make_pass_early_ipa_sra): ...new function.
2719 (pass_tail_recursion): Replace declaration with that of...
2720 (make_pass_tail_recursion): ...new function.
2721 (pass_tail_calls): Replace declaration with that of...
2722 (make_pass_tail_calls): ...new function.
2723 (pass_tree_loop): Replace declaration with that of...
2724 (make_pass_tree_loop): ...new function.
2725 (pass_tree_loop_init): Replace declaration with that of...
2726 (make_pass_tree_loop_init): ...new function.
2727 (pass_lim): Replace declaration with that of...
2728 (make_pass_lim): ...new function.
2729 (pass_tree_unswitch): Replace declaration with that of...
2730 (make_pass_tree_unswitch): ...new function.
2731 (pass_predcom): Replace declaration with that of...
2732 (make_pass_predcom): ...new function.
2733 (pass_iv_canon): Replace declaration with that of...
2734 (make_pass_iv_canon): ...new function.
2735 (pass_scev_cprop): Replace declaration with that of...
2736 (make_pass_scev_cprop): ...new function.
2737 (pass_empty_loop): Replace declaration with that of...
2738 (make_pass_empty_loop): ...new function.
2739 (pass_record_bounds): Replace declaration with that of...
2740 (make_pass_record_bounds): ...new function.
2741 (pass_graphite): Replace declaration with that of...
2742 (make_pass_graphite): ...new function.
2743 (pass_graphite_transforms): Replace declaration with that of...
2744 (make_pass_graphite_transforms): ...new function.
2745 (pass_if_conversion): Replace declaration with that of...
2746 (make_pass_if_conversion): ...new function.
2747 (pass_loop_distribution): Replace declaration with that of...
2748 (make_pass_loop_distribution): ...new function.
2749 (pass_vectorize): Replace declaration with that of...
2750 (make_pass_vectorize): ...new function.
2751 (pass_slp_vectorize): Replace declaration with that of...
2752 (make_pass_slp_vectorize): ...new function.
2753 (pass_complete_unroll): Replace declaration with that of...
2754 (make_pass_complete_unroll): ...new function.
2755 (pass_complete_unrolli): Replace declaration with that of...
2756 (make_pass_complete_unrolli): ...new function.
2757 (pass_parallelize_loops): Replace declaration with that of...
2758 (make_pass_parallelize_loops): ...new function.
2759 (pass_loop_prefetch): Replace declaration with that of...
2760 (make_pass_loop_prefetch): ...new function.
2761 (pass_iv_optimize): Replace declaration with that of...
2762 (make_pass_iv_optimize): ...new function.
2763 (pass_tree_loop_done): Replace declaration with that of...
2764 (make_pass_tree_loop_done): ...new function.
2765 (pass_ch): Replace declaration with that of...
2766 (make_pass_ch): ...new function.
2767 (pass_ccp): Replace declaration with that of...
2768 (make_pass_ccp): ...new function.
2769 (pass_phi_only_cprop): Replace declaration with that of...
2770 (make_pass_phi_only_cprop): ...new function.
2771 (pass_build_ssa): Replace declaration with that of...
2772 (make_pass_build_ssa): ...new function.
2773 (pass_build_alias): Replace declaration with that of...
2774 (make_pass_build_alias): ...new function.
2775 (pass_build_ealias): Replace declaration with that of...
2776 (make_pass_build_ealias): ...new function.
2777 (pass_dominator): Replace declaration with that of...
2778 (make_pass_dominator): ...new function.
2779 (pass_dce): Replace declaration with that of...
2780 (make_pass_dce): ...new function.
2781 (pass_dce_loop): Replace declaration with that of...
2782 (make_pass_dce_loop): ...new function.
2783 (pass_cd_dce): Replace declaration with that of...
2784 (make_pass_cd_dce): ...new function.
2785 (pass_call_cdce): Replace declaration with that of...
2786 (make_pass_call_cdce): ...new function.
2787 (pass_merge_phi): Replace declaration with that of...
2788 (make_pass_merge_phi): ...new function.
2789 (pass_split_crit_edges): Replace declaration with that of...
2790 (make_pass_split_crit_edges): ...new function.
2791 (pass_pre): Replace declaration with that of...
2792 (make_pass_pre): ...new function.
2793 (pass_profile): Replace declaration with that of...
2794 (make_pass_profile): ...new function.
2795 (pass_strip_predict_hints): Replace declaration with that of...
2796 (make_pass_strip_predict_hints): ...new function.
2797 (pass_lower_complex_O0): Replace declaration with that of...
2798 (make_pass_lower_complex_O0): ...new function.
2799 (pass_lower_complex): Replace declaration with that of...
2800 (make_pass_lower_complex): ...new function.
2801 (pass_lower_vector): Replace declaration with that of...
2802 (make_pass_lower_vector): ...new function.
2803 (pass_lower_vector_ssa): Replace declaration with that of...
2804 (make_pass_lower_vector_ssa): ...new function.
2805 (pass_lower_omp): Replace declaration with that of...
2806 (make_pass_lower_omp): ...new function.
2807 (pass_diagnose_omp_blocks): Replace declaration with that of...
2808 (make_pass_diagnose_omp_blocks): ...new function.
2809 (pass_expand_omp): Replace declaration with that of...
2810 (make_pass_expand_omp): ...new function.
2811 (pass_expand_omp_ssa): Replace declaration with that of...
2812 (make_pass_expand_omp_ssa): ...new function.
2813 (pass_object_sizes): Replace declaration with that of...
2814 (make_pass_object_sizes): ...new function.
2815 (pass_strlen): Replace declaration with that of...
2816 (make_pass_strlen): ...new function.
2817 (pass_fold_builtins): Replace declaration with that of...
2818 (make_pass_fold_builtins): ...new function.
2819 (pass_stdarg): Replace declaration with that of...
2820 (make_pass_stdarg): ...new function.
2821 (pass_early_warn_uninitialized): Replace declaration with that of...
2822 (make_pass_early_warn_uninitialized): ...new function.
2823 (pass_late_warn_uninitialized): Replace declaration with that of...
2824 (make_pass_late_warn_uninitialized): ...new function.
2825 (pass_cse_reciprocals): Replace declaration with that of...
2826 (make_pass_cse_reciprocals): ...new function.
2827 (pass_cse_sincos): Replace declaration with that of...
2828 (make_pass_cse_sincos): ...new function.
2829 (pass_optimize_bswap): Replace declaration with that of...
2830 (make_pass_optimize_bswap): ...new function.
2831 (pass_optimize_widening_mul): Replace declaration with that of...
2832 (make_pass_optimize_widening_mul): ...new function.
2833 (pass_warn_function_return): Replace declaration with that of...
2834 (make_pass_warn_function_return): ...new function.
2835 (pass_warn_function_noreturn): Replace declaration with that of...
2836 (make_pass_warn_function_noreturn): ...new function.
2837 (pass_cselim): Replace declaration with that of...
2838 (make_pass_cselim): ...new function.
2839 (pass_phiopt): Replace declaration with that of...
2840 (make_pass_phiopt): ...new function.
2841 (pass_forwprop): Replace declaration with that of...
2842 (make_pass_forwprop): ...new function.
2843 (pass_phiprop): Replace declaration with that of...
2844 (make_pass_phiprop): ...new function.
2845 (pass_tree_ifcombine): Replace declaration with that of...
2846 (make_pass_tree_ifcombine): ...new function.
2847 (pass_dse): Replace declaration with that of...
2848 (make_pass_dse): ...new function.
2849 (pass_nrv): Replace declaration with that of...
2850 (make_pass_nrv): ...new function.
2851 (pass_rename_ssa_copies): Replace declaration with that of...
2852 (make_pass_rename_ssa_copies): ...new function.
2853 (pass_sink_code): Replace declaration with that of...
2854 (make_pass_sink_code): ...new function.
2855 (pass_fre): Replace declaration with that of...
2856 (make_pass_fre): ...new function.
2857 (pass_check_data_deps): Replace declaration with that of...
2858 (make_pass_check_data_deps): ...new function.
2859 (pass_copy_prop): Replace declaration with that of...
2860 (make_pass_copy_prop): ...new function.
2861 (pass_vrp): Replace declaration with that of...
2862 (make_pass_vrp): ...new function.
2863 (pass_uncprop): Replace declaration with that of...
2864 (make_pass_uncprop): ...new function.
2865 (pass_return_slot): Replace declaration with that of...
2866 (make_pass_return_slot): ...new function.
2867 (pass_reassoc): Replace declaration with that of...
2868 (make_pass_reassoc): ...new function.
2869 (pass_rebuild_cgraph_edges): Replace declaration with that of...
2870 (make_pass_rebuild_cgraph_edges): ...new function.
2871 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2872 (make_pass_remove_cgraph_callee_edges): ...new function.
2873 (pass_build_cgraph_edges): Replace declaration with that of...
2874 (make_pass_build_cgraph_edges): ...new function.
2875 (pass_local_pure_const): Replace declaration with that of...
2876 (make_pass_local_pure_const): ...new function.
2877 (pass_tracer): Replace declaration with that of...
2878 (make_pass_tracer): ...new function.
2879 (pass_warn_unused_result): Replace declaration with that of...
2880 (make_pass_warn_unused_result): ...new function.
2881 (pass_diagnose_tm_blocks): Replace declaration with that of...
2882 (make_pass_diagnose_tm_blocks): ...new function.
2883 (pass_lower_tm): Replace declaration with that of...
2884 (make_pass_lower_tm): ...new function.
2885 (pass_tm_init): Replace declaration with that of...
2886 (make_pass_tm_init): ...new function.
2887 (pass_tm_mark): Replace declaration with that of...
2888 (make_pass_tm_mark): ...new function.
2889 (pass_tm_memopt): Replace declaration with that of...
2890 (make_pass_tm_memopt): ...new function.
2891 (pass_tm_edges): Replace declaration with that of...
2892 (make_pass_tm_edges): ...new function.
2893 (pass_split_functions): Replace declaration with that of...
2894 (make_pass_split_functions): ...new function.
2895 (pass_feedback_split_functions): Replace declaration with that of...
2896 (make_pass_feedback_split_functions): ...new function.
2897 (pass_strength_reduction): Replace declaration with that of...
2898 (make_pass_strength_reduction): ...new function.
2899 (pass_ipa_lower_emutls): Replace declaration with that of...
2900 (make_pass_ipa_lower_emutls): ...new function.
2901 (pass_ipa_function_and_variable_visibility): Replace declaration with
2903 (make_pass_ipa_function_and_variable_visibility): ...new function.
2904 (pass_ipa_tree_profile): Replace declaration with that of...
2905 (make_pass_ipa_tree_profile): ...new function.
2906 (pass_early_local_passes): Replace declaration with that of...
2907 (make_pass_early_local_passes): ...new function.
2908 (pass_ipa_whole_program_visibility): Replace declaration with that
2910 (make_pass_ipa_whole_program_visibility): ...new function.
2911 (pass_ipa_lto_gimple_out): Replace declaration with that of...
2912 (make_pass_ipa_lto_gimple_out): ...new function.
2913 (pass_ipa_increase_alignment): Replace declaration with that of...
2914 (make_pass_ipa_increase_alignment): ...new function.
2915 (pass_ipa_inline): Replace declaration with that of...
2916 (make_pass_ipa_inline): ...new function.
2917 (pass_ipa_free_lang_data): Replace declaration with that of...
2918 (make_pass_ipa_free_lang_data): ...new function.
2919 (pass_ipa_free_inline_summary): Replace declaration with that of...
2920 (make_pass_ipa_free_inline_summary): ...new function.
2921 (pass_ipa_cp): Replace declaration with that of...
2922 (make_pass_ipa_cp): ...new function.
2923 (pass_ipa_reference): Replace declaration with that of...
2924 (make_pass_ipa_reference): ...new function.
2925 (pass_ipa_pure_const): Replace declaration with that of...
2926 (make_pass_ipa_pure_const): ...new function.
2927 (pass_ipa_pta): Replace declaration with that of...
2928 (make_pass_ipa_pta): ...new function.
2929 (pass_ipa_lto_finish_out): Replace declaration with that of...
2930 (make_pass_ipa_lto_finish_out): ...new function.
2931 (pass_ipa_tm): Replace declaration with that of...
2932 (make_pass_ipa_tm): ...new function.
2933 (pass_ipa_profile): Replace declaration with that of...
2934 (make_pass_ipa_profile): ...new function.
2935 (pass_ipa_cdtor_merge): Replace declaration with that of...
2936 (make_pass_ipa_cdtor_merge): ...new function.
2937 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2939 (make_pass_cleanup_cfg_post_optimizing): ...new function.
2940 (pass_init_datastructures): Replace declaration with that of...
2941 (make_pass_init_datastructures): ...new function.
2942 (pass_fixup_cfg): Replace declaration with that of...
2943 (make_pass_fixup_cfg): ...new function.
2944 (pass_expand): Replace declaration with that of...
2945 (make_pass_expand): ...new function.
2946 (pass_instantiate_virtual_regs): Replace declaration with that of...
2947 (make_pass_instantiate_virtual_regs): ...new function.
2948 (pass_rtl_fwprop): Replace declaration with that of...
2949 (make_pass_rtl_fwprop): ...new function.
2950 (pass_rtl_fwprop_addr): Replace declaration with that of...
2951 (make_pass_rtl_fwprop_addr): ...new function.
2952 (pass_jump): Replace declaration with that of...
2953 (make_pass_jump): ...new function.
2954 (pass_jump2): Replace declaration with that of...
2955 (make_pass_jump2): ...new function.
2956 (pass_lower_subreg): Replace declaration with that of...
2957 (make_pass_lower_subreg): ...new function.
2958 (pass_cse): Replace declaration with that of...
2959 (make_pass_cse): ...new function.
2960 (pass_fast_rtl_dce): Replace declaration with that of...
2961 (make_pass_fast_rtl_dce): ...new function.
2962 (pass_ud_rtl_dce): Replace declaration with that of...
2963 (make_pass_ud_rtl_dce): ...new function.
2964 (pass_rtl_dce): Replace declaration with that of...
2965 (make_pass_rtl_dce): ...new function.
2966 (pass_rtl_dse1): Replace declaration with that of...
2967 (make_pass_rtl_dse1): ...new function.
2968 (pass_rtl_dse2): Replace declaration with that of...
2969 (make_pass_rtl_dse2): ...new function.
2970 (pass_rtl_dse3): Replace declaration with that of...
2971 (make_pass_rtl_dse3): ...new function.
2972 (pass_rtl_cprop): Replace declaration with that of...
2973 (make_pass_rtl_cprop): ...new function.
2974 (pass_rtl_pre): Replace declaration with that of...
2975 (make_pass_rtl_pre): ...new function.
2976 (pass_rtl_hoist): Replace declaration with that of...
2977 (make_pass_rtl_hoist): ...new function.
2978 (pass_rtl_store_motion): Replace declaration with that of...
2979 (make_pass_rtl_store_motion): ...new function.
2980 (pass_cse_after_global_opts): Replace declaration with that of...
2981 (make_pass_cse_after_global_opts): ...new function.
2982 (pass_rtl_ifcvt): Replace declaration with that of...
2983 (make_pass_rtl_ifcvt): ...new function.
2984 (pass_into_cfg_layout_mode): Replace declaration with that of...
2985 (make_pass_into_cfg_layout_mode): ...new function.
2986 (pass_outof_cfg_layout_mode): Replace declaration with that of...
2987 (make_pass_outof_cfg_layout_mode): ...new function.
2988 (pass_loop2): Replace declaration with that of...
2989 (make_pass_loop2): ...new function.
2990 (pass_rtl_loop_init): Replace declaration with that of...
2991 (make_pass_rtl_loop_init): ...new function.
2992 (pass_rtl_move_loop_invariants): Replace declaration with that of...
2993 (make_pass_rtl_move_loop_invariants): ...new function.
2994 (pass_rtl_unswitch): Replace declaration with that of...
2995 (make_pass_rtl_unswitch): ...new function.
2996 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2997 (make_pass_rtl_unroll_and_peel_loops): ...new function.
2998 (pass_rtl_doloop): Replace declaration with that of...
2999 (make_pass_rtl_doloop): ...new function.
3000 (pass_rtl_loop_done): Replace declaration with that of...
3001 (make_pass_rtl_loop_done): ...new function.
3002 (pass_web): Replace declaration with that of...
3003 (make_pass_web): ...new function.
3004 (pass_cse2): Replace declaration with that of...
3005 (make_pass_cse2): ...new function.
3006 (pass_df_initialize_opt): Replace declaration with that of...
3007 (make_pass_df_initialize_opt): ...new function.
3008 (pass_df_initialize_no_opt): Replace declaration with that of...
3009 (make_pass_df_initialize_no_opt): ...new function.
3010 (pass_reginfo_init): Replace declaration with that of...
3011 (make_pass_reginfo_init): ...new function.
3012 (pass_inc_dec): Replace declaration with that of...
3013 (make_pass_inc_dec): ...new function.
3014 (pass_stack_ptr_mod): Replace declaration with that of...
3015 (make_pass_stack_ptr_mod): ...new function.
3016 (pass_initialize_regs): Replace declaration with that of...
3017 (make_pass_initialize_regs): ...new function.
3018 (pass_combine): Replace declaration with that of...
3019 (make_pass_combine): ...new function.
3020 (pass_if_after_combine): Replace declaration with that of...
3021 (make_pass_if_after_combine): ...new function.
3022 (pass_ree): Replace declaration with that of...
3023 (make_pass_ree): ...new function.
3024 (pass_partition_blocks): Replace declaration with that of...
3025 (make_pass_partition_blocks): ...new function.
3026 (pass_match_asm_constraints): Replace declaration with that of...
3027 (make_pass_match_asm_constraints): ...new function.
3028 (pass_regmove): Replace declaration with that of...
3029 (make_pass_regmove): ...new function.
3030 (pass_split_all_insns): Replace declaration with that of...
3031 (make_pass_split_all_insns): ...new function.
3032 (pass_fast_rtl_byte_dce): Replace declaration with that of...
3033 (make_pass_fast_rtl_byte_dce): ...new function.
3034 (pass_lower_subreg2): Replace declaration with that of...
3035 (make_pass_lower_subreg2): ...new function.
3036 (pass_mode_switching): Replace declaration with that of...
3037 (make_pass_mode_switching): ...new function.
3038 (pass_sms): Replace declaration with that of...
3039 (make_pass_sms): ...new function.
3040 (pass_sched): Replace declaration with that of...
3041 (make_pass_sched): ...new function.
3042 (pass_ira): Replace declaration with that of...
3043 (make_pass_ira): ...new function.
3044 (pass_reload): Replace declaration with that of...
3045 (make_pass_reload): ...new function.
3046 (pass_clean_state): Replace declaration with that of...
3047 (make_pass_clean_state): ...new function.
3048 (pass_branch_prob): Replace declaration with that of...
3049 (make_pass_branch_prob): ...new function.
3050 (pass_value_profile_transformations): Replace declaration with that
3052 (make_pass_value_profile_transformations): ...new function.
3053 (pass_postreload_cse): Replace declaration with that of...
3054 (make_pass_postreload_cse): ...new function.
3055 (pass_gcse2): Replace declaration with that of...
3056 (make_pass_gcse2): ...new function.
3057 (pass_split_after_reload): Replace declaration with that of...
3058 (make_pass_split_after_reload): ...new function.
3059 (pass_branch_target_load_optimize1): Replace declaration with that
3061 (make_pass_branch_target_load_optimize1): ...new function.
3062 (pass_thread_prologue_and_epilogue): Replace declaration with that
3064 (make_pass_thread_prologue_and_epilogue): ...new function.
3065 (pass_stack_adjustments): Replace declaration with that of...
3066 (make_pass_stack_adjustments): ...new function.
3067 (pass_peephole2): Replace declaration with that of...
3068 (make_pass_peephole2): ...new function.
3069 (pass_if_after_reload): Replace declaration with that of...
3070 (make_pass_if_after_reload): ...new function.
3071 (pass_regrename): Replace declaration with that of...
3072 (make_pass_regrename): ...new function.
3073 (pass_cprop_hardreg): Replace declaration with that of...
3074 (make_pass_cprop_hardreg): ...new function.
3075 (pass_reorder_blocks): Replace declaration with that of...
3076 (make_pass_reorder_blocks): ...new function.
3077 (pass_branch_target_load_optimize2): Replace declaration with that
3079 (make_pass_branch_target_load_optimize2): ...new function.
3080 (pass_leaf_regs): Replace declaration with that of...
3081 (make_pass_leaf_regs): ...new function.
3082 (pass_split_before_sched2): Replace declaration with that of...
3083 (make_pass_split_before_sched2): ...new function.
3084 (pass_compare_elim_after_reload): Replace declaration with that of...
3085 (make_pass_compare_elim_after_reload): ...new function.
3086 (pass_sched2): Replace declaration with that of...
3087 (make_pass_sched2): ...new function.
3088 (pass_stack_regs): Replace declaration with that of...
3089 (make_pass_stack_regs): ...new function.
3090 (pass_stack_regs_run): Replace declaration with that of...
3091 (make_pass_stack_regs_run): ...new function.
3092 (pass_df_finish): Replace declaration with that of...
3093 (make_pass_df_finish): ...new function.
3094 (pass_compute_alignments): Replace declaration with that of...
3095 (make_pass_compute_alignments): ...new function.
3096 (pass_duplicate_computed_gotos): Replace declaration with that of...
3097 (make_pass_duplicate_computed_gotos): ...new function.
3098 (pass_variable_tracking): Replace declaration with that of...
3099 (make_pass_variable_tracking): ...new function.
3100 (pass_free_cfg): Replace declaration with that of...
3101 (make_pass_free_cfg): ...new function.
3102 (pass_machine_reorg): Replace declaration with that of...
3103 (make_pass_machine_reorg): ...new function.
3104 (pass_cleanup_barriers): Replace declaration with that of...
3105 (make_pass_cleanup_barriers): ...new function.
3106 (pass_delay_slots): Replace declaration with that of...
3107 (make_pass_delay_slots): ...new function.
3108 (pass_split_for_shorten_branches): Replace declaration with that of...
3109 (make_pass_split_for_shorten_branches): ...new function.
3110 (pass_split_before_regstack): Replace declaration with that of...
3111 (make_pass_split_before_regstack): ...new function.
3112 (pass_convert_to_eh_region_ranges): Replace declaration with that
3114 (make_pass_convert_to_eh_region_ranges): ...new function.
3115 (pass_shorten_branches): Replace declaration with that of...
3116 (make_pass_shorten_branches): ...new function.
3117 (pass_set_nothrow_function_flags): Replace declaration with that of...
3118 (make_pass_set_nothrow_function_flags): ...new function.
3119 (pass_dwarf2_frame): Replace declaration with that of...
3120 (make_pass_dwarf2_frame): ...new function.
3121 (pass_final): Replace declaration with that of...
3122 (make_pass_final): ...new function.
3123 (pass_rtl_seqabstr): Replace declaration with that of...
3124 (make_pass_rtl_seqabstr): ...new function.
3125 (pass_release_ssa_names): Replace declaration with that of...
3126 (make_pass_release_ssa_names): ...new function.
3127 (pass_early_inline): Replace declaration with that of...
3128 (make_pass_early_inline): ...new function.
3129 (pass_inline_parameters): Replace declaration with that of...
3130 (make_pass_inline_parameters): ...new function.
3131 (pass_update_address_taken): Replace declaration with that of...
3132 (make_pass_update_address_taken): ...new function.
3133 (pass_convert_switch): Replace declaration with that of...
3134 (make_pass_convert_switch): ...new function.
3135 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
3136 to a subclass of simple_ipa_opt_pass along with...
3137 (pass_data_ipa_tree_profile): ...new pass_data instance and...
3138 (make_pass_ipa_tree_profile): ...new function.
3139 * tree-sra.c (pass_sra_early): Convert from a global struct to a
3140 subclass of gimple_opt_pass along with...
3141 (pass_data_sra_early): ...new pass_data instance and...
3142 (make_pass_sra_early): ...new function.
3143 (pass_sra): Convert from a global struct to a subclass of
3144 gimple_opt_pass along with...
3145 (pass_data_sra): ...new pass_data instance and...
3146 (make_pass_sra): ...new function.
3147 (pass_early_ipa_sra): Convert from a global struct to a subclass of
3148 gimple_opt_pass along with...
3149 (pass_data_early_ipa_sra): ...new pass_data instance and...
3150 (make_pass_early_ipa_sra): ...new function.
3151 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
3152 subclass of gimple_opt_pass along with...
3153 (pass_data_ccp): ...new pass_data instance and...
3154 (make_pass_ccp): ...new function.
3155 (pass_fold_builtins): Convert from a global struct to a subclass of
3156 gimple_opt_pass along with...
3157 (pass_data_fold_builtins): ...new pass_data instance and...
3158 (make_pass_fold_builtins): ...new function.
3159 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
3160 subclass of gimple_opt_pass along with...
3161 (pass_data_copy_prop): ...new pass_data instance and...
3162 (make_pass_copy_prop): ...new function.
3163 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
3164 global struct to a subclass of gimple_opt_pass along with...
3165 (pass_data_rename_ssa_copies): ...new pass_data instance and...
3166 (make_pass_rename_ssa_copies): ...new function.
3167 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
3168 subclass of gimple_opt_pass along with...
3169 (pass_data_dce): ...new pass_data instance and...
3170 (make_pass_dce): ...new function.
3171 (pass_dce_loop): Convert from a global struct to a subclass of
3172 gimple_opt_pass along with...
3173 (pass_data_dce_loop): ...new pass_data instance and...
3174 (make_pass_dce_loop): ...new function.
3175 (pass_cd_dce): Convert from a global struct to a subclass of
3176 gimple_opt_pass along with...
3177 (pass_data_cd_dce): ...new pass_data instance and...
3178 (make_pass_cd_dce): ...new function.
3179 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
3180 subclass of gimple_opt_pass along with...
3181 (pass_data_dominator): ...new pass_data instance and...
3182 (make_pass_dominator): ...new function.
3183 (pass_phi_only_cprop): Convert from a global struct to a subclass of
3184 gimple_opt_pass along with...
3185 (pass_data_phi_only_cprop): ...new pass_data instance and...
3186 (make_pass_phi_only_cprop): ...new function.
3187 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
3188 subclass of gimple_opt_pass along with...
3189 (pass_data_dse): ...new pass_data instance and...
3190 (make_pass_dse): ...new function.
3191 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
3192 a subclass of gimple_opt_pass along with...
3193 (pass_data_forwprop): ...new pass_data instance and...
3194 (make_pass_forwprop): ...new function.
3195 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
3196 struct to a subclass of gimple_opt_pass along with...
3197 (pass_data_tree_ifcombine): ...new pass_data instance and...
3198 (make_pass_tree_ifcombine): ...new function.
3199 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
3200 subclass of gimple_opt_pass along with...
3201 (pass_data_ch): ...new pass_data instance and...
3202 (make_pass_ch): ...new function.
3203 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
3204 subclass of gimple_opt_pass along with...
3205 (pass_data_tree_loop): ...new pass_data instance and...
3206 (make_pass_tree_loop): ...new function.
3207 (pass_tree_loop_init): Convert from a global struct to a subclass of
3208 gimple_opt_pass along with...
3209 (pass_data_tree_loop_init): ...new pass_data instance and...
3210 (make_pass_tree_loop_init): ...new function.
3211 (pass_lim): Convert from a global struct to a subclass of
3212 gimple_opt_pass along with...
3213 (pass_data_lim): ...new pass_data instance and...
3214 (make_pass_lim): ...new function.
3215 (pass_tree_unswitch): Convert from a global struct to a subclass of
3216 gimple_opt_pass along with...
3217 (pass_data_tree_unswitch): ...new pass_data instance and...
3218 (make_pass_tree_unswitch): ...new function.
3219 (pass_predcom): Convert from a global struct to a subclass of
3220 gimple_opt_pass along with...
3221 (pass_data_predcom): ...new pass_data instance and...
3222 (make_pass_predcom): ...new function.
3223 (pass_vectorize): Convert from a global struct to a subclass of
3224 gimple_opt_pass along with...
3225 (pass_data_vectorize): ...new pass_data instance and...
3226 (make_pass_vectorize): ...new function.
3227 (pass_graphite): Convert from a global struct to a subclass of
3228 gimple_opt_pass along with...
3229 (pass_data_graphite): ...new pass_data instance and...
3230 (make_pass_graphite): ...new function.
3231 (pass_graphite_transforms): Convert from a global struct to a subclass
3232 of gimple_opt_pass along with...
3233 (pass_data_graphite_transforms): ...new pass_data instance and...
3234 (make_pass_graphite_transforms): ...new function.
3235 (pass_check_data_deps): Convert from a global struct to a subclass of
3236 gimple_opt_pass along with...
3237 (pass_data_check_data_deps): ...new pass_data instance and...
3238 (make_pass_check_data_deps): ...new function.
3239 (pass_iv_canon): Convert from a global struct to a subclass of
3240 gimple_opt_pass along with...
3241 (pass_data_iv_canon): ...new pass_data instance and...
3242 (make_pass_iv_canon): ...new function.
3243 (pass_scev_cprop): Convert from a global struct to a subclass of
3244 gimple_opt_pass along with...
3245 (pass_data_scev_cprop): ...new pass_data instance and...
3246 (make_pass_scev_cprop): ...new function.
3247 (pass_record_bounds): Convert from a global struct to a subclass of
3248 gimple_opt_pass along with...
3249 (pass_data_record_bounds): ...new pass_data instance and...
3250 (make_pass_record_bounds): ...new function.
3251 (pass_complete_unroll): Convert from a global struct to a subclass of
3252 gimple_opt_pass along with...
3253 (pass_data_complete_unroll): ...new pass_data instance and...
3254 (make_pass_complete_unroll): ...new function.
3255 (pass_complete_unrolli): Convert from a global struct to a subclass of
3256 gimple_opt_pass along with...
3257 (pass_data_complete_unrolli): ...new pass_data instance and...
3258 (make_pass_complete_unrolli): ...new function.
3259 (pass_parallelize_loops): Convert from a global struct to a subclass
3260 of gimple_opt_pass along with...
3261 (pass_data_parallelize_loops): ...new pass_data instance and...
3262 (make_pass_parallelize_loops): ...new function.
3263 (pass_loop_prefetch): Convert from a global struct to a subclass of
3264 gimple_opt_pass along with...
3265 (pass_data_loop_prefetch): ...new pass_data instance and...
3266 (make_pass_loop_prefetch): ...new function.
3267 (pass_iv_optimize): Convert from a global struct to a subclass of
3268 gimple_opt_pass along with...
3269 (pass_data_iv_optimize): ...new pass_data instance and...
3270 (make_pass_iv_optimize): ...new function.
3271 (pass_tree_loop_done): Convert from a global struct to a subclass of
3272 gimple_opt_pass along with...
3273 (pass_data_tree_loop_done): ...new pass_data instance and...
3274 (make_pass_tree_loop_done): ...new function.
3275 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
3276 struct to a subclass of gimple_opt_pass along with...
3277 (pass_data_cse_reciprocals): ...new pass_data instance and...
3278 (make_pass_cse_reciprocals): ...new function.
3279 (pass_cse_sincos): Convert from a global struct to a subclass of
3280 gimple_opt_pass along with...
3281 (pass_data_cse_sincos): ...new pass_data instance and...
3282 (make_pass_cse_sincos): ...new function.
3283 (pass_optimize_bswap): Convert from a global struct to a subclass of
3284 gimple_opt_pass along with...
3285 (pass_data_optimize_bswap): ...new pass_data instance and...
3286 (make_pass_optimize_bswap): ...new function.
3287 (pass_optimize_widening_mul): Convert from a global struct to a
3288 subclass of gimple_opt_pass along with...
3289 (pass_data_optimize_widening_mul): ...new pass_data instance and...
3290 (make_pass_optimize_widening_mul): ...new function.
3291 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
3292 subclass of gimple_opt_pass along with...
3293 (pass_data_phiopt): ...new pass_data instance and...
3294 (make_pass_phiopt): ...new function.
3295 (pass_cselim): Convert from a global struct to a subclass of
3296 gimple_opt_pass along with...
3297 (pass_data_cselim): ...new pass_data instance and...
3298 (make_pass_cselim): ...new function.
3299 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
3300 subclass of gimple_opt_pass along with...
3301 (pass_data_phiprop): ...new pass_data instance and...
3302 (make_pass_phiprop): ...new function.
3303 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
3304 subclass of gimple_opt_pass along with...
3305 (pass_data_pre): ...new pass_data instance and...
3306 (make_pass_pre): ...new function.
3307 (pass_fre): Convert from a global struct to a subclass of
3308 gimple_opt_pass along with...
3309 (pass_data_fre): ...new pass_data instance and...
3310 (make_pass_fre): ...new function.
3311 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
3312 subclass of gimple_opt_pass along with...
3313 (pass_data_reassoc): ...new pass_data instance and...
3314 (make_pass_reassoc): ...new function.
3315 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
3316 subclass of gimple_opt_pass along with...
3317 (pass_data_sink_code): ...new pass_data instance and...
3318 (make_pass_sink_code): ...new function.
3319 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
3320 subclass of gimple_opt_pass along with...
3321 (pass_data_strlen): ...new pass_data instance and...
3322 (make_pass_strlen): ...new function.
3323 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
3324 struct to a subclass of gimple_opt_pass along with...
3325 (pass_data_build_alias): ...new pass_data instance and...
3326 (make_pass_build_alias): ...new function.
3327 (pass_build_ealias): Convert from a global struct to a subclass of
3328 gimple_opt_pass along with...
3329 (pass_data_build_ealias): ...new pass_data instance and...
3330 (make_pass_build_ealias): ...new function.
3331 (pass_ipa_pta): Convert from a global struct to a subclass of
3332 simple_ipa_opt_pass along with...
3333 (pass_data_ipa_pta): ...new pass_data instance and...
3334 (make_pass_ipa_pta): ...new function.
3335 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
3336 subclass of gimple_opt_pass along with...
3337 (pass_data_uncprop): ...new pass_data instance and...
3338 (make_pass_uncprop): ...new function.
3339 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
3340 global struct to a subclass of gimple_opt_pass along with...
3341 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
3342 (make_pass_late_warn_uninitialized): ...new function.
3343 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
3344 to a subclass of gimple_opt_pass along with...
3345 (pass_data_init_datastructures): ...new pass_data instance and...
3346 (make_pass_init_datastructures): ...new function.
3347 (pass_early_warn_uninitialized): Convert from a global struct to a
3348 subclass of gimple_opt_pass along with...
3349 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
3350 (make_pass_early_warn_uninitialized): ...new function.
3351 (pass_update_address_taken): Convert from a global struct to a
3352 subclass of gimple_opt_pass along with...
3353 (pass_data_update_address_taken): ...new pass_data instance and...
3354 (make_pass_update_address_taken): ...new function.
3355 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
3356 struct to a subclass of gimple_opt_pass along with...
3357 (pass_data_release_ssa_names): ...new pass_data instance and...
3358 (make_pass_release_ssa_names): ...new function.
3359 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
3360 subclass of gimple_opt_pass along with...
3361 (pass_data_stdarg): ...new pass_data instance and...
3362 (make_pass_stdarg): ...new function.
3363 * tree-switch-conversion.c (pass_convert_switch): Convert from a
3364 global struct to a subclass of gimple_opt_pass along with...
3365 (pass_data_convert_switch): ...new pass_data instance and...
3366 (make_pass_convert_switch): ...new function.
3367 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
3368 to a subclass of gimple_opt_pass along with...
3369 (pass_data_tail_recursion): ...new pass_data instance and...
3370 (make_pass_tail_recursion): ...new function.
3371 (pass_tail_calls): Convert from a global struct to a subclass of
3372 gimple_opt_pass along with...
3373 (pass_data_tail_calls): ...new pass_data instance and...
3374 (make_pass_tail_calls): ...new function.
3375 * tree-vect-generic.c (pass_lower_vector): Convert from a global
3376 struct to a subclass of gimple_opt_pass along with...
3377 (pass_data_lower_vector): ...new pass_data instance and...
3378 (make_pass_lower_vector): ...new function.
3379 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
3380 gimple_opt_pass along with...
3381 (pass_data_lower_vector_ssa): ...new pass_data instance and...
3382 (make_pass_lower_vector_ssa): ...new function.
3383 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
3384 to a subclass of gimple_opt_pass along with...
3385 (pass_data_slp_vectorize): ...new pass_data instance and...
3386 (make_pass_slp_vectorize): ...new function.
3387 (pass_ipa_increase_alignment): Convert from a global struct to a
3388 subclass of simple_ipa_opt_pass along with...
3389 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
3390 (make_pass_ipa_increase_alignment): ...new function.
3391 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
3392 gimple_opt_pass along with...
3393 (pass_data_vrp): ...new pass_data instance and...
3394 (make_pass_vrp): ...new function.
3395 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
3396 subclass of simple_ipa_opt_pass along with...
3397 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
3398 (make_pass_ipa_free_lang_data): ...new function.
3399 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
3400 gimple_opt_pass along with...
3401 (pass_data_tsan): ...new pass_data instance and...
3402 (make_pass_tsan): ...new function.
3403 (pass_tsan_O0): Convert from a global struct to a subclass of
3404 gimple_opt_pass along with...
3405 (pass_data_tsan_O0): ...new pass_data instance and...
3406 (make_pass_tsan_O0): ...new function.
3407 * var-tracking.c (pass_variable_tracking): Convert from a global
3408 struct to a subclass of rtl_opt_pass along with...
3409 (pass_data_variable_tracking): ...new pass_data instance and...
3410 (make_pass_variable_tracking): ...new function.
3411 * web.c (pass_web): Convert from a global struct to a subclass of
3412 rtl_opt_pass along with...
3413 (pass_data_web): ...new pass_data instance and...
3414 (make_pass_web): ...new function.
3415 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3416 declaration with that of...
3417 (make_pass_mode_switch_use): ...new function.
3418 (pass_resolve_sw_modes): Replace declaration with that of...
3419 (make_pass_resolve_sw_modes): ...new function.
3420 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3421 from a global struct to a subclass of rtl_opt_pass along with...
3422 (pass_data_mode_switch_use): ...new pass_data instance and...
3423 (make_pass_mode_switch_use): ...new function.
3424 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3425 from a global struct to a subclass of rtl_opt_pass along with...
3426 (pass_data_resolve_sw_modes): ...new pass_data instance and...
3427 (make_pass_resolve_sw_modes): ...new function.
3428 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3429 struct to a subclass of rtl_opt_pass along with...
3430 (pass_data_insert_vzeroupper): ...new pass_data instance and...
3431 (make_pass_insert_vzeroupper): ...new function.
3432 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3433 global struct to a subclass of rtl_opt_pass along with...
3434 (pass_data_work_around_errata): ...new pass_data instance and...
3435 (make_pass_work_around_errata): ...new function.
3436 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3437 struct to a subclass of rtl_opt_pass along with...
3438 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3439 (make_pass_mips_machine_reorg2): ...new function.
3441 2013-08-05 David Malcolm <dmalcolm@redhat.com>
3443 * passes.c (pass_manager::operator new): New.
3445 2013-08-05 David Malcolm <dmalcolm@redhat.com>
3447 Handwritten part of conversion of passes to C++ classes.
3449 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3450 (toplev.o): Add dep on PASS_MANAGER_H.
3451 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3452 of early local pases to reflect this moving from a global to a
3453 member of gcc::pass_manager.
3454 (cgraph_add_new_function): Likewise.
3455 * lto-cgraph.c (lto_output_node): Update for conversion of
3456 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3457 * passes.c (opt_pass::clone): New.
3458 (opt_pass::gate): New.
3459 (opt_pass::execute): New.
3460 (opt_pass::opt_pass): New.
3461 (pass_manager::execute_early_local_passes): New.
3462 (pass_manager::execute_pass_mode_switching): new.
3463 (finish_optimization_passes): Convert to...
3464 (pass_manager::finish_optimization_passes): ...this.
3465 (finish_optimization_passes): Update for conversion of passes to
3467 (register_dump_files_1): Use has_gate since we cannot portably
3468 check a vtable entry against NULL.
3469 (dump_one_pass): Likewise.
3470 (ipa_write_summaries_2): Likewise.
3471 (ipa_write_optimization_summaries_1): Likewise.
3472 (ipa_read_summaries_1): Likewise.
3473 (ipa_read_optimization_summaries_1): Likewise.
3474 (execute_ipa_stmt_fixups): Likewise.
3475 (pass_manager::pass_manager): Rewrite pass-creation, invoking
3476 pass-creation functions rather than wiring up globals, and
3477 storing the results in fields of pass_manager generated using
3479 (pass_manager::dump_profile_report): Update for conversion of
3480 passes to C++ classes.
3481 (pass_manager::execute_ipa_summary_passes): Likewise.
3482 (execute_one_ipa_transform_pass): Likewise.
3483 (execute_one_pass): Use has_gate and has_execute since we cannot
3484 portably check a vtable entry against NULL.
3485 * pass_manager.h (pass_manager::finish_optimization_passes): New.
3486 (pass_manager): Use pass-instances.def to add fields for the
3487 various pass instances.
3488 * toplev.c (finalize): Update for move of
3489 finish_optimization_passes to a method of gcc::pass_manager.
3490 * toplev.h (finish_optimization_passes): Move to method of class
3492 * tree-pass.h (struct pass_data): New.
3493 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3494 (opt_pass::gate): Convert to virtual function.
3495 (opt_pass::~opt_pass): New.
3496 (opt_pass::clone): New.
3497 (opt_pass::execute): Convert to virtual function.
3498 (opt_pass::opt_pass): New.
3499 (opt_pass::ctxt_): new.
3500 (gimple_opt_pass): Convert to subclass of opt_pass.
3501 (gimple_opt_pass::gimple_opt_pass): New.
3502 (rtl_opt_pass): Convert to subclass of opt_pass.
3503 (rtl_opt_pass::rtl_opt_pass): New.
3504 (ipa_opt_pass_d): Convert to subclass of opt_pass.
3505 (ipa_opt_pass_d::ipa_opt_pass_d): New.
3506 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3507 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3508 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3509 invocation of pass_mode_switching to reflect this moving from a
3510 global to a member of gcc::pass_manager.
3511 (ix86_option_override): Rework how pass_insert_vzeroupper is
3512 added to the pass_manager to reflect autogenerated changes.
3513 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3515 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
3517 PR rtl-optimization/57708
3518 * recog.c (peep2_find_free_register): Validate all regs in a
3521 2013-08-05 Jan Hubicka <jh@suse.cz>
3524 * cgraph.c (verify_cgraph_node): Accept local flags from other
3526 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3527 (function_and_variable_visibility): Likewise.
3528 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3530 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3532 * graph.c (init_graph_slim_pretty_print): Remove.
3533 (print_graph_cfg): Do not call it. Use local pretty printer.
3534 (start_graph_dump): Likewise.
3536 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3538 * gimple-pretty-print.c (buffer): Remove.
3539 (initialized): Likewise.
3540 (maybe_init_pretty_print): Likewise.
3541 (print_gimple_stmt): Do not call it. Use non-static local
3542 pretty_printer variable.
3543 (print_gimple_expr): Likewise.
3544 (print_gimple_seq): Likewise.
3545 (gimple_dump_bb): Likewise.
3547 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3549 * asan.c (asan_pp): Remove.
3550 (asan_pp_initialized): Likewise.
3551 (asan_pp_initialize): Likewise.
3552 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
3553 (asan_emit_stack_protection): Tidy. Use local pretty printer.
3554 (asan_add_global): Likewise.
3556 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
3558 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
3559 * diagnostic.h (diagnostic_flush_buffer): Adjust.
3560 * pretty-print.c (pp_formatted_text_data): Likewise.
3561 (pp_indent): Rename from pp_base_indent.
3562 (pp_format): Rename from pp_base_format.
3563 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3564 (pp_format_verbatim): Rename from pp_base_format_verbatim.
3565 (pp_flush): Rename from pp_base_flush.
3566 (pp_set_line_maximum_length): Rename from
3567 pp_base_set_line_maximum_length.
3568 (pp_clear_output_area): Rename from pp_base_clear_output_area.
3569 (pp_set_prefix): Rename from pp_base_set_prefix.
3570 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3571 (pp_emit_prefix): Rename from pp_base_emit_prefix.
3572 (pp_append_text): Rename from pp_base_append_text.
3573 (pp_formatted_text): Rename from pp_base_formatted_text.
3574 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3575 (pp_remaining_character_count_for_line): Rename from
3576 pp_base_remaining_character_count_for_line.
3577 (pp_newline): Rename from pp_base_newline.
3578 (pp_character): Rename from pp_base_character.
3579 (pp_string): Rename from pp_base_string.
3580 (pp_maybe_space): Rename from pp_base_maybe_space.
3581 * asan.c (asan_pp_string): Adjust.
3582 (asan_emit_stack_protection): Likewise.
3583 (asan_add_global): Likewise.
3584 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3585 * tree-mudflap.c (mf_varname_tree): Likewise.
3586 * tree-pretty-print.c (pp_tree_identifier): Rename from
3587 pp_base_tree_identifier.
3588 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3589 Declare as function.
3591 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3593 * pretty-print.h (pp_bar_bar): New.
3594 (pp_ampersand_ampersand): Likewise.
3595 (pp_less_equal): Likewise.
3596 (pp_greater_equal): Likewise.
3597 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3598 printer functions instead of pp_string or operators and punctuators.
3599 (dump_gimple_call): Likewise.
3600 (dump_gimple_omp_for): Likewise.
3601 (dump_gimple_transaction): Likewise.
3602 (dump_gimple_phi): Likewise.
3603 (pp_gimple_stmt_1): Likewise.
3604 * sched-vis.c (print_insn): Likewise.
3605 * tree-mudflap.c (mf_varname_tree): Likewise.
3606 * tree-pretty-print.c (dump_block_node): Likewise.
3607 (dump_generic_node): Likewise.
3609 2013-08-02 Jan Hubicka <jh@suse.cz>
3611 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3613 * lto-streamer-out.c (tree_is_indexable): Results decls and
3614 parm decls are not indexable.
3615 (DFS_write_tree_body): Do not follow args and results.
3616 (hash_tree): Likewise.
3617 (output_functions): Rearrange so struct function is needed
3618 only when real body is output; be able to also ouptut abstract
3619 functions; output DECL_ARGUMENTS and DECL_RESULT.
3620 (lto_output): When not in WPA, ale store abstract functions.
3621 (write_symbol): Do not care about RESULT_DECL.
3622 (output_symbol_p): Handle correctly sbtract decls.
3623 * lto-streamer-in.c (input_function): Rearrange so struct
3624 function can be NULL at entry; allow streaming of
3625 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3626 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3627 sanity check during LTO.
3628 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3629 RESULT_DECl and DECL_ARGUMENTS.
3630 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3633 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3635 * pretty-print.h (pp_underscore): New.
3637 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3638 printer functions instead of pp_character.
3639 (dump_binary_rhs): Likewise.
3640 (dump_ternary_rhs): Likewise.
3641 (dump_gimple_call_args): Likewise.
3642 (pp_points_to_solution): Likewise.
3643 (dump_gimple_call): Likewise.
3644 (dump_gimple_switch): Likewise.
3645 (dump_gimple_cond): Likewise.
3646 (dump_gimple_bind): Likewise.
3647 (dump_gimple_try): Likewise.
3648 (dump_gimple_omp_for): Likewise.
3649 (dump_gimple_omp_continue): Likewise.
3650 (dump_gimple_omp_single): Likewise.
3651 (dump_gimple_omp_sections): Likewise.
3652 (dump_gimple_omp_block): Likewise.
3653 (dump_gimple_omp_critical): Likewise.
3654 (dump_gimple_transaction): Likewise.
3655 (dump_gimple_asm): Likewise.
3656 (dump_gimple_phi): Likewise.
3657 (dump_gimple_omp_parallel): Likewise.
3658 (dump_gimple_omp_task): Likewise.
3659 (dump_gimple_omp_atomic_load): Likewise.
3660 (dump_gimple_omp_atomic_store): Likewise.
3661 (dump_gimple_mem_ops): Likewise.
3662 (pp_gimple_stmt_1): Likewise.
3663 (pp_cfg_jump): Likewise.
3664 (dump_implicit_edges): Likewise.
3665 (gimple_dump_bb_for_graph): Likewise.
3666 * graph.c (draw_cfg_node): Likewise.
3667 * langhooks.c (lhd_print_error_function): Likewise.
3668 * sched-vis.c (print_exp): Likewise.
3669 (print_value): Likewise.
3670 (print_pattern): Likewise.
3671 (print_insn): Likewise.
3672 (rtl_dump_bb_for_graph): Likewise.
3673 * tree-pretty-print.c (dump_function_declaration): Likewise.
3674 (dump_array_domain): Likewise.
3675 (dump_omp_clause): Likewise.
3676 (dump_location): Likewise.
3677 (dump_generic_node): Likewise.
3678 (print_struct_decl): Likewise.
3679 * diagnostic.c (diagnostic_show_locus): Use pp_space.
3681 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3683 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3684 candidate table when replacing a candidate statement.
3685 (replace_rhs_if_not_dup): Likewise.
3686 (replace_one_candidate): Likewise.
3688 2013-08-02 Jan Hubicka <jh@suse.cz>
3689 Martin Liska <marxin.liska@gmail.com>
3691 * cgraphunit.c (add_new_function): Fix logic when adding from
3693 (assemble_thunk): Rename to ...
3694 (expand_thunk); .. this one; export; get it working with
3695 general functions; make produced gimple valid.
3696 * cgraph.h (expand_thunk): Declare.
3698 2013-08-02 Jan Hubicka <jh@suse.cz>
3700 * ipa-cp.c (gather_context_independent_values): Use
3701 ipa_get_param_move_cost.
3702 (get_replacement_map): Remove PARAM; move parameter folding
3704 (create_specialized_node): Update.
3705 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3706 assert that we have gimple body; update move_cost.
3707 (count_formal_params): Assert that we have gimple body.
3708 (ipa_dump_param): New function.
3709 (ipa_alloc_node_params): Break out from ...
3710 (ipa_initialize_node_params): ... here.
3711 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3712 (ipa_write_node_info): Stream move costs.
3713 (ipa_read_node_info): Read move costs.
3714 (ipa_update_after_lto_read): Do not recompute node params.
3715 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3716 (ipa_get_param): Check we are not in WPA.
3717 (ipa_get_param_move_cost): New.
3718 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3719 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3720 parm numbers to be present.
3722 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
3724 PR rtl-optimization/58048
3725 * lra-constraints.c (process_alt_operands): Don't check asm
3726 operand on register.
3728 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
3730 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3731 the implied StoreLoad barrier for atomic operations if before.
3733 2013-08-02 Jan Hubicka <jh@suse.cz>
3734 Martin Liska <marxin.liska@gmail.com>
3736 * cgraph.c (cgraph_function_body_availability): Do not check
3738 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3739 symtab_node_availability): Declare.
3740 * ipa.c (can_replace_by_local_alias): New.
3741 (function_and_variable_visibility): Use it.
3742 * symtab.c (symtab_for_node_and_aliases,
3743 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3745 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
3747 PR rtl-optimization/57963
3748 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3749 (lra_constraints): Use them.
3751 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
3753 * config/arm/types.md (define_attr "type"): Add "load_acq"
3755 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3757 (cortex_a53_store1): Likewise.
3759 2013-08-01 Jan Hubicka <jh@suse.cz>
3761 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3762 partitions are not needed.
3764 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
3766 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3767 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3768 MAYBE_NON_Q_CLASS_P where appropriate.
3770 2013-08-01 Jan Hubicka <jh@suse.cz>
3772 * cgraph.h (release_function_body): Declare.
3773 * tree.c (free_lang_data_in_decl): Free, parameters and return values
3774 of unused delcarations.
3776 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3778 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3779 RTL form when subtracting a constant.
3781 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3783 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3784 Generate canonical plus rtx with negated immediate instead of minus
3786 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3788 2013-08-01 Jan Hubicka <jh@suse.cz>
3790 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3791 (cgraph_release_function_body): Likewise.
3792 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3793 * cgraph.h (cgrpah_node): Rename abstract_and_needed
3794 to used_as_abstract_origin.
3795 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3796 symbols used as abstract origins.
3797 * cgraphunit.c (analyze_functions): Update.
3798 * ipa.c (symtab_remove_unreachable_nodes): Recompute
3799 used_as_abstract_origin.
3800 * tree-inline.c (tree_function_versioning): Update
3801 used_as_abstract_origin; be ready for DECL_RESULT and
3802 DECL_ARGUMENTS to be NULL.
3804 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3805 for abstract functions.
3806 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3809 2013-08-01 Jan Hubicka <jh@suse.cz>
3811 * profile.c (compute_value_histograms): Fix thinko.
3813 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
3815 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
3816 aarch-common-protos.h to extra_headers.
3817 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3818 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3819 * config/aarch64/t-aarch64 (aarch-common.o): Define.
3821 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
3823 * config/aarch64/aarch64.md (define_attr "type"): Delete.
3824 Include "../arm/types.md". Define "type" attribute for all patterns.
3825 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3828 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3830 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3831 to support power8 load fusion.
3832 (fusion_gpr_mem_load): Likewise.
3834 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3836 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3837 declarations for power8 load fusion.
3838 (emit_fusion_gpr_load): Likewise.
3840 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3841 tuning for power8, turn on fusion mode by default. Turn on sign
3842 extending fusion mode if normal fusion mode is on, and we are at
3844 (fusion_gpr_load_p): New function, return true if we can fuse an
3845 addis instruction with a dependent load to a GPR.
3846 (emit_fusion_gpr_load): Emit the instructions for power8 load
3849 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3850 (VSX load fusion peepholes): New peepholes to fuse together an
3851 addi instruction with a VSX load instruction.
3853 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3854 peepholes to fuse an addis instruction with a load to a GPR base
3855 register. If we are supporting sign extending fusions, convert
3856 sign extending loads to zero extending loads and add an explicit
3859 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
3861 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
3862 aarch-common-protos.h to extra_headers.
3863 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3864 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3865 (arm_early_store_addr_dep): Likewise.
3866 (arm_no_early_alu_shift_dep): Likewise.
3867 (arm_no_early_alu_shift_value_dep): Likewise.
3868 (arm_no_early_mul_dep): Likewise.
3869 (arm_no_early_store_addr_dep): Likewise.
3870 (arm_mac_accumulator_is_mul_result): Likewise.
3871 (arm_mac_accumulator_is_result): Likewise.
3872 * config/arm/aarch-common.c: ... here. New file.
3873 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3875 (arm_early_store_addr_dep): Likewise.
3876 (arm_no_early_alu_shift_dep): Likewise.
3877 (arm_no_early_alu_shift_value_dep): Likewise.
3878 (arm_no_early_mul_dep): Likewise.
3879 (arm_no_early_store_addr_dep): Likewise.
3880 (arm_mac_accumulator_is_mul_result): Likewise.
3881 (arm_mac_accumulator_is_result): Likewise.
3882 * config/arm/aarch-common-protos.h: ... here. New file.
3883 * config/arm/t-arm (aarch-common.o): Define.
3885 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
3887 * config/arm/arm.md: Include new file "types.md".
3888 (define_attr "type"): Move from here to ...
3889 (define_attr "mul32"): Likewise.
3890 (define_attr "mul64"): Likewise.
3891 * config/arm/types.md: ... here. New file.
3893 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
3895 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3896 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3898 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3900 * gen-pass-instances.awk: Fix offset of substr().
3902 2013-07-31 David Malcolm <dmalcolm@redhat.com>
3904 * Makefile.in (pass-instances.def): New.
3905 (passes.o): Replace dependency on passes.def with one on
3908 * gen-pass-instances.awk: New.
3910 * passes.c (pass_manager::pass_manager): Use pass-instances.def
3911 rather than passes.def, updating local definition of NEXT_PASS
3912 macro to add an extra NUM parameter (currently unused).
3914 2013-07-30 David Malcolm <dmalcolm@redhat.com>
3916 * Makefile.in (PASS_MANAGER_H): New.
3917 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3918 (passes.o): Likewise.
3919 (statistics.o): Likewise.
3920 (cgraphunit.o): Likewise.
3921 (context.o): Depend on PASS_MANAGER_H.
3923 * pass_manager.h: New.
3925 * cgraphunit.c (cgraph_add_new_function): Update for moves
3926 of globals to fields of pass_manager.
3927 (analyze_function): Likewise.
3928 (expand_function): Likewise.
3929 (ipa_passes): Likewise.
3930 (compile): Likewise.
3932 * context.c (context::context): New.
3933 * context.h (context::context): New.
3934 (context::get_passes): New.
3935 (context::passes_): New.
3937 * lto-cgraph.c (input_node): Update for moves of globals to
3938 fields of pass_manager.
3940 * passes.c (all_passes): Remove, in favor of a field of the
3941 same name within the new class pass_manager.
3942 (all_small_ipa_passes): Likewise.
3943 (all_lowering_passes): Likewise.
3944 (all_regular_ipa_passes): Likewise.
3945 (all_late_ipa_passes): Likewise.
3946 (all_lto_gen_passes): Likewise.
3947 (passes_by_id): Likewise.
3948 (passes_by_id_size): Likewise.
3949 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3950 the new class pass_manager.
3951 (set_pass_for_id): Convert to...
3952 (pass_manager::set_pass_for_id): ...method.
3953 (get_pass_for_id): Convert to...
3954 (pass_manager::get_pass_for_id): ...method.
3955 (register_one_dump_file): Move body of implementation into...
3956 (pass_manager::register_one_dump_file): ...here.
3957 (register_dump_files_1): Convert to...
3958 (pass_manager::register_dump_files_1): ...method.
3959 (register_dump_files): Convert to...
3960 (pass_manager::register_dump_files): ...method.
3961 (create_pass_tab): Update for moves of globals to fields of
3963 (dump_passes): Move body of implementation into...
3964 (pass_manager::dump_passes): ...here.
3965 (register_pass): Move body of implementation into...
3966 (pass_manager::register_pass): ...here.
3967 (init_optimization_passes): Convert into...
3968 (pass_manager::pass_manager): ...constructor for new
3969 pass_manager class, and initialize the pass_lists array.
3970 (check_profile_consistency): Update for moves of globals to
3971 fields of pass_manager.
3972 (dump_profile_report): Move body of implementation into...
3973 (pass_manager::dump_profile_report): ...here.
3974 (ipa_write_summaries_1): Update for moves of pass lists from
3975 being globals to fields of pass_manager.
3976 (ipa_write_optimization_summaries): Likewise.
3977 (ipa_read_summaries): Likewise.
3978 (ipa_read_optimization_summaries): Likewise.
3979 (execute_all_ipa_stmt_fixups): Likewise.
3981 * statistics.c (statistics_fini): Update for moves of globals to
3982 fields of pass_manager.
3984 * toplev.c (general_init): Replace call to
3985 init_optimization_passes with construction of the pass_manager
3988 * tree-pass.h (all_passes): Remove, in favor of a field of the
3989 same name within the new class pass_manager.
3990 (all_small_ipa_passes): Likewise.
3991 (all_lowering_passes): Likewise.
3992 (all_regular_ipa_passes): Likewise.
3993 (all_lto_gen_passes): Likewise.
3994 (all_late_ipa_passes): Likewise.
3995 (passes_by_id): Likewise.
3996 (passes_by_id_size): Likewise.
3997 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3998 the new class pass_manager.
3999 (get_pass_for_id): Remove.
4001 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
4003 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
4006 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
4008 * arm.md (mulhi3): New expand pattern.
4010 2013-07-30 Jan Hubicka <jh@suse.cz>
4011 Martin Liska <marxin.liska@gmail.com>
4013 * profile.c (compute_value_histograms): Do not ICE when
4014 there is mismatch only on some counters.
4016 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
4018 PR rtl-optimization/57637
4019 * function.c (move_insn_for_shrink_wrap): Also check the
4020 GEN set of the LIVE problem for the liveness analysis
4021 if it exists, otherwise give up.
4023 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4025 PR tree-optimization/57993
4026 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
4027 replaced statement in the candidate table.
4028 (phi_add_costs): Return infinite cost when the hidden basis does
4029 not dominate all phis on which the candidate is dependent.
4030 (replace_one_candidate): Record replaced statement in the
4033 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
4035 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
4036 (ashlv2si3): New expander.
4037 (*ashlv2si3_i): New define_insn_and_split.
4038 * predicates.md (float_operation): Allow patterns with three
4041 PR rtl-optimization/58021
4042 * mode-switching.c (create_pre_exit): Always split off preceding
4043 insns if we are not at the basic block head.
4045 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
4047 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
4048 (UCLIBC_DYNAMIC_LINKER): New macro.
4049 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
4051 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
4052 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
4054 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
4055 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
4056 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
4057 for SF and DF modes. Use ieee_quad_format for TF mode.
4058 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
4059 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
4060 (mips_option_override): Handle `-mnan=legacy'.
4061 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
4062 `-mabs=2008' and `-mnan=2008'.
4063 (OPTION_DEFAULT_SPECS): Add "nan" default.
4064 (ASM_SPEC): Handle `-mnan='.
4065 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
4066 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
4067 comment accordingly.
4068 (neg<mode>2): Likewise.
4069 * config/mips/mips.opt (mabs, mnan): New options.
4070 * doc/install.texi (Configuration): Document `--with-nan=' option.
4071 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
4073 (MIPS Options): Document them.
4074 * config.gcc <mips*-*-*>: Handle `--with-nan='.
4075 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
4076 * configure: Regenerate.
4077 * config.in: Regenerate.
4079 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
4081 * config/i386/i386.md (float post-reload splitters): Do not check
4082 for subregs of SSE registers.
4084 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
4085 H.J. Lu <hongjiu.lu@intel.com>
4089 * config/i386/i386.md (post-reload splitter
4090 to avoid partial SSE reg dependency stalls): New pattern.
4092 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
4094 * config/s390/s390.md ("movcc"): Swap load and store instructions.
4096 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
4098 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
4099 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
4101 2013-07-26 Cary Coutant <ccoutant@google.com>
4103 * dwarf2out.c (die_checksum_ordered): Don't include template
4104 instantiations in signature.
4105 (is_template_parameter): New function.
4106 (is_template_instantiation): New function.
4107 (generate_skeleton_bottom_up): Don't include template instantiations
4109 (generate_skeleton): Likewise.
4110 (break_out_comdat_types): Move recursive call to break out nested
4112 (prune_unused_types_mark_generic_parms_dies): Call
4113 is_template_parameter.
4115 2013-07-26 Ian Bolton <ian.bolton@arm.com>
4117 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
4118 uses vector registers.
4119 * config/aarch64/iterators.md: Add attributes rtn and vas.
4121 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4122 Richard Earnshaw <richard.earnshaw@arm.com>
4124 * combine.c (simplify_comparison): Re-canonicalize operands
4126 * config/arm/arm.md (movcond_addsi): New splitter.
4128 2013-07-25 Sterling Augustine <saugustine@google.com>
4130 * dwarf2out.c (size_of_pubnames): Move code to...
4131 (include_pubname_in_output): ...here. New.
4132 (want_pubnames): Rearrange.
4133 (output_pubnames): Call include_pubname_in_output. Move assertion.
4135 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
4137 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
4139 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
4142 * doc/extend.texi: Remove obsolete builtins. Fix
4143 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
4145 2013-07-25 Jan Hubicka <jh@suse.cz>
4147 * cgraph.c (release_function_body): Break out from ...
4148 (cgraph_release_function_body): ... this one; also release DECL_RESULT
4150 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
4151 old_tree in the map.
4152 (create_specialized_node): Update.
4153 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
4155 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
4156 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
4157 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
4158 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
4161 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4163 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
4164 addsi3_carryin_alt2_<optab>): Correct output template.
4166 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4168 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
4169 Adjust for arm_restrict_it.
4170 Remove trailing whitespace.
4172 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
4174 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
4175 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
4177 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
4179 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
4181 PR rtl-optimization/57960
4182 * lra-constraints.c (process_alt_operands): Use the right mode
4183 when checking strict_low.
4185 2013-07-25 Jan Hubicka <jh@suse.cz>
4187 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
4188 * cgraph.c (cgraph_remove_node): Do not release function body
4189 when in cgraph streaming.
4190 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
4191 in other partitions are not considered reachable; fix handling of
4194 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4196 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
4198 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4203 * config/arm/arm.md ("*sibcall_insn): Replace use of
4204 Ss with US. Adjust output for v5 and v4t.
4205 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
4207 * config/arm/constraints.md ("Ss"): Rename to US.
4209 2013-07-25 Terry Guo <terry.guo@arm.com>
4211 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
4212 shift_add/shift_sub0/shift_sub1 RTXs.
4214 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
4215 Anton Blanchard <anton@au1.ibm.com>
4217 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
4218 (altivec_vpks<VI_char>ss): Likewise.
4219 (altivec_vpks<VI_char>us): Likewise.
4220 (altivec_vpku<VI_char>us): Likewise.
4221 (altivec_vpku<VI_char>um): Likewise.
4223 2013-07-24 David Malcolm <dmalcolm@redhat.com>
4225 Introduce context class.
4227 * Makefile.in (CONTEXT_H): New.
4228 (OBJS): Add context.o.
4229 (toplev.o): Add CONTEXT_H to dependencies.
4232 * toplev.c (general_init): Create the singleton gcc::context instance.
4238 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
4240 PR rtl-optimization/57968
4241 * mode-switching.c (create_pre_exit): Allow instructions that
4242 don't set a return register to need a non-exit mode.
4244 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4245 Anton Blanchard <anton@au1.ibm.com>
4247 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
4248 operands to vperm for little endian.
4249 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
4250 of lvsl to create the control mask for a vperm for little endian.
4252 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4253 Anton Blanchard <anton@au1.ibm.com>
4255 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
4256 two operands for little-endian.
4258 2013-07-23 Steve Ellcey <sellcey@mips.com>
4260 * config/mips/mips.c (mips_case_values_threshold): New.
4261 (TARGET_CASE_VALUES_THRESHOLD): Define.
4263 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4264 Anton Blanchard <anton@au1.ibm.com>
4266 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
4267 selection of field for vector splat in little endian mode.
4269 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4271 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
4272 expanders to rs6000.md.
4273 (ior<mode>3): Likewise.
4274 (and<mode>3): Likewise.
4275 (one_cmpl<mode>2): Likewise.
4276 (nor<mode>3): Likewise.
4277 (andc<mode>3): Likewise.
4278 (eqv<mode>3): Likewise.
4279 (nand<mode>3): Likewise.
4280 (orc<mode>3): Likewise.
4282 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
4285 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
4286 to split multi-word logical operations.
4287 (rs6000_split_logical_di): Likewise.
4288 (rs6000_split_logical): Likewise.
4290 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
4291 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
4292 and allow TImode operations in 32-bit.
4293 (vsx_and<mode>3_64bit): Likewise.
4294 (vsx_ior<mode>3_32bit): Likewise.
4295 (vsx_ior<mode>3_64bit): Likewise.
4296 (vsx_xor<mode>3_32bit): Likewise.
4297 (vsx_xor<mode>3_64bit): Likewise.
4298 (vsx_one_cmpl<mode>2_32bit): Likewise.
4299 (vsx_one_cmpl<mode>2_64bit): Likewise.
4300 (vsx_nor<mode>3_32bit): Likewise.
4301 (vsx_nor<mode>3_64bit): Likewise.
4302 (vsx_andc<mode>3_32bit): Likewise.
4303 (vsx_andc<mode>3_64bit): Likewise.
4304 (vsx_eqv<mode>3_32bit): Likewise.
4305 (vsx_eqv<mode>3_64bit): Likewise.
4306 (vsx_nand<mode>3_32bit): Likewise.
4307 (vsx_nand<mode>3_64bit): Likewise.
4308 (vsx_orc<mode>3_32bit): Likewise.
4309 (vsx_orc<mode>3_64bit): Likewise.
4311 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
4312 logical types in GPRs.
4314 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
4315 logical insns to rs6000.md, and allow TImode operations in
4317 (altivec_ior<mode>3): Likewise.
4318 (altivec_xor<mode>3): Likewise.
4319 (altivec_one_cmpl<mode>2): Likewise.
4320 (altivec_nor<mode>3): Likewise.
4321 (altivec_andc<mode>3): Likewise.
4323 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
4324 attributes for moving the 128-bit logical operations into
4326 (BOOL_REGS_OUTPUT): Likewise.
4327 (BOOL_REGS_OP1): Likewise.
4328 (BOOL_REGS_OP2): Likewise.
4329 (BOOL_REGS_UNARY): Likewise.
4330 (BOOL_REGS_AND_CR0): Likewise.
4331 (one_cmpl<mode>2): Add support for DI logical operations on
4332 32-bit, splitting the operations to 32-bit.
4336 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
4337 changes to combine the 32/64-bit code, allow logical operations on
4338 TI mode in 32-bit, and to use similar match_operator patterns like
4339 scalar mode uses. Combine the Altivec and VSX code for logical
4340 operations, and move it here.
4341 (ior<mode>3, 128-bit types): Likewise.
4342 (xor<mode>3, 128-bit types): Likewise.
4343 (one_cmpl<mode>3, 128-bit types): Likewise.
4344 (nor<mode>3, 128-bit types): Likewise.
4345 (andc<mode>3, 128-bit types): Likewise.
4346 (eqv<mode>3, 128-bit types): Likewise.
4347 (nand<mode>3, 128-bit types): Likewise.
4348 (orc<mode>3, 128-bit types): Likewise.
4349 (and<mode>3_internal): Likewise.
4350 (bool<mode>3_internal): Likewise.
4351 (boolc<mode>3_internal1): Likewise.
4352 (boolc<mode>3_internal2): Likewise.
4353 (boolcc<mode>3_internal1): Likewise.
4354 (boolcc<mode>3_internal2): Likewise.
4355 (eqv<mode>3_internal1): Likewise.
4356 (eqv<mode>3_internal2): Likewise.
4357 (one_cmpl1<mode>3_internal): Likewise.
4359 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
4361 * config/microblaze/microblaze.c (microblaze_expand_prologue):
4362 Rename flag_stack_usage to flag_stack_usage_info.
4364 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
4366 * config/microblaze/sync.md: New file.
4367 * config/microblaze/microblaze.md: Include sync.md
4368 * config/microblaze/microblaze.c: Add print_operand 'y'.
4369 * config/microblaze/constraints.md: Add memory_contraint
4370 'Q' which is a single register.
4372 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
4374 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
4376 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
4378 * reload.c (find_reloads): Exit loop once we find this operand
4379 cannot be reloaded somehow for this alternative.
4381 * reload.c (find_reloads): Exit loop once we find a hard register.
4383 * rtlanal.c (computed_jump_p): Exit loop once we find label
4386 * i386.c (ix86_pad_returns): Exit loop after setting replace.
4388 * cfgloopmanip.c (remove_path): Exit loop after setting
4391 * gensupport.c (subst_dup): Avoid loop if code is not
4392 MATCH_DUP nor MATCH_OP_DUP.
4394 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
4396 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
4398 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4400 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
4401 true for SP_REGNUM if mode == ptr_mode.
4402 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
4403 with value R0_REGNUM + 31.
4405 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4407 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4408 pad pointer-typed argument downward.
4410 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4412 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4413 and __ILP32__ when the ILP32 model is in use.
4415 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4417 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4418 (aarch64_load_symref_appropriately): In the case of
4419 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4420 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4421 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4422 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4423 if 'mode' doesn't equal to 'ptr_mode'.
4424 (aarch64_output_mi_thunk): Add an assertion on the alignment of
4425 'vcall_offset'; change to call aarch64_emit_move differently depending
4426 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4427 to calculate the upper bound of 'vcall_offset'.
4428 (aarch64_cannot_force_const_mem): Change to also return true if
4430 (aarch64_legitimize_reload_address): In the case of large
4431 displacements, add new local variable 'xmode' and an assertion
4432 based on it; change to use 'xmode' to generate the new rtx and
4434 (aarch64_asm_trampoline_template): Change to generate the template
4435 differently depending on TARGET_ILP32 or not; change to use
4436 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4437 (aarch64_trampoline_size): Removed.
4438 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4439 and replace immediate literals with it. Change to use 'ptr_mode'
4440 instead of 'DImode' and call convert_memory_address if the mode
4441 of 'fnaddr' doesn't equal to 'ptr_mode'.
4442 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4444 (aarch64_elf_asm_destructor): Likewise.
4445 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4446 on TARGET_ILP32 instead of aarch64_trampoline_size.
4447 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4448 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4449 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4450 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
4451 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4452 (storewb_pair<GPI:mode>_<P:mode>): ... this.
4453 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4454 depending on the value of 'mode'.
4455 (add_losym_<mode>): New.
4456 (ldr_got_small_<mode>): New, based on ldr_got_small.
4457 (ldr_got_small): Remove.
4458 (ldr_got_small_sidi): New.
4459 * config/aarch64/iterators.md (P): New.
4460 (PTR): Change to 'ptr_mode' in the condition.
4462 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4464 * config.gcc (aarch64*-*-*): Support --with-abi.
4465 (aarch64*-*-elf): Support --with-multilib-list.
4466 (aarch64*-*-linux*): Likewise.
4467 (supported_defaults): Add abi to aarch64*-*-*.
4468 * configure.ac: Mention AArch64 for --with-multilib-list.
4469 * configure: Re-generated.
4470 * config/aarch64/biarchilp32.h: New file.
4471 * config/aarch64/biarchlp64.h: New file.
4472 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4474 (MULTILIB_DEFAULTS): Ditto.
4475 (DRIVER_SELF_SPECS): Ditto.
4476 (ASM_SPEC): Update to also substitute -mabi.
4477 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4478 file whose name depends on -mabi= and -mbig-endian.
4479 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4481 (POINTER_SIZE): New define.
4482 (POINTERS_EXTEND_UNSIGNED): Ditto.
4483 (enum aarch64_abi_type): New enumeration tag.
4484 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4485 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4486 (TARGET_ILP32): New define.
4487 * config/aarch64/aarch64.opt (mabi): New.
4489 (ilp32, lp64): New values for -mabi.
4490 * config/aarch64/t-aarch64 (comma): New define.
4491 (MULTILIB_OPTIONS): Ditto.
4492 (MULTILIB_DIRNAMES): Ditto.
4493 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4494 * doc/invoke.texi: Document -mabi for AArch64.
4496 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
4498 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4500 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4501 Anton Blanchard <anton@au1.ibm.com>
4503 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4504 endianness when selecting field to splat.
4506 2013-07-22 Eric Christopher <echristo@gmail.com>
4508 * dwarf2out.c (die_odr_checksum): New function to use
4509 CHECKSUM_ macros and ULEB128 for DIE tag.
4510 (generate_type_signature): Use.
4512 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
4514 * config.gcc (sparc*-*-*): Accept leon3 processor.
4515 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4516 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4517 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4518 * config/sparc/sparc.opt (enum processor_type): Add leon3.
4519 (mfix-ut699): Adjust comment.
4520 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4521 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4522 (CPP_CPU_SPEC): Likewise.
4523 (ASM_CPU_SPEC): Likewise.
4524 * config/sparc/sparc.c (leon3_cost): New constant.
4525 (sparc_option_override): Add leon3 support.
4526 (mem_ref): New function.
4527 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4528 (sparc_do_work_around_errata): Look into the instruction in the delay
4529 slot and adjust accordingly. Add fix for the data cache nullify issues
4530 of the UT699. Change insertion position for the NOP.
4531 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4532 (leon3_load): New reservation.
4533 (leon_store): Bump latency to 2.
4534 (grfpu): New automaton.
4535 (grfpu_alu): New unit.
4536 (grfpu_ds): Likewise.
4537 (leon_fp_alu): Adjust.
4538 (leon_fp_mult): Delete.
4539 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4540 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4541 * config/sparc/sparc.md (cpu): Add leon3.
4542 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4544 (atomic_test_and_set): Likewise.
4547 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
4549 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
4550 default for R5900 targets.
4551 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4552 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4553 * config/mips/mips.c (mips_option_override): Report an error for
4554 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
4555 for -march=r5900 -mhard-float.
4557 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
4559 * df-problems.c (can_move_insns_across): Exit loop once we
4560 find a non-fixed, non-global register.
4562 * ipa-pure-const.c (propagate_nothrow): Exit loop after
4565 * omega.c (omega_eliminate_red): Break after setting red_found.
4566 (omega_problem_has_red_equations): Similarly after setting found.
4567 (omega_query_variable): Similarly after setting coupled.
4569 2013-07-22 Marek Polacek <polacek@redhat.com>
4571 * gimplify.c: Don't include gimple.h twice.
4573 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4575 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4576 instead of TARGET_THUMB1.
4577 (Pz): New constraint.
4578 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4580 (compare_negsi_si): Likewise.
4581 (compare_addsi2_op0): Likewise.
4582 (compare_addsi2_op1): Likewise.
4583 (addsi3_carryin_<optab>): Likewise.
4584 (addsi3_carryin_alt2_<optab>): Likewise.
4585 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4586 for arm_restrict_it.
4587 (subsi3_carryin): Likewise.
4588 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4589 (minmax_arithsi): Disable for arm_restrict_it.
4590 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4591 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4592 (satsi_<SAT:code>_shift): Likewise.
4593 (arm_shiftsi3): Add alternative for 16-bit encoding.
4594 (arm32_movhf): Disable for arm_restrict_it.
4595 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4596 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4598 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
4600 * config/arm/arm.md (attribute "insn"): Delete.
4601 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4602 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4603 (not_shiftsi): Update for attribute change.
4604 (not_shiftsi_compare0): Likewise.
4605 (not_shiftsi_compare0_scratch): Likewise.
4606 (arm_one_cmplsi2): Likewise.
4607 (thumb1_one_cmplsi2): Likewise.
4608 (notsi_compare0): Likewise.
4609 (notsi_compare0_scratch): Likewise.
4610 (thumb1_movdi_insn): Likewise.
4611 (arm_movsi_insn): Likewise.
4612 (movhi_insn_arch4): Likewise.
4613 (movhi_bytes): Likewise.
4614 (arm_movqi_insn): Likewise.
4615 (thumb1_movqi_insn): Likewise.
4616 (arm32_movhf): Likewise.
4617 (thumb1_movhf): Likewise.
4618 (arm_movsf_soft_insn): Likewise.
4619 (thumb1_movsf_insn): Likewise.
4620 (thumb_movdf_insn): Likewise.
4621 (movsicc_insn): Likewise.
4622 (movsfcc_soft_insn): Likewise.
4623 (and_scc): Likewise.
4624 (cond_move): Likewise.
4625 (if_move_not): Likewise.
4626 (if_not_move): Likewise.
4627 (if_shift_move): Likewise.
4628 (if_move_shift): Likewise.
4629 (if_shift_shift): Likewise.
4630 (if_not_arith): Likewise.
4631 (if_arith_not): Likewise.
4632 (cond_move_not): Likewise.
4633 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4634 (neon_mov<mode>): Likewise.
4635 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4636 (thumb2_movsi_vfp): Likewise.
4637 (movsf_vfp): Likewise.
4638 (thumb2_movsf_vfp): Likewise.
4639 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4641 (cortexa7_older_only): Likewise.
4642 (cortexa7_younger): Likewise.
4643 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4644 (1020alu_shift_op): Likewise.
4645 (1020alu_shift_reg_op): Likewise.
4646 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4647 (alu_shift_op): Likewise.
4648 (alu_shift_reg_op): Likewise.
4649 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4650 (11_alu_shift_op): Likewise.
4651 (11_alu_shift_reg_op): Likewise.
4652 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4653 (9_alu_shift_reg_op): Likewise.
4654 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4656 (cortex_a15_alu_shift): Likewise.
4657 (cortex_a15_alu_shift_reg): Likewise.
4658 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4659 (cortex_a5_alu_shift): Likewise.
4660 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4662 (cortex_a53_alu_shift): Likewise.
4663 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4665 (cortex_a7_alu_reg): Likewise.
4666 (cortex_a7_alu_shift): Likewise.
4667 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4668 (cortex_a8_alu_shift): Likewise.
4669 (cortex_a8_alu_shift_reg): Likewise.
4670 (cortex_a8_mov): Likewise.
4671 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4672 (cortex_a9_dp_shift): Likewise.
4673 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4674 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4675 (cortex_r4_mov): Likewise.
4676 (cortex_r4_alu_shift): Likewise.
4677 (cortex_r4_alu_shift_reg): Likewise.
4678 * config/arm/fa526.md (526_alu_op): Update for attribute change.
4679 (526_alu_shift_op): Likewise.
4680 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4681 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4682 (626te_alu_shift_op): Likewise.
4683 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4684 (726te_alu_op): Likewise.
4685 (726te_alu_shift_op): Likewise.
4686 (726te_alu_shift_reg_op): Likewise.
4687 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4688 (mp626_alu_shift_op): Likewise.
4689 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4690 (pj4_alu_e1_conds): Likewise.
4691 (pj4_alu): Likewise.
4692 (pj4_alu_conds): Likewise.
4693 (pj4_shift): Likewise.
4694 (pj4_shift_conds): Likewise.
4695 (pj4_alu_shift): Likewise.
4696 (pj4_alu_shift_conds): Likewise.
4698 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4700 * config/arm/predicates.md (shiftable_operator_strict_it):
4702 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4703 Disable cond_exec version for arm_restrict_it.
4704 (thumb2_smaxsi3): Convert to generate cond_exec.
4705 (thumb2_sminsi3): Likewise.
4706 (thumb32_umaxsi3): Likewise.
4707 (thumb2_uminsi3): Likewise.
4708 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4709 (thumb2_neg_abssi2): Likewise.
4710 (thumb2_mov_scc): Add alternative for 16-bit encoding.
4711 (thumb2_movsicc_insn): Adjust alternatives.
4712 (thumb2_mov_negscc): Disable for arm_restrict_it.
4713 (thumb2_mov_negscc_strict_it): New pattern.
4714 (thumb2_mov_notscc_strict_it): New pattern.
4715 (thumb2_mov_notscc): Disable for arm_restrict_it.
4716 (thumb2_ior_scc): Likewise.
4717 (thumb2_ior_scc_strict_it): New pattern.
4718 (thumb2_cond_move): Adjust for arm_restrict_it.
4719 (thumb2_cond_arith): Disable for arm_restrict_it.
4720 (thumb2_cond_arith_strict_it): New pattern.
4721 (thumb2_cond_sub): Adjust for arm_restrict_it.
4722 (thumb2_movcond): Likewise.
4723 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4724 (thumb2_zero_extendhisi2_v6): Likewise.
4725 (thumb2_zero_extendqisi2_v6): Likewise.
4726 (orsi_notsi_si): Likewise.
4727 (orsi_not_shiftsi_si): Likewise.
4729 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
4731 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4732 instruction sequence is 1 byte shorter.
4734 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
4736 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4737 it is not needed after split.
4739 2013-07-20 Iain Sandoe <iain@codesourcery.com>
4742 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4743 second label for nonlocal goto receivers. Don't output pic base labels
4744 unless we're producing PIC; mark that action unreachable().
4745 (ix86_save_reg): If the function contains a nonlocal label, save the
4747 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4748 * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4749 (update_pic_label_number_if_needed): New.
4750 (machopic_output_function_base_name): Adjust for nonlocal receiver
4752 (machopic_should_output_picbase_label): New.
4753 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4754 (nonlocal_goto_receiver): New insn and split.
4756 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
4758 * config/aarch64/aarch64-builtins.c
4759 (aarch64_fold_builtin): Fold abs in all modes.
4760 * config/aarch64/aarch64-simd-builtins.def
4761 (abs): Enable for all modes.
4762 * config/aarch64/arm_neon.h
4763 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4764 (vabs_f64): Add missing intrinsic.
4766 2013-07-19 Ian Bolton <ian.bolton@arm.com>
4768 * config/aarch64/arm_neon.h (vabs_s64): New function
4770 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
4773 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4774 * config/avr/avr.md (adjust_len): Add `round'.
4775 * config/avr/avr-protos.h (avr_out_round): New prototype.
4776 (avr_out_plus): Add `out_label' argument.
4777 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4778 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4779 Handle the case where `insn' is just a pattern.
4780 (avr_out_bitop): Handle the case where `insn' is just a pattern.
4781 (avr_out_round): New function.
4782 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4784 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
4786 * config/microblaze/microblaze.c (microblaze_expand_prologue):
4787 Add check for flag_stack_usage to handle -fstack-usage support
4789 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
4791 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4792 interaction for new Power8 flags and VSX.
4794 2013-07-18 Sriraman Tallam <tmsriram@google.com>
4797 * tree-inline.c (expand_call_inline): Emit errors during
4798 early_inlining only if optimization is not turned on.
4800 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4804 * passes.c (init_optimization_passes): Move the construction of
4805 the pass hierarchy into a new passes.def file.
4807 * Makefile.in (passes.o): Add dependency on passes.def.
4809 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4811 * passes.c (init_optimization_passes): Introduce macros for
4812 constructing the tree of passes (INSERT_PASSES_AFTER,
4813 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4814 TERMINATE_PASS_LIST).
4816 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
4817 Wei Mi <wmi@google.com>
4819 PR rtl-optimization/57878
4820 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4822 (reload_pseudo_compare_func): Check nregs first for reload
4825 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4827 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4829 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
4831 * read-rtl.c (validate_const_int): Once an invalid character is
4832 seen, quit the loop.
4834 * gengtype.c (write_roots): Similarly once we find the "deletable"
4835 or "if_marked" option.
4837 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
4839 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4840 "xtab" and "sat". Move value "clz" from here to ...
4841 (attriubte "type"): ... here.
4842 (satsi_<SAT:code>): Delete "insn" attribute.
4843 (satsi_<SAT:code>_shift): Likewise.
4844 (arm_zero_extendqisi2addsi): Likewise.
4845 (arm_extendqisi2addsi): Likewise.
4846 (clzsi2): Update for attribute changes.
4847 (rbitsi2): Likewise.
4848 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4850 (arm_usatsihi): Likewise.
4851 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4853 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
4855 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4856 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
4857 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
4858 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
4859 in alphabetical order.
4860 (attribute "core_cycles"): Update for attribute changes.
4861 (arm_addsi3): Likewise.
4862 (addsi3_compare0): Likewise.
4863 (addsi3_compare0_scratch): Likewise.
4864 (addsi3_compare_op1): Likewise.
4865 (addsi3_compare_op2): Likewise.
4866 (compare_addsi2_op0): Likewise.
4867 (compare_addsi2_op1): Likewise.
4868 (addsi3_carryin_shift_<optab>): Likewise.
4869 (subsi3_carryin_shift): Likewise.
4870 (rsbsi3_carryin_shift): Likewise.
4871 (arm_subsi3_insn): Likewise.
4872 (subsi3_compare0): Likewise.
4873 (subsi3_compare): Likewise.
4874 (arm_andsi3_insn): Likewise.
4875 (thumb1_andsi3_insn): Likewise.
4876 (andsi3_compare0): Likewise.
4877 (andsi3_compare0_scratch): Likewise.
4878 (zeroextractsi_compare0_scratch
4879 (andsi_not_shiftsi_si): Likewise.
4880 (iorsi3_insn): Likewise.
4881 (iorsi3_compare0): Likewise.
4882 (iorsi3_compare0_scratch): Likewise.
4883 (arm_xorsi3): Likewise.
4884 (thumb1_xorsi3_insn): Likewise.
4885 (xorsi3_compare0): Likewise.
4886 (xorsi3_compare0_scratch): Likewise.
4887 (satsi_<SAT:code>_shift): Likewise.
4889 (arm_shiftsi3): Likewise.
4890 (shiftsi3_compare0): Likewise.
4891 (not_shiftsi): Likewise.
4892 (not_shiftsi_compare0): Likewise.
4893 (not_shiftsi_compare0_scratch): Likewise.
4894 (arm_one_cmplsi2): Likewise.
4895 (thumb_one_complsi2): Likewise.
4896 (notsi_compare0): Likewise.
4897 (notsi_compare0_scratch): Likewise.
4898 (thumb1_zero_extendhisi2): Likewise.
4899 (arm_zero_extendhisi2): Likewise.
4900 (arm_zero_extendhisi2_v6): Likewise.
4901 (arm_zero_extendhisi2addsi): Likewise.
4902 (thumb1_zero_extendqisi2): Likewise.
4903 (thumb1_zero_extendqisi2_v6): Likewise.
4904 (arm_zero_extendqisi2): Likewise.
4905 (arm_zero_extendqisi2_v6): Likewise.
4906 (arm_zero_extendqisi2addsi): Likewise.
4907 (thumb1_extendhisi2): Likewise.
4908 (arm_extendhisi2): Likewise.
4909 (arm_extendhisi2_v6): Likewise.
4910 (arm_extendqisi): Likewise.
4911 (arm_extendqisi_v6): Likewise.
4912 (arm_extendqisi2addsi): Likewise.
4913 (thumb1_extendqisi2): Likewise.
4914 (thumb1_movdi_insn): Likewise.
4915 (arm_movsi_insn): Likewise.
4916 (movsi_compare0): Likewise.
4917 (movhi_insn_arch4): Likewise.
4918 (movhi_bytes): Likewise.
4919 (arm_movqi_insn): Likewise.
4920 (thumb1_movqi_insn): Likewise.
4921 (arm32_movhf): Likewise.
4922 (thumb1_movhf): Likewise.
4923 (arm_movsf_soft_insn): Likewise.
4924 (thumb1_movsf_insn): Likewise.
4925 (movdf_soft_insn): Likewise.
4926 (thumb_movdf_insn): Likewise.
4927 (arm_cmpsi_insn): Likewise.
4928 (cmpsi_shiftsi): Likewise.
4929 (cmpsi_shiftsi_swp): Likewise.
4930 (arm_cmpsi_negshiftsi_si): Likewise.
4931 (movsicc_insn): Likewise.
4932 (movsfcc_soft_insn): Likewise.
4933 (arith_shiftsi): Likewise.
4934 (arith_shiftsi_compare0
4935 (arith_shiftsi_compare0_scratch
4936 (sub_shiftsi): Likewise.
4937 (sub_shiftsi_compare0
4938 (sub_shiftsi_compare0_scratch
4939 (and_scc): Likewise.
4940 (cond_move): Likewise.
4941 (if_plus_move): Likewise.
4942 (if_move_plus): Likewise.
4943 (if_move_not): Likewise.
4944 (if_not_move): Likewise.
4945 (if_shift_move): Likewise.
4946 (if_move_shift): Likewise.
4947 (if_shift_shift): Likewise.
4948 (if_not_arith): Likewise.
4949 (if_arith_not): Likewise.
4950 (cond_move_not): Likewise.
4951 (thumb1_ashlsi3): Set type attribute.
4952 (thumb1_ashrsi3): Likewise.
4953 (thumb1_lshrsi3): Likewise.
4954 (thumb1_rotrsi3): Likewise.
4955 (shiftsi3_compare0_scratch): Likewise.
4956 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4957 (neon_mov<mode>): Likewise.
4958 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4960 (thumb2_movsi_insn): Likewise.
4961 (thumb2_cmpsi_neg_shiftsi): Likewise.
4962 (thumb2_extendqisi_v6): Likewise.
4963 (thumb2_zero_extendhisi2_v6): Likewise.
4964 (thumb2_zero_extendqisi2_v6): Likewise.
4965 (thumb2_shiftsi3_short): Likewise.
4966 (thumb2_addsi3_compare0_scratch): Likewise.
4967 (orsi_not_shiftsi_si): Likewise.
4968 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4969 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4971 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4972 (1020alu_shift_op): Likewise.
4973 (1020alu_shift_reg_op): Likewise.
4974 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4975 (alu_shift_op): Likewise.
4976 (alu_shift_reg_op): Likewise.
4977 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4978 (11_alu_shift_op): Likewise.
4979 (11_alu_shift_reg_op): Likewise.
4980 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4981 (9_alu_shift_reg_op): Likewise.
4982 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4984 (cortex_a15_alu_shift): Likewise.
4985 (cortex_a15_alu_shift_reg): Likewise.
4986 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4988 (cortex_a5_alu_shift): Likewise.
4989 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4991 (cortex_a53_alu_shift): Likewise.
4992 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4994 (cortex_a7_alu_reg): Likewise.
4995 (cortex_a7_alu_shift): Likewise.
4996 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4998 (cortex_a8_alu_shift): Likewise.
4999 (cortex_a8_alu_shift_reg): Likewise.
5000 (cortex_a8_mov): Likewise.
5001 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
5002 (cortex_a9_dp_shift): Likewise.
5003 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
5005 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
5007 (cortex_r4_mov): Likewise.
5008 (cortex_r4_alu_shift): Likewise.
5009 (cortex_r4_alu_shift_reg): Likewise.
5010 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
5011 (526_alu_shift_op): Likewise.
5012 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
5013 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
5014 (626te_alu_shift_op): Likewise.
5015 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
5016 (726te_alu_op): Likewise.
5017 (726te_alu_shift_op): Likewise.
5018 (726te_alu_shift_reg_op): Likewise.
5019 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
5020 (mp626_alu_shift_op): Likewise.
5021 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
5022 (pj4_alu_e1_conds): Likewise.
5023 (pj4_alu): Likewise.
5024 (pj4_alu_conds): Likewise.
5025 (pj4_shift): Likewise.
5026 (pj4_shift_conds): Likewise.
5027 (pj4_alu_shift): Likewise.
5028 (pj4_alu_shift_conds): Likewise.
5029 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5031 (cortexa7_older_only): Likewise.
5032 (cortexa7_younger): Likewise.
5034 2013-07-18 David Malcolm <dmalcolm@redhat.com>
5036 * ipa-pure-const.c (generate_summary): Rename to...
5037 (pure_const_generate_summary): ... this.
5039 2013-07-17 Iain Sandoe <iain@codesourcery.com>
5041 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
5043 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
5046 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
5049 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5051 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
5052 enabled without -march=zEC12.
5053 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
5056 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5058 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
5059 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
5060 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
5061 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
5062 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
5063 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
5064 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
5066 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
5067 and ISA_HAS_NMADD3_NMSUB3.
5068 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
5069 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
5070 (nmsub4<mode>, nmsub3<mode>): Likewise.
5071 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
5073 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5075 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
5076 TARGET_MIPS5400 checking.
5078 2013-07-16 Jakub Jelinek <jakub@redhat.com>
5079 Peter Bergner <bergner@vnet.ibm.com>
5081 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
5082 registers in the comment.
5083 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
5084 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
5085 rather than FIRST_PSEUDO_REGISTERS.
5087 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
5089 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
5090 enable extra ISA flags with TARGET_HTM.
5092 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5094 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
5097 2013-07-15 Cong Hou <congh@google.com>
5099 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
5100 in compare function for sorting.
5102 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
5104 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
5105 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
5106 * config/rs6000/rs6000.opt: Add -mhtm option.
5107 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
5108 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
5109 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
5110 __HTM__ if the HTM instructions are available.
5111 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
5112 htm_spr_reg_operand): New define_predicates.
5113 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
5114 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
5116 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
5117 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
5118 HTM builtin functions.
5119 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
5120 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
5121 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
5122 (rs6000_builtin_mask_calculate): Likewise.
5123 (rs6000_option_override_internal): Likewise.
5124 (bdesc_htm): Add new HTM builtin support.
5125 (htm_spr_num): New function.
5126 (htm_spr_regno): Likewise.
5127 (rs6000_htm_spr_icode): Likewise.
5128 (htm_expand_builtin): Likewise.
5129 (htm_init_builtins): Likewise.
5130 (rs6000_expand_builtin): Add support for HTM builtin functions.
5131 (rs6000_init_builtins): Likewise.
5132 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
5134 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
5135 (TARGET_HTM, MASK_HTM): Define macros.
5136 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
5137 (FIXED_REGISTERS): Likewise.
5138 (CALL_USED_REGISTERS): Likewise.
5139 (CALL_REALLY_USED_REGISTERS): Likewise.
5140 (REG_ALLOC_ORDER): Likewise.
5141 (enum reg_class): Likewise.
5142 (REG_CLASS_NAMES): Likewise.
5143 (REG_CLASS_CONTENTS): Likewise.
5144 (REGISTER_NAMES): Likewise.
5145 (ADDITIONAL_REGISTER_NAMES): Likewise.
5146 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
5147 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
5148 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
5149 * config/rs6000/htm.md: New file.
5150 * config/rs6000/htmintrin.h: New file.
5151 * config/rs6000/htmxlintrin.h: New file.
5153 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
5155 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5156 Define SYMBOL_TINY_GOT, update comment.
5157 * config/aarch64/aarch64.c
5158 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
5159 (aarch64_expand_mov_immediate): Likewise.
5160 (aarch64_print_operand): Likewise.
5161 (aarch64_classify_symbol): Likewise.
5162 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
5163 (ldr_got_tiny): Define.
5165 2013-07-13 Tobias Grosser <tobias@grosser.es>
5167 PR tree-optimization/54094
5168 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
5169 scheduling dimension for the parallelism check from the polyhedral
5170 information in the AST.
5171 * graphite-dependences.c (carries_deps): Do not assume the schedule is
5174 2013-07-13 Jason Merrill <jason@redhat.com>
5176 * print-tree.c (debug_vec_tree): Use debug_raw.
5177 (debug_raw (vec<tree, va_gc> &)): New.
5178 (debug_raw (vec<tree, va_gc> *)): New.
5179 * tree.h: Declare them.
5181 2013-07-13 Bin Cheng <bin.cheng@arm.com>
5183 * ifcvt.c (ifcvt_after_combine): New static variable.
5184 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
5186 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
5187 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
5188 rest_of_handle_if_after_reload): Pass new argument for if_convert.
5190 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
5192 * config/mips/mips.c (mips_expand_call): Remove empty statement.
5194 2013-07-12 Michael Matz <matz@suse.de>
5197 * convert.c (convert_to_real): Reject non-float inner types.
5199 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
5201 * config/aarch64/aarch64-protos.h
5202 (aarch64_simd_immediate_valid_for_move): Remove.
5203 * config/aarch64/aarch64.c (simd_immediate_info): New member.
5204 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
5206 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
5208 2013-07-11 Steve Ellcey <sellcey@mips.com>
5210 * config/mips/mips.c (mips_conditional_register_usage): Do not
5211 use t[0-7] registers in MIPS16 mode when optimizing for size.
5213 2013-07-11 Sriraman Tallam <tmsriram@google.com>
5215 * config/i386/i386.c (dispatch_function_versions): Fix array
5216 indexing of function_version_info to match actual_versions.
5218 2013-07-11 Teresa Johnson <tejohnson@google.com>
5220 * vec.h (struct va_gc): Move release out-of-line.
5221 (va_gc::release): Call ggc_free on released vec.
5223 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
5225 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
5226 Require GOT register as additional operand in UNSPEC.
5227 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
5228 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5229 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5230 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
5231 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
5232 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
5233 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
5235 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
5238 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
5239 name seen by assembler/linker rather if available.
5241 2013-07-11 Andreas Schwab <schwab@suse.de>
5243 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
5245 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
5247 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
5249 2013-07-10 Joseph Myers <joseph@codesourcery.com>
5251 * doc/tm.texi.in: Move hook documentation to ....
5252 * target.def: ... here.
5254 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
5256 * doc/tm.texi: Regenerate.
5258 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
5261 * doc/invoke.texi: Document -Wconditionally-supported.
5263 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5266 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5269 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5272 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
5273 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
5274 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
5275 Remove duplicate devices.
5276 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
5277 * config/avr/t-multilib: Regenerate.
5278 * config/avr/avr-tables.opt: Regenerate.
5279 * doc/avr-mmcu.texi: Regenerate.
5281 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5284 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
5286 2013-07-10 Graham Stott <graham.stott@btinternet.com>
5288 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
5289 the cost of MULT when optimizing for size.
5291 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5293 * config/cr16/cr16-protos.h: Don't include target.h.
5295 2013-07-09 Joseph Myers <joseph@codesourcery.com>
5297 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
5298 adjust register size for TDmode and TFmode for VSX registers.
5300 2013-07-08 Kai Tietz <ktietz@redhat.com>
5303 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
5304 hook_bool_const_tree_true.
5306 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5308 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
5309 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
5310 * config/s390/s390.md: Define FPR*_REGNUM constants.
5311 Fix FPR2_REGNUM constant (18 -> 17).
5312 ("*trunc<BFP:mode><DFP_ALL:mode>2")
5313 ("*trunc<DFP_ALL:mode><BFP:mode>2")
5314 ("trunc<BFP:mode><DFP_ALL:mode>2")
5315 ("trunc<DFP_ALL:mode><BFP:mode>2")
5316 ("*extend<BFP:mode><DFP_ALL:mode>2")
5317 ("*extend<DFP_ALL:mode><BFP:mode>2")
5318 ("extend<BFP:mode><DFP_ALL:mode>2")
5319 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
5322 2013-07-08 Graham Stott <graham.stott@btinternet.com>
5324 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
5326 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5328 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
5329 and cfun_fpr_bit_p to cfun_fpr_save_p.
5330 (s390_frame_area, s390_register_info, s390_frame_info)
5331 (s390_emit_prologue, s390_emit_epilogue)
5332 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
5334 * config/s390/s390.h: Define *_REGNUM macros for floating point
5337 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
5339 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
5341 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
5343 PR rtl-optimization/57786
5344 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
5345 and break out of the loop when it is set to false.
5347 2013-07-08 Jakub Jelinek <jakub@redhat.com>
5350 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
5351 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
5352 (const_int 63)) 0)).
5353 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
5354 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
5355 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
5357 PR rtl-optimization/57829
5358 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
5359 mask bits outside of mode are just sign-extension from mode to HWI.
5361 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
5363 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
5364 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
5365 adjust_address instead of change_address to keep info about alignment.
5366 (emit_strmov): Remove.
5367 (emit_memmov): New function.
5368 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
5369 (expand_movmem_epilogue): Likewise and return updated rtx for
5371 (expand_constant_movmem_prologue): Likewise and return updated rtx for
5372 destination and source.
5373 (decide_alignment): Refactor, handle vector_loop.
5374 (ix86_expand_movmem): Likewise.
5375 (ix86_expand_setmem): Likewise.
5376 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
5378 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
5380 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
5381 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
5383 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
5385 * config/i386/sse.md (sse_movlhps): Change alternative 3
5386 of operand 2 to "m".
5388 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
5391 * config/i386/sse.md (iptr): New mode attribute.
5392 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
5393 (<sse>_vm<plusminus_insn><mode>3): Ditto.
5394 (<sse>_vmmul<mode>3): Ditto.
5395 (<sse>_vmdiv<mode>3): Ditto.
5396 (sse_vmrcpv4sf2): Ditto.
5397 (<sse>_vmsqrt<mode>2): Ditto.
5398 (sse_vmrsqrtv4sf2): Ditto.
5399 (<sse>_vm<code><mode>3): Ditto.
5400 (avx_vmcmp<mode>3): Ditto.
5401 (<sse>_vmmaskcmp<mode>3): Ditto.
5402 (<sse>_comi): Ditto.
5403 (<sse>_ucomi): Ditto.
5404 (*xop_vmfrcz_<mode>): Ditto.
5405 (*fmai_fmadd_<mode>): Ditto.
5406 (*fmai_fmsub_<mode>): Ditto.
5407 (*fmai_fnmadd_<mode>): Ditto.
5408 (*fmai_fnmsub_<mode>): Ditto.
5409 (*fma4i_vmfmadd_<mode>): Ditto.
5410 (*fma4i_vmfmsub_<mode>): Ditto.
5411 (*fma4i_vmfnmadd_<mode>): Ditto.
5412 (*fma4i_vmfnmsub_<mode>): Ditto.
5413 (*xop_vmfrcz_<mode>): Ditto.
5414 (sse_cvtps2pi): Ditto.
5415 (sse_cvttps2pi): Ditto.
5416 (sse_cvtss2si): Ditto.
5417 (sse_cvtss2si_2): Ditto.
5418 (sse_cvtss2siq_2): Ditto.
5419 (sse_cvttss2si): Ditto.
5420 (sse_cvttss2siq): Ditto.
5421 (sse_cvtsd2si): Ditto.
5422 (sse_cvtsd2si_2): Ditto.
5423 (sse_cvtsd2siq_2): Ditto.
5424 (sse_cvttsd2si): Ditto.
5425 (sse_cvttsd2siq): Ditto.
5426 (sse_cvtsd2ss): Ditto.
5427 (sse_cvtss2sd): Ditto.
5428 (avx2_pbroadcast<mode>): Ditto.
5429 (avx2_pbroadcast<mode>_1): Ditto.
5430 (*avx_vperm_broadcast_v4sf): Ditto.
5432 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5433 (sse_movlhps): Ditto.
5434 (sse_storehps): Ditto.
5435 (sse_loadhps): Ditto.
5436 (sse_storelps): Ditto.
5437 (sse_loadlps): Ditto.
5438 (*vec_concatv4sf): Ditto.
5439 (*vec_interleave_highv2df): Ditto.
5440 (*vec_interleave_lowv2df): Ditto.
5441 (*vec_extractv2df_1_sse): Ditto.
5442 (*vec_extractv2df_0_sse): Ditto.
5443 (sse2_storelpd): Ditto.
5444 (sse2_loadlpd): Ditto.
5445 (sse2_movsd): Ditto.
5446 (*vec_concatv4si): Ditto.
5447 (vec_concatv2di): Ditto.
5449 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5450 for Intel asm dialect.
5451 (mmx_punpcklwd): Ditto.
5452 (mmx_punpckldq): Ditto.
5454 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5455 for intel assembler dialect.
5457 2013-07-06 Jakub Jelinek <jakub@redhat.com>
5460 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5461 for BUILT_IN_C{LZ,LRSB}*.
5462 * tree.h (CASE_INT_FN): Add FN##IMAX case.
5463 * tree-vrp.c (extract_range_basic): Handle
5464 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
5465 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5466 fall thru to code calling set_value*.
5467 * builtins.c (expand_builtin): Remove *IMAX cases.
5468 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5469 if width is bigger than 2*HWI.
5471 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
5473 PR rtl-optimization/55342
5474 * lra-int.h (lra_subreg_reload_pseudos): New.
5475 * lra.c: Add undoing optional reloads to the block diagram.
5476 (lra_subreg_reload_pseudos): New.
5477 (lra_optional_reload_pseudos): Change comments.
5478 (lra): Init and clear lra_subreg_reload_pseudos. Clear
5479 lra_optional_reload_pseudos after undo transformations.
5480 * lra-assigns.c (pseudo_prefix_title): New.
5481 (lra_setup_reg_renumber): Use it.
5482 (spill_for): Ditto. Check subreg reload pseudos too.
5483 (assign_by_spills): Consider subreg reload pseudos too.
5484 * lra-constraints.c (simplify_operand_subreg): Use
5485 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5486 (curr_insn_transform): Recognize and do optional reloads.
5487 (undo_optional_reloads): New.
5488 (lra_undo_inheritance): Call undo_optional_reloads.
5490 2013-07-05 Thomas Quinot <quinot@adacore.com>
5492 * tree-complex.c (expand_complex_operations_1): Fix typo.
5494 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
5496 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5497 (tune_params): New member 'const vec_costs'.
5498 * config/aarch64/aarch64.c (generic_vector_cost): New.
5499 (generic_tunings): New member 'generic_vector_cost'.
5500 (aarch64_builtin_vectorization_cost): New.
5501 (aarch64_add_stmt_cost): New.
5502 (TARGET_VECTORIZE_ADD_STMT_COST): New.
5503 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5505 2013-07-03 Jakub Jelinek <jakub@redhat.com>
5508 * config/i386/predicates.md (vsib_address_operand): Disallow
5509 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5511 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
5514 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5515 expand_builtin_setjmp_receiver.
5516 (expand_label): Adjust, call expand_builtin_setjmp_receiver
5517 with NULL for the label parameter.
5518 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5519 the frame-pointer. Adjust comments.
5520 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5521 only if LABEL is non-NULL.
5523 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
5525 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5526 (ARM_ABI_AAPCS64): Ditto.
5528 (ARM_DEFAULT_ABI): Ditto.
5530 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
5532 * config/aarch64/aarch64-builtins.c
5533 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5534 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5536 * config/aarch64/aarch64-simd.md
5537 (aarch64_ld1<VALL:mode>): New.
5538 (aarch64_st1<VALL:mode>): Likewise.
5539 * config/aarch64/arm_neon.h
5540 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5542 2013-07-02 Sriraman Tallam <tmsriram@google.com>
5544 * config/i386/i386.c (gate_insert_vzeroupper): Check if
5546 (ix86_option_override_internal):Turn on all -mavx target flags by
5547 default as they are dependent on AVX anyway.
5549 2013-07-02 Cary Coutant <ccoutant@google.com>
5551 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5553 (loc_checksum_ordered): Likewise.
5554 (hash_loc_operands): Remove inline keyword.
5556 2013-07-02 Jakub Jelinek <jakub@redhat.com>
5558 PR tree-optimization/57741
5559 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5560 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5561 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5562 Allow REAL_CST step_exprs if flag_associative_math.
5563 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5565 2013-07-02 Ian Bolton <ian.bolton@arm.com>
5567 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5569 2013-07-02 Ian Bolton <ian.bolton@arm.com>
5571 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5573 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5575 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5577 (iorsi3_insn): Likewise.
5578 (arm_xorsi3): Likewise.
5580 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
5582 * arm.md (attribute "wtype"): Delete. Move attribute values from here
5584 (attribute "type"): ... here, and prefix with "wmmx_".
5585 (attribute "core_cycles"): Update for attribute changes.
5586 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5587 (tbcstv4hi): Likewise.
5588 (tbcstv2si): Likewise.
5589 (iwmmxt_iordi3): Likewise.
5590 (iwmmxt_xordi3): Likewise.
5591 (iwmmxt_anddi3): Likewise.
5592 (iwmmxt_nanddi3): Likewise.
5593 (iwmmxt_arm_movdi): Likewise.
5594 (iwmmxt_movsi_insn): Likewise.
5595 (mov<mode>_internal): Likewise.
5596 (and<mode>3_iwmmxt): Likewise.
5597 (ior<mode>3_iwmmxt): Likewise.
5598 (xor<mode>3_iwmmxt): Likewise.
5599 (add<mode>3_iwmmxt): Likewise.
5600 (ssaddv8qi3): Likewise.
5601 (ssaddv4hi3): Likewise.
5602 (ssaddv2si3): Likewise.
5603 (usaddv8qi3): Likewise.
5604 (usaddv4hi3): Likewise.
5605 (usaddv2si3): Likewise.
5606 (sub<mode>3_iwmmxt): Likewise.
5607 (sssubv8qi3): Likewise.
5608 (sssubv4hi3): Likewise.
5609 (sssubv2si3): Likewise.
5610 (ussubv8qi3): Likewise.
5611 (ussubv4hi3): Likewise.
5612 (ussubv2si3): Likewise.
5613 (mulv4hi3_iwmmxt): Likewise.
5614 (smulv4hi3_highpart): Likewise.
5615 (umulv4hi3_highpart): Likewise.
5616 (iwmmxt_wmacs): Likewise.
5617 (iwmmxt_wmacsz): Likewise.
5618 (iwmmxt_wmacu): Likewise.
5619 (iwmmxt_wmacuz): Likewise.
5620 (iwmmxt_clrdi): Likewise.
5621 (iwmmxt_clrv8qi): Likewise.
5622 (iwmmxt_clr4hi): Likewise.
5623 (iwmmxt_clr2si): Likewise.
5624 (iwmmxt_uavgrndv8qi3): Likewise.
5625 (iwmmxt_uavgrndv4hi3): Likewise.
5626 (iwmmxt_uavgv8qi3): Likewise.
5627 (iwmmxt_uavgv4hi3): Likewise.
5628 (iwmmxt_tinsrb): Likewise.
5629 (iwmmxt_tinsrh): Likewise.
5630 (iwmmxt_tinsrw): Likewise.
5631 (iwmmxt_textrmub): Likewise.
5632 (iwmmxt_textrmsb): Likewise.
5633 (iwmmxt_textrmuh): Likewise.
5634 (iwmmxt_textrmsh): Likewise.
5635 (iwmmxt_textrmw): Likewise.
5636 (iwmxxt_wshufh): Likewise.
5637 (eqv8qi3): Likewise.
5638 (eqv4hi3): Likewise.
5639 (eqv2si3): Likewise.
5640 (gtuv8qi3): Likewise.
5641 (gtuv4hi3): Likewise.
5642 (gtuv2si3): Likewise.
5643 (gtv8qi3): Likewise.
5644 (gtv4hi3): Likewise.
5645 (gtv2si3): Likewise.
5646 (smax<mode>3_iwmmxt): Likewise.
5647 (umax<mode>3_iwmmxt): Likewise.
5648 (smin<mode>3_iwmmxt): Likewise.
5649 (umin<mode>3_iwmmxt): Likewise.
5650 (iwmmxt_wpackhss): Likewise.
5651 (iwmmxt_wpackwss): Likewise.
5652 (iwmmxt_wpackdss): Likewise.
5653 (iwmmxt_wpackhus): Likewise.
5654 (iwmmxt_wpackwus): Likewise.
5655 (iwmmxt_wpackdus): Likewise.
5656 (iwmmxt_wunpckihb): Likewise.
5657 (iwmmxt_wunpckihh): Likewise.
5658 (iwmmxt_wunpckihw): Likewise.
5659 (iwmmxt_wunpckilb): Likewise.
5660 (iwmmxt_wunpckilh): Likewise.
5661 (iwmmxt_wunpckilw): Likewise.
5662 (iwmmxt_wunpckehub): Likewise.
5663 (iwmmxt_wunpckehuh): Likewise.
5664 (iwmmxt_wunpckehuw): Likewise.
5665 (iwmmxt_wunpckehsb): Likewise.
5666 (iwmmxt_wunpckehsh): Likewise.
5667 (iwmmxt_wunpckehsw): Likewise.
5668 (iwmmxt_wunpckelub): Likewise.
5669 (iwmmxt_wunpckeluh): Likewise.
5670 (iwmmxt_wunpckeluw): Likewise.
5671 (iwmmxt_wunpckelsb): Likewise.
5672 (iwmmxt_wunpckelsh): Likewise.
5673 (iwmmxt_wunpckelsw): Likewise.
5674 (ror<mode>3): Likewise.
5675 (ashr<mode>3_iwmmxt): Likewise.
5676 (lshr<mode>3_iwmmxt): Likewise.
5677 (ashl<mode>3_iwmmxt): Likewise.
5678 (ror<mode>3_di): Likewise.
5679 (ashr<mode>3_di): Likewise.
5680 (lshr<mode>3_di): Likewise.
5681 (ashl<mode>3_di): Likewise.
5682 (iwmmxt_wmadds): Likewise.
5683 (iwmmxt_wmaddu): Likewise.
5684 (iwmmxt_tmia): Likewise.
5685 (iwmmxt_tmiaph): Likewise.
5686 (iwmmxt_tmiabb): Likewise.
5687 (iwmmxt_tmiatb): Likewise.
5688 (iwmmxt_tmiabt): Likewise.
5689 (iwmmxt_tmiatt): Likewise.
5690 (iwmmxt_tmovmskb): Likewise.
5691 (iwmmxt_tmovmskh): Likewise.
5692 (iwmmxt_tmovmskw): Likewise.
5693 (iwmmxt_waccb): Likewise.
5694 (iwmmxt_wacch): Likewise.
5695 (iwmmxt_waccw): Likewise.
5696 (iwmmxt_waligni): Likewise.
5697 (iwmmxt_walignr): Likewise.
5698 (iwmmxt_walignr0): Likewise.
5699 (iwmmxt_walignr1): Likewise.
5700 (iwmmxt_walignr2): Likewise.
5701 (iwmmxt_walignr3): Likewise.
5702 (iwmmxt_wsadb): Likewise.
5703 (iwmmxt_wsadh): Likewise.
5704 (iwmmxt_wsadbz): Likewise.
5705 (iwmmxt_wsadhz): Likewise.
5706 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5707 (iwmmxt_wabsdiffb): Likewise.
5708 (iwmmxt_wabsdiffh): Likewise.
5709 (iwmmxt_wabsdiffw): Likewise.
5710 (iwmmxt_waddsubhx): Likewise
5711 (iwmmxt_wsubaddhx): Likewise.
5712 (addc<mode>3): Likewise.
5713 (iwmmxt_avg4): Likewise.
5714 (iwmmxt_avg4r): Likewise.
5715 (iwmmxt_wmaddsx): Likewise.
5716 (iwmmxt_wmaddux): Likewise.
5717 (iwmmxt_wmaddsn): Likewise.
5718 (iwmmxt_wmaddun): Likewise.
5719 (iwmmxt_wmulwsm): Likewise.
5720 (iwmmxt_wmulwum): Likewise.
5721 (iwmmxt_wmulsmr): Likewise.
5722 (iwmmxt_wmulumr): Likewise.
5723 (iwmmxt_wmulwsmr): Likewise.
5724 (iwmmxt_wmulwumr): Likewise.
5725 (iwmmxt_wmulwl): Likewise.
5726 (iwmmxt_wqmulm): Likewise.
5727 (iwmmxt_wqmulwm): Likewise.
5728 (iwmmxt_wqmulmr): Likewise.
5729 (iwmmxt_wqmulwmr): Likewise.
5730 (iwmmxt_waddbhusm): Likewise.
5731 (iwmmxt_waddbhusl): Likewise.
5732 (iwmmxt_wqmiabb): Likewise.
5733 (iwmmxt_wqmiabt): Likewise.
5734 (iwmmxt_wqmiatb): Likewise.
5735 (iwmmxt_wqmiatt): Likewise.
5736 (iwmmxt_wqmiabbn): Likewise.
5737 (iwmmxt_wqmiabtn): Likewise.
5738 (iwmmxt_wqmiatbn): Likewise.
5739 (iwmmxt_wqmiattn): Likewise.
5740 (iwmmxt_wmiabb): Likewise.
5741 (iwmmxt_wmiabt): Likewise.
5742 (iwmmxt_wmiatb): Likewise.
5743 (iwmmxt_wmiatt): Likewise.
5744 (iwmmxt_wmiabbn): Likewise.
5745 (iwmmxt_wmiabtn): Likewise.
5746 (iwmmxt_wmiatbn): Likewise.
5747 (iwmmxt_wmiattn): Likewise.
5748 (iwmmxt_wmiawbb): Likewise.
5749 (iwmmxt_wmiawbt): Likewise.
5750 (iwmmxt_wmiawtb): Likewise.
5751 (iwmmxt_wmiawtt): Likewise.
5752 (iwmmxt_wmiawbbn): Likewise.
5753 (iwmmxt_wmiawbtn): Likewise.
5754 (iwmmxt_wmiawtbn): Likewise.
5755 (iwmmxt_wmiawttn): Likewise.
5756 (iwmmxt_wmerge): Likewise.
5757 (iwmmxt_tandc<mode>3): Likewise.
5758 (iwmmxt_torc<mode>3): Likewise.
5759 (iwmmxt_torvsc<mode>3): Likewise.
5760 (iwmmxt_textrc<mode>3): Likewise.
5761 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5762 (wmmxt_pack): Likewise.
5763 (wmmxt_mult_c1): Likewise.
5764 (wmmxt_mult_c2): Likewise.
5765 (wmmxt_alu_c1): Likewise.
5766 (wmmxt_alu_c2): Likewise.
5767 (wmmxt_alu_c3): Likewise.
5768 (wmmxt_transfer_c1): Likewise.
5769 (wmmxt_transfer_c2): Likewise.
5770 (wmmxt_transfer_c3): Likewise.
5771 (marvell_f_iwmmxt_wstr): Likewise.
5772 (marvell_f_iwmmxt_wldr): Likewise.
5774 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
5776 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5778 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5781 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5782 * lra-constraints.c (need_for_split_p): Check call used hard regs
5783 living through calls.
5785 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5786 call used regs for call insn.
5788 2013-06-28 Jakub Jelinek <jakub@redhat.com>
5791 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5792 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5794 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
5796 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5798 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5800 * lra-constraints.c (need_for_split_p): Check call used hard regs
5801 living through calls.
5803 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
5806 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5807 to tie with any other modes. Eliminate Altivec vector mode tests,
5808 since these are a subset of ALTIVEC or VSX vector modes. Simplify
5809 code, to return 0 if testing MODE2 for a condition, if we've
5810 already tested MODE1 for the same condition.
5812 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5814 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5817 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5819 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5820 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5822 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5824 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5826 (aarch64_symbolic_constant_p): Remove.
5827 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5828 static. Fix line length and white space.
5829 (aarch64_symbolic_constant_p): Remove.
5830 * config/aarch64/predicates.md (aarch64_valid_symref):
5831 Use aarch64_classify_symbol_expression.
5833 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5835 * config/arm/constraints.md (Ts): New constraint.
5836 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5838 (compare_scc): Use "Ts" constraint for operand 0.
5839 (ior_scc_scc): Likewise.
5840 (and_scc_scc): Likewise.
5841 (and_scc_scc_nodom): Likewise.
5842 (ior_scc_scc_cmp): Likewise for operand 7.
5843 (and_scc_scc_cmp): Likewise.
5844 * config/arm/thumb2.md (thumb2_movsi_insn):
5845 Add alternatives for 16-bit encodings.
5846 (thumb2_movhi_insn): Likewise.
5847 (thumb2_movsicc_insn): Likewise.
5848 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
5849 (thumb2_negscc): Use "Ts" constraint.
5850 Move mvn instruction outside cond_exec block.
5851 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5852 for 16-bit encodings.
5854 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5856 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5858 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5859 (mulsi3subsi): Likewise.
5860 (mulsidi3adddi): Likewise.
5861 (mulsidi3_v6): Likewise.
5862 (umulsidi3_v6): Likewise.
5863 (umulsidi3adddi_v6): Likewise.
5864 (smulsi3_highpart_v6): Likewise.
5865 (umulsi3_highpart_v6): Likewise.
5866 (mulhisi3tb): Likewise.
5867 (mulhisi3bt): Likewise.
5868 (mulhisi3tt): Likewise.
5869 (maddhisi4): Likewise.
5870 (maddhisi4tb): Likewise.
5871 (maddhisi4tt): Likewise.
5872 (maddhidi4): Likewise.
5873 (maddhidi4tb): Likewise.
5874 (maddhidi4tt): Likewise.
5875 (zeroextractsi_compare0_scratch): Likewise.
5876 (insv_zero): Likewise.
5877 (insv_t2): Likewise.
5878 (anddi_notzesidi_di): Likewise.
5879 (anddi_notsesidi_di): Likewise.
5880 (andsi_notsi_si): Likewise.
5881 (iordi_zesidi_di): Likewise.
5882 (xordi_zesidi_di): Likewise.
5883 (andsi_iorsi3_notsi): Likewise.
5885 (smax_m1): Likewise.
5887 (not_shiftsi): Likewise.
5888 (unaligned_loadsi): Likewise.
5889 (unaligned_loadhis): Likewise.
5890 (unaligned_loadhiu): Likewise.
5891 (unaligned_storesi): Likewise.
5892 (unaligned_storehi): Likewise.
5893 (extv_reg): Likewise.
5894 (extzv_t2): Likewise.
5896 (udivsi3): Likewise.
5897 (arm_zero_extendhisi2addsi): Likewise.
5898 (arm_zero_extendqisi2addsi): Likewise.
5899 (compareqi_eq0): Likewise.
5900 (arm_extendhisi2_v6): Likewise.
5901 (arm_extendqisi2addsi): Likewise.
5902 (arm_movt): Likewise.
5903 (thumb2_ldrd): Likewise.
5904 (thumb2_ldrd_base): Likewise.
5905 (thumb2_ldrd_base_neg): Likewise.
5906 (thumb2_strd): Likewise.
5907 (thumb2_strd_base): Likewise.
5908 (thumb2_strd_base_neg): Likewise.
5909 (arm_negsi2): Add alternative for 16-bit encoding.
5910 (arm_one_cmplsi2): Likewise.
5912 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5914 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5915 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5916 (movdfcc): Likewise.
5917 * config/arm/vfp.md (*thumb2_movsf_vfp):
5918 Disable predication for arm_restrict_it.
5919 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5920 (*thumb2_movdfcc_vfp): Likewise.
5921 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5922 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5923 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5924 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5925 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5926 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5927 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5928 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5929 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5930 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5931 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5932 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5933 Disable predication for arm_restrict_it.
5935 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
5937 * config/i386/bmiintrin.h (_bextr_u32): New.
5938 (_bextr_u64): Ditto.
5940 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
5942 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5943 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5944 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5945 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5946 * config/mips/n32-elf.h: ...this new file.
5948 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
5951 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5952 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5954 2013-06-27 Catherine Moore <clm@codesourcery.com>
5956 * config/mips/mips-tables.opt: Regenerate.
5957 * config/mips/mips-cpus.def: Add m14ke and m14kec.
5958 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5959 * doc/invoke.texi: Add -m14kc.
5961 2013-06-27 Jakub Jelinek <jakub@redhat.com>
5964 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5965 constraints of operand 1 and 2.
5968 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5969 to match RTL canonicalization. Swap predicates and
5970 constraints of operand 1 and 2.
5972 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
5974 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5975 Process OP_INOUT regs for splitting too.
5977 2013-06-27 Jakub Jelinek <jakub@redhat.com>
5979 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5980 decl before the loop, initialize to NULL.
5981 (vectorizable_load): Initialize ptr_incr to NULL.
5983 2013-06-27 Martin Jambor <mjambor@suse.cz>
5986 * ipa-ref.h (ipa_maybe_record_reference): Declare.
5987 * ipa-ref.c (ipa_maybe_record_reference): New function.
5988 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5989 * ipa-cp.c (create_specialized_node): Record potential references from
5991 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5993 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
5995 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5996 parameter 'mode' of type 'enum machine_mode mode'; change to pass
5997 'mode' to force_reg.
5998 (aarch64_add_offset): Update calls to aarch64_force_temporary.
5999 (aarch64_expand_mov_immediate): Likewise.
6001 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
6003 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
6004 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
6006 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6008 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
6009 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6010 (struct machine_function): Add tbegin_p.
6011 (s390_canonicalize_comparison): Fold CC mode compares to
6012 conditional jump if possible.
6013 (s390_emit_jump): Return the emitted jump.
6014 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
6015 Handle CCRAWmode compares.
6016 (s390_option_override): Default to -mhtm if available.
6017 (s390_reg_clobbered_rtx): Handle floating point regs as well.
6018 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
6019 FPRs instead of df_regs_ever_live_p.
6020 (s390_optimize_nonescaping_tx): New function.
6021 (s390_init_frame_layout): Extend clobbered_regs array to cover
6023 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
6024 (s390_expand_tbegin): New function.
6025 (enum s390_builtin): New enum definition.
6026 (code_for_builtin): New array definition.
6027 (s390_init_builtins): New function.
6028 (s390_expand_builtin): New function.
6029 (TARGET_INIT_BUILTINS): Define.
6030 (TARGET_EXPAND_BUILTIN): Define.
6031 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
6032 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
6033 (s390_alc_comparison): Likewise.
6034 * config/s390/s390-modes.def: Add CCRAWmode.
6035 * config/s390/s390.h (processor_flags): Add PF_TX.
6036 (TARGET_CPU_HTM): Define macro.
6037 (TARGET_HTM): Define macro.
6038 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
6039 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
6040 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6041 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
6042 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
6043 (TBEGIN_MASK, TBEGINC_MASK): New constants.
6044 ("*cc_to_int"): Move up.
6045 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
6046 constants other than 0.
6047 ("*ccraw_to_int"): New insn and splitter definition.
6048 ("tbegin", "tbegin_nofloat", "tbegin_retry")
6049 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
6050 ("tx_assist"): New expander.
6051 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
6052 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
6053 * config/s390/s390.opt: Add -mhtm option.
6054 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
6055 * config/s390/htmxlintrin.h: New file.
6056 * config/s390/htmintrin.h: New file.
6057 * config/s390/s390intrin.h: New file.
6058 * doc/extend.texi: Document htm builtins.
6059 * config.gcc: Add the new header files to extra_headers.
6061 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
6063 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
6064 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
6066 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6067 Pat Haugen <pthaugen@us.ibm.com>
6068 Peter Bergner <bergner@vnet.ibm.com>
6070 * config/rs6000/power8.md: New.
6071 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
6072 setting for power8 entry.
6073 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
6074 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
6075 test for Power4/Power5 only.
6076 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
6078 (force_new_group): Adjust comment.
6079 * config/rs6000/rs6000.md: Include power8.md.
6081 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
6083 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
6084 * config/arm/arm-protos.h (arm_max_conditional_execute): New
6086 (tune_params): Update comment.
6087 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
6088 (arm_max_conditional_execute): New function.
6089 (thumb2_final_prescan_insn): Use max_insn_skipped and
6090 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
6092 2013-06-25 Jakub Jelinek <jakub@redhat.com>
6094 PR tree-optimization/57705
6095 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
6096 SSA_NAME step, provided that it is not defined inside the loop.
6097 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
6098 (get_initial_def_for_induction): Handle SSA_NAME IV step.
6100 2013-06-25 Martin Jambor <mjambor@suse.cz>
6103 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
6104 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
6106 (ipa_note_param_call): Initialize member_ptr flag.
6107 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
6108 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
6109 (ipa_write_indirect_edge_info): Stream member_ptr flag.
6110 (ipa_read_indirect_edge_info): Likewise.
6112 2013-06-25 Richard Biener <rguenther@suse.de>
6115 * passes.c (init_optimization_passes): Move pass_fold_builtins
6116 and pass_dce earlier with -Og.
6118 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
6120 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
6121 <BIT_FIELD_REF>: Remove trailing TAB.
6122 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
6125 2013-06-24 Martin Jambor <mjambor@suse.cz>
6127 PR tree-optimization/57358
6128 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
6129 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
6130 (ipa_analyze_params_uses): Generate pessimistic info when true.
6132 2013-06-24 Martin Jambor <mjambor@suse.cz>
6134 PR tree-optimization/57539
6135 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
6136 global.inlined_to of the new node to it. All callers changed.
6137 * ipa-inline-transform.c (clone_inlined_nodes): New variable
6138 inlining_into, pass it to cgraph_clone_node.
6139 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
6140 ipa_free_edge_args_substructures.
6141 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
6142 rdesc linked list. Do not assert rdesc edges have inlined caller.
6143 Assert we have found an rdesc in the rdesc list.
6145 2013-06-24 Richard Biener <rguenther@suse.de>
6147 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
6148 (pointer_set_lookup): Declare.
6149 (class pointer_map): New template class implementing a
6150 generic pointer to T map.
6151 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
6152 pointer_map<T>::contains, pointer_map<T>::insert,
6153 pointer_map<T>::traverse): New functions.
6154 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
6155 (pointer_set_lookup): New function.
6156 (pointer_set_contains): Use pointer_set_lookup.
6157 (pointer_set_insert): Likewise.
6158 (insert_aux): Remove.
6159 (struct pointer_map_t): Embed a pointer_set_t.
6160 (pointer_map_create): Adjust.
6161 (pointer_map_destroy): Likewise.
6162 (pointer_map_contains): Likewise.
6163 (pointer_map_insert): Likewise.
6164 (pointer_map_traverse): Likewise.
6165 * tree-streamer.h (struct streamer_tree_cache_d): Use a
6166 pointer_map<unsigned> instead of a pointer_map_t.
6167 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
6168 (streamer_tree_cache_lookup): Likewise.
6169 (streamer_tree_cache_create): Likewise.
6170 (streamer_tree_cache_delete): Likewise.
6171 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
6172 pointer_map<unsigned> instead of a pointer_map_t.
6173 (lto_init_tree_ref_encoder): Adjust.
6174 (lto_destroy_tree_ref_encoder): Likewise.
6175 * lto-section-out.c (lto_output_decl_index): Likewise.
6176 (lto_record_function_out_decl_state): Likewise.
6177 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
6179 2013-06-24 Richard Biener <rguenther@suse.de>
6181 PR tree-optimization/57488
6182 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
6184 2013-06-24 Alan Modra <amodra@gmail.com>
6186 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
6187 (gen_easy_altivec_constant): Likewise.
6188 * config/rs6000/predicates.md (easy_vector_constant_add_self,
6189 easy_vector_constant_msb): Likewise.
6191 2013-06-23 Jakub Jelinek <jakub@redhat.com>
6194 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
6195 add missing return true.
6197 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
6200 * config/sh/predicates.md (general_extend_operand): Invoke
6201 general_movsrc_operand for memory operands.
6202 (general_movsrc_operand): Allow reg+reg addressing, do not use
6203 general_operand for memory operands.
6205 2013-06-23 Sriraman Tallam <tmsriram@google.com>
6207 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
6208 when current target options does not apply.
6209 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
6210 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
6211 * config/i386/bmiintrin.h: Pass appropriate target
6212 attributes to header.
6213 * config/i386/mmintrin.h: Ditto.
6214 * config/i386/nmmintrin.h: Ditto.
6215 * config/i386/avx2intrin.h: Ditto.
6216 * config/i386/fxsrintrin.h: Ditto.
6217 * config/i386/tbmintrin.h: Ditto.
6218 * config/i386/xsaveintrin.h: Ditto.
6219 * config/i386/f16cintrin.h: Ditto.
6220 * config/i386/xtestintrin.h: Ditto.
6221 * config/i386/xsaveoptintrin.h: Ditto.
6222 * config/i386/bmi2intrin.h: Ditto.
6223 * config/i386/lzcntintrin.h: Ditto.
6224 * config/i386/smmintrin.h: Ditto.
6225 * config/i386/wmmintrin.h: Ditto.
6226 * config/i386/x86intrin.h: Remove all header include guards.
6227 * config/i386/prfchwintrin.h: Ditto.
6228 * config/i386/pmmintrin.h: Ditto.
6229 * config/i386/tmmintrin.h: Ditto.
6230 * config/i386/xmmintrin.h: Ditto.
6231 * config/i386/popcntintrin.h: Ditto.
6232 * config/i386/rdseedintrin.h: Ditto.
6233 * config/i386/ammintrin.h: Ditto.
6234 * config/i386/emmintrin.h: Ditto.
6235 * config/i386/immintrin.h: Remove all header include guards.
6236 * config/i386/fma4intrin.h: Ditto.
6237 * config/i386/lwpintrin.h: Ditto.
6238 * config/i386/xopintrin.h: Ditto.
6239 * config/i386/ia32intrin.h: Ditto.
6240 * config/i386/avxintrin.h: Ditto.
6241 * config/i386/rtmintrin.h: Ditto.
6242 * config/i386/fmaintrin.h: Ditto.
6243 * config/i386/mm3dnow.h: Ditto.
6245 2013-06-22 Sriraman Tallam <tmsriram@google.com>
6247 * common/config/i386/i386-common.c: Handle LZCNT.
6249 2013-06-22 Andi Kleen <ak@linux.intel.com>
6251 * doc/extend.texi: Use __atomic_store_n instead of
6252 __atomic_store in HLE example.
6254 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
6256 * config/sh/sh.c: Remove <cstdlib> workaround.
6258 2013-06-21 Andi Kleen <ak@linux.intel.com>
6260 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
6262 2013-06-21 Andi Kleen <ak@linux.intel.com>
6264 * doc/extend.texi: Document that __atomic_clear and
6265 __atomic_test_and_set should only be used with bool.
6267 2013-06-20 Jan Hubicka <jh@suse.cz>
6269 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
6271 * tree.c (decls_same_for_odr): New function.
6272 (same_for_edr): New function.
6273 (types_same_for_odr): New function.
6274 (get_binfo_at_offset): Use it.
6275 * tree.h (types_same_for_odr): Declare.
6277 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
6278 Jason Merrill <jason@redhat.com>
6280 * system.h: Include <cstdlib> as well as <stdlib.h>.
6282 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
6285 * config/i386/i386.c (construct_container): Report error if
6286 long double is used with disabled x87 float returns.
6288 2013-06-20 Jan Hubicka <jh@suse.cz>
6290 * lto-cgraph.c (input_symtab): Do not set cgraph state.
6292 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
6294 PR rtl-optimization/57425
6295 PR rtl-optimization/57569
6296 * alias.c (write_dependence_p): Remove parameters mem_mode and
6297 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
6298 Changed all callers.
6299 (canon_anti_dependence): Get comments and semantics in sync.
6300 Add parameter mem_canonicalized. Changed all callers.
6301 * rtl.h (canon_anti_dependence): Update prototype.
6303 2013-06-20 Richard Biener <rguenther@suse.de>
6305 * data-streamer-in.c (streamer_read_uhwi): Optimize single
6306 byte case, inline streamer_read_uchar and defer section
6309 2013-06-20 Richard Biener <rguenther@suse.de>
6311 PR tree-optimization/57584
6312 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
6313 SSA names into the expanded expression that take part in
6314 abnormal coalescing.
6316 2013-06-19 Sharad Singhai <singhai@google.com>
6318 * gcov.c (print_usage): Handle new option.
6319 (process_args): Ditto.
6320 (get_gcov_intermediate_filename): New function.
6321 (output_intermediate_file): New function.
6322 (output_gcov_file): New function
6323 (generate_results): Handle new option.
6324 (release_function): Relase demangled name.
6325 (read_graph_file): Handle demangled name.
6326 (output_lines): Ditto.
6327 * doc/gcov.texi: Document gcov intermediate format.
6329 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
6332 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6333 (lra_emit_add): Use the functions. Add comment about Y as an
6336 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
6339 * collect2.c (collect_atexit): New.
6340 (collect_exit): Delete.
6341 (main): Register collect_atexit with atexit.
6342 (collect_wait): Change collect_exit to exit.
6344 * collect2.h (collect_exit): Delete.
6345 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
6347 2013-06-19 Wei Mi <wmi@google.com>
6349 PR rtl-optimization/57518
6350 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
6351 if regno is used in paradoxical subreg.
6352 (update_equiv_regs): Check pdx_subregs[regno] before
6353 set a reg to be equivalent with a mem.
6355 2013-06-19 Matthias Klose <doko@ubuntu.com>
6358 * file-find.h (find_a_file): Add a mode parameter.
6359 * file-find.c (find_a_file): Likewise.
6360 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
6361 with X_OK for the executables.
6362 * collect2.c (main): Call find_a_file with X_OK.
6364 2013-06-19 Steve Ellcey <sellcey@mips.com>
6367 * config/mips/mips.md (casesi_internal_mips16_<mode>):
6368 Use NEXT_INSN instead of next_real_insn.
6370 2013-06-19 Jan Hubicka <jh@suse.cz>
6372 * cgraph.h (const_value_known_p): Replace by ...
6373 (ctor_for_folding): .. this one.
6374 * cgraphunit.c (process_function_and_variable_attributes): Use it.
6375 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
6376 * expr.c (expand_expr_real_1): Likewise.
6377 (string_constant): Likewise.
6378 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
6379 * ipa.c (process_references): Likewise.
6380 (symtab_remove_unreachable_nodes): Likewise.
6381 * ipa-inline-analysis.c (param_change_prob): Likewise.
6382 * gimple-fold.c (canonicalize_constructor_val): Likewise.
6383 (get_base_constructor): Likwise.
6384 * varpool.c (varpool_remove_node): Likewise.
6385 (varpool_remove_initializer): LIkewise.
6386 (dump_varpool_node): LIkwise.
6387 (const_value_known_p): Rewrite to ...
6388 (ctor_for_folding): ... this one.
6390 2013-06-19 Jakub Jelinek <jakub@redhat.com>
6393 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
6394 PERSONALITY in $PATH derived prefixes.
6396 2013-06-19 Jeff Law <law@redhat.com>
6398 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
6401 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6402 (simplify_bitwise_binary): Use it to simpify certain binary ops on
6405 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
6407 * config/arm/vfp.md: Move VFP instruction classification documentation
6409 * config/arm/arm.md: ... here. Update instruction classification
6412 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
6414 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6415 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6416 pattern. Use more efficient sequences on ARMv5 and Thumb2.
6418 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
6421 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6422 with NEXT_INSN. Use tablejump_p to check for jump table data
6425 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
6428 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6429 that now in C++ the value is correct per the C++ standards.
6431 2013-06-19 Richard Biener <rguenther@suse.de>
6433 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6436 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6439 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6442 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6443 with next_active_insn.
6445 2013-06-18 Sriraman Tallam <tmsriram@google.com>
6447 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6448 functions are inlined during failures to flag an error.
6449 * tree-inline.c (expand_call_inline): Allow the error to be flagged
6450 in early inline pass.
6452 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
6454 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6457 2013-06-18 Julian Brown <julian@codesourcery.com>
6459 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6460 Permit virtual register pre-reload if !strict.
6461 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6463 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6465 * config/arm/neon.md (movmisalign<mode>): Use
6466 neon_perm_struct_or_reg_operand instead of
6467 neon_struct_or_register_operand.
6468 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6469 neon_permissive_struct_operand instead of neon_struct_operand.
6470 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6471 neon_vector_mem_operand.
6472 * config/arm/predicates.md (neon_struct_operand): Adjust call to
6473 neon_vector_mem_operand.
6474 (neon_permissive_struct_operand): New.
6475 (neon_struct_or_register_operand): Rename to...
6476 (neon_perm_struct_or_reg_operand): This. Adjust call to
6477 neon_vector_mem_operand.
6479 2013-06-18 Richard Biener <rguenther@suse.de>
6481 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6482 * lto-streamer.h: Include pointer-set.h.
6483 (struct lto_decl_slot): Remove.
6484 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6485 Remove next_index entry.
6486 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6487 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6488 (lto_init_tree_ref_encoder): Adjust.
6489 (lto_destroy_tree_ref_encoder): Likewise.
6490 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6491 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6492 (lto_output_decl_index): Adjust.
6493 (lto_new_out_decl_state): Likewise.
6494 (lto_record_function_out_decl_state): Likewise.
6495 * lto-streamer-out.c (copy_function): Likewise.
6497 2013-06-18 Richard Biener <rguenther@suse.de>
6499 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6500 * cgraphunit.c: Include cfgloop.h.
6501 (init_lowered_empty_function): Initialize the loop tree.
6502 (assemble_thunk): Insert new BBs into loops.
6504 2013-06-18 Richard Biener <rguenther@suse.de>
6506 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6507 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6508 the map from cache entry to cache index optional.
6509 (streamer_tree_cache_replace_tree): Adjust accordingly.
6510 (streamer_tree_cache_append): Likewise.
6511 (streamer_tree_cache_delete): Likewise.
6512 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6513 streamer cache map from cache entry to cache index.
6514 * lto-streamer-out.c (create_output_block): Adjust.
6516 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
6518 * config/arm/arm.md (attribute "insn"): Move multiplication and
6519 division attributes to...
6520 (attribute "type"): ... here. Remove mult.
6521 (attribute "mul32"): New attribute.
6522 (attribute "mul64"): Add umaal.
6523 (*arm_mulsi3): Update attributes.
6524 (*arm_mulsi3_v6): Likewise.
6525 (*thumb_mulsi3): Likewise.
6526 (*thumb_mulsi3_v6): Likewise.
6527 (*mulsi3_compare0): Likewise.
6528 (*mulsi3_compare0_v6): Likewise.
6529 (*mulsi_compare0_scratch): Likewise.
6530 (*mulsi_compare0_scratch_v6): Likewise.
6531 (*mulsi3addsi): Likewise.
6532 (*mulsi3addsi_v6): Likewise.
6533 (*mulsi3addsi_compare0): Likewise.
6534 (*mulsi3addsi_compare0_v6): Likewise.
6535 (*mulsi3addsi_compare0_scratch): Likewise.
6536 (*mulsi3addsi_compare0_scratch_v6): Likewise.
6537 (*mulsi3subsi): Likewise.
6538 (*mulsidi3adddi): Likewise.
6539 (*mulsi3addsi_v6): Likewise.
6540 (*mulsidi3adddi_v6): Likewise.
6541 (*mulsidi3_nov6): Likewise.
6542 (*mulsidi3_v6): Likewise.
6543 (*umulsidi3_nov6): Likewise.
6544 (*umulsidi3_v6): Likewise.
6545 (*umulsidi3adddi): Likewise.
6546 (*umulsidi3adddi_v6): Likewise.
6547 (*smulsi3_highpart_nov6): Likewise.
6548 (*smulsi3_highpart_v6): Likewise.
6549 (*umulsi3_highpart_nov6): Likewise.
6550 (*umulsi3_highpart_v6): Likewise.
6551 (mulhisi3): Likewise.
6552 (*mulhisi3tb): Likewise.
6553 (*mulhisi3bt): Likewise.
6554 (*mulhisi3tt): Likewise.
6555 (maddhisi4): Likewise.
6556 (*maddhisi4tb): Likewise.
6557 (*maddhisi4tt): Likewise.
6558 (maddhidi4): Likewise.
6559 (*maddhidi4tb): Likewise.
6560 (*maddhidi4tt): Likewise.
6562 (udivsi3): Likewise.
6563 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6564 (thumb2_mulsi_short_compare0): Likewise.
6565 (thumb2_mulsi_short_compare0_scratch): Likewise.
6566 * config/arm/arm1020e.md (1020mult1): Update attribute change.
6567 (1020mult2): Likewise.
6568 (1020mult3): Likewise.
6569 (1020mult4): Likewise.
6570 (1020mult5): Likewise.
6571 (1020mult6): Likewise.
6572 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6574 (cortex_a15_mult64): Likewise.
6575 (cortex_a15_sdiv): Likewise.
6576 (cortex_a15_udiv): Likewise.
6577 * config/arm/arm1026ejs.md (mult1): Update attribute change.
6583 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6584 (pj4_ir_div): Likewise.
6585 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6586 (11_mult2): Likewise.
6587 (11_mult3): Likewise.
6588 (11_mult4): Likewise.
6589 (11_mult5): Likewise.
6590 (11_mult6): Likewise.
6591 (11_mult7): Likewise.
6592 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6593 (cortex_a8_mla): Likewise.
6594 (cortex_a8_mull): Likewise.
6595 (cortex_a8_smulwy): Likewise.
6596 (cortex_a8_smlald): Likewise.
6597 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6598 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6599 (cortex_r4_mul_3): Likewise.
6600 (cortex_r4_mla_4): Likewise.
6601 (cortex_r4_mla_3): Likewise.
6602 (cortex_r4_smlald): Likewise.
6603 (cortex_r4_mull): Likewise.
6604 (cortex_r4_sdiv): Likewise.
6605 (cortex_r4_udiv): Likewise.
6606 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6607 (cortex_a7_idiv): Likewise.
6608 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6609 (9_mult2): Likewise.
6610 (9_mult3): Likewise.
6611 (9_mult4): Likewise.
6612 (9_mult5): Likewise.
6613 (9_mult6): Likewise.
6614 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6615 (cortex_a53_sdiv): Likewise.
6616 (cortex_a53_udiv): Likewise.
6617 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6618 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6619 (mp626_mult2): Likewise.
6620 (mp626_mult3): Likewise.
6621 (mp626_mult4): Likewise.
6622 * config/arm/fa526.md (526_mult1): Update attribute change.
6623 (526_mult2): Likewise.
6624 * config/arm/arm-generic.md (mult): Update attribute change.
6625 (mult_ldsched_strongarm): Likewise.
6626 (mult_ldsched): Likewise.
6627 (multi_cycle): Likewise.
6628 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6629 * config/arm/fa606te.md (606te_mult1): Update attribute change.
6630 (606te_mult2): Likewise.
6631 (606te_mult3): Likewise.
6632 (606te_mult4): Likewise.
6633 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6634 (cortex_a9_mac16): Likewise.
6635 (cortex_a9_multiply): Likewise.
6636 (cortex_a9_mac): Likewise.
6637 (cortex_a9_multiply_long): Likewise.
6638 * config/arm/fa626te.md (626te_mult1): Update attribute change.
6639 (626te_mult2): Likewise.
6640 (626te_mult3): Likewise.
6641 (626te_mult4): Likewise.
6643 2013-06-18 Richard Biener <rguenther@suse.de>
6646 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6648 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6651 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6652 with next_active_insn.
6654 2013-06-18 Alan Modra <amodra@gmail.com>
6656 * config/rs6000/rs6000.h (enum data_align): New.
6657 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6658 (DATA_ABI_ALIGNMENT): Define.
6659 (CONSTANT_ALIGNMENT): Correct comment.
6660 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6661 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6663 2013-06-17 David Malcolm <dmalcolm@redhat.com>
6665 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6666 ATTRIBUTE_UNUSED marking.
6668 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
6670 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6671 alternative and update.
6672 (aarch64_dup_lanedi): Delete.
6673 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6674 * config/aarch64/aarch64-simd-builtins.def: Update.
6676 2013-06-17 Richard Biener <rguenther@suse.de>
6678 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6679 (lto_input_scc): Declare.
6680 (lto_input_tree_1): Likewise.
6681 (struct lto_stats_d): Add num_tree_bodies_output and
6682 num_pickle_refs_output.
6683 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6684 (lto_read_tree_1): Split out from ...
6685 (lto_read_tree): ... this.
6686 (lto_input_scc): New function.
6687 (lto_input_tree_1): Split out from ...
6688 (lto_input_tree): ... this. Handle LTO_tree_scc.
6689 (lto_data_in_create): Create the streamer cache without hashes.
6690 * lto-streamer-out.c (create_output_block): Create the streamer
6691 cache with hashes when not doing WPA.
6692 (lto_write_tree_1): Split out from ...
6693 (lto_write_tree): ... this.
6694 (get_symbol_initial_value): New function.
6695 (lto_output_tree_1): Split out from ...
6696 (lto_output_tree): ... this. Write trees as series of SCCs
6697 using a DFS walk via DFS_write_tree.
6698 (struct sccs, struct scc_entry): New types.
6699 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6700 (DFS_write_tree_body): New function.
6701 (DFS_write_tree): Likewise.
6702 (hash_tree): Likewise.
6703 (scc_entry_compare): Likewise.
6704 (hash_scc): Likewise.
6705 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6706 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6707 TREE_CHAIN as regular reference.
6708 (streamer_read_integer_cst): Remove.
6709 (streamer_get_pickled_tree): Adjust.
6710 * tree-streamer-out.c (streamer_write_chain): Disable streaming
6711 of DECL_EXTERNALs in BLOCK_VARS for now.
6712 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6714 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6715 Add hash value argument and record that if hashes are recorded
6717 (streamer_tree_cache_insert_1): Adjust.
6718 (streamer_tree_cache_insert): Likewise.
6719 (streamer_tree_cache_insert_at): Rename to ...
6720 (streamer_tree_cache_replace_tree): ... this and adjust.
6721 (streamer_tree_cache_append): Adjust.
6722 (record_common_node): Likewise.
6723 (streamer_tree_cache_create): Add argument whether to
6724 record hash values together with trees.
6725 (streamer_tree_cache_delete): Adjust.
6726 * tree-streamer.h (struct streamer_tree_cache_d): Add
6728 (streamer_read_integer_cst): Remove.
6729 (streamer_tree_cache_insert): Adjust.
6730 (streamer_tree_cache_append): Likewise.
6731 (streamer_tree_cache_insert_at): Rename to ...
6732 (streamer_tree_cache_replace_tree): ... this and adjust.
6733 (streamer_tree_cache_create): Add argument whether to record hashes.
6734 (streamer_tree_cache_get): Rename to ...
6735 (streamer_tree_cache_get_tree): ... this.
6736 (streamer_tree_cache_get_hash): New function.
6737 * tree.c (cache_integer_cst): New function.
6738 * tree.h (cache_integer_cst): Declare.
6739 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6740 * lto-symtab.c (lto_varpool_replace_node): Only release
6741 DECL_INITIAL of non-prevailing decls.
6742 * varpool.c (varpool_remove_initializer): Do not release
6743 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6745 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
6747 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6748 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6749 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6750 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6751 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6752 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6753 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6754 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6755 instead of TARGET_64BIT.
6756 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6757 Require ISA_HAS_<D>DIV.
6759 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
6761 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6762 (mips*-*-linux*): Move default with_llsc setting to where other
6764 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6766 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6767 Likewise. Remove default with_tune setting. Move default float
6768 setting to its own block. Handle with_llsc in the same block as above.
6770 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
6772 PR rtl-optimization/57425
6773 PR rtl-optimization/57569
6774 * alias.c (write_dependence_p): Add new parameters mem_mode,
6775 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
6776 Changed all callers.
6777 (canon_anti_dependence): New function.
6778 * cse.c (check_dependence): Use canon_anti_dependence.
6779 * cselib.c (cselib_invalidate_mem): Likewise.
6780 * rtl.h (canon_anti_dependence): Declare.
6782 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
6784 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6785 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6786 ".set mips3" for 64-bit targets.
6788 2013-06-15 Dehao Chen <dehao@google.com>
6790 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6791 * gimple-low.c (gimple_check_call_matching_types): Likewise.
6792 (gimple_check_call_args): Likewise.
6793 * value-prof.c (check_ic_target): Likewise.
6794 * ipa-inline.c (early_inliner): Likewise.
6795 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6796 * cgraph.c (cgraph_create_edge_1): Likewise.
6797 (cgraph_make_edge_direct): Likewise.
6799 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6802 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6803 rs6000_output_move_128bit to handle emitting quad memory
6804 operations. Set attribute length to 8 bytes.
6806 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
6808 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6810 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6811 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6812 (aarch64_<su>mlsl<mode>): Likewise.
6814 2013-06-14 Mike Stump <mikestump@comcast.net>
6816 * Makefile.in (TARGET_H): Add insn-codes.h.
6818 2013-06-14 Alan Modra <amodra@gmail.com>
6822 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6823 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
6824 bitfield expansion when EXPAND_MEMORY.
6825 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6827 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6829 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6830 test for clearing quad memory on 32-bit later.
6832 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
6834 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6835 (fold_negate_expr): Likewise.
6836 (fold_real_zero_addition_p): Handle vectors.
6837 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6839 2013-06-14 Alan Modra <amodra@gmail.com>
6841 * varasm.c (force_const_mem): Revert 2013-06-07 change.
6843 2013-06-13 Jan Hubicka <jh@suse.cz>
6845 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6846 Local comdats are not externally visible.
6847 * symtab.c (dump_symtab_base): Dump externally visible.
6848 (verify_symtab_base): Verify back links in the symtab hash.
6850 2013-06-13 Bin Cheng <bin.cheng@arm.com>
6852 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6853 CONVERT_EXPR as equal nodes.
6855 2013-06-13 Bin Cheng <bin.cheng@arm.com>
6857 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6859 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
6861 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6862 Generalize to complex and vector.
6863 * tree.c (build_all_ones_cst): New function.
6864 * tree.h (build_all_ones_cst): Declare it.
6866 2013-06-13 Alan Modra <amodra@gmail.com>
6868 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6869 * config/rs6000/rs6000.md (signbittf2): New insn.
6870 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6871 (abstf2_internal, cmptf_internal2): Likewise.
6872 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6874 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6875 Pat Haugen <pthaugen@us.ibm.com>
6876 Peter Bergner <bergner@vnet.ibm.com>
6878 * config/rs6000/rs6000.c (emit_load_locked): Add support for
6879 power8 byte, half-word, and quad-word atomic instructions.
6880 (emit_store_conditional): Likewise.
6881 (rs6000_expand_atomic_compare_and_swap): Likewise.
6882 (rs6000_expand_atomic_op): Likewise.
6884 * config/rs6000/sync.md (larx): Add new modes for power8.
6886 (AINT): New mode iterator to include TImode as well as normal
6887 integer modes on power8.
6888 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6889 that VSX registers are not considered. Use AINT mode iterator
6890 instead of INT1 to allow inclusion of quad word atomic operations
6892 (load_locked<mode>): Likewise.
6893 (store_conditional<mode>): Likewise.
6894 (atomic_compare_and_swap<mode>): Likewise.
6895 (atomic_exchange<mode>): Likewise.
6896 (atomic_nand<mode>): Likewise.
6897 (atomic_fetch_<fetchop_name><mode>): Likewise.
6898 (atomic_nand_fetch<mode>): Likewise.
6899 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6901 (ATOMIC): On power8, add QImode, HImode modes.
6902 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6903 modes that promote to SImode.
6904 (load_lockedti): Convert TImode arguments to PTImode, so that we
6905 get a guaranteed even/odd register pair.
6906 (load_lockedpti): Likewise.
6907 (store_conditionalti): Likewise.
6908 (store_conditionalpti): Likewise.
6910 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6911 atomic load/store instructions.
6914 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
6916 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6918 (insn_count): New attribute, with most cases extracted from...
6919 (length): ...here. Redefine most cases in terms of insn_count.
6920 (single_insn): Delete.
6921 (can_delay): Use insn_count to check for single instructions.
6922 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6923 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6924 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6925 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6926 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6927 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6928 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6929 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6930 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6931 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6932 rather than "length".
6933 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
6934 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6935 Use "insn_count" rather than "length".
6936 * config/mips/mips-dsp.md
6937 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6938 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6941 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
6943 PR tree-optimization/57361
6944 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6946 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
6948 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6950 (aarch64_simd_combine<mode>): New instruction expansion.
6951 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6953 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6954 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6956 2013-06-12 Jan Hubicka <jh@suse.cz>
6958 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6959 decl has when in streaming stage.
6960 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6961 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6963 2013-06-12 Roland Stigge <stigge@antcom.de>
6966 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6968 2013-06-12 Jakub Jelinek <jakub@redhat.com>
6970 PR tree-optimization/57537
6971 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6972 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6974 2013-06-12 Richard Biener <rguenther@suse.de>
6976 * data-streamer.h (streamer_write_char_stream): CSE
6977 obs->current_pointer.
6978 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6979 streamer_write_char_stream manually and optimize the resulting loop.
6980 (streamer_write_hwi_stream): Likewise.
6982 2013-06-12 Jan Hubicka <jh@suse.cz>
6984 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6985 * cgraph.h (varpool_create_empty_node): Declare.
6986 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6988 * symtab.c (symtab_unregister_node): Be lax about missin entries
6990 (symtab_get_node): Update comment.
6991 * varpool.c (varpool_create_empty_node): Break out from ...
6992 (varpool_node_for_decl): ... here.
6993 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6995 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
6997 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6998 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6999 part. Use straight-line flow at the end.
7000 <COMPONENT_REF>: Remove superfluous else.
7001 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
7003 2013-06-12 Jakub Jelinek <jakub@redhat.com>
7006 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
7007 target hook even for !TREE_PUBLIC decls. If no resolution info
7008 is available, return false for common and external decls.
7010 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
7012 * config/rl78/constraints.md (U): New constraint.
7013 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
7016 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
7019 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
7020 to allow returning address to AT_PLATFORM name.
7022 2013-06-11 Jan Hubicka <jh@suse.cz>
7024 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
7025 * cgraph.h (symtab_node_base): Add weakref flag.
7026 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
7027 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
7028 (output_weakrefs): Use weakref flag.
7029 * fold-const.c (simple_operand_p): Handle WEAK.
7030 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
7031 * ipa.c (varpool_externally_visible_p): Drop weakref.
7032 (function_and_variable_visibility): Update comment; fix weakref
7033 sanity checks; do not clear DECL_WEAK on them.
7034 * lto-cgraph.c (lto_output_node): update.
7035 (lto_output_varpool_node): Update.
7036 (input_overwrite_node): Update.
7037 (input_node): Update.
7038 (input_varpool_node): Update.
7039 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
7040 (lto_symtab_merge_symbols): Add sanity check.
7041 (lto_symtab_prevailing_decl): Do not special case weakrefs.
7042 * passes.c (rest_of_decl_compilation): Set static flag, too.
7043 * symtab.c (dump_symtab_base): Dump weakref.
7044 (verify_symtab_base): Sanity check weakrefs.
7045 (symtab_make_decl_local): Remove duplicated code.
7046 (symtab_alias_ultimate_target): Simplify.
7047 * varpool.c (varpool_create_variable_alias): Set weakref flag.
7049 2013-06-11 DJ Delorie <dj@redhat.com>
7051 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
7052 (rl78_unwind_word_mode): New.
7054 2013-06-11 David Malcolm <dmalcolm@redhat.com>
7056 * final.c (debug_prefix_maps): Make static.
7058 2013-06-11 David Malcolm <dmalcolm@redhat.com>
7060 * function.c (initial_trampoline): Remove stray copy.
7062 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
7064 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
7066 2013-06-11 Martin Jambor <mjambor@suse.cz>
7068 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
7069 within bounds at the beginning of the function.
7071 2013-06-11 Alan Modra <amodra@gmail.com>
7073 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
7075 (get_named_section): Don't NULL !DECL_P decl.
7077 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
7079 * doc/invoke.texi (core-avx2): Document.
7081 (atom): Updated with MOVBE.
7083 2013-06-11 Richard Biener <rguenther@suse.de>
7085 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
7087 2013-06-11 Anton Blanchard <anton@samba.org>
7089 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
7090 correct shift value in little-endian mode.
7092 2013-06-11 Jakub Jelinek <jakub@redhat.com>
7095 * varasm.c (get_variable_align): Move #endif to the right place.
7097 2013-06-10 Cary Coutant <ccoutant@google.com>
7099 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
7100 for hash so that hash table traversal order is deterministic.
7102 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7103 Pat Haugen <pthaugen@us.ibm.com>
7104 Peter Bergner <bergner@vnet.ibm.com>
7106 * config/rs6000/vector.md (GPR move splitter): Do not split moves
7107 of vectors in GPRS if they are direct moves or quad word load or
7110 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
7112 (direct_move_p): Likewise.
7113 (quad_load_store_p): Likewise.
7115 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
7116 classes into bins based on the physical register type.
7117 (reg_class_to_reg_type): Likewise.
7118 (IS_STD_REG_TYPE): Likewise.
7119 (IS_FP_VECT_REG_TYPE): Likewise.
7120 (reload_fpr_gpr): Arrays to determine what insn to use if we can
7121 use direct move instructions.
7122 (reload_gpr_vsx): Likewise.
7123 (reload_vsx_gpr): Likewise.
7124 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
7125 information that is a simplification of register classes. Also
7126 precalculate direct move reload helpers.
7127 (direct_move_p): New function to return true if the operation can
7128 be done as a direct move instruciton.
7129 (quad_load_store_p): New function to return true if the operation
7130 is a quad memory operation.
7131 (rs6000_legitimize_address): If quad memory, only allow register
7132 indirect for TImode addresses.
7133 (rs6000_legitimate_address_p): Likewise.
7134 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
7135 (rs6000_reload_register_type): Likewise.
7136 (register_to_reg_type): Return register type.
7137 (rs6000_secondary_reload_simple_move): New helper function for
7138 secondary reload and secondary memory needed to identify anything
7139 that is a simple move, and does not need reloading.
7140 (rs6000_secondary_reload_direct_move): New helper function for
7141 secondary reload to identify cases that can be done with several
7142 instructions via the direct move instructions.
7143 (rs6000_secondary_reload_move): New helper function for secondary
7144 reload to identify moves between register types that can be done.
7145 (rs6000_secondary_reload): Add support for quad memory operations
7146 and for direct move.
7147 (rs6000_secondary_memory_needed): Likewise.
7148 (rs6000_debug_secondary_memory_needed): Change argument names.
7149 (rs6000_output_move_128bit): New function to return the move to
7150 use for 128-bit moves, including knowing about the various
7151 limitations of quad memory operations.
7153 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
7154 memory operations. call rs6000_output_move_128bit for the actual
7155 instruciton(s) to generate.
7156 (vsx_movti_64bit): Likewise.
7158 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
7159 (UNSPEC_P8V_MTVSRWZ): Likewise.
7160 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
7161 (UNSPEC_P8V_MTVSRD): Likewise.
7162 (UNSPEC_P8V_XXPERMDI): Likewise.
7163 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
7164 (UNSPEC_FUSION_GPR): Likewise.
7165 (FMOVE128_GPR): New iterator for direct move.
7166 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
7169 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
7170 loads and direct move instructions.
7171 (zero_extendsidi2_lfiwzx): Likewise.
7172 (extendsidi2_lfiwax): Likewise.
7173 (extendsidi2_nocell): Likewise.
7174 (floatsi<mode>2_lfiwax): Likewise.
7176 (floatunssi<mode>2_lfiwzx): Likewise.
7178 (fix_trunc<mode>_stfiwx): Likewise.
7179 (fixuns_trunc<mode>_stfiwx): Likewise.
7180 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
7181 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
7182 (parity<mode>2_cmpb): Set length/type attr.
7183 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
7184 for 'mr.' to fast_compare.
7185 (bpermd_<mode>): Change type attr to popcnt.
7186 (p8_fmrgow_<mode>): New insns for power8 direct move support.
7187 (p8_mtvsrwz_1): Likewise.
7188 (p8_mtvsrwz_2): Likewise.
7189 (reload_fpr_from_gpr<mode>): Likewise.
7190 (p8_mtvsrd_1): Likewise.
7191 (p8_mtvsrd_2): Likewise.
7192 (p8_xxpermdi_<mode>): Likewise.
7193 (reload_vsx_from_gpr<mode>): Likewise.
7194 (reload_vsx_from_gprsf): Likewise.
7195 (p8_mfvsrd_3_<mode>): LIkewise.
7196 (reload_gpr_from_vsx<mode>): Likewise.
7197 (reload_gpr_from_vsxsf): Likewise.
7198 (p8_mfvsrd_4_disf): Likewise.
7199 (multi-word GPR splits): Do not split direct moves or quad memory
7202 2013-06-10 David Malcolm <dmalcolm@redhat.com>
7204 * tree-into-ssa.c (interesting_blocks): Make static.
7206 2013-06-10 Jakub Jelinek <jakub@redhat.com>
7209 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
7210 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
7211 Use DATA_ABI_ALIGNMENT for that case instead if defined.
7212 (get_variable_align): New function.
7213 (get_variable_section, emit_bss, emit_common,
7214 assemble_variable_contents, place_block_symbol): Use
7215 get_variable_align instead of DECL_ALIGN.
7216 (assemble_noswitch_variable): Add align argument, use it
7217 instead of DECL_ALIGN.
7218 (assemble_variable): Adjust caller. Use get_variable_align
7219 instead of DECL_ALIGN.
7220 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
7222 (DATA_ABI_ALIGNMENT): Define.
7223 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
7224 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
7225 opt is false, only return the psABI mandated alignment increase.
7226 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
7227 (DATA_ABI_ALIGNMENT): ... this.
7228 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
7229 (DATA_ABI_ALIGNMENT): ... this.
7230 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
7231 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
7232 (DATA_ABI_ALIGNMENT): ... this.
7233 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
7234 * doc/tm.texi: Regenerated.
7236 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
7238 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
7239 cmp_code to construct REG_EQUAL note.
7241 2013-06-09 Jakub Jelinek <jakub@redhat.com>
7244 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
7245 that operands[2] doesn't overlap with operands[0].
7247 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
7248 Jan Hubicka <jh@suse.cz>
7250 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
7251 hack to mark symbols as used.
7253 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
7255 PR rtl-optimization/57559
7256 * lra-constraints.c (process_alt_operands): Don't discourage
7257 memory with known offset for offsetable memory constraint.
7258 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
7260 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
7262 * varasm.c (struct oc_local_state): Reorder fields.
7263 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
7264 and adjust accordingly.
7265 (output_constructor): Reorder initialization code and adjust call to
7266 output_constructor_bitfield.
7268 2013-06-07 Jan Hubicka <jh@suse.cz>
7270 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
7272 2013-06-07 David Malcolm <dmalcolm@redhat.com>
7274 * tree-object-size.c (unknown): Make const.
7276 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
7278 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
7279 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
7280 for last alternative in the cpu_facility attribute.
7282 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7285 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
7286 (xordi3): Change operand 2 constraint to arm_xordi_operand.
7287 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
7288 * config/arm/constraints.md (Dg): New constraint.
7289 * config/arm/neon.md (xordi3_neon): Remove.
7290 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
7291 * config/arm/predicates.md (arm_xordi_operand): New predicate.
7293 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7295 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
7296 Clean up alternatives.
7298 2013-06-07 Alan Modra <amodra@gmail.com>
7300 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
7303 2013-06-07 Alan Modra <amodra@gmail.com>
7305 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
7307 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7309 * config/arm/constraints.md (Df): New constraint.
7310 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
7311 Correct length attribute for last two alternatives.
7313 2013-06-07 Alan Modra <amodra@gmail.com>
7315 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7316 override user -mfp-in-toc.
7317 (offsettable_ok_by_alignment): Consider just the current access
7318 rather than the whole object, unless BLKmode. Handle
7319 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
7320 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
7321 for -mcmodel=medium.
7322 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
7323 override user -mfp-in-toc or -msum-in-toc. Default to
7324 -mno-fp-in-toc for -mcmodel=medium.
7326 2013-06-06 DJ Delorie <dj@redhat.com>
7328 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
7329 TARGET_VALID_POINTER_MODE.
7331 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7332 Pat Haugen <pthaugen@us.ibm.com>
7333 Peter Bergner <bergner@vnet.ibm.com>
7335 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7336 Document new power8 builtins.
7338 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
7339 condition code register, to allow 128-bit logical operations to be
7340 done in the VSX or GPR registers.
7341 (nor<mode>3): Use the canonical form for nor.
7342 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
7343 vclz*, and vpopcnt* vector instructions.
7344 (nand<mode>3): Likewise.
7345 (orc<mode>3): Likewise.
7346 (clz<mode>2): LIkewise.
7347 (popcount<mode>2): Likewise.
7349 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
7350 that only the GPRs are recognized.
7352 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7353 support for new power8 builtins.
7355 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
7357 (xscvdpspn): Likewise.
7363 (vpopcnt): Likewise.
7364 (vpopcntb): Likewise.
7365 (vpopcnth): Likewise.
7366 (vpopcntw): Likewise.
7367 (vpopcntd): Likewise.
7372 (eqv_v16qi3): Likewise.
7373 (eqv_v8hi3): Likewise.
7374 (eqv_v4si3): Likewise.
7375 (eqv_v2di3): Likewise.
7376 (eqv_v4sf3): Likewise.
7377 (eqv_v2df3): Likewise.
7379 (nand_v16qi3): Likewise.
7380 (nand_v8hi3): Likewise.
7381 (nand_v4si3): Likewise.
7382 (nand_v2di3): Likewise.
7383 (nand_v4sf3): Likewise.
7384 (nand_v2df3): Likewise.
7386 (orc_v16qi3): Likewise.
7387 (orc_v8hi3): Likewise.
7388 (orc_v4si3): Likewise.
7389 (orc_v2di3): Likewise.
7390 (orc_v4sf3): Likewise.
7391 (orc_v2df3): Likewise.
7393 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
7394 allow power8 quad mode in 64-bit.
7395 (rs6000_builtin_vectorized_function): Add support to vectorize
7396 ISA 2.07 count leading zeros, population count builtins.
7397 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
7398 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
7399 (builtin_function_type): Add vgbbd builtin function which takes an
7401 (altivec_expand_vec_perm_const): Add support for new power8 merge
7404 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7405 that does not include TImdoe for use with 32-bit.
7406 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7408 (UNSPEC_VSX_CVDPSPN): Likewise.
7409 (vsx_xscvdpspn): Likewise.
7410 (vsx_xscvspdpn): Likewise.
7411 (vsx_xscvdpspn_scalar): Likewise.
7412 (vsx_xscvspdpn_directmove): Likewise.
7413 (vsx_and<mode>3): Split logical operations into 32-bit and
7414 64-bit. Add support to do logical operations on TImode as well as
7415 VSX vector types. Allow logical operations to be done in either
7416 VSX registers or in general purpose registers in 64-bit mode. Add
7417 splitters if GPRs were used. For AND, add clobber of CCmode to
7418 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
7420 (vsx_and<mode>3_32bit): Likewise.
7421 (vsx_and<mode>3_64bit): Likewise.
7422 (vsx_ior<mode>3): Likewise.
7423 (vsx_ior<mode>3_32bit): Likewise.
7424 (vsx_ior<mode>3_64bit): Likewise.
7425 (vsx_xor<mode>3): Likewise.
7426 (vsx_xor<mode>3_32bit): Likewise.
7427 (vsx_xor<mode>3_64bit): Likewise.
7428 (vsx_one_cmpl<mode>2): Likewise.
7429 (vsx_one_cmpl<mode>2_32bit): Likewise.
7430 (vsx_one_cmpl<mode>2_64bit): Likewise.
7431 (vsx_nor<mode>3): Likewise.
7432 (vsx_nor<mode>3_32bit): Likewise.
7433 (vsx_nor<mode>3_64bit): Likewise.
7434 (vsx_andc<mode>3): Likewise.
7435 (vsx_andc<mode>3_32bit): Likewise.
7436 (vsx_andc<mode>3_64bit): Likewise.
7437 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7438 and xxlorc instructions.
7439 (vsx_eqv<mode>3_64bit): Likewise.
7440 (vsx_nand<mode>3_32bit): Likewise.
7441 (vsx_nand<mode>3_64bit): Likewise.
7442 (vsx_orc<mode>3_32bit): Likewise.
7443 (vsx_orc<mode>3_64bit): Likewise.
7445 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7447 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7449 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7450 (p8_vmrgow): Likewise.
7451 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7452 GPRs to be split under VSX.
7453 (p8v_clz<mode>2): Add power8 count leading zero support.
7454 (p8v_popcount<mode>2): Add power8 population count support.
7455 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7458 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7461 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7463 (vec_nand): Likewise.
7464 (vec_vclz): Likewise.
7465 (vec_vclzb): Likewise.
7466 (vec_vclzd): Likewise.
7467 (vec_vclzh): Likewise.
7468 (vec_vclzw): Likewise.
7469 (vec_vgbbd): Likewise.
7470 (vec_vmrgew): Likewise.
7471 (vec_vmrgow): Likewise.
7472 (vec_vpopcnt): Likewise.
7473 (vec_vpopcntb): Likewise.
7474 (vec_vpopcntd): Likewise.
7475 (vec_vpopcnth): Likewise.
7476 (vec_vpopcntw): Likewise.
7478 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7480 PR rtl-optimization/57468
7481 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7484 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7486 PR rtl-optimization/57459
7487 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7488 type when setting live regs.
7490 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7492 * config/s390/s390.opt (mlra): New option.
7493 * config/s390/s390.c (s390_decompose_address): Check displacement
7494 for all registers for LRA.
7495 (s390_secondary_reload): Don't used secondary reloads for LRA.
7496 (s390_lra_p): New function.
7497 (TARGET_LRA_P): Define.
7498 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7499 of attribute cpu_facility to zarch for the last alternative.
7500 (*cmpmem_short): Ditto.
7502 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
7504 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7505 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
7506 (arm_expand_prologue): Likewise.
7508 2013-06-06 Teresa Johnson <tejohnson@google.com>
7511 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7512 as this is now done by redirect_edge_and_branch_force.
7513 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7514 barriers, and fix interaction with splitting.
7515 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7516 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7517 reflect changes made in the routine.
7518 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7519 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7520 since this is called in cfglayout mode, and replace partition fixup
7521 with assert as that is now done by force_nonfallthru_and_redirect.
7522 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7523 already be marked with region crossing note.
7524 (insert_section_boundary_note): Make non-static, gate on flag
7525 has_bb_partition, rewrite to also check for multiple partitions.
7526 (rest_of_handle_reorder_blocks): Remove call to
7527 insert_section_boundary_note, now done later during free_cfg.
7528 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7529 * bb-reorder.h (insert_section_boundary_note): Declare.
7530 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7531 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7532 invoke insert_section_boundary_note.
7533 (try_redirect_by_replacing_jump): Remove unnecessary
7534 check for region crossing note.
7535 (fixup_partition_crossing): New function.
7536 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7537 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7538 in non-cfglayout mode.
7539 (force_nonfallthru_and_redirect): Fixup partition boundaries,
7540 remove old code that tried to do this. Emit barrier correctly
7541 when we are in cfglayout mode.
7542 (last_bb_in_partition): New function.
7543 (rtl_split_edge): Correctly fixup partition boundaries.
7544 (commit_one_edge_insertion): Remove old code that tried to
7545 fixup region crossing edge since this is now handled in
7546 split_block, and set up insertion point correctly since
7547 block may now end in a jump.
7548 (verify_hot_cold_block_grouping): Guard against checking when not in
7549 linearized RTL mode.
7550 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7552 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7553 rtl_verify_flow_info, so not called in cfglayout mode.
7554 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7555 (fixup_reorder_chain): Remove old code that attempted to fixup region
7556 crossing note as this is now handled in force_nonfallthru_and_redirect.
7557 (duplicate_insn_chain): Don't duplicate switch section notes.
7558 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7560 * basic-block.h (emit_barrier_after_bb): Declare.
7562 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7564 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7565 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7566 arm_usatsihi): Adjust alternatives for arm_restrict_it.
7568 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7570 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7572 * config/arm/ldmstm.md: Regenerate.
7574 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7576 * config/arm/sync.md (atomic_loaddi_1):
7577 Disable predication for arm_restrict_it.
7578 (arm_load_exclusive<mode>): Likewise.
7579 (arm_load_exclusivesi): Likewise.
7580 (arm_load_exclusivedi): Likewise.
7581 (arm_load_acquire_exclusive<mode>): Likewise.
7582 (arm_load_acquire_exclusivesi): Likewise.
7583 (arm_load_acquire_exclusivedi): Likewise.
7584 (arm_store_exclusive<mode>): Likewise.
7585 (arm_store_exclusive<mode>): Likewise.
7586 (arm_store_release_exclusivedi): Likewise.
7587 (arm_store_release_exclusive<mode>): Likewise.
7589 2013-06-06 Richard Biener <rguenther@suse.de>
7591 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7593 (lto_tag_is_tree_code_p): Adjust.
7594 (lto_tag_is_gimple_code_p): Likewise.
7595 (lto_gimple_code_to_tag): Likewise.
7596 (lto_tag_to_gimple_code): Likewise.
7597 (lto_tree_code_to_tag): Likewise.
7598 (lto_tag_to_tree_code): Likewise.
7599 * data-streamer.h (streamer_write_hwi_in_range): Use
7600 uhwi streaming to stream the normalized range.
7601 (streamer_read_hwi_in_range): Likewise.
7603 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7605 * config/arm/arm.md (enabled_for_depr_it): New attribute.
7606 (predicable_short_it): Likewise.
7607 (predicated): Likewise.
7608 (enabled): Handle above.
7609 (define_cond_exec): Set predicated attribute to yes.
7611 2013-06-05 Mike Stump <mikestump@comcast.net>
7613 * gdbinit.in (__FUNCTION__): Add.
7615 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
7617 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7618 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7620 2013-06-05 Jan Hubicka <jh@suse.cz>
7622 * varasm.c (mark_decl_referenced): Revert the removal until targets
7625 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
7627 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7628 instead of mark_decl_referenced.
7630 2013-06-05 Jan Hubicka <jh@suse.cz>
7632 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7633 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7634 and symtab_used_from_object_file_p.
7635 (cgraph_make_node_local_1): Clear forced_by_abi.
7636 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7637 * cgraph.h (symtab_node_base): Add forced_by_abi.
7638 (decide_is_variable_needed): Remove.
7639 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7640 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7641 (decide_is_symbol_needed): ... this one; handle symbols in general;
7642 always analyze virtuals; honnor forced_by_abi.
7643 (cgraph_finalize_function): Update.
7644 (varpool_finalize_decl): Update.
7645 (symbol_defined_and_needed): Remove.
7646 (analyze_functions): Update.
7647 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7648 output_refs, input_overwrite_node): Handle forced_by_abi.
7649 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7650 (address_taken_from_non_vtable_p): ... this one.
7651 (comdat_can_be_unshared_p_1): New function.
7652 (cgraph_comdat_can_be_unshared_p): Rename to ...
7653 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7654 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7655 (function_and_variable_visibility): Clear forced_by_abi as needed.
7656 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7657 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7658 * symtab.c (dump_symtab_base): Dump forced_by_abi.
7659 * varpool.c (decide_is_variable_needed): Remove.
7661 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7663 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7664 (arm_option_override): Override arm_restrict_it where appropriate.
7665 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7666 * config/arm/arm.opt (mrestrict-it): New command-line option.
7667 * doc/invoke.texi: Document -mrestrict-it.
7669 2013-06-05 David Malcolm <dmalcolm@redhat.com>
7671 * tsan.c (tsan_atomic_table): Make const.
7673 2013-06-05 Richard Biener <rguenther@suse.de>
7675 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7676 index associated with the tree we are supposed to replace.
7677 * tree-streamer-out.c (pack_ts_base_value_fields): Output
7678 TREE_ASM_WRITTEN as zero for everything but SSA names.
7680 2013-06-05 David Malcolm <dmalcolm@redhat.com>
7682 * tree-ssa-structalias.c (call_stmt_vars): Make static.
7684 2013-06-04 Jan Hubicka <jh@suse.cz>
7686 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7687 (input_node, input_varpool_node): Handle correctly external same
7689 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7690 nodes at ltrans stage.
7692 2013-06-04 Jan Hubicka <jh@suse.cz>
7694 * ipa-inline.c (update_caller_keys): Fix availability test.
7695 (update_callee_keys): Likewise.
7696 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7697 to follow ELF standard.
7699 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
7701 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7702 (mips64r5900el-*-elf*): New configurations.
7703 * config/mips/mips-cpus.def (r5900): New processor.
7704 * config/mips/mips-tables.opt: Regenerate.
7705 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7706 (mips_issue_rate): Handle PROCESSOR_R5900.
7707 (mips_reorg_process_insns): Force reorder mode for the R5900.
7708 * config/mips/mips.h (TARGET_MIPS5900): Define.
7709 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7711 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7713 * config/mips/mips.md (processor): Add r5900.
7714 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7716 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7718 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7719 into function to generate MOVI instruction.
7720 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7721 (aarch64_preferred_simd_mode): Turn into wrapper.
7722 (aarch64_output_scalar_simd_mov_immediate): New function.
7723 * config/aarch64/aarch64-protos.h: Add prototype for above.
7725 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7727 * config/aarch64/aarch64.c (simd_immediate_info): Remove
7728 element_char member.
7729 (sizetochar): Return signed char.
7730 (aarch64_simd_valid_immediate): Remove elchar and other
7731 unnecessary variables.
7732 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7733 Calculate element_char as required.
7734 * config/aarch64/aarch64-protos.h: Update and move prototype
7735 for aarch64_output_simd_mov_immediate.
7736 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7739 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7741 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7742 information completed by aarch64_simd_valid_immediate.
7743 (aarch64_legitimate_constant_p): Update arguments.
7744 (aarch64_simd_valid_immediate): Work with struct rather than many
7746 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7747 (aarch64_simd_make_constant): Update arguments.
7748 (aarch64_output_simd_mov_immediate): Work with struct rather than
7749 many pointers. Output immediate directly rather than as operand.
7750 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7752 * config/aarch64/constraints.md (Dn): Update arguments.
7754 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7756 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7758 (aarch64_simd_immediate_valid_for_move): Remove.
7759 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7760 (aarch64_simd_make_constant): Update call.
7761 (aarch64_output_simd_mov_immediate): Update call.
7762 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7764 * config/aarch64/constraints.md (Dn): Update call.
7766 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7768 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7769 return type to bool for prototype.
7770 (aarch64_legitimate_constant_p): Check for true instead of not -1.
7771 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7772 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7774 2013-06-04 Catherine Moore <clm@codesourcery.com>
7776 * config/mips/mips.opt (meva): New.
7777 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7778 (ASM_SPEC): Handle -meva.
7779 * doc/invoke.texi (meva): Document.
7781 2013-06-04 Alan Modra <amodra@gmail.com>
7783 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7786 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7788 * rtl.def: Add extra fourth optional field to define_cond_exec.
7789 * gensupport.c (process_one_cond_exec): Process attributes from
7791 * doc/md.texi: Document fourth field in define_cond_exec.
7793 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
7795 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7796 out the processing order as in store_bit_field_1.
7798 2013-06-04 Jan Hubicka <jh@suse.cz>
7801 * cgraphunit.c (cgraph_process_same_body_aliases): Create
7802 non-VAR_DECL node if it does not exist yet.
7804 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
7806 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7807 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7808 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7809 target_cpu_default setting.
7811 2013-06-03 Teresa Johnson <tejohnson@google.com>
7813 * dumpfile.c (opt_info_switch_p): Change -fopt-info
7814 default to -fopt-info=optimized instead of all.
7815 * doc/invoke.texi: Ditto.
7816 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7817 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7818 (execute_vect_slp): Emit BB vectorization success under
7819 MSG_OPTIMIZED_LOCATIONS.
7820 * tree-vect-slp.c (vect_slp_transform_bb): Change
7821 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7822 * tree-vect-loop.c (vect_transform_loop): Ditto.
7824 2013-06-03 Jason Merrill <jason@redhat.com>
7827 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7828 Use TARGET_EXPR for C++.
7830 2013-06-03 Jakub Jelinek <jakub@redhat.com>
7832 PR rtl-optimization/57268
7833 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7834 if DEBUG_INSN_P (insn).
7837 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
7839 PR rtl-optimization/57268
7840 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7841 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7843 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
7845 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7846 (ix86_avoid_lea_for_addr): Likewise.
7847 (exact_dependency_1): Likewise.
7848 (ix86_adjust_cost): Likewise.
7849 (swap_top_of_ready_list): Fix formatting and !reload_completed check
7851 (do_reorder_for_imul): Fix typo, formatting and
7852 !reload_completed check removed.
7853 (ix86_sched_reorder): Fix typo and formatting.
7854 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7857 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
7859 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7861 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
7863 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7864 <STRING_CST>: Likewise.
7865 <VECTOR_CST>: Likewise.
7867 2013-06-01 Janus Weil <janus@gcc.gnu.org>
7868 Mikael Morin <mikael@gcc.gnu.org>
7870 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7871 * config.in: Regenerated.
7872 * configure: Regenerated.
7874 2013-06-01 Jan Hubicka <jh@suse.cz>
7877 * cgraphunit.c (compile): When weakref is not supported,
7878 set up transparent aliases before final output pass.
7879 * varasm.c (assemble_alias): Do not try to do it here.
7881 2013-06-01 Jan Hubicka <jh@suse.cz>
7884 * passes.c (for_per_function): Skip unanalyzed functions.
7886 2013-06-01 Jan Hubicka <jh@suse.cz>
7888 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7889 (lto_symtab_merge_symbols_1): ... this one.
7890 (lto_symtab_merge_cgraph_nodes): Rename to ...
7891 (lto_symtab_merge_symbols): ... this one; simplify.
7892 * cgraph.c (same_body_aliases_done): Rename to ...
7893 (cpp_implicit_aliases_done): ... this one.
7894 (cgraph_create_function_alias): Update.
7895 (cgraph_same_body_alias): Update.
7896 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7897 (verify_edge_corresponds_to_fndecl): Simplify.
7898 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7899 (cgraph_node): Remove same_body_alias.
7900 (varpool_node): Remove alias_of and extra_name_alias.
7901 (same_body_aliases_done): Rename to ..
7902 (cpp_implicit_aliases_done): ... this one.
7903 (symtab_alias_ultimate_target): Add default parameter.
7904 (symtab_resolve_alias): New function.
7905 (fixup_same_cpp_alias_visibility): Declare.
7906 (cgraph_function_node): Add default parameter.
7907 (cgraph_node_asm_name): Likewise.
7908 (cgraph_function_or_thunk_node): Add default parameter; do
7909 not ICE when it is NULL.
7910 (varpool_variable_node): Likewise.
7911 * tree-emutls.c (create_emultls_var): Update.
7912 (ipa_lower_emutls): Update.
7913 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7914 (cgraph_reset_node): Reset alias info.
7915 (cgraph_finalize_function): Update.
7916 (fixup_same_cpp_alias_visibility): Move to symtab.c.
7917 (analyze_function): Simplify.
7918 (cgraph_process_same_body_aliases): Simplify.
7919 (analyze_functions): Fixup same body aliases.
7920 (handle_alias_pairs): Simplify.
7921 (assemble_thunk): Update.
7922 (assemble_thunks_and_aliases): Update.
7923 (output_weakrefs): Rewrite.
7924 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7925 (lto_output_varpool_node): Likewise.
7926 (compute_ltrans_boundary): Remve assert.
7927 (get_alias_symbol): New functoin.
7928 (input_node): Rewrite alias handling.
7929 (input_varpool_node): Likewise.
7930 * ipa-pure-const.c (propagate_pure_const): Fix formating.
7931 * ipa.c (process_references): Handle weakrefs correctly.
7932 (symtab_remove_unreachable_nodes): Likewise.
7933 * trans-mem.c (get_cg_data): Update.
7934 (ipa_tm_create_version_alias): Update.
7935 (ipa_tm_execute): Update.
7936 * symtab.c (dump_symtab_base): Dump aliases.
7937 (verify_symtab_base): Verify aliases.
7938 (symtab_node_availability): New function.
7939 (symtab_alias_ultimate_target): Simplify.
7940 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7941 handle all the fixup cases.
7942 (symtab_resolve_alias): New function.
7943 * passes.c (ipa_write_summaries): Handle weakrefs.
7944 * varpool.c (varpool_analyze_node): Simplify.
7945 (assemble_aliases): Update.
7946 (varpool_create_variable_alias): Simplify.
7947 (varpool_extra_name_alias): Simplify.
7948 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7949 (lto_symtab_merge_symbols): ... this one.
7951 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
7954 PR rtl-optimization/57268
7955 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7956 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7958 2013-06-01 Tobias Burnus <burnus@net-b.de>
7961 2013-05-31 Tobias Burnus <burnus@net-b.de>
7964 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7967 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
7969 PR rtl-optimization/57268
7970 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7971 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7973 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
7975 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7976 unordered comparison operators when -fno-trapping-math is in effect
7978 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7979 and implement unordered comparison operators properly on the e500.
7981 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
7983 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7984 for constant scalar integers.
7985 (simplify_relational_operation_1): Likewise.
7987 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7989 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7990 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
7993 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
7994 Igor Zamyatin <igor.zamyatin@intel.com>
7996 Silvermont (SLM) architecture performance tuning.
7997 * config/i386/i386.h (enum ix86_tune_indices): Add
7998 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7999 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
8001 * config/i386/i386.c (initial_ix86_tune_features)
8002 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
8003 (ix86_lea_outperforms): Handle Silvermont tuning.
8004 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
8006 (ix86_use_lea_for_mov): Likewise.
8007 (ix86_avoid_lea_for_addr): Likewise.
8008 (ix86_lea_for_add_ok): Likewise.
8009 (exact_dependency_1): New function.
8010 (exact_store_load_dependency): Likewise.
8011 (ix86_adjust_cost): Handle Silvermont tuning.
8012 (do_reoder_for_imul): Likewise.
8013 (swap_top_of_ready_list): New function.
8014 (ix86_sched_reorder): Changed to handle Silvermont tuning.
8016 * config/i386/i386.md (peepholes that split memory operand in fp
8019 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
8021 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8022 Remove un-necessary braces.
8024 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
8026 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8027 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
8029 2013-05-31 Tobias Burnus <burnus@net-b.de>
8032 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
8034 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8037 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
8038 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
8039 * config/arm/neon.md (iordi3_neon): Remove.
8040 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
8041 * config/arm/predicates.md (imm_for_neon_logic_operand):
8042 Move to earlier in the file.
8043 (neon_logic_op2): Likewise.
8044 (arm_iordi_operand_neon): New predicate.
8046 2013-05-31 Richard Biener <rguenther@suse.de>
8048 PR tree-optimization/57478
8049 PR tree-optimization/57453
8050 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
8053 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
8055 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
8056 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
8058 2013-05-30 Tobias Burnus <burnus@net-b.de>
8059 Thomas Koenig <tkoenig@gcc.gnu.org>
8062 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
8063 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
8065 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
8067 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
8069 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
8071 * target.def (register_usage_leveling_p): New hook.
8072 * targhooks.c (default_register_usage_leveling_p): New.
8073 * targhooks.h (default_register_usage_leveling_p): New prototype.
8074 * lra-assigns.c (register_usage_leveling_p): Use the hook.
8075 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
8076 * doc/tm.texi: Update.
8077 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
8079 2013-05-30 Ian Bolton <ian.bolton@arm.com>
8081 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
8082 (*insv_reg<mode>): New define_insn.
8084 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
8086 PR rtl-optimization/57439
8087 * postreload.c (move2add_valid_value_p): Check that we have
8088 a zero subreg_regno_offset when accessing the register in
8091 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
8092 Igor Zamyatin <igor.zamyatin@intel.com>
8094 Silvermont (SLM) architecture pipeline model, tuning and
8096 * config.gcc: Add slm config options and target.
8098 * config/i386/slm.md: New.
8100 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
8102 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): New case
8104 (ix86_target_macros_internal): Likewise.
8106 * gcc/config/i386/i386.c (slm_cost): New cost.
8107 (m_SLM): New macro flag.
8108 (initial_ix86_tune_features): Set m_SLM.
8109 (x86_accumulate_outgoing_args): Likewise.
8110 (x86_arch_always_fancy_math_387): Likewise.
8111 (processor_target_table): Add slm cost.
8112 (cpu_names): Add slm cpu name.
8113 (x86_option_override_internal): Set SLM ISA.
8114 (ix86_issue_rate): New case PROCESSOR_SLM.
8115 (ia32_multipass_dfa_lookahead): Likewise.
8116 (fold_builtin_cpu): Add slm.
8118 * config/i386/i386.h (TARGET_SLM): New target macro.
8119 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
8120 (processor_type): Add PROCESSOR_SLM.
8122 * config/i386/i386.md (cpu): Add new value "slm".
8123 (slm.md): Include slm.md.
8125 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
8126 Zhenqiang Chen <zhenqiang.chen@linaro.org>
8128 * config/arm/arm-protos.h: Add and update function protos.
8129 * config/arm/arm.c (use_simple_return_p): New added.
8130 (thumb2_expand_return): Check simple_return flag.
8131 * config/arm/arm.md: Add simple_return and conditional simple_return.
8132 * config/arm/iterators.md: Add iterator for return and simple_return.
8134 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
8136 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
8137 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
8138 (arm_emit_vfp_multi_reg_pop): Likewise.
8139 (thumb2_emit_ldrd_pop): Likewise.
8140 (arm_expand_epilogue): Add misc REG_CFA notes.
8141 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
8143 2013-05-29 Lawrence Crowl <crowl@google.com>
8145 * config/arm/t-arm: Update for below.
8147 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
8148 Change type to hash_table. Update dependent calls and types.
8150 * config/i386/t-cygming: Update for below.
8152 * config/i386/t-interix: Update for below.
8154 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
8155 Change type to hash_table. Update dependent calls and types.
8156 (i386_find_on_wrapper_list::wrappers): Likewise.
8158 * config/ia64/t-ia64: Update for below.
8160 * config/ia64/ia64.c (bundle_state_table):
8161 Change type to hash_table. Update dependent calls and types.
8163 * config/mips/mips.c (mips_reorg_process_insns::htab):
8164 Change type to hash_table. Update dependent calls and types.
8166 * config/sol2.c (solaris_comdat_htab):
8167 Change type to hash_table. Update dependent calls and types.
8169 * config/t-sol2: Update for above.
8171 2013-05-29 Teresa Johnson <tejohnson@google.com>
8173 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
8174 functions are not yet marked as defined.
8176 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8177 Pat Haugen <pthaugen@us.ibm.com>
8178 Peter Bergner <bergner@vnet.ibm.com>
8180 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
8184 (vrotl<mode>3): Likewise.
8185 (vashl<mode>3): Likewise.
8186 (vlshr<mode>3): Likewise.
8187 (vashr<mode>3): Likewise.
8189 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8190 support for power8 V2DI builtins.
8192 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
8193 power8 V2DI builtins.
8194 (vupkhsw): Likewise.
8195 (vupklsw): Likewise.
8196 (vaddudm): Likewise.
8201 (vpkudum): Likewise.
8202 (vpksdss): Likewise.
8203 (vpkudus): Likewise.
8204 (vpksdus): Likewise.
8209 (vsubudm): Likewise.
8210 (vcmpequd): Likewise.
8211 (vcmpgtsd): Likewise.
8212 (vcmpgtud): Likewise.
8213 (vcmpequd_p): Likewise.
8214 (vcmpgtsd_p): Likewise.
8215 (vcmpgtud_p): Likewise.
8216 (vupkhsw): Likewise.
8217 (vupklsw): Likewise.
8218 (vaddudm): Likewise.
8223 (vpksdss): Likewise.
8224 (vpksdus): Likewise.
8225 (vpkudum): Likewise.
8226 (vpkudus): Likewise.
8231 (vsubudm): Likewise.
8233 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
8234 support for power8 V2DI instructions.
8236 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
8237 power8 V2DI instructions. Combine pack and unpack insns to use an
8238 iterator for each mode. Check whether a particular mode supports
8239 Altivec instructions instead of just checking TARGET_ALTIVEC.
8240 (UNSPEC_VPKUWUM): Likewise.
8241 (UNSPEC_VPKSHSS): Likewise.
8242 (UNSPEC_VPKSWSS): Likewise.
8243 (UNSPEC_VPKUHUS): Likewise.
8244 (UNSPEC_VPKSHUS): Likewise.
8245 (UNSPEC_VPKUWUS): Likewise.
8246 (UNSPEC_VPKSWUS): Likewise.
8247 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
8248 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
8249 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
8250 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
8251 (UNSPEC_VUPKHSB): Likewise.
8252 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
8253 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
8254 (UNSPEC_VUPKHSH): Likewise.
8255 (UNSPEC_VUPKLSB): Likewise.
8256 (UNSPEC_VUPKLSH): Likewise.
8258 (VI_char): Likewise.
8259 (VI_scalar): Likewise.
8260 (VI_unit): Likewise.
8262 (VP_small): Likewise.
8263 (VP_small_lc): Likewise.
8264 (VU_char): Likewise.
8265 (add<mode>3): Likewise.
8266 (altivec_vaddcuw): Likewise.
8267 (altivec_vaddu<VI_char>s): Likewise.
8268 (altivec_vadds<VI_char>s): Likewise.
8269 (sub<mode>3): Likewise.
8270 (altivec_vsubcuw): Likewise.
8271 (altivec_vsubu<VI_char>s): Likewise.
8272 (altivec_vsubs<VI_char>s): Likewise.
8273 (altivec_vavgs<VI_char>): Likewise.
8274 (altivec_vcmpbfp): Likewise.
8275 (altivec_eq<mode>): Likewise.
8276 (altivec_gt<mode>): Likewise.
8277 (altivec_gtu<mode>): Likewise.
8278 (umax<mode>3): Likewise.
8279 (smax<mode>3): Likewise.
8280 (umin<mode>3): Likewise.
8281 (smin<mode>3): Likewise.
8282 (altivec_vpkuhum): Likewise.
8283 (altivec_vpkuwum): Likewise.
8284 (altivec_vpkshss): Likewise.
8285 (altivec_vpkswss): Likewise.
8286 (altivec_vpkuhus): Likewise.
8287 (altivec_vpkshus): Likewise.
8288 (altivec_vpkuwus): Likewise.
8289 (altivec_vpkswus): Likewise.
8290 (altivec_vpks<VI_char>ss): Likewise.
8291 (altivec_vpks<VI_char>us): Likewise.
8292 (altivec_vpku<VI_char>us): Likewise.
8293 (altivec_vpku<VI_char>um): Likewise.
8294 (altivec_vrl<VI_char>): Likewise.
8295 (altivec_vsl<VI_char>): Likewise.
8296 (altivec_vsr<VI_char>): Likewise.
8297 (altivec_vsra<VI_char>): Likewise.
8298 (altivec_vsldoi_<mode>): Likewise.
8299 (altivec_vupkhsb): Likewise.
8300 (altivec_vupkhs<VU_char>): Likewise.
8301 (altivec_vupkls<VU_char>): Likewise.
8302 (altivec_vupkhsh): Likewise.
8303 (altivec_vupklsb): Likewise.
8304 (altivec_vupklsh): Likewise.
8305 (altivec_vcmpequ<VI_char>_p): Likewise.
8306 (altivec_vcmpgts<VI_char>_p): Likewise.
8307 (altivec_vcmpgtu<VI_char>_p): Likewise.
8308 (abs<mode>2): Likewise.
8309 (vec_unpacks_hi_v16qi): Likewise.
8310 (vec_unpacks_hi_v8hi): Likewise.
8311 (vec_unpacks_lo_v16qi): Likewise.
8312 (vec_unpacks_hi_<VP_small_lc>): Likewise.
8313 (vec_unpacks_lo_v8hi): Likewise.
8314 (vec_unpacks_lo_<VP_small_lc>): Likewise.
8315 (vec_pack_trunc_v8h): Likewise.
8316 (vec_pack_trunc_v4si): Likewise.
8317 (vec_pack_trunc_<mode>): Likewise.
8319 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
8321 (vec_vmaxsd): Likewise.
8322 (vec_vmaxud): Likewise.
8323 (vec_vminsd): Likewise.
8324 (vec_vminud): Likewise.
8325 (vec_vpksdss): Likewise.
8326 (vec_vpksdus): Likewise.
8327 (vec_vpkudum): Likewise.
8328 (vec_vpkudus): Likewise.
8329 (vec_vrld): Likewise.
8330 (vec_vsld): Likewise.
8331 (vec_vsrad): Likewise.
8332 (vec_vsrd): Likewise.
8333 (vec_vsubudm): Likewise.
8334 (vec_vupkhsw): Likewise.
8335 (vec_vupklsw): Likewise.
8337 2013-05-29 Jan Hubicka <jh@suse.cz>
8339 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
8340 flags; reorder rest of fields in more consistent way.
8341 (varpool_node): Remove analyzed, finalized and alias.
8342 (cgraph_ndoe): Likewise.
8343 (symtab_alias_ultimate_target): New function.
8344 (cgraph_function_node): Move offline.
8345 (cgraph_reset_node): Declare.
8346 (cgraph_comdat_can_be_unshared_p): Remove.
8347 (varpool_remove_initializer): Declare.
8348 (varpool_first_defined_variable, varpool_next_defined_variable
8349 cgraph_first_defined_function, cgraph_next_defined_function): Update.
8350 (cgraph_function_with_gimple_body_p): Update.
8351 (varpool_all_refs_explicit_p): Update.
8352 (symtab_alias_target): New function.
8353 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
8354 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
8355 (cgraph_function_or_thunk_node): Simplify using
8356 symtab_alias_ultimate_target.
8357 (varpool_variable_node): Likewise.
8358 * cgraph.c (cgraph_create_function_alias): Update.
8359 (cgraph_add_thunk): Update.
8360 (cgraph_remove_node): Update.
8361 (dump_cgraph_node): Do not dump removed flags.
8362 (cgraph_function_body_availability): Update.
8363 (cgraph_propagate_frequency): Update.
8364 (verify_cgraph_node): Check sanity of local flag.
8365 (cgraph_function_node): Move here from cgraph.h; revamp for
8366 cgraph_function_or_thunk_node.
8367 * lto-symtab.c (lto_varpool_replace_node): Update.
8368 (lto_symtab_resolve_can_prevail_p): Update.
8369 (lto_symtab_merge_cgraph_nodes): Update.
8370 * ipa-cp.c (determine_versionability, initialize_node_lattices,
8371 propagate_constants_accross_call, devirtualization_time_bonus,
8372 ipcp_propagate_stage): Update.
8373 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
8374 * ipa-inline-transform.c (clone_inlined_nodes,
8375 preserve_function_body_p): Update.
8376 * ipa-reference.c (propagate): Update.
8377 (write_node_summary_p): Update.
8378 * toplev.c (wrapup_global_declaration_2): Update.
8379 * cgraphunit.c (cgraph_analyze_function): Rename to ...
8380 (analyze_function) ... this one.
8381 (cgraph_process_new_functions): Update.
8382 (cgraph_reset_node): Export.
8383 (cgraph_finalize_function): Update.
8384 (cgraph_add_new_function): Update.
8385 (process_function_and_variable_attributes): Update.
8386 (varpool_finalize_decl): Update.
8387 (symbol_finalized): Remove.
8388 (symbol_finalized_and_needed): Rename to ...
8389 (symbol_defined_and_needed): ... update.
8390 (cgraph_analyze_functions): Update.
8391 (handle_alias_pairs): Update.
8392 (mark_functions_to_output): Update.
8393 (assemble_thunk): Update.
8394 (output_in_order): Update.
8395 (output_weakrefs): Update.
8396 (finalize_compilation_unit): Update.
8397 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
8398 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
8399 input_node, input_varpool_node): Update.
8400 * dbxout.c (dbxout_expand_expr): Update.
8401 * cgraphclones.c (cgraph_clone_node): Update.
8402 (cgraph_copy_node_for_versioning): Update.
8403 (cgraph_materialize_clone): Update.
8404 (cgraph_materialize_all_clones): Update.
8405 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8406 propagate_pure_const, propagate_nothrow): Update.
8407 * lto-streamer-out.c (lto_output, write_symbol): Update.
8408 * ipa-utils.c (ipa_reverse_postorder): Update.
8409 * ipa-inline.c (can_inline_edge_p): Update.
8410 (update_caller_keys, ipa_inline): Update.
8411 * dwarf2out.c (reference_to_unused,
8412 premark_types_used_by_global_vars_helper): Update.
8413 * tree-eh.c (tree_could_trap_p): Update.
8414 * ipa-split.c (consider_split, execute_split_functions): Update.
8415 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8416 has_addr_references_p): Update; move ahead in file for better
8418 (process_references): Simplify.
8419 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8421 (cgraph_comdat_can_be_unshared_p): Make static.
8422 (cgraph_externally_visible_p): Update.
8423 (varpool_externally_visible_p): Update.
8424 (function_and_variable_visibility): Update.
8425 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8426 ipa_tm_mark_force_output_node): Update.
8427 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8428 estimate_edge_devirt_benefit, inline_generate_summary,
8429 inline_write_summary): Update.
8430 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8431 * ipa-prop.c (ipa_compute_jump_functions): Update.
8432 (ipa_print_node_params, ipa_prop_read_section,
8433 ipa_update_after_lto_read, read_replacements_section): Update.
8434 * varasm.c (mark_decl_referenced): Update.
8435 (assemble_alias, dump_tm_clone_pairs): Update.
8436 * tree-inline.c (copy_bb): Update.
8437 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8439 * symtab.c (dump_symtab_base): Print new flags.
8440 (verify_symtab_base): Verify new flags.
8441 (symtab_alias_ultimate_target): New function.
8442 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8443 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8445 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8447 * i386.c (ix86_get_function_versions_dispatcher,
8448 ix86_generate_version_dispatcher_body): Update.
8449 (fold_builtin_cpu): Use varpool_add_new_variable.
8450 * varpool.c (varpool_remove_initializer): Break out from ...
8451 (varpool_remove_node): ... this one.
8452 (dump_varpool_node, varpool_node_for_asm,
8453 cgraph_variable_initializer_availability, varpool_analyze_node,
8454 varpool_assemble_decl, varpool_remove_unreferenced_decls,
8455 varpool_finalize_named_section_flags, varpool_create_variable_alias):
8458 2013-05-29 Jan Hubicka <jh@suse.cz>
8460 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8462 2013-05-29 Easwaran Raman <eraman@google.com>
8464 PR tree-optimization/57442
8465 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8466 when control exits the main loop.
8468 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
8470 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8472 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8473 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8474 * rx/t-rx: Add rx100 under multi library matches option for nofpu
8477 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8479 PR tree-optimization/57441
8480 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8481 Don't limit size of incr_vec to number of candidates.
8483 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
8485 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8486 and mips16 directories.
8487 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8488 (MULTILIB_DIRNAMES): Ditto.
8489 (MULTILIB_EXCEPTIONS): Add new exceptions.
8490 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8491 (MULTILIB_DIRNAMES): Ditto.
8492 (MULTILIB_EXCEPTIONS): Add new exceptions.
8494 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8495 Marcus Shawcroft <marcus.shawcroft@arm.com>
8497 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8498 SYMBOL_TINY_ABSOLUTE.
8499 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8500 SYMBOL_TINY_ABSOLUTE.
8501 (aarch64_expand_mov_immediate): Likewise.
8502 (aarch64_classify_symbol): Likewise.
8503 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8504 Permit SYMBOL_TINY_ABSOLUTE.
8505 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8507 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8508 Marcus Shawcroft <marcus.shawcroft@arm.com>
8510 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8511 Refactor if/switch. Replace gcc_assert with if.
8513 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8515 * config/i386/i386.c (initial_ix86_tune_features): Enable
8516 FP Reassociation for AMD bdver1 and bdver2.
8518 2013-05-29 Martin Jambor <mjambor@suse.cz>
8520 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8521 and IMAGPART_EXPR do not occur within other handled_components.
8523 2013-05-29 Richard Biener <rguenther@suse.de>
8525 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8526 access on whether the use is in the BB we currently try to
8528 (vect_bb_vectorization_profitable_p): Pass the BB we currently
8529 vectorize to vect_bb_slp_scalar_cost.
8531 2013-05-29 Richard Biener <rguenther@suse.de>
8533 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8534 computing scalar cost offsetted by stmts that are kept live
8536 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8537 for computation of scalar cost.
8539 2013-05-28 Steve Ellcey <sellcey@mips.com>
8541 * config/mips/mips-cpus.def (mips32r2): Change processor type.
8543 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
8545 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8546 array notation built-in reduction functions.
8547 * doc/passes.texi (Passes): Added documentation about changes done
8549 * doc/invoke.texi (C Dialect Options): Added documentation about
8550 the -fcilkplus flag.
8551 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8552 (BUILTINS_DEF): Depend on cilkplus.def.
8553 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
8554 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8555 * cilkplus.def: New file.
8557 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
8559 PR rtl-optimization/57439
8560 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8562 2013-05-28 Easwaran Raman <eraman@google.com>
8564 PR tree-optimization/57337
8565 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8566 (find_insert_point): Correctly identify the insertion point
8567 when two statements with the same UID is compared.
8569 2013-05-28 Richard Biener <rguenther@suse.de>
8571 PR tree-optimization/56787
8572 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8573 from the list of data references.
8574 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8576 (vect_analyze_loop_operations): Likewise.
8577 (vect_transform_loop): Remove clobbers.
8579 2013-05-28 Martin Jambor <mjambor@suse.cz>
8581 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8582 and REALPART_EXPRs have scalar type.
8584 2013-05-28 Richard Biener <rguenther@suse.de>
8586 PR tree-optimization/57411
8587 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8589 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8590 virtual operand propagation.
8592 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
8594 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8595 destination register for bmasksi_vis.
8596 (vector_init_bshuffle): Likewise.
8597 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8599 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
8601 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8602 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8603 mode if the instruction isn't available in the original mode.
8604 * config/sparc/sparc.opt (mfix-ut699): New option.
8605 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8606 (divdf3): Turn into expander.
8607 (divdf3_nofix): New insn.
8608 (divdf3_fix): Likewise.
8609 (divsf3): Disable if -mfix-ut699.
8610 (sqrtdf2): Turn into expander.
8611 (sqrtdf2_nofix): New insn.
8612 (sqrtdf2_fix): Likewise.
8613 (sqrtsf2): Disable if -mfix-ut699.
8615 2013-05-27 Richard Biener <rguenther@suse.de>
8618 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8619 block for the new loop.
8621 2013-05-27 Richard Biener <rguenther@suse.de>
8623 PR tree-optimization/57343
8624 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8625 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8626 (number_of_iterations_cond): Do not build the folded tree.
8628 2013-05-27 Richard Biener <rguenther@suse.de>
8632 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8633 OEP_CONSTANT_ADDRESS_OF retained.
8635 PR tree-optimization/57417
8636 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8638 (set_ssa_val_to): Compare addresses using
8639 get_addr_base_and_unit_offset.
8641 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
8643 PR rtl-optimization/56833
8644 * postreload.c (move2add_record_mode): New function.
8645 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8646 (move2add_use_add2_insn): Use move2add_record_sym_value.
8647 (move2add_use_add3_insn): Likewise.
8648 (reload_cse_move2add): Use move2add_valid_value_p and
8649 move2add_record_mode. Invalidate call-clobbered and REG_INC
8650 affected regs by setting reg_mode to VOIDmode.
8651 (move2add_note_store): Don't pretend the inside of a SUBREG is
8652 the actual destination. Invalidate single/leading registers by
8653 setting reg_mode to VOIDmode.
8654 Use move2add_record_sym_value, move2add_valid_value_p and
8655 move2add_record_mode.
8657 2013-05-27 Richard Biener <rguenther@suse.de>
8659 PR tree-optimization/57396
8660 * tree-affine.c (double_int_constant_multiple_p): Properly
8661 return false for val == 0 and div != 0.
8663 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8665 * config/mips/mips.h: Use #elif in preprocessor conditions.
8667 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8670 * config/mips/constraints.md (kl): New constraint.
8671 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8672 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
8673 constraint for operand 0. Split after CSE for MIPS16. Emit a move
8675 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8677 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8680 * config/mips/mips.c (mips_can_inline_p): New function.
8681 (TARGET_CAN_INLINE_P): Define.
8683 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
8685 * sched-int.h (ds_t, dw_t): Make unsigned int.
8686 Fix documentation that describes how all the ds_t bits are used.
8687 Reserve the last bit for delayed-branch scheduling.
8688 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8689 (BITS_PER_DEP_WEAK): Fix definition and documentation.
8690 (gen_dep_weak_1): Remove prototype.
8691 * sched-deps.c (get_dep_weak_1): Make static.
8692 * target.def (speculate_insn, needs_block_p, gen_spec_check,
8693 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8694 * doc/tm.texi: Regenerate.
8695 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8697 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
8700 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8702 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
8703 Sandra Loosemore <sandra@codesourcery.com>
8705 * config.gcc (powerpc-*): Allow native for with-cpu.
8707 2013-05-24 Jeff Law <law@redhat.com>
8709 PR tree-optimization/57124
8710 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8711 conversion feeding a condition if the range has an overflow
8712 if -fstrict-overflow. Add warnings for when we do make the
8715 2013-05-24 Dehao Chen <dehao@google.com>
8717 * tree-cfg.c (locus_discrim_map): Fix the typo.
8718 (locus_discrim_hasher): Likewise.
8719 (locus_discrim_hasher::hash): Likewise.
8720 (locus_discrim_hasher::equal): Likewise.
8722 2013-05-24 Martin Jambor <mjambor@suse.cz>
8724 PR tree-optimization/57294
8725 * cgraph.h (ipa_record_stmt_references): Declare.
8726 * cgraphbuild.c (ipa_record_stmt_references): New function.
8727 (build_cgraph_edges): Use ipa_record_stmt_references.
8728 (rebuild_cgraph_edges): Likewise.
8729 (cgraph_rebuild_references): Likewise.
8730 * ipa-prop.c (ipa_modify_call_arguments): Discard references
8731 associated with the old statement and build references from the
8732 newly built statements.
8733 * ipa-ref.c (ipa_remove_stmt_references): New function.
8734 * ipa-ref.h (ipa_remove_stmt_references): Declare.
8736 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
8738 * lra-constraints.c (emit_spill_move): Use smaller mode for
8740 (check_and_process_move): Consider mem-reg moves for secondary
8742 (curr_insn_transform): Don't lose insns emitted before for
8743 secondary memory moves.
8744 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
8745 reg set up in the current insn.
8747 2013-05-24 Dehao Chen <dehao@google.com>
8749 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8751 (locus_descrim_hasher::equal): Likewise.
8752 (build_gimple_cfg): New discriminator assignment algorithm.
8753 (make_edges): Likewise.
8754 (next_discriminator_for_locus): Likewise.
8755 (same_line_p): Likewise.
8756 (assign_discriminators): Likewise.
8757 (make_cond_expr_edges): Likewise.
8758 (make_gimple_switch_edges): Likewise.
8759 (make_goto_expr_edges): Likewise.
8760 (make_gimple_asm_edges): Likewise.
8762 2013-05-24 Ian Bolton <ian.bolton@arm.com>
8764 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8765 X format specifier to only display bottom 16 bits.
8766 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8767 immediate to match for operand 2, since it will be masked.
8769 2013-05-24 Richard Biener <rguenther@suse.de>
8771 PR tree-optimization/57287
8772 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8773 all SSA names that occur in abnormal PHIs.
8775 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
8778 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8779 that index is not negative.
8781 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
8783 PR rtl-optimization/55177
8784 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8785 (simplify_byte_swapping_operation): New.
8786 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8787 (simplify_relational_operation_1): Deal with BSWAP.
8789 2013-05-23 Richard Henderson <rth@redhat.com>
8792 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8793 (ix86_reorg): Call it.
8795 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
8798 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8799 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8800 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8802 2013-05-23 Christian Bruel <christian.bruel@st.com>
8805 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8807 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8808 Marcus Shawcroft <marcus.shawcroft@arm.com>
8810 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8811 * config/aarch64/constraints.md (Usa): Remove.
8812 * doc/md.texi (AArch64 Usa): Remove.
8814 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8815 Marcus Shawcroft <marcus.shawcroft@arm.com>
8817 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8818 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8819 * config/aarch64/predicates.md (aarch64_const_address): Remove.
8820 (aarch64_mov_operand): Use aarch64_mov_operand_p.
8822 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
8824 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8825 instruction (AdvSIMD).
8826 * config/aarch64/aarch64-builtins.c
8827 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8828 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8830 2013-05-23 Martin Jambor <mjambor@suse.cz>
8833 * tree.h (contains_bitfld_component_ref_p): Declare.
8834 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8835 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
8837 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8838 not access a bit-field. Assert all final offsets are byte-aligned.
8840 2013-05-23 Richard Biener <rguenther@suse.de>
8842 PR tree-optimization/57380
8843 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8844 least one invariant or re-used load.
8845 * passes.c (init_optimization_passes): Move pass_phiprop before
8848 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
8850 * config/aarch64/aarch64-simd.md
8851 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8853 2013-05-23 Richard Biener <rguenther@suse.de>
8856 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8857 OEP_CONSTANT_ADDRESS_OF retained.
8859 2013-05-23 Jakub Jelinek <jakub@redhat.com>
8862 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8863 don't lower unit. Handle unit not being always BITS_PER_WORD.
8865 2013-05-23 Richard Biener <rguenther@suse.de>
8867 PR rtl-optimization/57341
8868 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8869 instead of true_dependence.
8871 2013-05-22 David Malcolm <dmalcolm@redhat.com>
8873 * bb-reorder.c (branch_threshold): Make const.
8874 (exec_threshold): Ditto.
8876 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8877 Pat Haugen <pthaugen@us.ibm.com>
8878 Peter Bergner <bergner@vnet.ibm.com>
8880 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8881 documentation for the power8 crypto builtins.
8883 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8885 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8886 macros for defining power8 builtin functions.
8887 (BU_P8V_AV_2): Likewise.
8888 (BU_P8V_AV_P): Likewise.
8889 (BU_P8V_VSX_1): Likewise.
8890 (BU_P8V_OVERLOAD_1): Likewise.
8891 (BU_P8V_OVERLOAD_2): Likewise.
8892 (BU_CRYPTO_1): Likewise.
8893 (BU_CRYPTO_2): Likewise.
8894 (BU_CRYPTO_3): Likewise.
8895 (BU_CRYPTO_OVERLOAD_1): Likewise.
8896 (BU_CRYPTO_OVERLOAD_2): Likewise.
8897 (XSCVSPDP): Fix typo, point to the correct instruction.
8898 (VCIPHER): Add power8 crypto builtins.
8899 (VCIPHERLAST): Likewise.
8900 (VNCIPHER): Likewise.
8901 (VNCIPHERLAST): Likewise.
8902 (VPMSUMB): Likewise.
8903 (VPMSUMH): Likewise.
8904 (VPMSUMW): Likewise.
8905 (VPERMXOR_V2DI): Likewise.
8906 (VPERMXOR_V4SI: Likewise.
8907 (VPERMXOR_V8HI: Likewise.
8908 (VPERMXOR_V16QI: Likewise.
8909 (VSHASIGMAW): Likewise.
8910 (VSHASIGMAD): Likewise.
8912 (VPERMXOR): Likewise.
8913 (VSHASIGMA): Likewise.
8915 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8916 __CRYPTO__ if the crypto instructions are available.
8917 (altivec_overloaded_builtins): Add support for overloaded power8
8920 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8921 support for power8 crypto builtins.
8922 (builtin_function_type): Likewise.
8923 (altivec_init_builtins): Add support for builtins that take vector
8924 long long (V2DI) arguments.
8926 * config/rs6000/crypto.md: New file, define power8 crypto
8929 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8930 Pat Haugen <pthaugen@us.ibm.com>
8931 Peter Bergner <bergner@vnet.ibm.com>
8933 * doc/invoke.texi (Option Summary): Add power8 options.
8934 (RS/6000 and PowerPC Options): Likewise.
8936 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8937 constraints.md instead of rs6000.h. Reorder w* constraints. Add
8938 wm, wn, wr documentation.
8940 * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8941 registers if direct move instructions are enabled.
8942 (wn): New constraint for no registers.
8943 (wq): New constraint for quad word even GPR registers.
8944 (wr): New constraint if 64-bit instructions are enabled.
8945 (wv): New constraint if power8 vector instructions are enabled.
8946 (wQ): New constraint for quad word memory locations.
8948 * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8949 constraint for 0..15 for crypto instructions.
8950 (gpc_reg_operand): If VSX allow registers in VSX registers as well
8951 as GPR and floating point registers.
8952 (int_reg_operand): New predicate to match only GPR registers.
8953 (base_reg_operand): New predicate to match base registers.
8954 (quad_int_reg_operand): New predicate to match even GPR registers
8955 for quad memory operations.
8956 (vsx_reg_or_cint_operand): New predicate to allow vector logical
8957 operations in both GPR and VSX registers.
8958 (quad_memory_operand): New predicate for quad memory operations.
8959 (reg_or_indexed_operand): New predicate for direct move support.
8961 * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8962 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8963 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8964 (POWERPC_MASKS): Add power8 options.
8965 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8968 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8969 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8971 * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8972 (-mpower8-fusion): New power8 options.
8973 (-mpower8-fusion-sign): Likewise.
8974 (-mpower8-vector): Likewise.
8975 (-mcrypto): Likewise.
8976 (-mdirect-move): Likewise.
8977 (-mquad-memory): Likewise.
8979 * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8981 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8983 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8984 (rs6000_debug_vector_unit): Add p8_vector.
8985 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8986 definitions. Also print fusion state.
8987 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8988 (rs6000_builtin_mask_calculate): Add power8 builtin support.
8989 (rs6000_option_override_internal): Add support for power8.
8990 (rs6000_common_init_builtins): Add debugging for skipped builtins
8992 (rs6000_adjust_cost): Add power8 support.
8993 (rs6000_issue_rate): Likewise.
8994 (insn_must_be_first_in_group): Likewise.
8995 (insn_must_be_last_in_group): Likewise.
8996 (force_new_group): Likewise.
8997 (rs6000_register_move_cost): Likewise.
8998 (rs6000_opt_masks): Likewise.
9000 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
9001 power8 capable assembler, default to power7 options.
9002 (TARGET_DIRECT_MOVE): Likewise.
9003 (TARGET_CRYPTO): Likewise.
9004 (TARGET_P8_VECTOR): Likewise.
9005 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
9006 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
9007 (VECTOR_MEM_P8_VECTOR_P): Likewise.
9008 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
9009 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
9010 (TARGET_XSCVDPSPN): Likewise.
9011 (TARGET_XSCVSPDPN): Likewsie.
9012 (TARGET_SYNC_HI_QI): Likewise.
9013 (TARGET_SYNC_TI): Likewise.
9014 (MASK_CRYPTO): Likewise.
9015 (MASK_DIRECT_MOVE): Likewise.
9016 (MASK_P8_FUSION): Likewise.
9017 (MASK_P8_VECTOR): Likewise.
9018 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
9019 temporary used by some of the direct move instructions to get two FP
9020 temporary registers does not force creation of a stack frame.
9021 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
9022 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
9023 that any VSX registers are tieable, even if they are also an
9024 Altivec vector mode.
9025 (r6000_reg_class_enum): Add wm, wr, wv constraints.
9026 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
9027 (RS6000_BTM_CRYPTO): Likewise.
9028 (RS6000_BTM_COMMON): Likewise.
9030 * config/rs6000/rs6000.md (cpu attribute): Add power8.
9031 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
9032 (enum rs6000_vector): Add power8 vector support.
9034 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9038 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
9039 (any_sibcall_could_use_r3): this and handle indirect calls.
9040 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
9042 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9044 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
9046 2013-05-22 Richard Biener <rguenther@suse.de>
9049 * profile.c (branch_prob): Do not split blocks that are
9050 abnormally receiving from ECF_RETURNS_TWICE functions.
9052 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
9054 * recog.c (offsettable_address_addr_space_p): Fix calculation of
9055 address mode. Move pointer mode initialization to the same place.
9057 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
9059 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
9060 while it has any effect.
9062 2013-05-21 Easwaran Raman <eraman@google.com>
9064 PR tree-optimization/57322
9065 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
9066 UID of the statement added to the BB to be 1.
9068 2013-05-21 Jakub Jelinek <jakub@redhat.com>
9070 PR tree-optimization/57331
9071 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
9072 of conversion from pointer type to integral type with integer.
9074 2013-05-21 Martin Jambor <mjambor@suse.cz>
9077 * ipa-prop.c (ipa_read_node_info): Process param_used and
9078 controlled_uses in the same order as when writing.
9080 2013-05-21 Magnus Granberg <baldrick@free.fr>
9083 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
9085 2013-05-21 Richard Biener <rguenther@suse.de>
9087 PR tree-optimization/57318
9088 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
9089 estimate stmts with side-effects as likely eliminated.
9091 2013-05-21 Richard Biener <rguenther@suse.de>
9093 PR tree-optimization/57330
9094 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
9095 preserve the call stmts fntype.
9097 2013-05-21 Richard Biener <rguenther@suse.de>
9099 PR tree-optimization/57303
9100 * tree-ssa-sink.c (statement_sink_location): Improve killing
9101 stmt detection and properly handle self-assignments.
9103 2013-05-21 Christian Bruel <christian.bruel@st.com>
9105 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
9106 spanning registers. LEAF_REG_REMAP is supported only for contiguous
9107 registers. Set register size out of the PARALLEL loop.
9109 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
9112 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
9113 (*fmasf4, *fmasf4_media): New insns.
9115 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9117 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
9118 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
9119 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
9120 (mips_idiv_insns): Update the comments to say that the returned
9121 instruction counts are in units of BASE_INSN_LENGTH.
9122 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
9123 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
9124 using 2 rather than 4 as the length of indirect MIPS16 and
9125 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
9126 length of a NOP. Don't divide MIPS16 lengths by 2.
9127 (mips16_split_long_branches): Assume a branch is long if the
9128 length is greater than 4 rather than 8.
9129 * config/mips/mips.md (length): Give MIPS16 lengths directly,
9130 rather than multiplying them by 2. Multiply instruction counts
9131 by BASE_INSN_LENGTH rather than 4.
9132 (*jump_mips16, tls_get_tp_mips16_<mode>)
9133 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
9135 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9137 * config/mips/mips.md (extended_mips16): Remove branch case.
9138 (length): Remove duplicated extended_mips16 test.
9140 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9142 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
9144 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
9146 * recog.h (Recog_data): Rename to...
9147 (recog_data_d): ...this.
9148 (recog_data): Update accordingly.
9149 * recog.c (recog_data): Likewise.
9150 * reload.c (save_recog_data): Likewise.
9151 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
9152 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
9154 2013-05-17 Julian Brown <julian@codesourcery.com>
9156 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
9157 found in a REG_EQUAL note, invalidate it.
9159 2013-05-17 Easwaran Raman <eraman@google.com>
9161 * tree-ssa-reassoc.c (find_insert_point): New function.
9162 (insert_stmt_after): Likewise.
9163 (get_def_stmt): Likewise.
9164 (ensure_ops_are_available): Likewise.
9165 (not_dominated_by): Likewise.
9166 (rewrite_expr_tree): Do not move statements beyond what is
9167 necessary. Remove call to swap_ops_for_binary_stmt...
9168 (reassociate_bb): ... and move it here.
9169 (build_and_add_sum): Assign UIDs for new statements.
9170 (linearize_expr): Likewise.
9171 (do_reassoc): Renumber gimple statement UIDs.
9173 2013-05-17 Jan Hubicka <jh@suse.cz>
9175 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
9177 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
9178 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
9180 (output_weakrefs): Update.
9182 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
9183 Martin Jambor <mjambor@suse.cz>
9186 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
9187 value that corresponds to the given aggval is found in values vector.
9189 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
9191 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
9192 sse, sse2, sse3, ssse3 and sse4a flags to options.
9194 2013-05-17 David Malcolm <dmalcolm@redhat.com>
9196 * gengtype-state.c: (s_expr_writer): New class, to handle
9197 prettifying of output layout of s-expressions.
9198 (state_writer): New class, to write out gtype.state.
9199 (state_written_type_count): Move this variable into member data of
9201 (s_expr_writer::s_expr_writer): New code: constructor for new class
9202 (state_writer::state_writer(): ditto
9203 (s_expr_writer::write_new_line): New function
9204 (s_expr_writer::write_any_indent): ditto
9205 (s_expr_writer::begin_s_expr): ditto
9206 (s_expr_writer::end_s_expr): ditto
9207 (write_state_fileloc): convert to method of state_writer...
9208 (state_writer:: write_state_fileloc): ...and use methods of
9209 s_expr_writer to write indentation into the gtype.state output file
9210 to visually represent the hierarchical structure of the list
9212 (write_state_fields): ditto, renaming to...
9213 (state_writer::write_state_fields)
9214 (write_state_a_string): ditto, renaming to...
9215 (state_writer::write_state_a_string)
9216 (write_state_string_option): ditto, renaming to...
9217 (state_writer::write_state_string_option)
9218 (write_state_type_option): ditto, renaming to...
9219 (state_writer::write_state_type_option)
9220 (write_state_nested_option): ditto, renaming to...
9221 (state_writer::write_state_nested_option)
9222 (write_state_option): ditto, renaming to...
9223 (state_writer::write_state_option)
9224 (write_state_options): ditto, renaming to...
9225 (state_writer::write_state_options)
9226 (write_state_lang_bitmap): ditto, renaming to...
9227 (state_writer::write_state_lang_bitmap)
9228 (write_state_version): ditto, renaming to...
9229 (state_writer::write_state_version)
9230 (write_state_scalar_type): ditto, renaming to...
9231 (state_writer::write_state_scalar_type)
9232 (write_state_string_type): ditto, renaming to...
9233 (state_writer::write_state_string_type)
9234 (write_state_undefined_type): ditto, renaming to...
9235 (state_writer::write_state_undefined_type)
9236 (write_state_struct_union_type): ditto, renaming to...
9237 (state_writer::write_state_struct_union_type)
9238 (write_state_struct_type): ditto, renaming to...
9239 (state_writer::write_state_struct_type)
9240 (write_state_user_struct_type): ditto, renaming to...
9241 (state_writer::write_state_user_struct_type)
9242 (write_state_lang_struct_type): ditto, renaming to...
9243 (state_writer::write_state_lang_struct_type)
9244 (write_state_param_struct_type): ditto, renaming to...
9245 (state_writer::write_state_param_struct_type)
9246 (write_state_pointer_type): ditto, renaming to...
9247 (state_writer::write_state_pointer_type)
9248 (write_state_array_type): ditto, renaming to...
9249 (state_writer::write_state_array_type)
9250 (write_state_gc_used): ditto, renaming to...
9251 (state_writer::write_state_gc_used)
9252 (write_state_common_type_content): ditto, renaming to...
9253 (state_writer::write_state_common_type_content)
9254 (write_state_type): ditto, renaming to...
9255 (state_writer::write_state_type)
9256 (write_state_pair_list): ditto, renaming to...
9257 (state_writer::write_state_pair_list)
9258 (write_state_pair): ditto, renaming to...
9259 (state_writer::write_state_pair)
9260 (write_state_typedefs): ditto, renaming to...
9261 (state_writer::write_state_typedefs)
9262 (write_state_structures): ditto, renaming to...
9263 (state_writer::write_state_structures)
9264 (write_state_param_structs): ditto, renaming to...
9265 (state_writer::write_state_param_structs)
9266 (write_state_variables): ditto, renaming to...
9267 (state_writer::write_state_variables)
9268 (write_state_srcdir): ditto, renaming to...
9269 (state_writer::write_state_srcdir)
9270 (write_state_files_list): ditto, renaming to...
9271 (state_writer::write_state_files_list)
9272 (write_state_languages): ditto, renaming to...
9273 (state_writer::write_state_languages)
9274 (write_state): create a state_writer instance and use it when
9275 writing out the state file
9277 2013-05-17 Mike Stump <mikestump@comcast.net>
9279 PR rtl-optimization/57304
9280 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
9281 accessing DF_REF_REAL_LOC.
9283 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9285 PR rtl-optimization/57281
9286 PR rtl-optimization/57300
9287 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
9288 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
9289 what the other splitter did if the registers are dead.
9291 2013-05-17 Richard Biener <rguenther@suse.de>
9293 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
9296 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9298 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
9301 2013-05-17 Marek Polacek <polacek@redhat.com>
9303 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
9304 length when doing non-zero store of storing '\0' to '\0'.
9306 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9308 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
9309 vect_external_def oprnd1 with loop_vinfo, try to emit
9310 optional cast, negation and and stmts on the loop preheader
9311 edge instead of into the pattern def seq.
9313 PR tree-optimization/57051
9314 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
9315 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
9317 2013-05-16 Nick Clifton <nickc@redhat.com>
9319 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
9320 (rl78_is_naked_func): New function.
9321 (rl78_expand_prologue): Skip prologue generation for naked functions.
9322 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
9323 * doc/extend.texi (naked): Add RL78 to the list of processors
9324 that supports this attribute.
9326 2013-05-16 Jeff Law <law@redhat.com>
9328 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
9330 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
9332 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
9333 cache parameters using detect_caches_amd also for CYRIX,
9334 NSC and TM2 signatures.
9336 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
9337 Dzianis Kahanovich <mahatma@eu.by>
9341 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9342 VIA/Centaur processors and determine their cache parameters
9343 using detect_caches_amd.
9345 2013-05-16 Teresa Johnson <tejohnson@google.com>
9347 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
9348 (rtl_verify_edges): New function.
9349 (rtl_verify_bb_insns): Ditto.
9350 (rtl_verify_bb_pointers): Ditto.
9351 (rtl_verify_bb_insn_chain): Ditto.
9352 (rtl_verify_fallthru): Ditto.
9353 (rtl_verify_bb_layout): Ditto.
9354 (rtl_verify_flow_info_1): Outline checks into new functions.
9355 (rtl_verify_flow_info): Ditto.
9357 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
9359 * cfghooks.c (copy_bbs): Add update_dominance argument.
9360 * cfghooks.h (copy_bbs): Update prototype.
9361 * tree-cfg.c (gimple_duplicate_sese_region):
9362 Add update_dominance argument.
9363 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
9364 * tree-ssa-loop-ch.c (copy_loop_headers): Update
9365 gimple_duplicate_sese_region call.
9366 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
9367 Update copy_bbs call.
9368 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
9369 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
9371 2013-05-16 Jakub Jelinek <jakub@redhat.com>
9373 * tree-vectorizer.h (NUM_PATTERNS): Increment.
9374 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
9375 vect_recog_rotate_pattern.
9376 (vect_recog_rotate_pattern): New function.
9378 2013-05-16 Jason Merrill <jason@redhat.com>
9380 * Makefile.in (LLINKER): New variable.
9381 (mostlyclean): Remove link mutex.
9382 * configure.ac: Handle --enable-link-mutex.
9383 * lock-and-run.sh: New script.
9385 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9388 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
9391 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9393 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
9395 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
9397 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
9398 * config/arm/arm.c (next_consecutive_mem): New function.
9399 (gen_movmem_ldrd_strd): Likewise.
9400 * config/arm/arm.md (movmemqi): Update condition and code.
9401 (unaligned_loaddi, unaligned_storedi): New patterns.
9403 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9405 * config.gcc: Obsolete *-*-solaris2.9*.
9406 * doc/install.texi (Specific, *-*-solaris2*): Document it.
9408 2013-05-16 Richard Biener <rguenther@suse.de>
9410 * passes.c (init_optimization_passes): Move pass_parallelize_loops
9411 earlier, after GRAPHITE transforms and IV canonicalization.
9413 2013-05-16 Jakub Jelinek <jakub@redhat.com>
9415 * omp-low.c (extract_omp_for_data): For collapsed loops,
9416 if at least one of the loops is known at compile time to
9417 iterate zero times, set count to 0.
9418 (expand_omp_regimplify_p): New function.
9419 (expand_omp_for_generic): For collapsed loops, if at least
9420 one of the loops isn't known to iterate at least once,
9421 add runtime check with setting count to 0.
9422 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9423 For unsigned types if it isn't known at compile time that
9424 the loop will iterate at least once, add runtime check to bypass
9425 the whole loop if initial condition isn't true.
9427 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
9429 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9431 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
9434 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9435 transformations to avoid an infinite loop.
9437 2013-05-16 Marek Polacek <polacek@redhat.com>
9439 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9441 2013-05-15 Leif Ekblad <leif@rdos.net>
9443 * config/i386/i386.c (ix86_decompose_address): Use
9444 DEFAULT_TLS_SEG_REG to access TLS segment register.
9445 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9446 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9447 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9449 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
9452 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9453 sibling calls to functions that would normally be lazily bound,
9454 unless $gp is call-clobbered.
9456 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
9458 * config/i386/i386.c (ix86_option_override_internal): Update
9459 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
9460 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9461 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9462 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9464 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9466 2013-05-15 Andreas Schwab <schwab@suse.de>
9468 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9469 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
9472 2013-05-15 Teresa Johnson <tejohnson@google.com>
9474 * loop-unroll.c (report_unroll_peel): Check decision before
9475 emitting unroll/peel message.
9477 2013-05-15 Teresa Johnson <tejohnson@google.com>
9479 * function.h (has_bb_partition): New rtl_data flag.
9480 (bb_reorder_complete): Ditto.
9481 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9482 instead of flag_reorder_blocks_and_partition.
9483 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9484 with some enhancements.
9485 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9486 * bb-reorder.c (connect_traces): Check for has_bb_partition
9487 instead of flag_reorder_blocks_and_partition.
9488 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9489 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9490 verify_hot_cold_block_grouping.
9491 (partition_hot_cold_basic_blocks): Set has_bb_partition.
9493 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9496 * config/arm/predicates.md (call_insn_operand): New predicate.
9497 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
9498 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9499 if insn is not a tail call.
9500 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9502 * config/arm/arm.h (enum reg_class): New caller save register class.
9503 (REG_CLASS_NAMES): Likewise.
9504 (REG_CLASS_CONTENTS): Likewise.
9505 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9508 2013-05-15 Richard Biener <rguenther@suse.de>
9510 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9511 of MSG_OPTIMIZED_LOCATIONS.
9512 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9513 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9515 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9516 of MSG_OPTIMIZED_LOCATIONS.
9517 (execute_vect_slp): Likewise.
9518 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9519 (vect_create_cond_for_alias_checks): Likewise.
9520 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9521 (vect_recog_widen_mult_pattern): Likewise.
9522 (vect_recog_widen_sum_pattern): Likewise.
9523 (vect_recog_over_widening_pattern): Likewise.
9524 (vect_recog_widen_shift_pattern): Likewise.
9525 (vect_recog_vector_vector_shift_pattern): Likewise.
9526 (vect_recog_divmod_pattern): Likewise.
9527 (vect_recog_mixed_size_cond_pattern): Likewise.
9528 (vect_recog_bool_pattern): Likewise.
9529 (vect_pattern_recog_1): Likewise.
9531 2013-05-15 Martin Jambor <mjambor@suse.cz>
9533 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9534 non-functions to builtin_unreachable.
9535 * ipa-inline-transform.c (inline_call): Do not assert estimates were
9536 correct when new direct edges were discovered.
9538 2013-05-15 Martin Jambor <mjambor@suse.cz>
9540 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9541 header, print symbol order instead of node uid, print more information
9542 about indirect edge targets.
9543 (ipa_make_edge_direct_to_target): Print symbol order instead of node
9545 (ipa_make_edge_direct_to_target): Likewise.
9546 (remove_described_reference): Likewise.
9547 (propagate_controlled_uses): Likewise.
9548 (ipa_print_node_params): Also print symbol order.
9549 (ipcp_transform_function): Print symbol order instead of node uids.
9550 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9551 (cgraph_get_create_real_symbol_node): Likewise.
9552 * ipa-cp.c (print_lattice): Likewise.
9553 (print_all_lattices): Likewise.
9554 (determine_versionability): Likewise.
9555 (initialize_node_lattices): Likewise.
9556 (estimate_local_effects): Likewise.
9557 (update_profiling_info): Likewise.
9558 (create_specialized_node): Likewise.
9559 (perhaps_add_new_callers): Likewise.
9560 (decide_about_value): Likewise.
9561 (decide_whether_version_node): Likewise.
9562 (identify_dead_nodes): Likewise.
9563 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9564 (dump_inline_summary): Likewise.
9565 (estimate_node_size_and_time): Likewise.
9566 (inline_analyze_function): Likewise.
9567 * ipa-inline.c (report_inline_failed_reason): Likewise.
9568 (want_early_inline_function_p): Likewise.
9569 (edge_badness): Likewise.
9570 (update_edge_key): Likewise.
9571 (inline_small_functions): Likewise. Add dumping of order to two other
9573 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9574 instead of node uids.
9575 (propagate_pure_const): Likewise.
9576 (propagate_pure_const): Likewise.
9577 * ipa-utils.c (dump_cgraph_node_set): Likewise.
9578 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9579 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9581 * tree-pretty-print.c (dump_function_header): Likewise.
9582 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9583 Print symbol order instead of node uids.
9585 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9587 * config/s390/s390.c (s390_register_move_cost): Don't impose the
9588 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9590 2013-05-15 Richard Biener <rguenther@suse.de>
9592 PR tree-optimization/57275
9593 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9594 return value for fail to do runtime alias checks for gather loads.
9596 2013-05-15 Jan Hubicka <jh@suse.cz>
9600 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9601 weakrefs are not external.
9602 (lto_symtab_merge_decls): Fix thinko when dealing with
9603 non-lto_symtab decls.
9604 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9605 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9606 * varpool.c (dump_varpool_node): Dump more flags.
9608 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
9610 * config/i386/i386.c (processor_alias_table): Add instruction
9611 FSGSBASE for AMD bdver3 architecture.
9613 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9615 * tree.c (warn_deprecated_use): Print file:line using locus color.
9616 * diagnostic.c (diagnostic_report_current_module): Print file:line
9617 and file:line:column using locus color.
9619 2013-05-14 Mike Stump <mikestump@comcast.net>
9621 * gdbinit.in: Add __null.
9623 2013-05-14 Mike Stump <mikestump@comcast.net>
9625 * recog.h: Rename struct recog_data to Recog_data.
9626 * recog.c: Likewise.
9627 * reload.c (can_reload_into): Likewise.
9628 * config/picochip/picochip.c: Likewise.
9630 2013-05-14 Mike Stump <mikestump@comcast.net>
9632 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9634 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
9636 * resource.h (struct resources): Remove unch_memory member.
9637 (CLEAR_RESOURCE): Don't clear unch_memory.
9638 * resource.c (mark_referenced_resources): Don't set it.
9639 (mark_set_resources): Likewise.
9640 (mark_target_live_regs): Don't clear it.
9641 (init_resource_info): Likewise.
9642 * reorg.c (resource_conflicts_p): Don't compare it.
9643 (redundant_insn): Don't set it.
9645 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9647 * emit-rtl.c (next_label): Remove unused function.
9648 (skip_consecutive_labels, link_cc0_insns): Move to ...
9649 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9650 only place where these functions are used, and make them static.
9652 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
9654 * fold-const.c (fold_negate_expr): Handle vectors.
9655 (fold_truth_not_expr): Make it static.
9656 (fold_invert_truthvalue): New static function.
9657 (invert_truthvalue_loc): Handle vectors. Do not call
9658 fold_truth_not_expr directly.
9659 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9660 <TRUTH_NOT_EXPR>: Do not cast to boolean.
9661 (fold_comparison): Handle vector constants.
9662 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9663 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9664 * tree.h (fold_truth_not_expr): Remove declaration.
9666 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
9668 * config/aarch64/aarch64-simd.md
9669 (aarch64_vcond_internal<mode>): Rename to...
9670 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9671 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9672 float modes. Clarify all iterator modes.
9673 (vcond<mode><mode>): Use new name for vcond expanders.
9674 (vcond<v_cmp_result><mode>): Likewise.
9675 (vcondu<mode><mode>: Likewise.
9676 * config/aarch64/iterators.md (VDQF_COND): New.
9678 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
9681 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9682 variable for the shift amount. Check that we shift by non-negative
9685 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
9688 * config/arm/arm.h (EPILOGUE_USES): Only return true
9689 for LR_REGNUM after epilogue_completed.
9691 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
9693 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9696 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9699 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9700 and Solaris 11+/x86 with gld.
9701 * configure: Regenerate.
9703 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9705 * expmed.c (expand_shift_1): Canonicalize rotates by
9706 constant bitsize / 2 to bitsize - 1.
9707 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9708 case ROTATERT>: Likewise.
9711 2013-05-10 Jakub Jelinek <jakub@redhat.com>
9713 * config/i386/i386.md (rotateinv): New code attr.
9714 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9715 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9716 roll $31, %eax, etc.
9718 2013-05-14 Richard Biener <rguenther@suse.de>
9721 * tree-eh.c (sink_clobbers): Give up for successors with
9722 multiple predecessors and no virtual uses.
9724 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
9726 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9727 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9729 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9732 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9733 the case when both op0 and op1 have VOIDmode.
9735 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
9737 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9738 in multiply-accumulate mode.
9740 2013-05-13 Guozhi Wei <carrot@google.com>
9742 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9744 2013-05-13 Kai Tietz <ktietz@redhat.com>
9747 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9748 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9749 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9750 * config/i386/i386.c (ix86_option_override_internal): Likewise.
9751 (ix86_expand_prologue): Likewise.
9752 (ix86_expand_split_stack_prologue): Likewise.
9753 (legitimate_pic_address_disp_p): Likewise.
9754 (legitimize_pic_address): Likewise.
9755 (legitimize_tls_address): Likewise.
9756 (legitimize_pe_coff_symbol): Likewise.
9757 (output_pic_addr_const): Likewise.
9758 (construct_plt_address): Likewise.
9759 (ix86_expand_call): Likewise.
9760 (x86_output_mi_thunk): Likewise.
9761 (x86_function_profiler): Likewise.
9763 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
9765 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9766 similar switch cases.
9767 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9768 (aarch64_simd_mov_to_<mode>low): Delete.
9769 (aarch64_simd_mov_to_<mode>high): Delete.
9770 (move_lo_quad_<mode>): Add w<-r alternative.
9771 (aarch64_simd_move_hi_quad_<mode>): Likewise.
9772 (aarch64_simd_mov_from_*): Update type attribute.
9773 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9776 2013-05-13 Jan Hubicka <jh@suse.cz>
9778 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9779 * config/i386/i386.c (ix86_compute_frame_layout,
9780 ix86_expand_epilogue, emit_i387_cw_initialization,
9781 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9782 ix86_local_alignment): Fix use of size/speed predicates.
9784 2013-05-13 Jakub Jelinek <jakub@redhat.com>
9786 PR tree-optimization/45216
9787 PR tree-optimization/57157
9788 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9789 the (-Y) & (B - 1) variant if OP is |.
9790 * expmed.c (expand_shift_1): For rotations by const0_rtx just
9791 return shifted. Use (-op1) & (prec - 1) as other_amount
9792 instead of prec - op1.
9794 2013-05-13 Martin Jambor <mjambor@suse.cz>
9797 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9798 (ipa_constant_data): New type.
9799 (ipa_jump_func): Use ipa_constant_data to hold information about
9800 constant jump functions.
9801 (ipa_get_jf_constant): Adjust to jump function type changes.
9802 (ipa_get_jf_constant_rdesc): New function.
9803 (ipa_param_descriptor): New field controlled_uses.
9804 (ipa_get_controlled_uses): New function.
9805 (ipa_set_controlled_uses): Likewise.
9806 * ipa-ref.h (ipa_find_reference): Declare.
9807 * ipa-prop.c (ipa_cst_ref_desc): New type.
9808 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9810 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
9811 New parameter cs. Adjust all callers.
9812 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9813 (remove_described_reference): New function.
9814 (jfunc_rdesc_usable): Likewise.
9815 (try_make_edge_direct_simple_call): Decrement controlled use count,
9816 attempt to remove reference if it hits zero.
9817 (combine_controlled_uses_counters): New function.
9818 (propagate_controlled_uses): Likewise.
9819 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9820 (ipa_edge_duplication_hook): Duplicate reference descriptions.
9821 (ipa_print_node_params): Print described use counter.
9822 (ipa_write_jump_function): Adjust to jump function type changes.
9823 (ipa_read_jump_function): New parameter CS, pass it to
9824 ipa_set_jf_constant. Adjust caller.
9825 (ipa_write_node_info): Stream controlled use count
9826 (ipa_read_node_info): Likewise.
9827 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9829 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9830 count. Remove cloning-added reference if it reaches zero.
9831 * ipa-ref.c (ipa_find_reference): New function.
9833 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
9835 * config/i386/i386.c (processor_target_table): Modified default
9836 alignment values for AMD BD and BT architectures.
9838 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
9840 * tree-vect-generic.c (uniform_vector_p): Move ...
9841 * tree.c (uniform_vector_p): ... here.
9842 * tree.h (uniform_vector_p): Declare it.
9843 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9846 2013-05-13 Jakub Jelinek <jakub@redhat.com>
9848 PR tree-optimization/57230
9849 * tree-ssa-strlen.c (handle_char_store): Record length for
9850 array store from STRING_CST.
9852 PR tree-optimization/57230
9853 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9856 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
9858 * config/epiphany/epiphany.c (epiphany_init): Check size of
9859 NUM_MODES_FOR_MODE_SWITCHING.
9860 (epiphany_expand_prologue):
9861 Remove CONFIG_REGNUM initial value handling code.
9862 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9863 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9864 (emit_set_fp_mode, epiphany_mode_after): Likewise.
9865 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9866 Don't return 1 for FP_MODE_NONE.
9867 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9868 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9869 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9870 * config/epiphany/epiphany.md (save_config): New pattern.
9872 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
9874 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9876 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
9878 * config/i386/i386.md (memory): Handle sseishft1.
9879 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9880 (*vec_extractv2di_1): Ditto.
9882 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
9884 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9887 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
9889 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9890 Add mthumb/march=armv7-a multilib.
9891 Add mthumb/march=armv7-r multilib.
9892 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9894 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
9896 * config/v850/t-rtems: Add more multilibs.
9898 2013-05-10 Richard Biener <rguenther@suse.de>
9900 PR tree-optimization/57214
9901 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9902 not propagate from SSA names that occur in abnormal PHI nodes.
9904 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
9906 * stor-layout.c (element_precision): New function.
9907 * machmode.h (element_precision): Declare it.
9908 * tree.c (build_minus_one_cst): New function.
9909 (element_precision): Likewise.
9910 * tree.h (build_minus_one_cst): Declare new function.
9911 (element_precision): Likewise.
9912 * fold-const.c (operand_equal_p): Use element_precision.
9913 (fold_binary_loc): Handle vector types.
9914 * convert.c (convert_to_integer): Use element_precision.
9915 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9918 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
9920 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9921 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9922 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9923 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9924 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9925 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9926 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9927 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9928 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9929 (Uuw8): New constraints.
9930 (Usb4): Move into alphabetical order.
9931 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9932 (sd8_operand, ub8_operand, uw8_operand): New predicates.
9933 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9934 previously unnamed patterns.
9935 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9936 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9937 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9938 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9939 of set_attr_alternative/if_then_else. Use extended_mips16 instead
9940 of specific lengths.
9942 2013-05-10 Jakub Jelinek <jakub@redhat.com>
9944 * config/i386/i386.md (rotateinv): New code attr.
9945 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9946 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9947 roll $31, %eax, etc.
9949 PR tree-optimization/45216
9950 PR tree-optimization/57157
9951 * tree-ssa-forwprop.c (simplify_rotate): New function.
9952 (ssa_forward_propagate_and_combine): Call it.
9954 2013-05-10 Richard Biener <rguenther@suse.de>
9956 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9957 disable peeling when we version for aliasing.
9958 (vector_alignment_reachable_p): Honor explicit user alignment.
9959 (vect_supportable_dr_alignment): Likewise.
9960 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9961 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9962 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9963 then peeling to arrange for the cost-model check to come first.
9965 2013-05-10 Alan Modra <amodra@gmail.com>
9967 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9968 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9969 * configure: Regenerate.
9971 2013-05-10 Alan Modra <amodra@gmail.com>
9974 * varasm.c (default_elf_select_section): Move !DECL_P check..
9975 (get_named_section): ..to here before calling get_section_name.
9977 (default_section_type_flags): Add DECL_P check.
9978 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9979 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9981 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
9983 * config/epiphany/epiphany.c (epiphany_expand_prologue):
9984 When using gen_stack_adjust_str with a register offset, add a
9985 REG_FRAME_RELATED_EXPR note.
9987 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
9989 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9990 (*vec_extractv4si_zext_mem): Ditto.
9991 (*vec_extractv2di): Add 0->x and x->x alternatives.
9992 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9993 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9995 2013-05-09 Jason Merrill <jason@redhat.com>
9997 N3639 C++1y VLA support
9998 * gimplify.c (gimplify_vla_decl): Don't touch an existing
10001 * tree.c (build_constructor_va): New.
10002 * tree.h: Declare it.
10004 2013-05-09 Martin Jambor <mjambor@suse.cz>
10007 * gimple-fold.c (canonicalize_constructor_val): Call
10008 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
10010 2013-05-09 Jan Hubicka <jh@suse.cz>
10011 Richard Biener <rguenther@suse.de>
10014 * symtab.c (symtab_make_decl_local): Do not add private names.
10016 2013-05-09 Jan Hubicka <jh@suse.cz>
10019 * symtab.c (insert_to_assembler_name_hash): Handle clones.
10020 (unlink_from_assembler_name_hash): Likewise.
10021 (symtab_prevail_in_asm_name_hash, symtab_register_node,
10022 symtab_unregister_node, symtab_initialize_asm_name_hash,
10023 change_decl_assembler_name): Update.
10025 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
10027 * config/aarch64/aarch64.md: New movtf split.
10028 (*movtf_aarch64): Update.
10029 (aarch64_movdi_tilow): Handle TF modes and rename to
10030 aarch64_movdi_<mode>low.
10031 (aarch64_movdi_tihigh): Handle TF modes and rename to
10032 aarch64_movdi_<mode>high
10033 (aarch64_movtihigh_di): Handle TF modes and rename to
10034 aarch64_mov<mode>high_di
10035 (aarch64_movtilow_di): Handle TF modes and rename to
10036 aarch64_mov<mode>low_di
10037 (aarch64_movtilow_tilow): Remove spurious whitespace.
10038 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
10040 (aarch64_print_operand): Update.
10042 2013-05-09 Alan Modra <amodra@gmail.com>
10044 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
10046 * configure: Regenerate.
10048 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
10050 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
10051 splitter preparation statements.
10052 * config/i386/sse.md (*vec_extract* splitters): Ditto.
10053 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
10056 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10058 * gimple-ssa-strength-reduction.c (count_candidates): Change
10059 return value to int.
10060 (analyze_candidates_and_replace): Change type of length to int.
10062 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
10064 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
10065 (*vec_extract<mode>): Use VI12_128 mode iterator.
10066 (*vec_extract<mode>_mem): Ditto.
10067 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
10070 2013-05-08 Diego Novillo <dnovillo@google.com>
10075 2012-08-17 Diego Novillo <dnovillo@google.com>
10078 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
10079 * config.in: Regenerate.
10080 * configure: Regenerate.
10081 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
10083 2013-05-08 Jan Hubicka <jh@suse.cz>
10086 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
10087 * cgraph.h (symtab_node_base): Add unique_name.
10088 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10089 input_overwrite_node, input_varpool_node): Stream unique_name.
10090 * cgraphclones.c (cgraph_create_virtual_clone,
10091 cgraph_function_versioning): Set unique_name.
10092 * ipa.c (function_and_variable_visibility): Set unique_name.
10094 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10096 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
10097 (alloc_cand_and_find_basis): Restrict conditional candidate
10098 processing to CAND_MULTs.
10100 2013-05-08 Jan Hubicka <jh@suse.cz>
10103 lto-symtab.c (lto_symtab_symbol_p): New function.
10104 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
10105 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
10106 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
10107 Skip static symbols.
10109 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
10111 PR tree-optimization/57200
10112 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
10113 Only call inform if the preceding warning_at returns true.
10115 2013-05-07 Han Shen <shenhan@google.com>
10117 * cfgexpand.c (record_or_union_type_has_array_p): New function.
10118 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
10119 * common.opt (fstack-protector-strong): New option.
10120 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
10121 * doc/invoke.texi (Optimization Options): Document
10122 "-fstack-protector-strong".
10123 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
10125 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
10127 * config/mips/mips.c (mips_machine_reorg2): Return 0.
10129 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
10131 * ira.c (update_equiv_regs): Add insn having equiv memory even if
10132 it is not lhs of the insn.
10133 (setup_reg_equiv): Remove insn having equiv memory which it is not
10135 * lra-constraints.c (process_address): Try to improve generation
10136 code for address base + disp.
10137 (lra_constraints): Make correct the code for checking insn setting
10138 up backward equivalence. Remove insn only if it is in the init
10140 * lra-eliminations.c (update_reg_eliminate): Change return value.
10141 (lra_eliminate): Use the result.
10143 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
10145 * config/i386/sse.md (ssescalarnummask): New mode attribute.
10146 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
10147 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
10148 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
10149 register target operands.
10150 (*vec_extractv8hi_sse2): New pattern.
10151 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
10152 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
10153 (*vec_extract<mode>_mem): New insn and split pattern.
10155 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
10157 * config/arm/arm.c (arm_asan_shadow_offset): New function.
10158 (TARGET_ASAN_SHADOW_OFFSET): Define.
10159 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
10160 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
10162 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10164 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
10165 (incr_vec_index): Return -1 if increment not found.
10166 (create_add_on_incoming_edge): Assert if increment not found.
10167 (record_increment): Limit number of increments recorded.
10168 (all_phi_incrs_profitable): Return false if an increment not found.
10169 (replace_profitable_candidates): Don't process increments that were
10171 (analyze_candidates_and_replace): Limit size of incr_vec.
10173 2013-05-07 Richard Biener <rguenther@suse.de>
10175 * calls.c (special_function_p): setjmp-like functions are leaf.
10176 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
10177 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
10179 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
10181 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
10182 (aarch64_simd_mov<mode>): New expander.
10183 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
10184 (aarch64_simd_mov_to_<mode>high): Likewise.
10185 (aarch64_simd_mov_from_<mode>low): Likewise.
10186 (aarch64_simd_mov_from_<mode>high): Likewise.
10187 (aarch64_dup_lane<mode>): Update.
10188 (aarch64_dup_lanedi): New instruction pattern.
10189 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
10190 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
10192 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10194 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
10195 (replace_mult_candidate): Remove unnecessary argument; remove
10196 unnecessary parameter from call to introduce_cast_before_cand.
10197 (replace_unconditional_candidate): Remove unnecessary parameter
10198 from call to replace_mult_candidate.
10199 (replace_conditional_candidate): Likewise.
10200 (insert_initializers): Use make_temp_ssa_name.
10201 (introduce_cast_before_cand): Remove unnecessary argument; use
10202 make_temp_ssa_name.
10203 (replace_one_candidate): Remove unnecessary argument; remove
10204 unnecessary parameter from calls to introduce_cast_before_cand.
10205 (replace_profitable_candidates): Remove unnecessary parameters
10206 from calls to replace_one_candidate.
10208 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10210 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10211 phi def as possibly hiding a basis for a CAND_ADD whose operands
10212 have been commuted in the analysis.
10213 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
10215 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10217 * config/aarch64/aarch64.md
10218 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
10219 shift value between 0-4.
10221 2013-05-07 Richard Biener <rguenther@suse.de>
10223 * double-int.h (rshift): New overload.
10224 * double-int.c (rshift): New function.
10225 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
10226 (create_reference_ops_from_ref): Remove.
10227 (vn_reference_insert): Use shared ops for constructing the
10228 reference and copy it.
10230 2013-05-07 Richard Biener <rguenther@suse.de>
10232 PR middle-end/57190
10233 * tree-eh.c (sink_clobbers): Properly propagate
10234 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
10236 2013-05-07 Jakub Jelinek <jakub@redhat.com>
10238 PR tree-optimization/57149
10239 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
10240 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
10241 collect_phi_def_edges, execute_late_warn_uninitialized): Use
10242 uninit_undefined_value_p instead of ssa_undefined_value_p.
10245 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
10246 for modifier == EXPAND_INITIALIZER.
10248 2013-05-07 Anton Blanchard <anton@samba.org>
10250 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
10251 for powerpc64 little endian.
10252 * configure: Regenerate.
10254 2013-05-06 Graham Stott <grahams@btinternet.com>
10256 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
10257 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
10258 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
10259 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
10261 2013-05-06 Graham Stott <grahams@btinternet.com>
10263 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
10264 codes which allow non-lvalues.
10266 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
10268 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
10269 components are all 1s.
10270 (integer_minus_onep): New function.
10271 * tree.h (integer_minus_onep): Declare it.
10272 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
10273 integer_minus_onep instead of integer_all_onesp.
10275 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10278 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
10279 variations of these patterns.
10281 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
10283 * config/i386/i386.md (isa): Add x64_sse4 member.
10284 (enabled): Handle x64_sse4.
10285 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
10286 instruction for 64bit SSE4_1 targets. Update insn attributes.
10287 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
10288 instruction for SSE4_1 targets. Update insn attributes.
10289 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10290 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
10291 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
10293 (*vec_extractv4si): Rename from *sse4_1_pextrd.
10294 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
10295 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
10297 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10300 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
10302 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
10304 * final.c (do_assembler_dialects): Don't handle curly braces and
10305 vertical bar escaped by % as dialect delimiters.
10306 (output_asm_insn): Print curly braces and vertical bar if escaped
10307 by % and ASSEMBLER_DIALECT defined.
10308 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
10309 * doc/tm.texi: Regenerated.
10311 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
10313 * config/mips/mips.c: Include tree-pass.h.
10314 (mips_reorg): Split in pre- and post-dbr_schedule parts.
10315 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
10316 (pass_mips_machine_reorg2): New machine specific pass.
10317 (insert_pass_mips_machine_reorg2): New pass plugin definition.
10318 (mips_option_override): Register the new pass.
10319 * rtl.h (cleanup_barriers): Remove prototype.
10320 (dbr_schedule): Likewise.
10321 * jump.c (cleanup_barriers): Make static.
10322 * reorg.c (dbr_schedule): Likewise.
10324 2013-05-06 Richard Biener <rguenther@suse.de>
10326 PR tree-optimization/57185
10327 * tree-parloops.c (add_field_for_reduction): Handle anonymous
10328 SSA names properly.
10330 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
10333 * config/i386/i386.c (add_parameter_dependencies): Add dependence
10334 between "first_arg" and "insn", not "last" and "insn".
10336 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10338 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
10339 (find_candidates_in_block): Re-enable slsr_process_phi.
10340 (create_phi_basis): Fix double counting of candidate adjustment.
10342 2013-05-06 Richard Biener <rguenther@suse.de>
10344 PR middle-end/57147
10345 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
10346 the edge is also fallthru, preserve it and just clear the
10348 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
10349 also complex, preserve that and just clear the fallthru flag.
10350 * tree-inline.c (update_ssa_across_abnormal_edges): Also
10351 update virtual operands.
10353 2013-05-06 Alan Modra <amodra@gmail.com>
10355 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
10356 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
10357 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
10358 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
10359 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10360 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
10362 2013-05-06 Alan Modra <amodra@gmail.com>
10364 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
10365 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
10366 (DEFAULT_ASM_ENDIAN): Define.
10367 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10368 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
10369 Update -K PIC clause from sysv4.h.
10370 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
10371 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
10373 2013-05-06 Alan Modra <amodra@gmail.com>
10375 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
10376 twice for little-endian.
10377 (ashrdi3_no_power, ashrdi3): Support little-endian.
10379 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10382 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
10383 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
10385 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
10386 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
10387 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
10390 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
10391 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
10393 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
10394 * config/pa/pa.opt: Make mbig-switch a no-op.
10395 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
10396 (CASE_VECTOR_MODE): Always return SImode.
10397 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
10398 for the !TARGET_BIG_SWITCH case.
10399 * config/pa/pa-linux.h: Likewise.
10400 * config/pa/pa-openbsd.h: Likewise.
10401 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
10402 * config/pa/pa.md (short_jump): Remove define_insn.
10403 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
10404 (casesi0): Remove define_insn.
10405 (type): Remove btable_branch.
10406 (pa_combine_type): Likewise.
10407 (in_nullified_branch_delay): Likewise.
10408 (in_call_delay): Likewise.
10409 (define_delay): Likewise.
10410 (define_insn_reservation "Z3"): Likewise.
10411 (define_insn_reservation "Z4"): Likewise.
10412 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10413 (pa_adjust_insn_length): Remove adjustment for btable branches.
10414 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10417 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
10419 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10420 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10421 Add m->r,x alternatives.
10422 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10423 splitters using SWI48x mode iterator.
10424 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
10425 TARGET_64BIT. Add m->x alternative.
10426 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10427 Add o->x alternative. Enable for TARGET_SSE.
10428 (sse_storeq): Remove expander.
10429 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10430 with memory input operand.
10431 (*vec_extractv2di_1 splitter): New.
10432 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10433 * config/i386/i386.md (ssevecmodelower): New mode attribute.
10435 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
10437 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
10438 (INT_LOWPART): Delete.
10439 (extract_MB): Adjust.
10440 (extract_ME): Adjust.
10441 (print_operand): Adjust.
10443 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
10445 * config/rs6000/predicates.md (reg_or_add_cint_operand,
10446 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10447 (reg_or_logical_cint_operand, easy_fp_constant,
10448 logical_const_operand): Delete "CONST_DOUBLE" case.
10449 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10450 "HOST_BITS_PER_WIDE_INT == 64" test.
10451 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
10452 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10453 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10454 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10456 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10457 CONST_DOUBLE DImode/VOIDmode case.
10458 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10459 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
10460 CONST_DOUBLE VOIDmode case.
10461 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10462 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10463 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10464 Delete CONST_DOUBLE case.
10465 (splitters for mov FMOVE64 const_double): Delete
10466 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
10467 "HOST_BITS_PER_WIDE_INT >= 64" test.
10468 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10470 (mov DI const_double): Delete.
10472 2013-05-04 Jakub Jelinek <jakub@redhat.com>
10474 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10475 on op shows all bits zero in mode of a lowpart subreg, return zero.
10477 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10480 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10481 to save TFmode registers and DImode to save TImode registers for
10482 caller save operations.
10483 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10484 mark being partially clobbered since they only use the first
10487 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10488 and TDmode only use the upper 64-bits of each VSX register.
10490 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10492 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10493 (find_candidates_in_block): Disable slsr_process_phi.
10495 2013-05-03 Guozhi Wei <carrot@google.com>
10497 * coverage.c (coverage_obj_init): Move the construction of gcov
10499 (build_init_ctor): ... here.
10501 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10503 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10504 (slsr_cand_d): Redefine def_phi.
10505 (stride_status, phi_adjust_status, count_phis_status): New enums.
10506 (find_phi_def): New.
10507 (find_basis_for_base_expr): New.
10508 (find_basis_for_candidate): Handle hidden bases.
10509 (alloc_cand_and_find_basis): Handle phi candidates.
10510 (slsr_process_phi): New.
10511 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10512 (create_mul_imm_cand): Likewise.
10513 (create_add_ssa_cand): Exclude phi base candidates.
10514 (create_add_imm_cand): Likewise.
10515 (slsr_process_cast): Likewise.
10516 (slsr_process_copy): Likewise.
10517 (find_candidates_in_block): Handle phi candidates.
10518 (dump_candidate): Likewise.
10519 (unconditional_cands): Delete.
10520 (unconditional_cands_with_known_stride_p): Delete.
10521 (phi_dependent_cand_p): New.
10522 (cand_increment): Handle phi-dependent candidates.
10523 (replace_dependent): Delete.
10524 (replace_mult_candidate): New.
10525 (replace_unconditional_candidate): New.
10526 (incr_vec_index): Move to avoid forward reference.
10527 (create_add_on_incoming_edge): New.
10528 (create_phi_basis): New.
10529 (replace_dependents): Delete.
10530 (replace_conditional_candidate): New.
10531 (phi_add_costs): New.
10532 (replace_uncond_cands_and_profitable_phis): New.
10533 (record_increment): Handle phi adjustments.
10534 (record_phi_increments): New.
10535 (record_increments): Handle phi adjustments.
10536 (phi_incr_cost): New.
10537 (lowest_cost_path): Handle phis.
10538 (total_savings): Likewise.
10539 (analyze_increments): Likewise.
10540 (ncd_with_phi): New.
10541 (ncd_of_cand_and_phis): New.
10542 (nearest_common_dominator_for_cands): Handle phi increments.
10543 (all_phi_incrs_profitable): New.
10544 (replace_profitable_candidates): Handle phi-dependent candidates.
10545 (analyze_candidates_and_replace): Likewise.
10547 2013-05-03 Teresa Johnson <tejohnson@google.com>
10550 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10551 do not exceed REG_BR_PROB_BASE.
10553 2013-05-03 Jeff Law <law@redhat.com>
10555 PR tree-optimization/57144
10556 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10557 operand of the condition will bit into the new type when eliminating
10558 a cast feeding a condition.
10560 2013-05-03 Jakub Jelinek <jakub@redhat.com>
10562 PR rtl-optimization/57130
10563 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10564 of COMPARE as in_code to the recursive call if needed.
10566 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
10568 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10569 (enabled): Handle new members.
10570 * config/i386/sse.md (*vec_concatv2si): Merge from
10571 *vec_concatv2si_sse2 and vec_concatv2si_sse.
10572 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10574 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
10576 PR tree-optimization/57027
10577 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10578 for fnms opportunity, check we got the prerequisite kind
10579 of tree / gimple before using accessor functions.
10581 2013-05-03 Richard Biener <rguenther@suse.de>
10583 * double-int.h (lshift): New overload without precision
10584 and arith argument.
10585 (operator *=, operator +=, operator -=): Move ...
10586 * double-int.c (operator *=, operator +=, operator -=): ... here
10587 and implement more efficiently.
10588 (mul_double_with_sign): Remove.
10589 (lshift_double): Adjust to take unsinged shift argument, push
10590 dispatching code to callers.
10591 (mul_double_wide_with_sign): Add early out for callers that
10592 are not interested in high parts or overflow.
10593 (lshift): New function.
10594 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10595 dispatch code here.
10596 (lrotate, rrotate): Use logical shifts.
10597 * expr.c (get_inner_reference): Use lshift.
10598 * fixed-value.c (do_fixed_divide): Likewise.
10599 * tree-dfa.c (get_ref_base_and_extent): Likewise.
10600 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10601 (indirect_refs_may_alias_p): Likewise.
10602 (stmt_kills_ref_p_1): Likewise.
10604 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
10606 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10608 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
10610 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10611 scalar form of FABD instruction.
10613 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
10615 * lra-constraints.c (process_alt_operands): Add checking alt
10616 number to choose the best alternative.
10618 2013-05-02 Richard Biener <rguenther@suse.de>
10620 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10621 bitmap and its handling.
10622 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10624 2013-05-02 Richard Biener <rguenther@suse.de>
10626 PR middle-end/57140
10627 * tree-inline.c (copy_loops): Properly handle removed loops.
10628 (copy_cfg_body): Mark destination loops for fixup if source
10629 loops needed fixup.
10631 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
10634 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10635 generating simple_return for naked functions.
10637 2013-05-02 Martin Jambor <mjambor@suse.cz>
10639 PR middle-end/56988
10640 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10641 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10643 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10644 ipa_agg_replacement_value structures.
10645 (known_aggs_to_agg_replacement_list): Likewise.
10646 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10647 (read_agg_replacement_chain): Likewise.
10648 (ipcp_transform_function): Also check that by_ref flags match.
10650 2013-05-02 Richard Biener <rguenther@suse.de>
10652 * graphds.h (struct graph): Add obstack member.
10653 * graphds.c (new_graph): Initialize obstack and allocate
10655 (add_edge): Allocate edge from the obstack.
10656 (free_graph): Free the obstack instead of all edges and vertices.
10658 2013-05-02 Teresa Johnson <tejohnson@google.com>
10660 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10662 * cfg.c (update_bb_profile_for_threading): Ditto.
10663 * tree-inline.c (copy_bb): Ditto.
10664 (copy_edges_for_bb): Ditto.
10665 (initialize_cfun): Ditto.
10666 (copy_cfg_body): Ditto.
10667 (expand_call_inline): Ditto.
10668 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10669 (estimate_node_size_and_time): Ditto.
10670 (inline_merge_summary): Ditto.
10671 * cgraphclones.c (cgraph_clone_edge): Ditto.
10672 (cgraph_clone_node): Ditto.
10673 * sched-rgn.c (compute_dom_prob_ps): Ditto.
10674 (compute_trg_info): Ditto.
10676 2013-05-02 Ian Bolton <ian.bolton@arm.com>
10678 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10679 S reg when fp attribute set.
10680 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10682 2013-05-02 Ian Bolton <ian.bolton@arm.com>
10684 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10686 (*and_one_cmplsi3_compare0_uxtw): Likewise.
10687 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10688 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10690 2013-05-02 Richard Biener <rguenther@suse.de>
10692 * tree-scalar-evolution.c (scev_info_hasher): Remove.
10693 (struct instantiate_cache_entry): New type.
10694 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10695 (struct instantiate_cache_type): New type.
10696 (set_instantiated_value, get_instantiated_value): Remove.
10697 (get_instantiated_value_entry): New function.
10698 (instantiate_scev_name): Use the new cache and adjust.
10699 (instantiate_scev_poly): Adjust.
10700 (instantiate_scev_binary): Likewise.
10701 (instantiate_array_ref): Likewise.
10702 (instantiate_scev_convert): Likewise.
10703 (instantiate_scev_not): Likewise.
10704 (instantiate_scev_3): Likewise.
10705 (instantiate_scev_2): Likewise.
10706 (instantiate_scev_r): Likewise.
10707 (instantiate_scev): Likewise.
10708 (resolve_mixers): Likewise.
10710 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
10713 * lra-constraints.c (best_small_class_operands_num): Remove.
10714 (process_alt_operands): Remove small_class_operands_num. Take
10715 small classes operands into losers and only if the operand is not
10716 matched. Modify debugging output.
10717 (curr_insn_transform): Remove best_small_class_operands_num.
10720 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10722 * config/aarch64/aarch64-builtins.c
10723 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10724 * config/aarch64/aarch64-simd-builtins.def
10725 (reduc_splus_): Add new modes.
10726 (reduc_uplus_): New.
10727 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10728 (reduc_uplus_v4sf): Likewise.
10729 (reduc_splus_v4sf): Likewise.
10730 (aarch64_addv<mode>): Likewise.
10731 (reduc_uplus_<mode>): Likewise.
10732 (reduc_splus_<mode>): Likewise.
10733 (aarch64_addvv2di): Likewise.
10734 (reduc_uplus_v2di): Likewise.
10735 (reduc_splus_v2di): Likewise.
10736 (aarch64_addvv2si): Likewise.
10737 (reduc_uplus_v2si): Likewise.
10738 (reduc_splus_v2si): Likewise.
10739 (reduc_<sur>plus_<mode>): New.
10740 (reduc_<sur>plus_v2di): Likewise.
10741 (reduc_<sur>plus_v2si): Likewise.
10742 (reduc_<sur>plus_v4sf): Likewise.
10743 (aarch64_addpv4sf): Likewise.
10744 * config/aarch64/arm_neon.h
10745 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10746 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10747 add UNSPEC_SADDV, UNSPEC_UADDV.
10749 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10751 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10753 * config/aarch64/arm_neon.h
10754 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10756 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10758 * config/aarch64/aarch64-builtins
10759 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10761 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10763 * config/aarch64/aarch64-simd-builtins.def
10764 (reduc_smax_): New.
10765 (reduc_smin_): Likewise.
10766 (reduc_umax_): Likewise.
10767 (reduc_umin_): Likewise.
10768 (reduc_smax_nan_): Likewise.
10769 (reduc_smin_nan_): Likewise.
10772 (smax): Update for V2SF, V4SF and V2DF modes.
10775 (smin_nan): Likewise.
10776 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10777 (<su><maxmin><mode>3): ...This, refactor.
10778 (s<maxmin><mode>3): New.
10779 (<maxmin_uns><mode>3): Likewise.
10780 (reduc_<maxmin_uns>_<mode>): Refactor.
10781 (reduc_<maxmin_uns>_v4sf): Likewise.
10782 (reduc_<maxmin_uns>_v2si): Likewise.
10783 (aarch64_<fmaxmin><mode>: Remove.
10784 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10786 (vmin<q>_f<32,64>): Likewise.
10787 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10789 (su): Add mappings for smax, smin, umax, umin.
10791 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10792 (FMAXMIN): Rename as...
10793 (FMAXMIN_UNS): ...This.
10795 (fmaxminv): Likewise.
10796 (fmaxmin): Likewise.
10798 (maxmin_uns_op): Likewise.
10800 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10802 * config/aarch64/arm_neon.h
10803 (vac<ge, gt><sd>_f<32, 64>): Rename to...
10804 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10805 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10807 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10809 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10810 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10812 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10814 * config/aarch64/aarch64-simd.md
10815 (vcond<mode>_internal): Handle special cases for constant masks.
10816 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10817 (vcondu<mode><mode>): Likewise.
10818 (vcond<v_cmp_result><mode>): New.
10820 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10822 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10823 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10824 * config/aarch64/aarch64-simd-builtins.def
10825 (cmeq): Update to BUILTIN_VALLDI.
10830 * config/aarch64/arm_neon.h
10831 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10832 to builtins or C as appropriate.
10834 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10836 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10838 (cmhi): Rename to...
10840 * config/aarch64/aarch64-simd.md
10841 (simd_mode): Add SF.
10842 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10843 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10844 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10845 (cstore<mode>_neg): ...This.
10846 * config/aarch64/iterators.md
10848 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10849 (COMPARISONS): New.
10850 (UCOMPARISONS): Likewise.
10851 (optab): Add missing comparisons.
10857 (VCMP_S): Likewise.
10858 (VCMP_U): Likewise.
10859 (V_cmp_result): Add DF, SF modes.
10860 (v_cmp_result): Likewise.
10862 (vmtype): Likewise.
10863 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10865 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
10867 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10868 define_insn to define_insn_and_split.
10869 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10870 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10871 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10872 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10873 (thumb2_negscc): Likewise.
10875 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
10877 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10879 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
10881 * config/arm/thumb2.md: Remove trailing whitespaces.
10883 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
10885 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10886 Use gen_int_mode rather than GEN_INT.
10888 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
10890 * value-prof.c (stream_in_histogram_value): Remove the strayed
10893 2013-04-30 Richard Biener <rguenther@suse.de>
10895 PR middle-end/57122
10896 * cfghooks.c (split_edge): Properly check for the loop latch edge.
10898 2013-04-30 Richard Biener <rguenther@suse.de>
10900 PR middle-end/57107
10901 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10903 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
10905 PR rtl-optimization/56957
10906 PR rtl-optimization/57105
10907 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10908 variable. Use just INSN_UID for determining whether an insn
10909 should be only disconnected from the insn stream.
10910 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10912 2013-04-30 Jakub Jelinek <jakub@redhat.com>
10914 PR tree-optimization/57104
10915 * tsan.c (instrument_expr): Don't instrument accesses to
10916 DECL_HARD_REGISTER VAR_DECLs.
10918 2013-04-30 Richard Biener <rguenther@suse.de>
10920 * function.h (loops_for_fn): New inline function.
10921 (set_loops_for_fn): Likewise.
10922 * cfgloop.h (place_new_loop): Add struct function parameter.
10923 (get_loop): Likewise.
10924 (get_loops): Likewise.
10925 (number_of_loops): Likewise.
10926 (fel_next): Adjust.
10927 (fel_init): Likewise.
10928 * cfg.c (get_loop_copy): Adjust.
10929 * cfgloop.c (flow_loops_dump): Likewise.
10930 (record_loop_exits): Likewise.
10931 (verify_loop_structure): Likewise.
10932 * cfgloopanal.c (mark_irreducible_loops): Likewise.
10933 (estimate_reg_pressure_cost): Likewise.
10934 (mark_loop_exit_edges): Likewise.
10935 * cfgloopmanip.c (place_new_loop): Likewise.
10936 (add_loop): Likewise.
10937 (duplicate_loop): Likewise.
10938 * graph.c (draw_cfg_nodes): Likewise.
10939 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10940 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10941 (extract_affine_chrec): Likewise.
10942 (build_scop_iteration_domain): Likewise.
10943 * graphite.c (graphite_initialize): Likewise.
10944 * ira-build.c (create_loop_tree_nodes): Likewise.
10945 (more_one_region_p): Likewise.
10946 (rebuild_regno_allocno_maps): Likewise.
10947 (mark_loops_for_removal): Likewise.
10948 (mark_all_loops_for_removal): Likewise.
10949 (remove_unnecessary_regions): Likewise.
10950 (ira_build): Likewise.
10951 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10952 * loop-init.c (fix_loop_structure): Likewise.
10953 (gate_rtl_move_loop_invariants): Likewise.
10954 (gate_rtl_unswitch): Likewise.
10955 (gate_rtl_unroll_and_peel_loops): Likewise.
10956 (rtl_doloop): Likewise.
10957 * lto-streamer-in.c (input_cfg): Likewise.
10958 * lto-streamer-out.c (output_cfg): Likewise.
10959 * modulo-sched.c (sms_schedule): Likewise.
10960 * predict.c (tree_estimate_probability): Likewise.
10961 (tree_estimate_probability_driver): Likewise.
10962 (estimate_loops): Likewise.
10963 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10964 (move_sese_region_to_fn): Likewise.
10965 (debug_loop_num): Likewise.
10966 * tree-chrec.c (chrec_evaluate): Likewise.
10967 (hide_evolution_in_other_loops_than_loop): Likewise.
10968 (chrec_component_in_loop_num): Likewise.
10969 (reset_evolution_in_loop): Likewise.
10970 (evolution_function_is_invariant_rec_p): Likewise.
10971 * tree-if-conv.c (main_tree_if_conversion): Likewise.
10972 * tree-inline.c (copy_loops): Likewise.
10973 (copy_cfg_body): Likewise.
10974 (tree_function_versioning): Likewise.
10975 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10976 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10978 (add_to_evolution_1): Likewise.
10979 (scev_const_prop): Likewise.
10980 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10981 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10982 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10983 (tree_ssa_lim_initialize): Likewise.
10984 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10985 (verify_loop_closed_ssa): Likewise.
10986 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
10987 (tree_ssa_loop_im): Likewise.
10988 (tree_ssa_loop_unswitch): Likewise.
10989 (tree_vectorize): Likewise.
10990 (check_data_deps): Likewise.
10991 (tree_ssa_loop_ivcanon): Likewise.
10992 (tree_ssa_loop_bounds): Likewise.
10993 (tree_complete_unroll): Likewise.
10994 (tree_complete_unroll_inner): Likewise.
10995 (tree_parallelize_loops): Likewise.
10996 (tree_ssa_loop_prefetch): Likewise.
10997 (tree_ssa_loop_ivopts): Likewise.
10998 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10999 * tree-vectorizer.c (vectorize_loops): Likewise.
11001 2013-04-29 Mike Frysinger <vapier@gentoo.org>
11003 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
11004 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
11005 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
11006 with EABI_LINK_SPEC.
11008 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11011 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
11014 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
11017 * lra-constraints.c (process_alt_operands): Discourage a bit more
11018 using memory for pseudos. Print cost dump for alternatives.
11019 Modify cost values for conflicts with early clobbers.
11020 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
11022 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11025 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
11027 2013-04-29 Ian Bolton <ian.bolton@arm.com>
11029 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
11030 from/to S register.
11031 (movdi_aarch64): Support LDR/STR from/to D register.
11033 2013-04-29 Ian Bolton <ian.bolton@arm.com>
11035 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
11036 or higher by default.
11038 2013-04-29 Richard Biener <rguenther@suse.de>
11040 PR middle-end/57075
11041 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
11042 even if not adding abnormal edges for calls that can make
11045 2013-04-29 Richard Biener <rguenther@suse.de>
11047 PR middle-end/57103
11048 * tree-cfg.c (move_stmt_op): Fix condition under which to update
11050 (move_stmt_r): Remove redundant checking.
11052 2013-04-29 Teresa Johnson <tejohnson@google.com>
11055 * basic-block.h (apply_scale): New function.
11056 (apply_probability): Use apply_scale.
11057 * gimple-streamer-in.c (input_bb): Ditto.
11058 * lto-streamer-in.c (input_cfg): Ditto.
11059 * lto-cgraph.c (merge_profile_summaries): Ditto.
11060 * tree-optimize.c (execute_fixup_cfg): Ditto.
11061 * tree-inline.c (copy_bb): Update comment to use apply_scale.
11062 (copy_edges_for_bb): Ditto.
11063 (copy_cfg_body): Ditto.
11065 2013-04-29 Tom de Vries <tom@codesourcery.com>
11067 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
11068 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
11069 (tail_merge_optimize): Handle current_loops == NULL.
11071 2013-04-26 Jeff Law <law@redhat.com>
11073 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
11074 (simplify_cond_using_ranges): Generalize code to simplify
11075 COND_EXPRs where one argument is a constant and the other
11076 is an SSA_NAME created by an integral type conversion.
11078 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11080 * config/arm/arm.md (store_minmaxsi): Use only when
11081 optimize_insn_for_size_p.
11083 2013-04-29 Christian Bruel <christian.bruel@st.com>
11086 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
11088 2013-04-29 Richard Biener <rguenther@suse.de>
11090 PR middle-end/57089
11091 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
11092 loop tree make sure to schedule a fixup for the child as well.
11093 (expand_omp_for_generic): Properly add loops.
11094 (expand_omp_for_static_nochunk): Likewise.
11095 (expand_omp_for_static_chunk): Likewise.
11096 (expand_omp_for): For the degenerate case fixup loops.
11097 (expand_omp_sections): Fix default bb placement in loops.
11098 (expand_omp_atomic_pipeline): Properly add loops.
11100 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11102 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
11104 2013-04-29 Tom de Vries <tom@codesourcery.com>
11106 * tree-ssa-tail-merge.c: Update header comment.
11108 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11110 * config/aarch64/arm_neon.h
11111 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
11112 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
11113 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
11114 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
11115 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
11116 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
11117 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
11118 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
11120 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11122 * config/aarch64/aarch64-simd.md
11123 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
11124 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
11125 fix_trunc, fixuns_trunc.
11126 (ftrunc<VDQF:mode>2): New.
11127 * config/aarch64/iterators.md (optab): Add fix, fixuns.
11128 (fix_trunc_optab): New.
11130 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11132 * config/aarch64/aarch64-builtins.c
11133 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
11134 iceilf, lround, iroundf.
11136 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11139 * config/i386/i386.h (enum ix86_tune_indices)
11140 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
11141 New, split from X86_TUNE_INTER_UNIT_MOVES.
11142 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
11143 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
11144 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
11145 (TARGET_INTER_UNIT_MOVES): Remove.
11146 * config/i386/i386.c (initial_ix86_tune_features): Update.
11147 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
11148 (ix86_expand_convert_uns_didf_sse): Use
11149 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11150 (ix86_expand_vector_init_one_nonzero): Ditto.
11151 (ix86_expand_vector_init_interleave): Ditto.
11152 (inline_secondary_memory_needed): Return true for moves from SSE class
11153 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
11154 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11155 * config/i386/constraints.md (Yi, Ym): Depend on
11156 TARGET_INTER_UNIT_MOVES_TO_VEC.
11157 (Yj, Yn): New constraints.
11158 * config/i386/i386.md (*movdi_internal): Change constraints of
11159 operand 1 from Yi to Yj and from Ym to Yn.
11160 (*movsi_internal): Ditto.
11161 (*movdf_internal): Ditto.
11162 (*movsf_internal): Ditto.
11163 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
11164 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11165 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
11166 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
11167 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
11168 * config/i386/sse.md (movdi_to_sse): Ditto.
11169 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
11170 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
11171 TARGET_INTER_UNIT_MOVES.
11172 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
11173 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
11174 instead of TARGET_INTER_UNIT_MOVES.
11175 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
11176 operand 1 from Yi to Yj and from Ym to Yn.
11178 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11180 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
11181 (float_truncate_hi_): Likewise.
11182 (float_extend_lo_): Likewise.
11183 (float_truncate_lo_): Likewise.
11184 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
11185 (aarch64_float_extend_lo_v2df): Likewise.
11186 (vec_unpacks_hi_v4sf): Likewise.
11187 (aarch64_float_truncate_lo_v2sf): Likewise.
11188 (aarch64_float_truncate_hi_v4sf): Likewise.
11189 (vec_pack_trunc_v2df): Likewise.
11190 (vec_pack_trunc_df): Likewise.
11192 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11194 * config/aarch64/aarch64-builtins.c
11195 (aarch64_fold_builtin): Fold float conversions.
11196 * config/aarch64/aarch64-simd-builtins.def
11197 (floatv2si, floatv4si, floatv2di): New.
11198 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
11199 * config/aarch64/aarch64-simd.md
11200 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
11201 * config/aarch64/iterators.md (FLOATUORS): New.
11202 (optab): Add float, floatuns.
11203 (su_optab): Likewise.
11205 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11207 * config/aarch64/aarch64-builtins.c
11208 (aarch64_builtin_vectorized_function): Use new names for
11210 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
11211 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
11212 (fcvtzu): Split as...
11213 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
11214 (fcvtas): Split as...
11215 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
11216 (fcvtau): Split as...
11217 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
11218 (fcvtps): Split as...
11219 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
11220 (fcvtpu): Split as...
11221 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
11222 (fcvtms): Split as...
11223 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
11224 (fcvtmu): Split as...
11225 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
11226 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
11227 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
11228 (lfrintnusf, lfrintnudf): Likewise.
11229 * config/aarch64/aarch64-simd.md
11230 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
11232 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
11233 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
11234 (fcvt_pattern): Likewise.
11236 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11238 * config/aarch64/aarch64-simd.md
11239 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
11240 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
11242 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11244 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
11245 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
11246 (vrnd<a,m,n,p>_f32): Implement using builtins.
11247 (vrnd<i,x><q>_f<32, 64>): New.
11249 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11251 * config/aarch64/aarch64-builtins.c
11252 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
11253 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
11254 (frintz): Rename to...
11256 (frintp): Rename to...
11258 (frintm): Rename to...
11260 (frinti): Rename to...
11261 (nearbyint): ...this.
11262 (frintx): Rename to...
11264 (frinta): Rename to...
11266 * config/aarch64/aarch64-simd.md
11267 (aarch64_frint<frint_suffix><mode>): Delete.
11268 (<frint_pattern><mode>2): Convert to insn.
11269 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
11270 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
11271 (frint_pattern): Likewise.
11272 (frint_suffix): Likewise.
11274 2013-04-29 Richard Biener <rguenther@suse.de>
11276 PR tree-optimization/57081
11277 * loop-init.c: Include tree-flow.h.
11278 (loop_optimizer_finalize): Free number of iteration estimates.
11279 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
11281 2013-04-29 Jakub Jelinek <jakub@redhat.com>
11283 PR tree-optimization/57083
11284 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
11285 non-singleton shift count range, zero extend low_bound for uns case.
11287 * config/i386/predicates.md (general_vector_operand): New predicate.
11288 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
11289 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
11290 if they aren't nonimmediate operands. If their original values
11291 satisfy const_vector_equal_evenodd_p, don't shift them.
11292 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
11293 predicates. For the SSE4.1 case force operands[{1,2}] into registers
11294 if not nonimmediate_operand.
11295 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
11296 instead of register_operand.
11297 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
11299 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
11301 * stor-layout.c (finalize_size_functions): Allocate a structure and
11302 reset cfun before dumping the functions.
11304 2013-04-27 Jakub Jelinek <jakub@redhat.com>
11306 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
11309 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
11310 use xop_pmacsdqh if uns_p.
11311 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
11312 the immediate rotate count.
11314 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
11316 * rtl.h (struct rtx_def): Add comment for field jump.
11317 (LRA_SUBREG_P): New macro.
11318 * recog.c (register_operand): Check LRA_SUBREG_P.
11319 * lra.c (lra): Add note at the end of RTL code. Align non-empty
11321 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11322 (lra_final_code_change): Skip subreg change for operators.
11323 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11324 if there are no operand changes.
11325 * lra-constraints.c (curr_insn_set): New.
11326 (match_reload): Set LRA_SUBREG_P.
11327 (emit_spill_move): Ditto.
11328 (check_and_process_move): Use curr_insn_set. Process only single
11329 set insns. Don't initialize sec_mem_p and change_p.
11330 (simplify_operand_subreg): Use LRA_SUBREG_P.
11331 (reg_in_class_p): New function.
11332 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
11333 of #ifdef. Add code to remove cycling.
11334 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11335 non-null disp. Reload inner instead of disp when base and index
11336 are null. Try to put lo_sum into register.
11337 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11338 (check_and_process_move): Move code for move cost check to
11339 simple_move_p. Remove equiv_substitution.
11340 (simple_move_p): New function.
11341 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
11342 curr_insn_set. Call check_and_process_move only for single set
11343 insns. Use the new function. Move call of check_and_process_move
11344 after operand equiv substitution and address process.
11346 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11349 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
11350 with nonlocal goto receivers or returns twice calls, ignore
11351 unininitialized values from abnormal edges to nl goto receiver
11352 or returns twice call.
11354 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11356 PR tree-optimization/57051
11357 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
11358 and VEC_RSHIFT_EXPR if shift count is a multiple of element
11361 2013-04-26 Richard Biener <rguenther@suse.de>
11363 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
11364 (expand_omp_taskreg): Likewise. Mark loops for fixup.
11365 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
11366 (fixup_loop_arrays_after_move): New function.
11367 (move_sese_region_to_fn): Properly outline the loop tree parts
11368 of the SESE region.
11370 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
11372 * config/i386/i386.md (type, unit): Fix long lines.
11374 2013-04-26 Richard Biener <rguenther@suse.de>
11376 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
11377 (lto-streamer-out.o): Likewise.
11378 * cfgloop.c (init_loops_structure): Export, add struct function
11379 argument and adjust.
11380 (flow_loops_find): Adjust.
11381 * cfgloop.h (enum loop_estimation): Add EST_LAST.
11382 (init_loops_structure): Declare.
11383 * lto-streamer-in.c: Include cfgloop.h.
11384 (input_cfg): Input the loop tree.
11385 * lto-streamer-out.c: Include cfgloop.h.
11386 (output_cfg): Output the loop tree.
11387 (output_struct_function_base): Do not drop PROP_loops.
11389 2013-03-26 Richard Biener <rguenther@suse.de>
11391 * tree-cfg.c (execute_build_cfg): Build the loop tree.
11392 (pass_build_cfg): Provide PROP_loops.
11393 (move_sese_region_to_fn): Remove loops that are outlined into fn
11395 * tree-inline.c: Include cfgloop.h.
11396 (initialize_cfun): Do not drop PROP_loops.
11397 (copy_loops): New function.
11398 (copy_cfg_body): Copy loop structure.
11399 (tree_function_versioning): Initialize destination loop tree.
11400 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
11401 (pass_parallelize_loops): Do IL verification.
11402 * loop-init.c (loop_optimizer_init): Fixup loops if required.
11403 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
11404 the CFG make sure we fixup loops as well.
11405 * tree-ssa-tail-merge.c: Include cfgloop.h.
11406 (replace_block_by): When merging loop latches mark loops for fixup.
11407 * lto-streamer-out.c (output_struct_function_base): Drop
11408 PROP_loops for now.
11409 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11410 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11411 * ipa-split.c: Include cfgloop.h.
11412 (split_function): Add the new return block to the loop tree root.
11413 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11414 whether we have removed the forwarder block.
11415 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11416 * cfgloop.h (place_new_loop): Declare.
11417 * cfgloopmanip.c (place_new_loop): Export.
11418 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11419 (tree-switch-conversion.o): Likewise.
11420 (tree-complex.o): Likewise.
11421 (tree-inline.o): Likewise.
11422 (tree-ssa-tailmerge.o): Likewise.
11423 (ipa-split.o): Likewise.
11424 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11425 (tree-ssa-copy.o): Likewise.
11426 * tree-switch-conversion.c: Include cfgloop.h
11427 (process_switch): If we emit a bit-test cascade, schedule loops
11429 * tree-complex.c: Include cfgloop.h.
11430 (expand_complex_div_wide): Properly add new basic-blocks to loops.
11431 * asan.c: Include cfgloop.h.
11432 (create_cond_insert_point): Properly add new basic-blocks to
11433 loops, schedule loop fixup.
11434 * cfgloop.c (verify_loop_structure): Check that looks are not
11436 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11438 (expand_omp_for_generic): Likewise.
11439 (expand_omp_sections): Likewise.
11440 (expand_omp_atomic_pipeline): Schedule loops for fixup.
11441 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11442 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11443 is initialized, not when loops are present.
11444 * tree-parloops.c (parallelize_loops): Remove checking here.
11445 * passes.c (init_optimization_passes): Schedule a copy-propagation
11446 pass before complete unrolling of inner loops.
11448 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11450 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11451 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11452 (colorize_init): Add argument to _WIN32 version.
11453 * toplev.c: Include diagnostic-color.h.
11454 (process_options): Default to -fdiagnostics-color=auto if
11455 GCC_COLORS env var is in the environment.
11456 * common.opt (fdiagnostics-color=): Add Var and Init.
11457 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11458 env var is in the environment, the default is auto rather than never.
11460 * diagnostic.h (file_name_as_prefix): Add context argument.
11461 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
11462 the string as locus.
11463 * langhooks.c (lhd_print_error_function): Adjust caller.
11465 2013-04-25 Lawrence Crowl <crowl@google.com>
11467 * var-tracking.c (shared_hash_def::htab):
11468 Change type to hash_table. Update dependent calls and types.
11470 2013-04-25 Lawrence Crowl <crowl@google.com>
11472 * Makefile.in: Update as needed below.
11474 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11475 Move declaration to after the type's method definitons.
11477 * attribs.c (htab_t scoped_attributes::attribute_hash):
11478 Change type to hash_table. Update dependent calls and types.
11480 * bitmap.c (htab_t bitmap_desc_hash):
11481 Change type to hash_table. Update dependent calls and types.
11483 * cselib.c (htab_t cselib_hash_table):
11484 Change type to hash_table. Update dependent calls and types.
11486 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11487 (hash_string_slot_node): Move implementation into lto-streamer.h
11488 struct string_slot_hasher.
11489 (eq_string_slot_node): Likewise.
11491 * data-streamer-out.c: Update output_block::string_hash_table
11492 dependent calls and types.
11494 * dwarf2cfi.c (htab_t trace_index):
11495 Change type to hash_table. Update dependent calls and types.
11497 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11498 Change type to hash_table. Update dependent calls and types.
11499 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11500 (htab_t optimize_external_refs::map): Likewise.
11501 (htab_t output_comp_unit::extern_map): Likewise.
11502 (htab_t output_comdat_type_unit::extern_map): Likewise.
11503 (htab_t output_macinfo::macinfo_htab): Likewise.
11504 (htab_t optimize_location_lists::htab): Likewise.
11505 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11507 * except.c (htab_t ehspec_hash_type):
11508 Change type to hash_table. Update dependent calls and types.
11509 (assign_filter_values::ttypes): Likewise.
11510 (assign_filter_values::ehspec): Likewise.
11511 (sjlj_assign_call_site_values::ar_hash): Likewise.
11512 (convert_to_eh_region_ranges::ar_hash): Likewise.
11514 * gcse.c (htab_t pre_ldst_table):
11515 Change type to hash_table. Update dependent calls and types.
11517 * ggc-common.c (htab_t saving_htab):
11518 Change type to hash_table. Update dependent calls and types.
11519 (htab_t loc_hash): Likewise.
11520 (htab_t ptr_hash): Likewise.
11521 (call_count): Rename ggc_call_count.
11522 (call_alloc): Rename ggc_call_alloc.
11523 (loc_descriptor): Rename make_loc_descriptor.
11524 (add_statistics): Rename ggc_add_statistics.
11526 * ggc-common.c (saving_htab):
11527 Change type to hash_table. Update dependent calls and types.
11529 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11530 (push_gimplify_context): Likewise.
11531 (pop_gimplify_context): Likewise.
11532 (struct gimple_temp_hash_elt): Added.
11533 (struct gimplify_hasher): Likewise.
11534 (struct gimplify_ctx.temp_htab):
11535 Change type to hash_table. Update dependent calls and types.
11537 * gimple-fold.c: Include gimplify-ctx.h.
11539 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11540 Change type to hash_table. Update dependent calls and types.
11541 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11542 avoid potential global name collision.
11544 * gimplify.c: Include gimplify-ctx.h.
11545 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11546 (htab_t gimplify_ctx::temp_htab):
11547 Update dependent calls and types for new type hash_table.
11548 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11549 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11551 * gimplify-ctx.h: New.
11552 (struct gimple_temp_hash_elt): Move from gimplify.c.
11553 (class gimplify_hasher): New.
11554 (struct gimplify_ctx): Move from gimple.h.
11555 (htab_t gimplify_ctx::temp_htab):
11556 Change type to hash_table. Update dependent calls and types.
11558 * graphite-clast-to-gimple.c: Include graphite-htab.h.
11559 (htab_t ivs_params::newivs_index):
11560 Change type to hash_table. Update dependent calls and types.
11561 (htab_t ivs_params::params_index): Likewise.
11562 (htab_t print_generated_program::params_index): Likewise.
11563 (htab_t gloog::newivs_index): Likewise.
11564 (htab_t gloog::params_index): Likewise.
11566 * graphite.c: Include graphite-htab.h.
11567 4htab_t graphite_transform_loops::bb_pbb_mapping):
11568 Change type to hash_table. Update dependent calls and types.
11570 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11571 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11572 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11574 * graphite-dependences.c: Include graphite-htab.h.
11575 (loop_is_parallel_p): Change hash table type of parameter.
11577 * graphite-htab.h: New.
11578 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11579 (extern find_pbb_via_hash): Move from graphite-poly.h.
11580 (extern loop_is_parallel_p): Move from graphite-poly.h.
11581 (extern get_loop_body_pbbs): Move from graphite-poly.h.
11583 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11584 (extern loop_is_parallel_p): Move to graphite-htab.h.
11585 (extern get_loop_body_pbbs): Move to graphite-htab.h.
11587 * haifa-sched.c (htab_t delay_htab):
11588 Change type to hash_table. Update dependent calls and types.
11589 (htab_t delay_htab_i2): Likewise.
11591 * ira-color.c (htab_t allocno_hard_regs_htab):
11592 Change type to hash_table. Update dependent calls and types.
11594 * ira-costs.c (htab_t cost_classes_htab):
11595 Change type to hash_table. Update dependent calls and types.
11597 * loop-invariant.c (htab_t merge_identical_invariants::eq):
11598 Change type to hash_table. Update dependent calls and types.
11600 * loop-iv.c (htab_t bivs):
11601 Change type to hash_table. Update dependent calls and types.
11603 * loop-unroll.c (htab_t opt_info::insns_to_split):
11604 Change type to hash_table. Update dependent calls and types.
11605 (htab_t opt_info::insns_with_var_to_expand): Likewise.
11607 * lto-streamer.h (struct string_slot): Move from data-streamer.h
11608 (struct string_slot_hasher): New.
11609 (htab_t output_block::string_hash_table):
11610 Change type to hash_table. Update dependent calls and types.
11612 * lto-streamer-in.c (freeing_string_slot_hasher): New.
11613 (htab_t file_name_hash_table):
11614 Change type to hash_table. Update dependent calls and types.
11616 * lto-streamer-out.c: Update output_block::string_hash_table dependent
11619 * lto-streamer.c (htab_t tree_htab):
11620 Change type to hash_table. Update dependent calls and types.
11622 * omp-low.c: Include gimplify-ctx.h.
11624 * passes.c (htab_t name_to_pass_map):
11625 Change type to hash_table. Update dependent calls and types.
11626 (pass_traverse): Rename to passes_pass_traverse.
11628 * plugin.c (htab_t event_tab):
11629 Change type to hash_table. Update dependent calls and types.
11631 * postreload-gcse.c (htab_t expr_table):
11632 Change type to hash_table. Update dependent calls and types.
11633 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11635 * sese.c (debug_rename_map_1): Make extern.
11636 (htab_t copy_bb_and_scalar_dependences::rename_map):
11637 Change type to hash_table. Update dependent calls and types.
11639 * sese.h (extern debug_rename_map): Move to .c file.
11641 * store-motion.c (htab_t store_motion_mems_table):
11642 Change type to hash_table. Update dependent calls and types.
11644 * trans-mem.c (htab_t tm_new_mem_hash):
11645 Change type to hash_table. Update dependent calls and types.
11647 * tree-browser.c (htab_t TB_up_ht):
11648 Change type to hash_table. Update dependent calls and types.
11650 * tree-cfg.c (htab_t discriminator_per_locus):
11651 Change type to hash_table. Update dependent calls and types.
11653 * tree-complex.c: Include tree-hasher.h
11654 (htab_t complex_variable_components):
11655 Change type to hash_table. Update dependent calls and types.
11657 * tree-eh.c (htab_t finally_tree):
11658 Change type to hash_table. Update dependent calls and types.
11660 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11661 struct int_tree_hasher.
11662 (extern int_tree_map_eq): Likewise.
11663 (uid_decl_map_hash): Removed.
11664 (extern decl_tree_map_eq): Likewise.
11666 * tree-hasher.h: New.
11667 (struct int_tree_hasher): New.
11668 (typedef int_tree_htab_type): New.
11670 * tree-inline.c: Include gimplify-ctx.h.
11672 * tree-mudflap.c: Include gimplify-ctx.h.
11674 * tree-parloops.c: Include tree-hasher.h.
11675 (htab_t eliminate_local_variables_stmt::decl_address):
11676 Change type to hash_table. Update dependent calls and types.
11677 (htab_t separate_decls_in_region::decl_copies): Likewise.
11679 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11680 Change type to hash_table. Update dependent calls and types.
11682 * tree-sra.c (candidates):
11683 Change type to hash_table. Update dependent calls and types.
11685 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11687 (int_tree_map_hash): Likewise.
11689 * tree-ssa-dom.c (htab_t avail_exprs):
11690 Change type to hash_table. Update dependent calls and types.
11692 * tree-ssa-live.c (var_map_base_init::tree_to_index):
11693 Change type to hash_table. Update dependent calls and types.
11695 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11696 Change type to hash_table. Update dependent calls and types.
11698 * tree-ssa-phiopt.c (seen_ssa_names):
11699 Change type to hash_table. Update dependent calls and types.
11701 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11702 Change type to hash_table. Update dependent calls and types.
11704 * tree-ssa-uncprop.c (equiv):
11705 Change type to hash_table. Update dependent calls and types.
11707 2013-04-25 Jakub Jelinek <jakub@redhat.com>
11709 PR rtl-optimization/57003
11710 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11711 call note_stores with kill_clobbered_value callback again after
11712 killing regs_invalidated_by_call.
11714 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11716 * config/aarch64/aarch64-simd.md
11717 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11718 (aarch64_simd_bsl<mode>): Likewise.
11719 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11721 2013-04-25 Marek Polacek <polacek@redhat.com>
11723 PR tree-optimization/57066
11724 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11726 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
11728 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11730 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11732 * config/aarch64/aarch64-builtins.c
11733 (aarch64_fold_builtin): New.
11734 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11735 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11736 * config/aarch64/aarch64-simd-builtins.def (abs): New.
11737 * config/aarch64/arm_neon.h
11738 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11740 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11741 Tejas Belagod <tejas.belagod@arm.com>
11743 * config/aarch64/aarch64-builtins.c
11744 (aarch64_gimple_fold_builtin): New.
11745 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11746 * config/aarch64/aarch64-simd-builtins.def (addv): New.
11747 * config/aarch64/aarch64-simd.md (addpv4sf): New.
11748 (addvv4sf): Update.
11749 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11751 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11753 * config/aarch64/aarch64.md
11754 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11756 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11758 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11759 (*ngcsi_uxtw): New pattern.
11761 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11762 Julian Brown <julian@codesourcery.com>
11764 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11765 (TB_DREG): Add T_V4HF.
11766 (v4hf_UP): New macro.
11767 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11768 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11769 Handle initialisation of V4HF. Adjust initialisation of reinterpret
11771 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11772 (arm_vector_mode_supported_p): Handle V4HF.
11773 (arm_mangle_map): Handle V4HFmode.
11774 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11775 * config/arm/arm_neon_builtins.def: Add entries for
11776 vcvtv4hfv4sf, vcvtv4sfv4hf.
11777 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11778 (neon_vcvtv4hfv4sf): Likewise.
11779 * config/arm/neon-gen.ml: Handle half-precision floating point
11781 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11782 * config/arm/arm_neon.h: Regenerate.
11783 * config/arm/neon.ml (type elts): Add F16.
11784 (type vectype): Add T_float16x4, T_floatHF.
11785 (type vecmode): Add V4HF.
11786 (type features): Add Requires_FP_bit feature.
11787 (elt_width): Handle F16.
11788 (elt_class): Likewise.
11789 (elt_of_class_width): Likewise.
11790 (mode_of_elt): Refactor.
11791 (type_for_elt): Handle F16, fix error messages.
11792 (vectype_size): Handle T_float16x4.
11793 (vcvt_sh): New function.
11794 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11795 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11796 (string_of_mode): Handle V4HF.
11797 * doc/arm-neon-intrinsics.texi: Regenerate.
11799 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11801 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11802 format specifier in 'X' case.
11804 2013-04-25 Alan Modra <amodra@gmail.com>
11807 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11808 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11809 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11810 Repeat for many other rotate/shift and mask patterns using subregs.
11811 Name lshiftrt insns.
11812 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11813 on WORDS_BIG_ENDIAN.
11815 2013-04-25 Alan Modra <amodra@gmail.com>
11817 * config.gcc: Support little-endian powerpc-linux targets.
11818 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11819 (LINK_OS_LINUX_SPEC): Define.
11820 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11821 Preserve MASK_LITTLE_ENDIAN.
11822 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11823 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11824 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11825 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11826 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11827 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11828 Correct fp word order for little-endian. Don't shift toc entries
11829 smaller than a word for little-endian.
11830 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11831 (bswapdi2 splits): Correct low-part subreg for little-endian.
11832 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11833 low/high where such is correct only for be.
11834 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11835 little-endian for -mcall-aixdesc.
11837 2013-04-25 Alan Modra <amodra@gmail.com>
11839 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11840 replace_equiv_address_nv.
11842 2013-04-25 Alan Modra <amodra@gmail.com>
11844 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11846 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11849 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11850 * rtl.h (struct rtx_def): ...
11852 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11854 PR rtl-optimizations/57046
11855 * lra-constraints (split_reg): Set up lra_risky_transformations_p
11856 for multi-reg splits.
11858 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
11860 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11862 2013-04-24 Sterling Augustine <saugustine@google.com>
11864 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11865 (comp_dir_string, debug_str_dwo_section): New.
11866 (DEBUG_STR_DWO_SECTION): Rename to ...
11867 (DEBUG_DWO_STR_SECTION): ... this.
11868 (DEBUG_NORM_STR_SECTION): Delete.
11869 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11870 (DEBUG_STR_DWO_SECTION_FLAGS): New.
11871 (find_AT_string): Move most logic to ...
11872 (find_AT_string_in_table): ... here. New.
11873 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11874 add_skeleton_AT_string. Delete logic.
11875 (output_skeleton_debug_sections): Remove call to
11876 add_top_level_skeleton_die_attrs.
11877 (add_comp_dir_attribute): Move logic to comp_dir_string.
11878 (dwarf2out_init): Initialize debug_str_dwo_section.
11879 (output_indirect_string): Call find_string_form.
11880 (output_indirect_strings): Rewrite.
11881 (prune_unused_types): Empty skeleton_debug_str_hash.
11882 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11883 (dwarf2out_finish): Call output_indirect_strings.
11885 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
11887 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11889 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11891 * rtl.h (struct rtx_def): Add comment for field jump.
11892 (LRA_SUBREG_P): New macro.
11893 * recog.c (register_operand): Check LRA_SUBREG_P.
11894 * lra.c (lra): Add note at the end of RTL code. Align non-empty
11896 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11897 (lra_final_code_change): Skip subreg change for operators.
11898 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11899 if there are no operand changes.
11900 * lra-constraints.c (curr_insn_set): New.
11901 (match_reload): Set LRA_SUBREG_P.
11902 (emit_spill_move): Ditto.
11903 (check_and_process_move): Use curr_insn_set. Process only single
11904 set insns. Don't initialize sec_mem_p and change_p.
11905 (simplify_operand_subreg): Use LRA_SUBREG_P.
11906 (reg_in_class_p): New function.
11907 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
11908 of #ifdef. Add code to remove cycling.
11909 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11910 non-null disp. Reload inner instead of disp when base and index
11911 are null. Try to put lo_sum into register.
11912 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11913 (check_and_process_move): Move code for move cost check to
11914 simple_move_p. Remove equiv_substitution.
11915 (simple_move_p): New function.
11916 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
11917 curr_insn_set. Call check_and_process_move only for single set
11918 insns. Use the new function. Move call of check_and_process_move
11919 after operand equiv substitution and address process.
11921 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
11923 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11924 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11925 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11927 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
11929 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11931 2013-04-24 Marek Polacek <polacek@redhat.com>
11933 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11934 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11935 (select_loops_exit_conditions): Likewise.
11936 (number_of_iterations_for_all_loops): Likewise.
11937 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11938 (scev_analysis): Likewise.
11940 2013-04-02 Catherine Moore <clm@codesourcery.com>
11941 Chao-ying Fu <fu@mips.com>
11943 * config/mips/micromips.md (jraddiusp): New pattern.
11944 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11945 instruction if possible.
11947 2013-04-24 Alan Modra <amodra@gmail.com>
11949 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11951 2013-04-24 Julian Brown <julian@codesourcery.com>
11952 Chung-Lin Tang <cltang@codesourcery.com>
11954 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11955 dependency behavior in enumeration type DIE generation. Add TODO note
11956 to comments about future DW_FORM_sdata/udata re-work of related code.
11958 2013-04-23 Lawrence Crowl <crowl@google.com>
11960 * Makefile.in: Update as needed below.
11962 * hash-table.h (class hash_table):
11963 Correct many methods with parameter types compare_type to the correct
11964 value_type. (Correct code was unlikely to notice the change.)
11965 (hash_table::elements_with_deleted) New.
11966 (class hashtable::iterator): New.
11967 (hashtable::begin()): New.
11968 (hashtable::end()): New.
11969 (FOR_EACH_HASH_TABLE_ELEMENT): New.
11971 * statistics.c (statistics_hashes):
11972 Change type to hash_table. Update dependent calls and types.
11974 * tree-into-ssa.c (var_infos):
11975 Change type to hash_table. Update dependent calls and types.
11977 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11978 Change type to hash_table. Update dependent calls and types.
11980 * tree-ssa-loop-im.c (struct mem_ref.refs):
11981 Change type to hash_table. Update dependent calls and types.
11983 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11984 Change type to hash_table. Update dependent calls and types.
11986 * tree-ssa-sccvn.c (vn_tables_s::nary):
11987 Change type to hash_table. Update dependent calls and types.
11988 (vn_tables_s::phis): Likewise.
11989 (vn_tables_s::references): Likewise.
11991 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11992 (vn_reference_eq): Update parameter and return types.
11994 * tree-ssa-structalias.c (pointer_equiv_class_table):
11995 Change type to hash_table. Update dependent calls and types.
11996 (location_equiv_class_table): Likewise.
11998 * tree-vect-data-refs.c: Consequential changes for making
11999 peeling a hash_table.
12001 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
12002 (destroy_loop_vec_info): Dependent hash_table update.
12004 * tree-vectorizer.h (peeling_htab):
12005 Change type to hash_table. Update dependent calls and types.
12007 2013-04-23 Shiva Chen <shiva0217@gmail.com>
12009 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
12010 to check the register content is equal or not.
12011 * lra-constraints.c (match_reload): Use lra_assign_reg_val
12012 to assign register content record.
12013 * lra-eliminations.c (update_reg_eliminate): Use
12014 lra_update_reg_val_offset to update register content offset.
12015 * lra-int.h (struct lra_reg): Add offset member.
12016 (lra_reg_val_equal_p): New static inline function.
12017 (lra_update_reg_val_offset): New static inline function.
12018 (lra_assign_reg_val): New static inline function.
12019 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
12020 to assign register content record.
12021 (initialize_lra_reg_info_element): Initial offset to zero.
12023 2013-04-23 Catherine Moore <clm@codesourcery.com>
12025 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
12026 operands. Record compression.
12028 2013-04-23 Xinliang David Li <davidxl@google.com>
12030 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
12032 2013-04-23 Richard Biener <rguenther@suse.de>
12034 PR middle-end/57036
12035 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
12036 parameter, only add abnormal goto edges from the copied body
12037 if the call could perform abnormal gotos.
12038 (copy_cfg_body): Adjust.
12040 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
12042 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
12044 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
12046 * coretypes.h (gimple_stmt_iterator): Add struct to make
12049 2013-04-23 Richard Biener <rguenther@suse.de>
12051 PR tree-optimization/57026
12052 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
12053 from SSA names occuring in abnormal PHI nodes.
12055 2013-04-22 Andi Kleen <ak@linux.intel.com>
12057 * lto/lto.c (print_lto_report_1): Fix LTO report names.
12059 2013-04-22 Andi Kleen <ak@linux.intel.com>
12061 * lto/lto.c (print_lto_report_1): Declare early.
12062 (read_cgraph_and_symbols): Call print_lto_report_1 early.
12064 2013-04-22 Andi Kleen <ak@linux.intel.com>
12066 * common.opt (-flto-report-wpa): Add.
12067 * doc/invoke.texi (-flto-report-wpa): Add.
12068 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
12071 2013-04-22 Xinliang David Li <davidxl@google.com>
12073 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
12074 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
12075 * Makefile.in: New dependency
12077 David Daney <ddaney.cavm@gmail.com>
12079 * configure.ac (gcc_cv_as_micromips_support): Use the
12080 --fatal-warnings option.
12081 * configure: Regenerate.
12083 2013-04-22 Marek Polacek <polacek@redhat.com>
12086 * tsan.c (instrument_expr): Don't instrument expression
12087 in case its size is zero.
12089 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
12093 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
12095 * config/alpha/alpha.c (TARGET_LRA_P): New define.
12097 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12099 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
12100 (gimple_stmt_iterator): New typedef.
12101 * gimple.h (gimple_stmt_iterator): Rename to...
12102 (gimple_stmt_iterator_d): ... This.
12103 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
12104 trees be valid for GIMPLE and GENERIC.
12105 (TARGET_GIMPLE_FOLD_BUILTIN): New.
12106 * gimple-fold.c (gimple_fold_call): Call target hook
12107 gimple_fold_builtin.
12108 * hooks.c (hook_bool_gsiptr_false): New.
12109 * hooks.h (hook_bool_gsiptr_false): New.
12110 * target.def (fold_stmt): New.
12111 * doc/tm.texi: Regenerate.
12113 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
12116 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
12117 a set sp if no stack realignment.
12119 2013-04-22 Nick Clifton <nickc@redhat.com>
12121 * config.gcc (tilegx-linux): Extend extra_objs rather than
12123 (tilepro-linux): Likewise.
12125 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12127 * config/aarch64/aarch64-builtins.c
12129 (CF0, CF1, CF2, CF3, CF4, CF10): New.
12130 (VAR<1-12>): Add MAP parameter.
12131 (BUILTIN_*): Likewise.
12132 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
12133 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
12134 (aarch64_ushl_n<mode>): Likewise.
12135 (aarch64_sshr_n<mode>): Likewise.
12136 (aarch64_ushr_n<mode>): Likewise.
12137 (aarch64_<maxmin><mode>): Likewise.
12138 (aarch64_sqrt<mode>): Likewise.
12139 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
12140 (vshr<q>_n_*): Likewise.
12142 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12144 * config/aarch64/aarch64-builtins.c
12145 (aarch64_simd_builtin_type_mode): Handle SF types.
12147 (BUILTIN_GPF): Define.
12148 (aarch64_init_simd_builtins): Handle SF types.
12149 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
12150 (frecps): Likewise.
12151 (frecpx): Likewise.
12152 * config/aarch64/aarch64-simd.md
12153 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
12154 (aarch64_frecpe<mode>): New.
12155 (aarch64_frecps<mode>): Likewise.
12156 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
12157 (v8type): Add frecp<esx>.
12158 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
12159 (aarch64_frecps<mode>): Likewise.
12160 * config/aarch64/iterators.md (FRECP): New.
12161 (frecp_suffix): Likewise.
12162 * config/aarch64/arm_neon.h
12163 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
12165 2013-04-22 Christian Bruel <christian.bruel@st.com>
12168 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
12169 (REG_CLASS_NAMES): Idem.
12170 (REG_CLASS_CONTENTS): Idem.
12171 (REGCLASS_HAS_FP_REG): Idem.
12172 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
12173 (sh_conditional_register_usage): Idem.
12175 2013-04-21 Jeff Law <law@redhat.com>
12177 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
12178 (ssa_forward_propagate_and_combine): Use it.
12180 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
12182 * lra.c: Update the flow chart diagram.
12184 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
12186 PR rtl-optimization/56847
12187 * lra-constraints.c (process_alt_operands): Discourage alternative
12188 with non-matche doffsettable memory constraint fro memory with
12191 2013-04-19 Richard Biener <rguenther@suse.de>
12193 PR tree-optimization/56982
12194 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
12196 * gimplify.c (gimplify_call_expr): Notice special calls.
12197 (gimplify_modify_expr): Likewise.
12198 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
12199 abnormal control flow receivers.
12200 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
12201 in the same way as cfun->has_nonlocal_labels.
12202 (gimple_purge_dead_abnormal_call_edges): Likewise.
12203 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
12204 receivers start a basic-block.
12206 2013-04-19 Richard Biener <rguenther@suse.de>
12208 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
12210 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
12211 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
12212 (SLP_TREE_LOAD_PERMUTATION): Add.
12213 (vect_transform_slp_perm_load): Adjust prototype.
12214 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
12215 (vect_free_slp_instance): Likewise.
12216 (vect_create_new_slp_node): Likewise.
12217 (vect_supported_slp_permutation_p): Remove.
12218 (vect_slp_rearrange_stmts): Adjust.
12219 (vect_supported_load_permutation_p): Likewise. Inline
12220 vect_supported_slp_permutation_p here.
12221 (vect_analyze_slp_instance): Compute load permutations per
12222 slp node instead of per instance.
12223 (vect_get_slp_defs): Adjust.
12224 (vect_transform_slp_perm_load): Likewise.
12225 (vect_schedule_slp_instance): Remove redundant code.
12226 (vect_schedule_slp): Remove hack for PR56270, add it ...
12227 * tree-vect-stmts.c (vectorizable_load): ... here, do not
12228 CSE loads for SLP. Adjust.
12230 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
12232 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
12233 spelling in two comments.
12235 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
12238 * config/arm/arm.c (load_multiple_sequence): Require SP
12239 as base register for loads if SP is in the register list.
12241 2013-04-19 Martin Jambor <mjambor@suse.cz>
12243 PR tree-optimization/56718
12244 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
12245 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
12246 and made public. Adjusted all callers.
12247 (ipa_intraprocedural_devirtualization): New function.
12248 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
12249 (ipa_intraprocedural_devirtualization): Likewise.
12250 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
12252 2013-04-19 Richard Biener <rguenther@suse.de>
12254 PR tree-optimization/57000
12255 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
12257 2013-04-19 Terry Guo <terry.guo@arm.com>
12259 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
12261 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
12262 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
12263 (cortex_m4_fmacs): Use new reservations.
12264 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
12266 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
12268 PR rtl-optimization/56999
12269 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
12271 (lra_coalesce): Remove split_origin_bitmap and related code.
12272 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
12273 ranges if necessary.
12275 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
12277 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
12279 (ix86_expand_call): Remove clobbered_registers array and use
12280 x86_64_ms_sysv_extra_clobbered_registers instead.
12281 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
12283 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
12285 * config/i386/i386.md (*call_rex64_ms_sysv): Use
12286 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
12287 (*call_value_rex64_ms_sysv): Ditto.
12289 2013-04-18 Cary Coutant <ccoutant@google.com>
12291 * dwarf2out.c (output_pubnames): Check die_perennial_p of
12292 parent instead of die_mark.
12294 2013-04-18 Diego Novillo <dnovillo@google.com>
12296 * gimple.c (create_gimple_tmp): New.
12297 (get_expr_type): New.
12298 (build_assign): New.
12299 (build_type_cast): New.
12300 * gimple.h (enum ssa_mode): Define.
12301 (gimple_seq_set_location): New.
12302 * asan.c (build_check_stmt): Change some gimple_build_* calls
12303 to use build_assign and build_type_cast.
12305 2013-04-18 Richard Biener <rguenther@suse.de>
12307 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
12308 handle negative step. Remove redundant checks.
12309 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
12310 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
12311 for negative step and grouped loads fail to vectorize.
12313 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
12315 * emit-rtl.c (reset_insn_used_flags): New function.
12316 (reset_all_used_flags): Use it.
12317 (verify_insn_sharing): New function.
12318 (verify_rtl_sharing): Fix verification for SEQUENCEs.
12320 2013-04-18 Jakub Jelinek <jakub@redhat.com>
12322 PR tree-optimization/56984
12323 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
12324 and (x >> M) >= N don't register any assertion if N << M is the
12327 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
12329 * lower-subreg.c (resolve_simple_move): If called self-recursive,
12330 do not delete_insn insns that have not yet been emitted, only
12331 unlink them with remove_insn.
12332 * df-scan.c (df_insn_delete): Revert r197492.
12334 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
12336 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
12337 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
12339 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
12341 * config/arm/arm.md (movsicc_insn): Convert define_insn into
12342 define_insn_and_split.
12343 (and_scc,ior_scc,negscc): Likewise.
12344 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
12346 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
12348 * config/arm/arm.c (use_return_insn): Return 0 for targets that
12349 can benefit from using a sequence of LDRD instructions in epilogue
12350 instead of a single LDM instruction.
12352 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
12355 * doc/extend.texi: Fix typo.
12357 2013-04-17 Richard Biener <rguenther@suse.de>
12359 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
12360 (vect_build_slp_tree): ... here.
12361 (vect_build_slp_tree_1): Compute which stmts of the SLP group
12362 match. Remove special-casing of mismatched complex loads.
12363 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
12364 re-try the match with swapped commutative operands.
12365 (vect_supported_load_permutation_p): Remove special-casing of
12366 mismatched complex loads.
12367 (vect_analyze_slp_instance): Adjust.
12369 2013-04-17 Richard Biener <rguenther@suse.de>
12371 PR rtl-optimization/56921
12372 * cfgloop.h (struct loop): Add simple_loop_desc member.
12373 (struct niter_desc): Mark with GTY(()).
12374 (simple_loop_desc): Do not use aux field but simple_loop_desc.
12375 * loop-iv.c (get_simple_loop_desc): Likewise.
12376 (free_simple_loop_desc): Likewise.
12379 2013-04-16 Richard Biener <rguenther@suse.de>
12381 PR rtl-optimization/56921
12382 * loop-init.c (pass_rtl_move_loop_invariants): Add
12383 TODO_do_not_ggc_collect to todo_flags_finish.
12384 (pass_rtl_unswitch): Same.
12385 (pass_rtl_unroll_and_peel_loops): Same.
12386 (pass_rtl_doloop): Same.
12388 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
12390 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
12391 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
12392 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
12394 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
12395 * tree-streamer.c (record_common_node): Adjust reference in comment.
12397 2013-04-17 Terry Guo <terry.guo@arm.com>
12399 * config/arm/cortex-m4.md: Add a new bypass.
12401 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12403 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12405 (*subs_<optab><mode>_multp2): New pattern.
12406 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12407 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12409 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12411 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12412 (*subs_mul_imm_<mode>): New pattern.
12414 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
12417 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12418 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
12419 (vsx_movti_32bit): Change j->wa to O->wa.
12421 2013-04-16 Richard Biener <rguenther@suse.de>
12423 PR rtl-optimization/56921
12424 * loop-init.c (pass_rtl_move_loop_invariants): Add
12425 TODO_do_not_ggc_collect to todo_flags_finish.
12426 (pass_rtl_unswitch): Same.
12427 (pass_rtl_unroll_and_peel_loops): Same.
12428 (pass_rtl_doloop): Same.
12430 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
12432 * config/arm/arm.c (emit_multi_reg_push): New declaration
12433 for an existing function.
12434 (arm_emit_strd_push): New function.
12435 (arm_expand_prologue): Used here.
12436 (arm_emit_ldrd_pop): New function.
12437 (arm_expand_epilogue): Used here.
12438 (arm_get_frame_offsets): Update condition.
12439 (arm_emit_multi_reg_pop): Add a special case for load of a single
12440 register with writeback.
12442 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
12444 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12447 2013-04-16 Richard Biener <rguenther@suse.de>
12449 PR tree-optimization/56756
12450 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12451 (first_mem_ref_loc): New.
12452 (execute_sm): Place the load temporarily before a previous
12453 access instead of in the latch edge to ensure its SSA dependencies
12454 are defined at points dominating the load.
12456 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
12458 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12459 correct fix by moving header and footer insn to the footer of
12460 the merged basic block. Clear BB_END of the merged-away block.
12462 PR middle-end/43631
12463 * emit-rtl.c (make_note_raw): New function.
12464 (link_insn_into_chain): New static inline function.
12465 (add_insn): Use it.
12466 (add_insn_before, add_insn_after): Factor insn chain linking code...
12467 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12468 using link_insn_into_chain.
12469 (note_outside_basic_block_p): New helper function for emit_note_after
12470 and emit_note_before.
12471 (emit_note_after): Use nobb variant of add_insn_after if the note
12472 should not be contained in a basic block.
12473 (emit_note_before): Use nobb variant of add_insn_before if the note
12474 should not be contained in a basic block.
12475 (emit_note_copy): Use make_note_raw.
12476 (emit_note): Likewise.
12477 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12478 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12479 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12480 the moved barrier the tail of the basic block it follows.
12481 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12483 2013-04-15 Jakub Jelinek <jakub@redhat.com>
12485 PR tree-optimization/56962
12486 * gimple-ssa-strength-reduction.c (record_increment): Only set
12487 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12488 either rhs1 or rhs2 is equal to c->base_expr.
12490 2013-04-15 Richard Biener <rguenther@suse.de>
12492 PR tree-optimization/56933
12493 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12495 (GROUP_READ_WRITE_DEPENDENCE): Remove.
12496 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12497 * tree-vect-data-refs.c (vect_analyze_group_access): Move
12498 dependence check ...
12499 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12501 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12502 GROUP_READ_WRITE_DEPENDENCE.
12504 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12506 * emit-rtl.c (reset_all_used_flags): New function.
12507 (verify_rtl_sharing): Call reset_all_used_flags before and after
12508 performing the checks.
12510 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12512 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12513 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12514 * config/arm/constraints.md (De): New constraint.
12515 * config/arm/neon.md (anddi3_neon): Delete.
12516 (neon_vand<mode>): Expand to standard anddi3 pattern.
12517 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12518 Move earlier in the file.
12519 (neon_inv_logic_op2): Likewise.
12520 (arm_anddi_operand_neon): New predicate.
12522 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12524 * configure.ac (gcc_cv_ld_as_needed): Set
12525 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12526 Use -z ignore, -z record on *-*-solaris2*.
12527 (HAVE_LD_AS_NEEDED): Update comment.
12528 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12529 * configure: Regenerate.
12530 * config.in: Regenerate.
12531 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12532 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12533 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12534 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12535 equivalents. Fix markup.
12536 * doc/tm.texi: Regenerate.
12538 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
12540 * config/i386/i386.opt: New option mstack-protector-guard=.
12541 * config/i386/i386-opts.h: Add enum stack_protector_guard.
12542 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12543 TARGET_SSP_TLS_GUARD.
12544 * config/i386/i386.c (ix86_option_override_internal): Set
12545 ix86_stack_protector_guard.
12546 * config/i386/i386.md (stack_protect_set): Enable for
12547 TARGET_SSP_TLS_GUARD only.
12548 (stack_protect_set_<mode>): Ditto.
12549 (stack_protect_test): Ditto.
12550 (stack_protect_test_<mode>): Ditto.
12551 * doc/invoke.texi (i386 Option): Document.
12553 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
12556 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12557 (S_MODES): Set H_MODE bit.
12558 (SF_MODES): Set only S_MODE and SF_MODE bits.
12559 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12560 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12561 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12562 <MODE_FLOAT>: Likewise.
12564 2013-04-15 Joey Ye <joey.ye@arm.com>
12566 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12568 2013-04-15 Joey Ye <joey.ye@arm.com>
12570 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12572 (thumb_far_jump_used_p): Count instruction size and set
12575 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
12577 * reorg.c (fill_simple_delay_slots): Reindent block of code.
12578 * resource.c (mark_target_live_regs): Reformat conditional block.
12580 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
12582 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12583 notes, they are emitted only just before final.
12584 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12586 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
12588 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12589 * cfgrtl.c (delete_insn): Call it here instead.
12590 * lra-spills.c (lra_final_code_change): Use delete_insn.
12591 * haifa-sched.c (sched_remove_insn): Likewise.
12592 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12593 returning to the nop pool.
12594 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12595 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
12597 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
12599 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12600 * doc/tm.texi: Regenerated.
12602 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
12604 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12607 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
12609 * df-core.c (df_find_def): Compare register numbers.
12610 (df_find_use): Likewise.
12612 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
12615 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12616 lra_in_progress for return.
12618 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
12620 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12621 define_insn into define_insn_and_split and emit movsicc patterns.
12623 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
12625 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12627 2013-04-12 Richard Biener <rguenther@suse.de>
12629 * tree-pass.h (TODO_do_not_ggc_collect): New.
12630 * passes.c (execute_one_ipa_transform_pass): Honor
12631 TODO_do_not_ggc_collect.
12632 (execute_one_pass): Likewise.
12635 2013-04-10 Richard Biener <rguenther@suse.de>
12637 * passes.c (init_optimization_passes): Remove reload pass.
12638 * ira.c (do_reload): Merge into ...
12640 (rest_of_handle_reload): Remove.
12641 (pass_reload): Likewise.
12642 * config/i386/i386.c (ix86_option_override): Refer to ira instead
12643 of reload for vzeroupper pass placement.
12645 2013-04-12 Jakub Jelinek <jakub@redhat.com>
12647 PR tree-optimization/56918
12648 PR tree-optimization/56920
12649 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12650 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
12651 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
12652 use wide_mul_with_sign method.
12654 2013-04-12 Richard Biener <rguenther@suse.de>
12656 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12657 not be considered a gimple constant.
12659 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
12661 * fold-const.c (const_binop): Handle vector shifts by a scalar.
12662 (fold_binary_loc): Call const_binop also for mixed vector-scalar
12665 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
12666 Jakub Jelinek <jakub@redhat.com>
12668 * opts.c: Include diagnostic-color.h.
12669 (common_handle_option): Handle OPT_fdiagnostics_color_.
12670 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12671 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12672 (diagnostic-color.o): New.
12673 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12674 (diagnostic_color_rule): New enum.
12675 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12676 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12677 the location string.
12678 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12679 either NULL, or color kind.
12680 * diagnostic-color.c: New file.
12681 * diagnostic-color.h: New file.
12682 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12684 * doc/invoke.texi (-fdiagnostics-color): Document.
12685 * pretty-print.h (pp_show_color): Define.
12686 (struct pretty_print_info): Add show_color field.
12687 * diagnostic.c: Include diagnostic-color.h.
12688 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12689 macros. Colorize error:, warning: etc. strings and also the location
12691 (diagnostic_show_locus): Colorize the caret line.
12692 * pretty-print.c: Include diagnostic-color.h.
12693 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
12694 inside of %< %> quotes or quoted through q format modifier.
12696 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12698 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12700 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12702 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12703 code in CC_NZ mode.
12704 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12707 2013-04-11 Marek Polacek <polacek@redhat.com>
12709 PR tree-optimization/48184
12710 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12712 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
12714 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12715 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12716 (skip_simple_arithmetic): Tidy up.
12717 * tree.h (skip_simple_constant_arithmetic): Declare.
12719 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12721 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12723 2013-04-11 Richard Biener <rguenther@suse.de>
12725 * tree-vect-loop.c (get_initial_def_for_induction): Properly
12726 generate vector constants.
12728 2013-04-11 Richard Biener <rguenther@suse.de>
12730 PR tree-optimization/56878
12731 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12732 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12734 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12735 Prefer to align the DR with the most invariant base address.
12737 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
12739 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12742 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
12744 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12745 floating-point vector comparisons against 0.
12747 2013-04-11 Jakub Jelinek <jakub@redhat.com>
12749 PR tree-optimization/56899
12750 * fold-const.c (extract_muldiv_1): Apply distributive law
12751 only if TYPE_OVERFLOW_WRAPS (ctype).
12753 2013-04-11 Bin Cheng <bin.cheng@arm.com>
12756 * ira-costs.c (scan_one_insn): Check whether the source rtx of
12757 loading has side effect.
12759 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
12761 * config/sparc/sparc.c: Include tree-pass.h.
12762 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12763 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
12764 head of file. Change return type. Split off gate function.
12765 (sparc_gate_work_around_errata): New function.
12766 (pass_work_around_errata): New pass definition.
12767 (insert_pass_work_around_errata) New pass insert definition to
12768 insert pass_work_around_errata just after delayed-branch scheduling.
12769 (sparc_option_override): Insert the pass.
12770 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12772 2013-04-10 David S. Miller <davem@davemloft.net>
12774 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12775 or -mcpu=hypersparc.
12777 * target.def (cstore_mode): New hook.
12778 * target.h: Include insn-codes.h
12779 * targhooks.c: Likewise.
12780 (default_cstore_mode): New function.
12781 * targhooks.h: Declare it.
12782 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12783 * doc/tm.texi: Rebuild.
12784 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12785 target hook, rather than inspecting the insn_data.
12786 * config/sparc/sparc.c (sparc_cstore_mode): New function.
12787 (TARGET_CSTORE_MODE): Redefine.
12788 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12790 * config/sparc/predicates.md (cstore_result_operand): New special
12792 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12793 Use it for operand 0.
12794 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12795 (*snesi_special): Likewise.
12796 (*snesi_zero): Likewise.
12797 (*seqsi_zero): Likewise.
12798 (*sltu_insn): Likewise.
12799 (*sgeu_insn): Likewise.
12800 (*seqdi_special): Make operand 0 and comparison operation be of
12802 (*snedi_special): Likewise.
12803 (*snedi_special_vis3): Likewise.
12804 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12805 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12806 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12807 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12808 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12809 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12810 (*sltu_extend_sp64): Likewise.
12811 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12812 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12813 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12814 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12815 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12817 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
12819 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12820 (aarch64_start_file): Use the new function.
12822 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
12823 Jason Merrill <jason@redhat.com>
12825 * common.opt: Add -gdwarf.
12826 * opts.c (common_handle_option): Handle it.
12827 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12829 2013-04-10 Richard Biener <rguenther@suse.de>
12831 * passes.c (execute_todo): Do not call ggc_collect conditional here.
12832 (execute_one_ipa_transform_pass): But unconditionally here.
12833 (execute_one_pass): And here.
12834 (init_optimization_passes): Remove reload pass.
12835 * tree-pass.h (TODO_ggc_collect): Remove.
12836 (pass_reload): Likewise.
12837 * ira.c (do_reload): Merge into ...
12839 (rest_of_handle_reload): Remove.
12840 (pass_reload): Likewise.
12841 * config/i386/i386.c (ix86_option_override): Refer to ira instead
12842 of reload for vzeroupper pass placement.
12843 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12844 and todo_flags_finish of all passes.
12846 2013-04-10 Richard Biener <rguenther@suse.de>
12848 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12849 first_const_oprnd field, rename first_def_type to first_op_type.
12850 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12851 (vect_get_and_check_slp_defs): Always use the type of the
12852 operand. Allow mixed vect_external_def, vect_constant_def types.
12853 (vect_get_constant_vectors): Handle mixed vect_external_def,
12854 vect_constant_def types.
12856 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
12858 PR tree-optimization/55524
12859 * tree-ssa-math-opts.c
12860 (convert_mult_to_fma): Don't use an fms construct
12861 when we don't have an fms operation, but fnma, and it looks
12862 likely that we'll be able to use the latter.
12864 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
12866 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12868 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12869 inline fail caused by overwritable functions.
12871 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
12873 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12874 unnecessary bits in the constant power of two case.
12876 2013-04-10 Richard Biener <rguenther@suse.de>
12878 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12879 broken code swapping operands.
12880 (vect_build_slp_tree): Do not compute load permutations here.
12881 (vect_analyze_slp_instance): Compute load permutations here,
12882 after building the SLP tree.
12884 2013-04-09 Christian Bruel <christian.bruel@st.com>
12886 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12887 of next/prev_real_insn.
12889 2013-04-09 Jan Hubicka <jh@suse.cz>
12891 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12892 Drop aliased parameter.
12893 (function_and_variable_visibility): Do not handle alias pairs.
12894 * cgraph.c (varpool_externally_visible_p): Update prototype.
12895 * varpool.c (varpool_add_new_variable): Update.
12897 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12899 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12901 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
12903 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12905 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12907 2013-04-09 Marek Polacek <polacek@redhat.com>
12909 PR tree-optimization/48762
12910 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12912 2013-04-09 Richard Biener <rguenther@suse.de>
12914 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12916 (vect_build_slp_tree): Likewise.
12917 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12918 calculating the cost of a SLP instance.
12919 (vect_analyze_slp_instance): Use it from here, after building
12922 2013-04-09 Jakub Jelinek <jakub@redhat.com>
12924 PR middle-end/56883
12925 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12926 expand_omp_for_static_chunk): Use simple_p = true in
12927 force_gimple_operand_gsi calls when assigning to addressable decls.
12929 2013-04-09 Jeff Law <law@redhat.com>
12931 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12932 when the boolean was created by converting a wider object which
12933 had a boolean range.
12935 2013-04-09 Richard Biener <rguenther@suse.de>
12937 * tree-vectorizer.h (slp_void_p): Remove.
12938 (slp_tree): Typedef before _slp_tree declaration.
12939 (struct _slp_tree): Use a vector of slp_tree as children.
12940 (vect_get_place_in_interleaving_chain): Remove.
12941 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12943 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12945 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12946 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12947 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12948 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12949 Use slp_node instead of slp_void_p and adjust.
12951 2013-04-09 Richard Biener <rguenther@suse.de>
12953 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12954 work that is not necessary.
12956 2013-04-09 Jakub Jelinek <jakub@redhat.com>
12958 PR tree-optimization/56854
12959 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12960 forward into clobber stmts if it would change MEM_REF lhs into
12963 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
12965 * tree.c (type_hash_lookup, type_hash_add): Make static.
12966 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12968 2013-04-09 Richard Biener <rguenther@suse.de>
12970 * tree.h (unsave_expr_now): Remove.
12971 * tree-inline.c (mark_local_for_remap_r): Remove.
12972 (unsave_expr_1): Likewise.
12973 (unsave_r): Likewise.
12974 (unsave_expr_now): Likewise.
12975 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12976 (propagate_tree_value): Likewise.
12978 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
12980 * doc/rtl.texi (sequence): Rewrite documentation to match the
12981 current use of SEQUENCE rtl objects.
12982 * rtl.def (SEQUENCE): Likewise.
12984 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12985 Update documentation.
12986 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12987 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12989 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12991 2013-04-08 Teresa Johnson <tejohnson@google.com>
12993 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12994 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12996 (estimate_edge_size_and_time): Add comment to suggest using rounding
12998 (estimate_node_size_and_time): Ditto.
12999 (remap_edge_change_prob): Use helper rounding divide methods.
13000 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
13001 (gimple_mod_pow2_value_transform): Ditto.
13002 (gimple_mod_subtract_transform): Ditto.
13003 (gimple_ic_transform): Ditto.
13004 (gimple_stringops_transform): Ditto.
13005 * stmt.c (conditional_probability): Ditto.
13006 (emit_case_dispatch_table): Ditto.
13007 * lto-cgraph.c (merge_profile_summaries): Ditto.
13008 * tree-optimize.c (execute_fixup_cfg): Ditto.
13009 * cfgcleanup.c (try_forward_edges): Ditto.
13010 * cfgloopmanip.c (scale_loop_profile): Ditto.
13012 (duplicate_loop_to_header_edge): Ditto.
13013 (lv_adjust_loop_entry_edge): Ditto.
13014 * tree-vect-loop.c (vect_transform_loop): Ditto.
13015 * profile.c (compute_branch_probabilities): Ditto.
13016 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
13017 * lto-streamer-in.c (input_cfg): Ditto.
13018 * gimple-streamer-in.c (input_bb): Ditto.
13019 * ipa-cp.c (update_profiling_info): Ditto.
13020 (update_specialized_profile): Ditto.
13021 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
13022 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
13024 * sched-rgn.c (compute_dom_prob_ps): Ditto.
13025 (compute_trg_info): Ditto.
13026 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
13027 (purge_dead_edges): Ditto.
13028 * loop-unswitch.c (unswitch_loop): Ditto.
13029 * cgraphclones.c (cgraph_clone_edge): Ditto.
13030 (cgraph_clone_node): Ditto.
13031 * tree-inline.c (copy_bb): Ditto.
13032 (copy_edges_for_bb): Ditto.
13033 (initialize_cfun): Ditto.
13034 (copy_cfg_body): Ditto.
13035 (expand_call_inline): Ditto.
13037 2013-04-08 Kai Tietz <ktietz@redhat.com>
13039 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
13040 TARGET_CYGWIN64 by TARGET_64BIT.
13042 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
13044 * config/epiphany/epiphany.md (GPR_1): New constant.
13045 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
13046 * config/epiphany/epiphany.c (gen_compare_reg):
13047 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
13048 is already in place.
13049 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
13050 Don't require being called during rtl expansion; If y operlaps r0,
13052 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
13053 (epiphany_expand_epilogue): Likewise.
13055 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
13056 Don't use CC_FPmode for ORDERED / UNORDERED.
13057 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
13059 * config/epiphany/constraints.md (CnL): New constraint.
13060 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
13061 * config/epiphany/predicates.md (add_operand): Allow 1024.
13063 * config/epiphany/epiphany.md (logical_op): New code iterator.
13064 (op_mnc): New code attribute.
13065 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
13066 (mov_f+1, mov_f+2): New peephole2 patterns.
13068 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
13069 (cstoresi4): Also allow re-use of zero result when doing a NE
13070 comparison to a non-zero operand.
13071 Use (clobber (scratch)) for first insn if the gpr output is not needed.
13073 * config/epiphany/epiphany.md (<insn_opname>v2si3):
13074 Use gen_addsi3_i / gen_subsi3_i.
13076 2013-04-08 Jakub Jelinek <jakub@redhat.com>
13080 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
13081 contain anything but clobbers, at most one __builtin_stack_restore,
13082 optionally debug stmts and final resx, and if it has at least one
13083 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
13084 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
13085 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
13086 which isn't defaut definition, remove them.
13087 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
13088 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
13089 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
13090 with MEM_REF LHS with SSA_NAME address.
13092 2013-04-08 Jeff Law <law@redhat.com>
13094 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
13096 2013-04-08 Richard Biener <rguenther@suse.de>
13098 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
13100 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
13101 determined vector type.
13102 (vect_analyze_data_refs): Likewise.
13103 (vect_get_new_vect_var): Adjust.
13104 (vect_create_destination_var): Preserve SSA name versions.
13105 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
13106 not dump anything here.
13108 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
13110 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
13111 Add member lr_slot_known.
13112 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
13114 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
13115 Remove code that sets lr_slot_offset according to what a previous
13116 version of epiphany_emit_save_restore used to do.
13117 (epiphany_emit_save_restore): When doing an lr save or restore,
13118 set/verify lr_slot_known and lr_slot_offset.
13120 2013-04-08 Xinyu Qi <xyqi@marvell.com>
13123 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
13126 2013-04-08 Richard Biener <rguenther@suse.de>
13128 * alias.c (find_base_term): Fix thinko in previous change.
13130 2013-04-08 Jakub Jelinek <jakub@redhat.com>
13132 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
13133 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
13134 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
13135 if possible to compute val.
13136 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
13137 For QImode integers don't require anything about precision. Use
13138 const_with_all_bytes_same to find out if the constant doesn't have
13139 repeated bytes in it.
13141 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
13143 * config/s390/s390.c (s390_expand_insv): Only accept insertions
13146 2013-04-08 Marek Polacek <polacek@redhat.com>
13148 PR rtl-optimization/48182
13149 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
13152 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
13155 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
13156 nuses, make sure we have a label.
13158 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13161 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
13162 (rs6000_emit_swdiv_low_precision): Remove.
13163 (rs6000_emit_swdiv): Rewrite to handle between one and four
13164 iterations of Newton-Raphson generally; modify required number of
13165 iterations for some cases.
13166 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
13168 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
13170 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
13171 set-but-unused variable.
13173 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
13174 basic blocks of released function bodies garbage-collectable.
13176 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
13177 (struct rtl_opt_pass): Add TODO_df_finish.
13179 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
13181 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13183 * config/arm/constraints.md (q): New constraint.
13184 * config/arm/ldrdstrd.md: New file.
13185 * config/arm/arm.md (ldrdstrd.md) New include.
13186 (arm_movdi): Use "q" instead of "r" constraint
13187 for double-word memory access.
13188 (movdf_soft_insn): Likewise.
13189 * config/arm/vfp.md (movdi_vfp): Likewise.
13190 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
13191 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
13192 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
13193 (mem_ok_for_ldrd_strd): Likewise.
13194 (output_move_double): Update assertion.
13196 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13198 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
13200 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13202 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
13203 define_insn_and_split.
13204 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
13206 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13208 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
13209 define_insn_and_split.
13210 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
13211 (shiftsi3_compare): New pattern.
13212 (rrx): New pattern.
13213 * config/arm/unspecs.md (UNSPEC_RRX): New.
13215 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13217 * config/arm/arm.md (negdi_extendsidi): New pattern.
13218 (negdi_zero_extendsidi): Likewise.
13220 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13222 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
13223 define_insn_and_split.
13224 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
13225 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
13227 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13229 * config/arm/arm.md (arm_subdi3): Convert define_insn into
13230 define_insn_and_split.
13231 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
13232 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
13234 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13236 * config/arm/arm.md (subsi3_carryin): New pattern.
13237 (subsi3_carryin_const): Likewise.
13238 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
13239 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
13241 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13243 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
13245 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13247 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
13248 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
13250 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13252 * config/arm/arm.c (arm_expand_builtin): Change fcode
13253 type to unsigned int.
13255 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13257 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
13259 2013-04-04 Ian Lance Taylor <iant@google.com>
13261 * doc/standards.texi (Standards): The Go frontend supports the Go 1
13264 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
13266 PR middle-end/56729
13267 * df-scan.c (df_insn_delete): Disable failing assert.
13269 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13271 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
13272 New function prototype.
13273 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
13274 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
13275 (arm_builtin_vectorized_function): New function.
13277 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13279 * config/arm/arm_neon_builtins.def: New file.
13280 * config/arm/arm.c (neon_builtin_data): Move contents to
13281 arm_neon_builtins.def.
13282 (enum arm_builtins): Include neon builtin definitions.
13283 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
13284 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
13286 2013-04-04 Marek Polacek <polacek@redhat.com>
13288 PR tree-optimization/48186
13289 * predict.c (maybe_hot_frequency_p): Return false if
13290 HOT_BB_FREQUENCY_FRACTION is 0.
13291 (cgraph_maybe_hot_edge_p): Likewise.
13293 2013-04-04 Richard Biener <rguenther@suse.de>
13295 PR tree-optimization/56826
13296 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
13299 2013-04-04 Richard Biener <rguenther@suse.de>
13301 PR tree-optimization/56213
13302 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
13303 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
13305 2013-04-04 Richard Biener <rguenther@suse.de>
13307 PR tree-optimization/56837
13308 * tree-loop-distribution.c (classify_partition): For non-zero
13309 values require that the value has the same precision as its
13310 mode to be useful as memset value.
13312 2013-04-03 Nick Clifton <nickc@redhat.com>
13314 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
13315 (fmssf4): Use fmsf.s on E3V5 architectures.
13316 (fnmasf4): Use fnmaf.s on E3V5 architectures.
13317 (fnmssf4): Use fnmsf.s on E3V5 architectures.
13319 2013-04-03 Jeff Law <law@redhat.com>
13321 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
13322 (lra-eliminations.o): Likewise.
13324 2013-04-03 Teresa Johnson <tejohnson@google.com>
13326 * gcov-io.c (compute_working_sets): Moved most of body of old
13327 compute_working_sets here from profile.c.
13328 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
13329 (gcov_working_set_t): Moved typedef here from basic-block.h
13330 (compute_working_set): Declare.
13331 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
13332 (get_working_sets): Renamed from compute_working_set,
13333 replace most of body with call to new compute_working_sets.
13334 (get_exec_counts): Replace call to compute_working_sets
13335 to get_working_sets.
13336 * profile.h (get_working_sets): Renamed from compute_working_set.
13337 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
13338 to get_working_sets.
13339 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
13340 * gcov-dump.c (dump_working_sets): New function.
13342 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
13344 * hwint.c (sext_hwi, zext_hwi): New functions.
13345 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
13346 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
13347 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
13348 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
13349 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
13350 (sext_hwi, zext_hwi): New functions.
13352 2013-04-03 Jeff Law <law@redhat.com>
13354 PR tree-optimization/56799
13355 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
13356 back test for widening conversion erroneously dropped in prior change.
13358 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13361 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
13362 instead of next_real_insn.
13364 2013-04-03 Marek Polacek <polacek@redhat.com>
13367 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13369 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13372 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
13374 (thumb1_output_casesi): Likewise.
13375 (thumb2_output_casesi): Likewise.
13377 2013-04-03 Richard Biener <rguenther@suse.de>
13379 PR tree-optimization/56817
13380 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
13382 (tree_unroll_loops_completely_1): ... new function to manually
13383 walk the loop tree, properly defering outer loops of unrolled
13384 loops to later iterations.
13386 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13388 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
13389 (vectorizable_load): Likewise.
13390 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
13391 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
13393 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13395 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
13398 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
13400 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
13402 2013-04-03 Bin Cheng <bin.cheng@arm.com>
13404 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13406 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13408 PR tree-optimization/56790
13409 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13412 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13414 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13416 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13417 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13420 2013-04-03 Jakub Jelinek <jakub@redhat.com>
13423 * tree.h (force_folding_builtin_constant_p): New decl.
13424 * builtins.c (force_folding_builtin_constant_p): New variable.
13425 (fold_builtin_constant_p): Fold immediately also if
13426 force_folding_builtin_constant_p.
13428 2013-04-03 Richard Biener <rguenther@suse.de>
13430 PR tree-optimization/56812
13431 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13432 DRs of the same interleaving chain are independent.
13434 2013-04-02 Jason Merrill <jason@redhat.com>
13436 * gdbinit.in (pbb): Use debug fn.
13438 2013-04-02 Lawrence Crowl <crowl@google.com>
13440 * sese.h (struct ivtype_map_elt_s): Remove unused.
13441 (extern debug_ivtype_map): Remove unused.
13442 (extern eq_ivtype_map_elts): Remove unused.
13443 * sese.c (debug_ivtype_map): Removed unused.
13444 (debug_ivtype_map_1): Removed unused.
13445 (debug_ivtype_elt): Remove unused.
13446 (eq_ivtype_map_elts): Remove unused.
13449 2013-04-02 Kai Tietz <ktietz@redhat.com>
13452 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13453 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13454 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13456 (legitimize_pe_coff_symbol): Likewise.
13457 (is_imported_p): New helper-function.
13458 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13459 for Windows x64 targets.
13460 (ix86_expand_prologue): Optimize for pe-coff targets.
13461 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13462 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13463 medium/large code-model.
13464 (legitimize_pic_address): Likewise.
13465 (legitimize_tls_address): Likewise.
13466 (ix86_expand_call): Likewise.
13467 (x86_output_mi_thunk): Likewise.
13468 (get_dllimport_decl): Add new beimport argument.
13469 (construct_plt_address): Don't assert for x64 pe-coff targets.
13470 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13472 (SYMBOL_FLAG_STUBVAR): New macro.
13473 (SYMBOL_REF_STUBVAR_P): Likewise.
13474 * config/i386/winnt.c (stub_list): New structure.
13475 (stub_head): New local variable.
13476 (i386_pe_record_stub): New function.
13477 (i386_pe_file_end): Emit refptr-stubs.
13479 2013-04-02 Jakub Jelinek <jakub@redhat.com>
13481 PR rtl-optimization/56745
13482 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13483 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13486 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13487 and both of them are MEM_REFs, just compare first argument for
13488 equality and attempt to deal even with differing offsets.
13491 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13492 of gimple_clobber_p to be MEM_REF.
13493 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13494 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
13495 after gimplification.
13496 * asan.c (get_mem_ref_of_assignment): Don't instrument
13497 gimple_clobber_p stmts.
13498 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13499 gimple_clobber_p stmt if they have MEM_REF lhs and
13500 are dead because of another gimple_clobber_p stmt.
13501 * tree-ssa-live.c (clear_unused_block_pointer): Treat
13502 gimple_clobber_p stmts like debug stmts.
13503 (remove_unused_locals): Remove clobbers with MEM_REF lhs
13504 that refer to unused VAR_DECLs or uninitialized values.
13505 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13506 gimple_clobber_p stmts if they refer to removed parameters.
13507 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13510 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
13512 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13513 using SWI48 mode attribute.
13515 2013-04-02 Wei Mi <wmi@google.com>
13517 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13518 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13519 *<rotate_insn><mode>3_mask in i386.md.
13521 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
13523 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13525 2013-04-02 Richard Biener <rguenther@suse.de>
13527 PR tree-optimization/56778
13528 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13529 Runtime alias tests are not supported for gather loads.
13530 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13531 stmts referenced from SSA operands before updating SSA form.
13533 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
13534 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13536 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13537 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13538 * config/arm/cortex-a53.md: New file.
13539 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13540 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13541 * config/arm/arm.c (arm_issue_rate): Likewise.
13542 * config/arm/arm-tune.md: Regenerate
13543 * config/arm/arm-tables.opt: Regenerate.
13544 * config/arm/arm-cores.def: Add cortex-a53.
13546 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
13548 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13551 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
13553 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13554 scalar load/store operations using B/H registers.
13555 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13557 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
13559 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13561 * config/aarch64/aarch64.c
13562 (aarch64_simd_scalar_immediate_valid_for_move): New.
13563 * config/aarch64/aarch64-protos.h
13564 (aarch64_simd_scalar_immediate_valid_for_move): New.
13565 * config/aarch64/constraints.md (Dh, Dq): New.
13566 * config/aarch64/iterators.md (hq): New.
13568 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
13570 * reorg.c (get_branch_condition): Deal with conditional returns.
13571 (fill_simple_delay_slots): Remove dead code dealing with jumps.
13573 2013-04-01 Wei Mi <wmi@google.com>
13575 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13576 Truncate operand 2 using %b asm operand modifier.
13577 (*<shift_insn><mode>3_mask): Ditto.
13578 (*<rotate_insn><mode>3_mask): Ditto.
13580 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
13582 PR middle-end/56798
13583 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13585 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
13587 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13589 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13591 2013-03-30 Lawrence Crowl <crowl@google.com>
13593 * dse.c (clear_alias_sets): Remove never set.
13594 (disqualified_clear_alias_sets): Remove never set.
13595 (clear_alias_mode_pool): Remove never set.
13596 (dse_step0): Remove condition that is never true.
13597 (canon_address): Remove condition that is never true.
13598 (dse_step7): Remove condition that is never true.
13599 (rest_of_handle_dse): Remove condition that is never true.
13600 (rest_of_handle_dse::did_global): Remove never read from above.
13601 (dse_step2_spill): Remove never called from above.
13602 (dse_step5_spill): Remove never called from above.
13604 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
13606 * doc/md.texi (Standard Names) <casesi>: Update documentation for
13607 JUMP_TABLE_DATA changes.
13608 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13609 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13610 (Insns) <jump_table_data>: New entry.
13611 * doc/tm.texi: Regenerate.
13613 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13615 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13616 for table jump at the end of a basic block using tablejump_p.
13617 * targhooks.c (default_invalid_within_doloop): Likewise.
13618 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13619 target hook implementation that is identical to the default hook.
13620 (rs6000_invalid_within_doloop): Remove.
13622 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13623 unused variable from tablejump_p call.
13625 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13626 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13627 (INSN_DELETED_P): Likewise.
13628 (emit_jump_table_data): New prototype.
13629 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13630 after 4th as unused.
13631 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13632 * sched-vis.c (print_insn): Likewise.
13633 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13634 insn for compatibility with back ends that use next_active_insn to
13635 identify jump table data.
13636 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13637 (remove_insn): Likewise.
13638 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13640 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13641 (emit_jump_table_data): New function.
13643 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13644 basic block, a JUMP_TABLE_DATA never is.
13645 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13646 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13647 off from code handling real insns.
13648 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13649 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13650 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13651 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13652 is not a NONDEBUG_INSN_P.
13653 * ira-costs.c (scan_one_insn): Likewise.
13654 * jump.c (mark_all_labels): Likewise.
13655 (mark_jump_label_1): Likewise.
13656 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13657 * lra.c (get_insn_freq): Expect all insns reaching here to be in
13659 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13660 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13661 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13662 JUMP_TABLE_DATA_P insns.
13663 (calculate_elim_costs_all_insns): Likewise.
13664 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13665 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13666 (delete_output_reload): Code style fixups.
13667 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13668 insn flags on this non-insn.
13669 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13670 as scheduling barriers, for pre-change compatibility.
13671 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13672 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13674 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13675 redundant JUMP_TABLE_DATA_P test.
13676 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13677 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13678 (frv_for_each_packet): Likewise.
13679 * config/i386/i386.c (min_insn_size): Likewise.
13680 (ix86_avoid_jump_mispredicts): Likewise.
13681 * config/m32r/m32r.c (m32r_is_insn): Likewise.
13682 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13683 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13684 (mips16_insn_length): Robustify.
13685 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13686 (mips16_split_long_branches): Likewise.
13687 * config/pa/pa.c (pa_combine_instructions): Likewise.
13688 * config/rs6000/rs6000.c (get_next_active_insn): Treat
13689 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13690 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13691 as contributing to pool range lengths.
13692 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13693 Remove redundant JUMP_TABLE_DATA_P test.
13694 (sh_loop_align): Likewise.
13695 (split_branches): Likewise.
13696 (sh_insn_length_adjustment): Likewise.
13697 * config/spu/spu.c (get_branch_target): Likewise.
13699 2013-03-29 Jan Hubicka <jh@suse.cz>
13701 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13702 gcov streaming; stream hot bb threshold to ltrans.
13703 * predict.c (get_hot_bb_threshold): Break out from ....
13704 (maybe_hot_count_p): ... here.
13705 (set_hot_bb_threshold): New function.
13706 * lto-section-in.c (lto_section_name): Add profile.
13707 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13708 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13709 and data-streamer.h
13710 (histogram_entry): New structure.
13711 (histogram, histogram_pool): New global vars.
13712 (histogram_hash): New structure.
13713 (histogram_hash::hash): New method.
13714 (histogram_hash::equal): Likewise.
13715 (account_time_size): New function.
13716 (cmp_counts): New function.
13717 (dump_histogram): New function.
13718 (ipa_profile_generate_summary): New function.
13719 (ipa_profile_write_summary): New function.
13720 (ipa_profile_read_summary): New function.
13721 (ipa_profile): Decide on threshold.
13722 (pass_ipa_profile): Add ipa_profile_write_summary and
13723 ipa_profile_read_summary.
13724 * Makefile.in (ipa.o): Update dependencies.
13725 * lto-streamer.h (LTO_section_ipa_profile): New section.
13727 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
13729 * tree.h (VAR_P): New.
13731 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
13734 * doc/invoke.texi ([-fwhole-program]): Fix typo.
13736 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
13738 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13739 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13740 (control_flow_insn_p): Likewise.
13741 * cfgrtl.c (duplicate_insn_chain): Likewise.
13742 * final.c (get_attr_length_1): Likewise.
13743 (shorten_branches): Likewise.
13744 (final_scan_insn): Likewise.
13745 * function.c (instantiate_virtual_regs): Likewise.
13746 * gcse.c (insert_insn_end_basic_block): Likewise.
13747 * ira-costs.c (scan_one_insn): Likewise.
13748 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13749 * lra.c (check_rtl): Likewise.
13750 * reload1.c (elimination_costs_in_insn): Likewise.
13751 * reorg.c (follow_jumps): Likewise.
13753 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13754 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13755 (thumb_far_jump_used_p): Likewise.
13756 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13757 (workaround_speculation): Likewise.
13758 (add_sched_insns_for_speculation): Likewise.
13759 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13760 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13761 (frv_for_each_packet): Likewise.
13762 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13763 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13764 (final_emit_insn_group_barriers): Likewise.
13765 * config/m32r/m32r.c (m32r_is_insn): Likewise.
13766 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13767 (mips16_insn_length): Likewise.
13768 * config/pa/pa.c (pa_reorg): Likewise.
13769 (pa_combine_instructions): Likewise.
13770 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13771 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13772 (sh_reorg): Likewise.
13773 (split_branches): Likewise.
13774 * config/spu/spu.c (get_branch_target): Likewise.
13776 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13779 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
13781 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13782 Fix declaration name.
13784 2013-03-28 Lawrence Crowl <crowl@google.com>
13786 * graphds.h (struct graph.indicies): Remove unused.
13787 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13788 (SCOP_ORIGINAL_PDDRS): Remove unused.
13789 * sese.h (extern insert_loop_close_phis): Removed unused.
13790 (extern insert_guard_phis): Removed unused.
13791 (extern ivtype_map_elt_info): Removed unused.
13792 (new_ivtype_map_elt): Removed unused.
13793 * sese.c (ivtype_map_elt_info): Removed unused.
13795 2013-03-28 Lawrence Crowl <crowl@google.com>
13797 * Makefile.in: Add several missing include dependences.
13799 (test-dump.o): New. This object is not added to any executable,
13800 but is present for ad-hoc testing.
13802 (debug (const bitmap_head_def &)): New.
13803 (debug (const bitmap_head_def *)): New.
13805 (extern debug (const bitmap_head_def &)): New.
13806 (extern debug (const bitmap_head_def *)): New.
13808 (debug (edge_def &)): New.
13809 (debug (edge_def *)): New.
13811 (debug (basic_block_def &)): New.
13812 (debug (basic_block_def *)): New.
13814 (dump_node (const_tree, int, FILE *)): Correct source file.
13816 (debug (die_struct &)): New.
13817 (debug (die_struct *)): New.
13819 (extern debug (die_struct &)): New.
13820 (extern debug (die_struct *)): New.
13821 * gimple-pretty-print.c
13822 (debug (gimple_statement_d &)): New.
13823 (debug (gimple_statement_d *)): New.
13824 * gimple-pretty-print.h
13825 (extern debug (gimple_statement_d &)): New.
13826 (extern debug (gimple_statement_d *)): New.
13828 (debug (ira_allocno_copy &)): New.
13829 (debug (ira_allocno_copy *)): New.
13830 (debug (ira_allocno &)): New.
13831 (debug (ira_allocno *)): New.
13833 (extern debug (ira_allocno_copy &)): New.
13834 (extern debug (ira_allocno_copy *)): New.
13835 (extern debug (ira_allocno &)): New.
13836 (extern debug (ira_allocno *)): New.
13838 (debug (live_range &)): New.
13839 (debug (live_range *)): New.
13841 (debug (lra_live_range &)): New.
13842 (debug (lra_live_range *)): New.
13844 (debug (lra_live_range &)): New.
13845 (debug (lra_live_range *)): New.
13847 (debug (omega_pb_d &)): New.
13848 (debug (omega_pb_d *)): New.
13850 (extern debug (omega_pb_d &)): New.
13851 (extern debug (omega_pb_d *)): New.
13853 (debug (const rtx_def &)): New.
13854 (debug (const rtx_def *)): New.
13856 (debug_tree (tree): Move within file.
13857 (debug_raw (const tree_node &)): New.
13858 (debug_raw (const tree_node *)): New.
13859 (dump_tree_via_hooks (const tree_node *, int)): New.
13860 (debug (const tree_node &)): New.
13861 (debug (const tree_node *)): New.
13862 (debug_verbose (const tree_node &)): New.
13863 (debug_verbose (const tree_node *)): New.
13864 (debug_head (const tree_node &)): New.
13865 (debug_head (const tree_node *)): New.
13866 (debug_body (const tree_node &)): New.
13867 (debug_body (const tree_node *)): New.
13868 (debug_vec_tree (tree): Move and reimplement in terms of dump.
13869 (debug (vec<tree, va_gc> &)): New.
13870 (debug (vec<tree, va_gc> *)): New.
13872 (extern debug (const rtx_def &)): New.
13873 (extern debug (const rtx_def *)): New.
13875 (debug_raw (simple_bitmap_def &)): New.
13876 (debug_raw (simple_bitmap_def *)): New.
13877 (debug (simple_bitmap_def &)): New.
13878 (debug (simple_bitmap_def *)): New.
13880 (extern debug (simple_bitmap_def &)): New.
13881 (extern debug (simple_bitmap_def *)): New.
13882 (extern debug_raw (simple_bitmap_def &)): New.
13883 (extern debug_raw (simple_bitmap_def *)): New.
13885 (debug (vinsn_def &)): New.
13886 (debug (vinsn_def *)): New.
13887 (debug_verbose (vinsn_def &)): New.
13888 (debug_verbose (vinsn_def *)): New.
13889 (debug (expr_def &)): New.
13890 (debug (expr_def *)): New.
13891 (debug_verbose (expr_def &)): New.
13892 (debug_verbose (expr_def *)): New.
13893 (debug (vec<rtx> &)): New.
13894 (debug (vec<rtx> *)): New.
13896 (extern debug (vinsn_def &)): New.
13897 (extern debug (vinsn_def *)): New.
13898 (extern debug_verbose (vinsn_def &)): New.
13899 (extern debug_verbose (vinsn_def *)): New.
13900 (extern debug (expr_def &)): New.
13901 (extern debug (expr_def *)): New.
13902 (extern debug_verbose (expr_def &)): New.
13903 (extern debug_verbose (expr_def *)): New.
13904 (extern debug (vec<rtx> &)): New.
13905 (extern debug (vec<rtx> *)): New.
13907 (_list_iter_cond_expr): Make inline instead of static.
13909 (debug (sreal &)): New.
13910 (debug (sreal *)): New.
13912 (extern debug (sreal &)): New.
13913 (extern debug (sreal *)): New.
13915 (extern debug_raw (const tree_node &)): New.
13916 (extern debug_raw (const tree_node *)): New.
13917 (extern debug (const tree_node &)): New.
13918 (extern debug (const tree_node *)): New.
13919 (extern debug_verbose (const tree_node &)): New.
13920 (extern debug_verbose (const tree_node *)): New.
13921 (extern debug_head (const tree_node &)): New.
13922 (extern debug_head (const tree_node *)): New.
13923 (extern debug_body (const tree_node &)): New.
13924 (extern debug_body (const tree_node *)): New.
13925 (extern debug (vec<tree, va_gc> &)): New.
13926 (extern debug (vec<tree, va_gc> *)): New.
13928 (debug (struct loop &)): New.
13929 (debug (struct loop *)): New.
13930 (debug_verbose (struct loop &)): New.
13931 (debug_verbose (struct loop *)): New.
13932 * tree-dump.c: Add header dependence.
13934 (extern debug (struct loop &)): New.
13935 (extern debug (struct loop *)): New.
13936 (extern debug_verbose (struct loop &)): New.
13937 (extern debug_verbose (struct loop *)): New.
13939 (debug (data_reference &)): New.
13940 (debug (data_reference *)): New.
13941 (debug (vec<data_reference_p> &)): New.
13942 (debug (vec<data_reference_p> *)): New.
13943 (debug (vec<ddr_p> &)): New.
13944 (debug (vec<ddr_p> *)): New.
13946 (extern debug (data_reference &)): New.
13947 (extern debug (data_reference *)): New.
13948 (extern debug (vec<data_reference_p> &)): New.
13949 (extern debug (vec<data_reference_p> *)): New.
13950 (extern debug (vec<ddr_p> &)): New.
13951 (extern debug (vec<ddr_p> *)): New.
13953 (debug (pt_solution &)): New.
13954 (debug (pt_solution *)): New.
13956 (extern debug (pt_solution &)): New.
13957 (extern debug (pt_solution *)): New.
13959 (debug (_var_map &)): New.
13960 (debug (_var_map *)): New.
13961 (debug (tree_live_info_d &)): New.
13962 (debug (tree_live_info_d *)): New.
13964 (extern debug (_var_map &)): New.
13965 (extern debug (_var_map *)): New.
13966 (extern debug (tree_live_info_d &)): New.
13967 (extern debug (tree_live_info_d *)): New.
13969 2013-03-28 Jan Hubicka <jh@suse.cz>
13971 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13973 2013-03-28 Ian Bolton <ian.bolton@arm.com>
13975 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13976 record only when desired or required.
13978 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
13980 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13981 *vec_extractv2di_1_rex64. Use x64 isa attribute.
13983 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
13985 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13986 (*andsi3_compare0_uxtw): New pattern.
13987 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13988 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13990 2013-03-28 Jan Hubicka <jh@suse.cz>
13992 * data-streamer-in.c (streamer_read_gcov_count): New function.
13993 * gimple-streamer-out.c: Include value-prof.h.
13994 (output_gimple_stmt): Output histogram.
13995 (output_bb): Use streamer_write_gcov_count.
13996 * value-prof.c: Include data-streamer.h
13997 (dump_histogram_value): Add HIST_TYPE_MAX.
13998 (stream_out_histogram_value): New function.
13999 (stream_in_histogram_value): New function.
14000 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
14001 (stream_out_histogram_value, stream_in_histogram_value): Declare.
14002 * data-streamer-out.c (streamer_write_gcov_count): New function.
14003 (streamer_write_gcov_count_stream): New function.
14004 * lto-cgraph.c (lto_output_edge): Update counter streaming.
14005 (lto_output_node): Likewise.
14006 (input_node, input_edge): Likewise.
14007 * lto-streamer-out.c (output_cfg): Update streaming.
14008 * lto-streamer-in.c (input_cfg): Likewise.
14009 * data-streamer.h (streamer_write_gcov_count,
14010 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
14011 * gimple-streamer-in.c: Include value-prof.h
14012 (input_gimple_stmt): Input histograms.
14013 (input_bb): Update profile streaming.
14015 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
14017 * genmodes.c (emit_max_int): New function.
14018 (emit_insn_modes_h): Added call to emit_max_function.
14019 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
14021 * machmode.def: Fixed comment.
14023 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
14025 * combine.c (try_combine): Removed useless assert.
14026 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
14028 2013-03-28 Marek Polacek <polacek@redhat.com>
14029 Richard Biener <rguenther@suse.de>
14031 PR tree-optimization/56695
14032 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
14033 build signed result of a vector comparison.
14034 * tree-cfg.c (verify_gimple_comparison): Check that a result
14035 of a vector comparison has signed type.
14037 2013-03-28 Richard Biener <rguenther@suse.de>
14039 PR tree-optimization/37021
14040 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
14041 do not restrict gaps between groups.
14042 * tree-vect-stmts.c (vectorizable_load): Properly account for
14043 a gap between groups.
14045 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
14047 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
14048 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
14051 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
14053 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
14054 * doc/extend.texi (Named Address Spaces): Ditto.
14055 (Variable Attributes): Ditto.
14057 2013-03-27 Kai Tietz <ktietz@redhat.com>
14059 * config.build: Add support for cygwin x64 target.
14060 * config.gcc: Likewise.
14061 * config.host: Likewise.
14062 * configure.ac: Likewise
14063 * configure: Regenerated.
14065 2013-03-27 Kai Tietz <ktietz@redhat.com>
14067 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
14068 * config/i386/t-cygwin-w64: New file.
14069 * config/i386/cygwin-w64.h: New file.
14070 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
14071 and add support for x64-cygwin target.
14072 (CPP_SPEC): Likewise.
14073 (CXX_WRAP_SPEC_LIST): Undefine before define.
14074 (LIBGCJ_SONAME): Use 15 as version.
14076 2013-03-27 Richard Biener <rguenther@suse.de>
14078 PR tree-optimization/56716
14079 * tree-ssa-structalias.c (perform_var_substitution): Adjust
14080 dumping for ref nodes.
14082 2013-03-27 Martin Jambor <mjambor@suse.cz>
14084 PR tree-optimization/55334
14085 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
14086 restricted pointers to arrays.
14088 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
14090 * Makefile.in (.SUFFIXES): Add .cc.
14091 (.c.o): Apply same recipe for implicit rule .cc.o.
14093 2013-03-27 Richard Biener <rguenther@suse.de>
14095 PR tree-optimization/37021
14096 * tree-vect-data-refs.c (vect_check_strided_load): Allow
14097 REALPART/IMAGPART_EXPRs around the supported refs.
14098 * tree-ssa-structalias.c (find_func_aliases): Assume that
14099 floating-point values are not used to transfer pointers.
14101 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
14103 * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
14104 * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
14105 * gcc/doc/tm.texi: Regenerate.
14106 * gcc/targhooks.h (default_has_ifunc_p): New.
14107 * gcc/targhooks.c (default_has_ifunc_p): Ditto.
14108 * gcc/config/linux-protos.h: New file.
14109 * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
14110 this hook for linux which disables support of indirect functions in
14112 * gcc/config/linux-android.c: New file.
14113 * gcc/config/t-linux-android.c: Ditto.
14114 * gcc/config.gcc: Added new object file linux-android.o.
14115 * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
14116 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
14117 * gcc/varasm.c (do_assemble_alias): Likewise.
14118 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
14119 doesn't support indirect functions.
14120 * configure: Regenerate.
14122 2013-03-27 Bin Cheng <bin.cheng@arm.com>
14125 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
14126 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
14129 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
14131 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
14133 2013-03-27 Terry Guo <terry.guo@arm.com>
14135 * config/arm/arm-cores.def: Added core cortex-r7.
14136 * config/arm/arm-tune.md: Regenerated.
14137 * config/arm/arm-tables.opt: Regenerated.
14138 * doc/invoke.texi: Added entry for core cortex-r7.
14140 2013-03-27 Walter Lee <walt@tilera.com>
14142 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
14143 double-decrement of next_scratch_regno.
14145 2013-03-27 Walter Lee <walt@tilera.com>
14147 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
14149 (insn_v1mulus): Ditto.
14150 (insn_v2muls): Ditto.
14152 2013-03-27 Walter Lee <walt@tilera.com>
14154 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
14155 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
14157 2013-03-27 Walter Lee <walt@tilera.com>
14159 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
14160 (*sibcall_value): Ditto.
14162 2013-03-27 Walter Lee <walt@tilera.com>
14164 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
14165 (insn_mnz_v8qi): ... this ...
14166 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
14168 (insn_v<n>mnz): Replaced by ...
14169 (insn_v1mnz): ... this ...
14170 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
14172 (insn_mz_<mode>): Replaced by ...
14173 (insn_mz_v8qi): ... this ...
14174 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
14176 (insn_v<n>mz): Replaced by ...
14177 (insn_v1mz): ... this ...
14178 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
14181 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
14183 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
14185 2013-03-26 Roland McGrath <mcgrathr@google.com>
14187 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
14188 than fprintf with a non-constant, non-format string.
14190 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
14192 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
14193 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
14194 operand 0 predicate.
14195 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
14196 attribute. Use general_x64nomem_operand as operand 1 predicate.
14197 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
14198 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
14199 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
14200 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
14201 (mov<mode>_insv_1): Remove expander. Merge insn with
14202 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
14203 Use general_x64nomem_operand as operand 1 predicate.
14204 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
14205 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
14206 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
14207 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
14208 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
14209 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
14210 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
14211 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
14212 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
14213 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
14214 (general_x64nomem_operand): Ditto.
14216 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14218 * config/rtems.opt: Add -pthread option.
14220 2013-03-26 Richard Biener <rguenther@suse.de>
14222 * alias.c (find_base_term): Avoid redundant and not used recursion.
14223 (base_alias_check): Get the initial base term from the caller.
14224 (true_dependence_1): Compute and pass base terms to base_alias_check.
14225 (write_dependence_p): Likewise.
14226 (may_alias_p): Likewise.
14228 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
14230 * config/aarch64/aarch64.c (aarch64_classify_address): Support
14231 PC-relative load in SI modes and above only.
14233 2013-03-26 Xinyu Qi <xyqi@marvell.com>
14235 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
14236 * config/arm/iwmmxt.md (WCGR0): Update.
14237 (WCGR1, WCGR2, WCGR3): Likewise.
14239 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
14241 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
14242 Use x64 and nox64 isa attributes.
14244 2013-03-26 Richard Biener <rguenther@suse.de>
14246 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
14247 alignment computations and rely on get_object_alignment_1
14248 for the !TYPE_P case.
14249 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
14251 2013-03-26 Walter Lee <walt@tilera.com>
14253 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
14254 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
14256 2013-03-25 Jeff Law <law@redhat.com>
14258 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
14259 check for INTEGRAL_TYPE_P that was missing due to checking in
14260 wrong version of prior patch.
14262 2013-03-25 Walter Lee <walt@tilera.com>
14264 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
14265 TILEGX_INSN_SHUFFLEBYTES1.
14266 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
14268 (tilegx_builtins): Ditto.
14269 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
14271 2013-03-25 Walter Lee <walt@tilera.com>
14273 * config/tilegx/tilegx.md (floatsisf2): New pattern.
14274 (floatunssisf2): New pattern.
14275 (floatsidf2): New pattern.
14276 (floatunssidf2): New pattern.
14278 2013-03-25 Walter Lee <walt@tilera.com>
14280 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
14281 tests for constraint J, K, N, P.
14283 2013-03-25 Walter Lee <walt@tilera.com>
14285 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
14286 Use indirect/pcrel encoding.
14287 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
14290 2013-03-25 Steve Ellcey <sellcey@mips.com>
14292 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
14293 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
14294 * config/mips/mips.c (mips_option_override): Set IMADD default.
14295 * config/mips/mips.h (PTF_AVOID_IMADD): New.
14296 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
14297 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
14298 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
14299 * doc/invoke.texi (-mimadd/-mno-imadd): New.
14301 2013-03-25 Jeff Law <law@redhat.com>
14303 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
14304 slightly to avoid creating and folding useless trees. Simplify
14305 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
14307 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
14309 * config/i386/i386.md (*zero_extendsidi2): Merge with
14310 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
14311 * config/i386/predicates.md (x86_64_zext_operand): Rename from
14312 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
14313 targets. Clarify comment.
14315 2013-03-25 Martin Jambor <mjambor@suse.cz>
14317 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
14318 pass-through jump functions differently.
14319 (ipa_read_jump_function): Likewise. Also use setter functions to set
14322 2013-03-25 Martin Jambor <mjambor@suse.cz>
14324 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
14325 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
14327 (ipa_get_indirect_edge_target): New function.
14328 (devirtualization_time_bonus): New parameter known_aggs, pass it to
14329 ipa_get_indirect_edge_target. Update all callers.
14330 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
14331 ipa_get_indirect_edge_target_1 instead of calling
14332 ipa_get_indirect_edge_target.
14333 (create_specialized_node): Pass aggvlas to
14334 ipcp_discover_new_direct_edges.
14336 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14338 * config/arm/arm.md (f_sels, f_seld): New types.
14339 (*cmov<mode>): New pattern.
14340 * config/arm/predicates.md (arm_vsel_comparison_operator): New
14343 2013-03-25 Kai Tietz <ktietz@redhat.com>
14345 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
14346 POSIX-printf for mingw-hosted builds.
14348 2013-03-25 Richard Biener <rguenther@suse.de>
14350 PR middle-end/56694
14351 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
14352 must-not-throw stmt location.
14354 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14356 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
14357 Emit load-acquire versions when acq is true.
14358 (arm_emit_store_exclusive): Add rel parameter.
14359 Emit store-release versions when rel is true.
14360 (arm_split_compare_and_swap): Use acquire-release instructions
14362 of barriers when appropriate.
14363 (arm_split_atomic_op): Likewise.
14364 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
14365 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
14366 (VUNSPEC_SLX): Likewise.
14367 (VUNSPEC_LDA): Likewise.
14368 (VUNSPEC_STL): Likewise.
14369 * config/arm/sync.md (atomic_load<mode>): New pattern.
14370 (atomic_store<mode>): Likewise.
14371 (arm_load_acquire_exclusive<mode>): Likewise.
14372 (arm_load_acquire_exclusivesi): Likewise.
14373 (arm_load_acquire_exclusivedi): Likewise.
14374 (arm_store_release_exclusive<mode>): Likewise.
14376 2013-03-25 Catherine Moore <clm@codesourcery.com>
14378 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
14379 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
14380 * config/mip/predicates.md (lwsp_swsp_operand,
14381 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
14382 sb16_operand, db4_operand, db7_operand, ib3_operand,
14383 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
14384 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
14385 andi16_operand): New predicates.
14386 * config/mips/mips.md (compression): New attribute.
14387 (enabled): New attribute.
14388 (length): Consider compression in computing length.
14389 (shift_compression): New code attribute.
14390 (*add<mode>3): New operands. Record compression.
14391 (sub<mode>3): Likewise.
14392 (one_cmpl<mode>2): Likewise.
14393 (*and<mode>3): Likewise.
14394 (*ior<mode>3): Likewise.
14395 (unnamed pattern for xor): Likewise.
14396 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
14397 (*<optab><mode>3): Likewise.
14398 (*mov<mode>_internal: Likewise.
14399 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
14400 (mips_unsigned_immediate_p): New.
14401 (umips_lwsp_swsp_address_p): New.
14402 (m16_based_address_p): New.
14403 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
14404 (mips_unsigned_immediate_p): New prototype.
14405 (lwsp_swsp_address_p): New prototype.
14406 (m16_based_address_p): New prototype.
14407 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14408 (mips_signed_immediate_p): New function.
14409 (m16_based_address_p): New function.
14410 (lwsp_swsp_address_p): New function.
14411 (mips_print_operand_punctuation): Recognize short delay slot insns
14412 for microMIPS.add<mode>3"
14414 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14417 * config/arm/iterators.md (v_cmp_result): New mode attribute.
14418 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14420 2013-03-25 Richard Biener <rguenther@suse.de>
14422 PR tree-optimization/56689
14423 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14426 2013-03-25 Richard Biener <rguenther@suse.de>
14428 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14430 (memory_references): Likewise.
14431 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14432 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14433 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14434 (gather_mem_refs_in_loops): Fold into ...
14435 (analyze_memory_references): ... this. Move initialization
14436 to tree_ssa_lim_initialize.
14437 (fill_always_executed_in): Rename to ...
14438 (fill_always_executed_in_1): ... this.
14439 (fill_always_executed_in): Move contains_call computation to
14440 this new function from ...
14441 (tree_ssa_lim_initialize): ... here.
14442 (tree_ssa_lim): Call fill_always_executed_in.
14444 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
14446 * postreload.c (reload_combine): Fix code detecting returns.
14448 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
14450 * function.c (emit_use_return_register_into_block): On cc0 targets,
14451 do not emit the sequence between cc0 setter and user.
14453 2013-03-25 Kai Tietz <ktietz@redhat.com>
14455 * config/i386/predicates.md (local_symbolic_operand): Interpret
14456 dll-imported symbols as none-local.
14458 2013-03-25 Richard Biener <rguenther@suse.de>
14460 * tree-ssa-loop-im.c (struct depend): Remove.
14461 (struct lim_aux_data): Make depends a vec of gimples.
14462 (free_lim_aux_data): Adjust.
14463 (add_dependency): Likewise.
14464 (set_level): Likewise.
14466 2013-03-25 Richard Biener <rguenther@suse.de>
14468 PR middle-end/56434
14469 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14470 the pointer returned by calls with ECF_MALLOC set.
14472 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
14474 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14476 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
14478 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14479 using MMXMODE mode iterator.
14480 (*move<mode>_internal): Merge with *movv2sf_internal and
14481 *movv2sf_internal_rex64 using MMXMODE mode iterator.
14483 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
14485 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14486 (record_last_mem_set_info): Likewise.
14488 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14489 of XNEWVEC followed by memset.
14490 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14492 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
14494 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14495 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14496 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14497 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14498 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14499 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14500 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14501 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14502 BARRIER_P instead of GET_CODE.
14504 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14506 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14507 inaccuracy in the probing code.
14509 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14510 (ctrapdi4): Likewise.
14512 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14514 * calls.c (expand_call): Add missing guard to code handling return
14515 of non-BLKmode structures in MSB.
14516 * function.c (expand_function_end): Likewise.
14518 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14520 * combine.c (try_combine): Adjust comment. Do not add the set of
14521 insn #0 if the destination indirectly is set or dies in insn #2.
14522 Tidy up code to distribute a new note.
14524 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
14526 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14527 also for alternatives 16 and 17.
14529 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
14531 * config/i386/sse.md (*mov<mode>_internal): Merge with
14532 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
14533 Emit insn template depending on type attribute. Use
14534 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14535 movd instead of movq mnemonic for interunit moves. Rewrite mode
14536 attribute calculation. Remove unit attribute calculation.
14537 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14538 Set prefix_data16 attribute for DImode ssemov types.
14539 Use Ym instead of y for SSE-MMX conversion alternatives.
14540 Reorder operand constraints.
14542 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
14544 * df.h (df_insn_delete): Adjust prototype.
14545 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14546 and let it decide whether mark the basic block dirty.
14547 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14548 * df-scan.c (df_insn_info_delete): New helper function, split
14549 off from df_insn_delete.
14550 (df_scan_free_bb_info): Use it.
14551 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14553 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
14554 that the insn is actually an insn and it has a non-NULL basic block.
14555 Do not mark basic block dirty if only deleting a DEBUG_INSN.
14557 2013-03-22 Richard Biener <rguenther@suse.de>
14559 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14561 (mem_ref_alloc): Do not allocate them.
14562 (refs_independent_p): Do not query or maintain a cache.
14564 2013-03-22 Richard Biener <rguenther@suse.de>
14566 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14567 (gather_mem_refs_in_loops): Do not compute it.
14568 (analyze_memory_references): Do not allocate it.
14569 (tree_ssa_lim_finalize): Do not free it.
14570 (for_all_locs_in_loop): Do not query all_refs_in_loop.
14572 2013-03-22 Richard Biener <rguenther@suse.de>
14574 * is-a.h (as_a): Use gcc_checking_assert.
14576 2013-03-22 Ian Bolton <ian.bolton@arm.com>
14578 * config/aarch64/aarch64.c (aarch64_print_operand): New
14579 format specifier for printing a constant in hex.
14580 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14581 format specifier for printing second operand.
14583 2013-03-22 Richard Biener <rguenther@suse.de>
14585 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14587 (gather_mem_refs_in_loops): Perform store accumulation here.
14588 (create_vop_ref_mapping_loop): Remove.
14589 (create_vop_ref_mapping): Likewise.
14590 (analyze_memory_references): Initialize refs_stored_in_loop.
14591 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14592 (record_indep_loop): Remove.
14593 (record_dep_loop): New function.
14594 (ref_indep_loop_p_1): Adjust to only walk over references
14595 in the loop, not its subloops.
14596 (ref_indep_loop_p): Rename to ...
14597 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14598 maintaining a more fine-grained cache.
14599 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14600 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14602 2013-03-22 Richard Biener <rguenther@suse.de>
14604 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14605 (struct mem_ref): Make accesses_in_loop a vec of a vec of
14606 aggregate mem_ref_loc.
14607 (free_mem_ref_locs): Inline into ...
14608 (memref_free): ... this and adjust.
14609 (mem_ref_alloc): Adjust.
14610 (mem_ref_locs_alloc): Remove.
14611 (record_mem_ref_loc): Adjust.
14612 (get_all_locs_in_loop): Rewrite into ...
14613 (for_all_locs_in_loop): ... this iterator.
14614 (rewrite_mem_ref_loc): New functor.
14615 (rewrite_mem_refs): Use for_all_locs_in_loop.
14616 (sm_set_flag_if_changed): New functor.
14617 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14618 (ref_always_accessed): New functor.
14619 (ref_always_accessed_p): Use for_all_locs_in_loop.
14621 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14623 * tree-pass.h (PROP_gimple_lvec): New.
14624 * passes.c (dump_properties): Handle PROP_gimple_lvec.
14625 (init_optimization_passes): Move pass_lower_vector.
14626 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14628 (pass_lower_vector): Provide PROP_gimple_lvec.
14629 (pass_lower_vector_ssa): Likewise.
14630 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14632 2013-03-21 Mark Wielaard <mjw@redhat.com>
14634 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14636 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
14638 * config/i386/i386.md (*movdi_internal): Disparage slightly
14639 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
14640 conversion alternatives.
14642 2013-03-21 Jakub Jelinek <jakub@redhat.com>
14644 PR middle-end/48087
14645 * diagnostic.def (DK_WERROR): New kind.
14646 * diagnostic.h (werrorcount): Define.
14647 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14648 promoted to DK_ERROR, increment DK_WERROR counter instead of
14650 * toplev.c (toplev_main): Call print_ignored_options even if
14651 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
14652 even if just werrorcount is non-zero.
14655 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14657 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14658 (string_cst_pool_decl): New function.
14659 (optimize_one_addr_into_implicit_ptr): New function.
14660 (resolve_addr_in_expr): Optimize DWARF location expression
14661 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14662 which doesn't live in memory, but has DW_AT_location or
14663 DW_AT_const_value, or refers to a string literal, into
14664 DW_OP_GNU_implicit_pointer.
14665 (optimize_location_into_implicit_ptr): New function.
14666 (resolve_addr): If removing DW_AT_location of a variable because
14667 it was DW_OP_addr of address of the variable, but the variable doesn't
14668 live in memory, try to emit const value attribute for the initializer.
14670 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14672 * tree.h (VECTOR_TYPE_P): New macro.
14673 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14674 TYPE_MODE): Use it.
14675 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14676 VEC_COND_EXPR cannot be lvalues.
14677 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14679 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14681 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14682 Restrict the transformation to equal modes.
14684 2013-03-21 Richard Biener <rguenther@suse.de>
14686 PR tree-optimization/39326
14687 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14688 (MEM_ANALYZABLE): Adjust.
14689 (record_mem_ref_loc): Move bitmap ops ...
14690 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
14691 unanalyzable refs, do not record locations for it.
14692 (analyze_memory_references): Allocate ref zero as shared
14694 (refs_independent_p): Do not test for unanalyzed mems here.
14695 (ref_indep_loop_p_1): Special-case disambiguation against
14696 the unanalyzed ref.
14697 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14699 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
14701 * config/arm/arm-protos.h (tune_params): Add
14702 prefer_neon_for_64bits field.
14703 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14704 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14705 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14706 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14707 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14708 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14709 (arm_option_override): Handle -mneon-for-64bits new option.
14710 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14711 (prefer_neon_for_64bits): Declare new variable.
14712 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14713 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14714 (arch_enabled): Handle new arch types. Remove support for onlya8
14716 (one_cmpldi2): Use new arch names.
14717 (zero_extend<mode>di2, extend<mode>di2): Ditto.
14718 * config/arm/arm.opt (mneon-for-64bits): Add option.
14719 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14720 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14721 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14723 * doc/invoke.texi (-mneon-for-64bits): Document.
14725 2013-03-21 Richard Biener <rguenther@suse.de>
14727 PR tree-optimization/39326
14728 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14729 (sort_bbs_in_loop_postorder_cmp): New function.
14730 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14732 2013-03-21 Richard Biener <rguenther@suse.de>
14734 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14735 (vect_insert_into_interleaving_chain): Likewise.
14736 (vect_drs_dependent_in_basic_block): Inline ...
14737 (vect_slp_analyze_data_ref_dependence): ... here. New function,
14739 (vect_analyze_data_ref_dependence): ... here. Simplify.
14740 (vect_check_interleaving): Simplify.
14741 (vect_analyze_data_ref_dependences): Likewise. Split out ...
14742 (vect_slp_analyze_data_ref_dependences): ... this new function.
14743 (dr_group_sort_cmp): New function.
14744 (vect_analyze_data_ref_accesses): Compute data-reference groups
14745 here instead of in vect_analyze_data_ref_dependence. Use
14746 a more efficient algorithm.
14747 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14748 vect_slp_analyze_data_ref_dependences. Call
14749 vect_analyze_data_ref_accesses earlier.
14750 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14751 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14752 (vect_slp_analyze_data_ref_dependences): New prototype.
14754 2013-03-21 Richard Biener <rguenther@suse.de>
14756 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14757 ref is stored in the loop.
14758 (find_refs_for_sm): Walk only over all stores.
14759 (store_motion_loop): Allocate from lim_bitmap_obstack.
14760 (store_motion): Likewise.
14762 2013-03-21 Richard Biener <rguenther@suse.de>
14764 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14765 Update virtual SSA form.
14767 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14769 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14770 * configure: Regenerate.
14771 * config.in: Regenerate.
14772 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14773 if !HAVE_LD_EH_FRAME_CIEV3.
14775 2013-03-21 Richard Biener <rguenther@suse.de>
14777 * tree-cfg.c (verify_expr_no_block): New function.
14778 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14779 nor DECL_VALUE_EXPR have locations with associated blocks.
14780 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14781 (clear_unused_block_pointer): Remove code dealing with
14782 blocks in DECL_DEBUG_EXPR locations.
14784 2013-03-21 Richard Biener <rguenther@suse.de>
14786 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14787 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
14788 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14789 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14790 instead of DECL_DEBUG_EXPR_IS_FROM.
14791 * gimplify.c (gimplify_modify_expr): Likewise.
14792 * tree-cfg.c (verify_expr_location_1): Likewise.
14793 * tree-complex.c (create_one_component_var): Likewise.
14794 * tree-sra.c (create_access_replacement): Likewise.
14795 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14796 (clear_unused_block_pointer): Likewise.
14797 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14798 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14799 * var-tracking.c (var_debug_decl): Likewise.
14800 (track_expr_p): Likewise.
14801 * tree-inline.c (add_local_variables): Likewise. Set
14802 DECL_HAS_DEBUG_EXPR_P after copying it.
14803 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14804 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
14806 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
14809 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14810 * configure: Regenerate.
14811 * config.in: Regenerate.
14812 * config/i386/i386.md (*movdf_internal): Use
14813 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14814 movd instead of movq mnemonic for interunit moves.
14815 (*movdi_internal): Ditto.
14817 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14819 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14820 (abd<mode>_3): New pattern.
14821 (aba<mode>_3): New pattern.
14822 (fabd<mode>_3): New pattern.
14824 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14826 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14827 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14828 occurrence of REGISTER_PREFIX as its empty string.
14830 2013-03-20 Jeff Law <law@redhat.com>
14832 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14833 addititional equivalences for equality comparisons between an SSA_NAME
14834 and a constant where the SSA_NAME was set from a widening conversion.
14836 2013-03-20 Walter Lee <walt@tilera.com>
14838 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14840 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14842 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14843 depending on type attribute.
14844 (*movti_internal): Ditto.
14845 (*movtf_internal): Ditto.
14846 (*movxf_internal): Ditto.
14847 (*movdf_internal): Ditto.
14848 (*movsf_internal): Ditto.
14850 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14852 * config/i386/i386.md (*movti_internal): Set prefix attribute to
14853 maybe_vex for sselog1 and ssemov types.
14854 (*movdi_internal): Reorder operand constraints.
14855 (*movsi_internal): Ditto. Set prefix attribute to
14856 maybe_vex for sselog1 and ssemov types.
14857 (*movtf_internal): Set prefix attribute to maybe_vex
14858 for sselog1 and ssemov types.
14859 (*movdf_internal): Ditto. Set prefix_data16 attribute for
14860 DImode ssemov types. Reorder operand constraints.
14861 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
14862 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
14863 attribute for SImode ssemov types. Reorder operand constraints.
14865 2013-03-20 Martin Jambor <mjambor@suse.cz>
14867 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14868 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14870 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
14872 * config/rs6000/predicates.md (indexed_address, update_address_mem
14873 update_indexed_address_mem): New predicates.
14874 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14875 attribute for load/store instructions.
14876 * config/rs6000/dfp.md (movsd_store): Likewise.
14877 (movsd_load): Likewise.
14878 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14879 (unnamed HI->DI extend define_insn): Likewise.
14880 (unnamed SI->DI extend define_insn): Likewise.
14881 (unnamed QI->SI extend define_insn): Likewise.
14882 (unnamed QI->HI extend define_insn): Likewise.
14883 (unnamed HI->SI extend define_insn): Likewise.
14884 (unnamed HI->SI extend define_insn): Likewise.
14885 (extendsfdf2_fpr): Likewise.
14886 (movsi_internal1): Likewise.
14887 (movsi_internal1_single): Likewise.
14888 (movhi_internal): Likewise.
14889 (movqi_internal): Likewise.
14890 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14891 attribute for load/store instructions.
14892 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14894 (mov<mode>_softfloat): Likewise.
14895 (mov<mode>_hardfloat32): Likewise.
14896 (mov<mode>_hardfloat64): Likewise.
14897 (mov<mode>_softfloat64): Likewise.
14898 (movdi_internal32): Likewise.
14899 (movdi_internal64): Likewise.
14900 (probe_stack_<mode>): Likewise.
14902 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14904 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14905 floating point, and decimal floating point to reload iterator.
14907 * config/rs6000/constraints.md (wl constraint): New constraints to
14908 return FLOAT_REGS if certain options are used to reduce the number
14909 of separate patterns that exist in the file.
14910 (wx constraint): Likewise.
14911 (wz constraint): Likewise.
14913 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14914 -mdebug=reg, print wg, wl, wx, and wz constraints.
14915 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14916 Initialize the reload functions for 64-bit binary/decimal floating
14918 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14919 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14920 create the buffer on the stack to overcome not having a 32-bit
14922 (rs6000_emit_move): Likewise.
14923 (rs6000_secondary_memory_needed_rtx): Likewise.
14924 (rs6000_alloc_sdmode_stack_slot): Likewise.
14925 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14926 via xxlxor, just like DFmode 0.0.
14928 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14929 define as 1 if we are running on a power7 or newer.
14930 (enum r6000_reg_class_enum): Add new constraints.
14932 * config/rs6000/dfp.md (movsd): Delete, combine with binary
14933 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
14934 with other moves by using conditional constraits (wg). Use LFIWZX
14935 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
14936 (movsd splitter): Likewise.
14937 (movsd_hardfloat): Likewise.
14938 (movsd_softfloat): Likewise.
14940 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14941 binary and decimal floating point moves.
14942 (fmove_ok): New attributes to combine binary and decimal floating
14943 point moves, and to combine power6x (mfpgpr) moves along normal
14945 (real_value_to_target): Likewise.
14946 (f32_lr): Likewise.
14947 (f32_lm): Likewise.
14948 (f32_li): Likewise.
14949 (f32_sr): Likewise.
14950 (f32_sm): Likewise.
14951 (f32_si): Likewise.
14952 (movsf): Combine binary and decimal floating point moves. Combine
14953 power6x (mfpgpr) moves with other moves by using conditional
14954 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
14955 (mov<mode> for SFmode/SDmode); Likewise.
14956 (SFmode/SDmode splitters): Likewise.
14957 (movsf_hardfloat): Likewise.
14958 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14959 (movsf_softfloat): Likewise.
14960 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14962 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14963 wx and wz constraints.
14965 * config/rs6000/constraints.md (wg constraint): New constraint to
14966 return FLOAT_REGS if -mmfpgpr (power6x) was used.
14968 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14971 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14972 -mdebug=reg, print wg, wl, wx, and wz constraints.
14973 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14974 Initialize the reload functions for 64-bit binary/decimal floating
14976 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14977 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14978 create the buffer on the stack to overcome not having a 32-bit
14980 (rs6000_emit_move): Likewise.
14981 (rs6000_secondary_memory_needed_rtx): Likewise.
14982 (rs6000_alloc_sdmode_stack_slot): Likewise.
14983 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14984 via xxlxor, just like DFmode 0.0.
14986 * config/rs6000/dfp.md (movdd): Delete, combine with binary
14987 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
14988 with other moves by using conditional constraits (wg). Use LFIWZX
14989 and STFIWX for loading SDmode on power7.
14990 (movdd splitters): Likewise.
14991 (movdd_hardfloat32): Likewise.
14992 (movdd_softfloat32): Likewise.
14993 (movdd_hardfloat64_mfpgpr): Likewise.
14994 (movdd_hardfloat64): Likewise.
14995 (movdd_softfloat64): Likewise.
14997 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14998 64-bit binary and decimal floating point moves.
14999 (FMOVE64X): Likewise.
15000 (movdf): Combine 64-bit binary and decimal floating point moves.
15001 Combine power6x (mfpgpr) moves with other moves by using
15002 conditional constraits (wg).
15003 (mov<mode> for DFmode/DDmode): Likewise.
15004 (DFmode/DDmode splitters): Likewise.
15005 (movdf_hardfloat32): Likewise.
15006 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
15007 (movdf_softfloat32): Likewise.
15008 (movdf_hardfloat64_mfpgpr): Likewise.
15009 (movdf_hardfloat64): Likewise.
15010 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
15011 (movdf_softfloat64): Likewise.
15012 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
15013 (reload_<mode>_load): Move to later in the file so they aren't in
15014 the middle of the floating point move insns.
15015 (reload_<mode>_store): Likewise.
15017 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
15020 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
15021 constraint if -mdebug=reg.
15022 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
15023 Enable using dd reload support if needed.
15025 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
15026 binary and decimal floating point moves in rs6000.md.
15027 (movtd_internal): Likewise.
15029 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
15030 decimal floating point moves.
15032 (movtf_internal): Likewise.
15033 (mov<mode>_internal, TDmode/TFmode): Likewise.
15034 (movtf_softfloat): Likewise.
15035 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
15037 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
15038 movdi_internal64, using wg constraint for move direct operations.
15039 (movdi_internal64): Likewise.
15041 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
15042 MODES_TIEABLE_P for selected modes. Print the numerical value of
15043 the various virtual registers. Use GPR/FPR first/last values,
15044 instead of hard coding the register numbers. Print which modes
15045 have reload functions registered.
15046 (rs6000_option_override_internal): If -mdebug=reg, trace the options
15047 settings before/after setting cpu, target and subtarget settings.
15048 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
15049 and for secondary reload failures in rs6000_secondary_reload_inner.
15050 (rs6000_secondary_reload_fail): Likewise.
15051 (rs6000_secondary_reload_inner): Likewise.
15053 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
15054 macros for first/last GPR and FPR registers.
15055 (LAST_GPR_REGNO): Likewise.
15056 (FIRST_FPR_REGNO): Likewise.
15057 (LAST_FPR_REGNO): Likewise.
15059 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
15060 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
15061 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
15062 (vcond<mode><mode>): Likewise.
15063 (vcondu<mode><mode>): Likewise.
15064 (vector_gtu<mode>): Likewise.
15065 (vector_gte<mode>): Likewise.
15066 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
15067 to prevent the compiler from converting DImode operations to TImode.
15068 (ior<mode>3): Likewise.
15069 (and<mode>3): Likewise.
15070 (one_cmpl<mode>2): Likewise.
15071 (nor<mode>3): Likewise.
15072 (andc<mode>3): Likewise.
15074 * config/rs6000/constraints.md (wt constraint): New constraint
15075 that returns VSX_REGS if TImode is allowed in VSX registers.
15077 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
15078 constant under VSX.
15080 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
15081 similar to TImode, but it is restricted to being in the GPRs.
15083 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
15084 TImode to occupy a single VSX register.
15086 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
15087 -mvsx-timode for power7/power8.
15088 (power7 cpu): Likewise.
15089 (power8 cpu): Likewise.
15091 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
15092 sure that TFmode/TDmode take up two registers if they are ever
15093 allowed in the upper VSX registers.
15094 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
15096 (rs6000_init_hard_regno_mode_ok): Likewise.
15097 (rs6000_debug_reg_global): Add debugging for PTImode and wt
15098 constraint. Print if LRA is turned on.
15099 (rs6000_option_override_internal): Give an error if -mvsx-timode
15100 and VSX is not enabled.
15101 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
15102 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
15103 to reg+offset addressing. Use PTImode when checking offset
15104 addresses for validity.
15105 (reg_offset_addressing_ok_p): Likewise.
15106 (rs6000_legitimate_offset_address_p): Likewise.
15107 (rs6000_legitimize_address): Likewise.
15108 (rs6000_legitimize_reload_address): Likewise.
15109 (rs6000_legitimate_address_p): Likewise.
15110 (rs6000_eliminate_indexed_memrefs): Likewise.
15111 (rs6000_emit_move): Likewise.
15112 (rs6000_secondary_reload): Likewise.
15113 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
15114 reloads to fpr registers to continue to use reg+offset addressing,
15115 but 64-bit reloads to altivec registers need reg+reg addressing.
15116 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
15117 it. Treat LO_SUM like a PLUS operation.
15118 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
15119 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
15120 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
15121 registers to share a register with a smaller sized type, since VSX
15122 puts scalars in the upper 64-bits.
15123 (print_operand): Add support for PTImode.
15124 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
15125 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
15126 registers, but don't have arithmetic support.
15127 (rs6000_memory_move_cost): Add test for VSX.
15128 (rs6000_opt_masks): Add -mvsx-timode.
15130 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
15133 (VSr): Use wt constraint for TImode.
15134 (VSv): Drop TImode support.
15135 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
15136 (vsx_movti_64bit): Likewise.
15137 (vsx_movti_32bit): Likewise.
15138 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
15139 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
15140 one '?' on the appropriate output constraint. Do not allow TImode
15141 logical operations on 32-bit systems.
15142 (vsx_ior<mode>3): Likewise.
15143 (vsx_xor<mode>3): Likewise.
15144 (vsx_one_cmpl<mode>2): Likewise.
15145 (vsx_nor<mode>3): Likewise.
15146 (vsx_andc<mode>3): Likewise.
15147 (vsx_concat_<mode>): Likewise.
15148 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
15150 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
15151 OPTION_MASK_VSX_TIMODE.
15152 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
15153 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
15155 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
15156 (TI2 iterator): New iterator for TImode, PTImode.
15157 (wd mode attribute): Add values for vector types.
15158 (movti_string): Replace TI move operations with operations for TImode
15159 and PTImode. Add support for TImode being allowed in VSX registers.
15160 (mov<mode>_string, TImode/PTImode): Likewise.
15161 (movti_ppc64): Likewise.
15162 (mov<mode>_ppc64, TImode/PTImode): Likewise.
15163 (TI mode splitters): Likewise.
15165 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
15168 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
15170 PR tree-optimization/56355
15171 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
15172 Also handle integers with undefined overflow.
15174 2013-03-20 Catherine Moore <clm@codesourcery.com>
15175 Maciej W. Rozycki <macro@codesourcery.com>
15176 Tom de Vries <tom@codesourcery.com>
15177 Nathan Sidwell <nathan@codesourcery.com>
15178 Iain Sandoe <iain@codesourcery.com>
15179 Nathan Froyd <froydnj@codesourcery.com>
15180 Chao-ying Fu <fu@mips.com>
15182 * doc/extend.texi: (micromips, nomicromips, nocompression):
15183 Document new function attributes.
15184 * doc/invoke.texi (minterlink-compressed, mmicromips,
15185 m14k, m14ke, m14kec): Document new options.
15186 (minterlink-mips16): Update documentation.
15187 * doc/md.texi (ZC, ZD): Document new constraints.
15188 * configure.ac (gcc_cv_as_micromips): Check if linker
15189 supports the .set micromips directive.
15190 * configure: Regenerate.
15191 * config.in: Regenerate.
15192 * config/mips/mips-tables.opt: Regenerate.
15193 * config/mips/micromips.md: New file.
15194 * constraints.md (ZC, ZD): New constraints.
15195 * config/mips/predicates.md (movep_src_register): New predicate.
15196 (movep_src_operand): New predicate.
15197 (non_volatile_mem_operand): New predicate.
15198 * config/mips/mips.md (multimem): New type.
15199 (length): Differentiate between 17-bit and 18-bit branch offsets.
15200 (MOVEP1, MOVEP2): New mode iterator.
15201 (mov_<load>l): Use ZC constraint.
15202 (mov_<load>r): Likewise.
15203 (mov_<store>l): Likewise.
15204 (mov_<store>r): Likewise.
15205 (*branch_equality<mode>_inverted): Add microMIPS support.
15206 (*branch_equality<mode>): Likewise.
15207 (*jump_absolute): Likewise.
15208 (indirect_jump_<mode>): Likewise.
15209 (tablejump_<mode>): Likewise.
15210 (<optab>_internal): Likewise.
15211 (sibcall_internal): Likewise.
15212 (sibcall_value_internal): Likewise.
15213 (prefetch): Use constraint ZD.
15214 * config/mips/mips.opt (minterlink-compressed): New option.
15215 (minterlink-mips16): Now an alias for minterlink-compressed.
15216 (mmicromips): New option.
15217 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15218 (compare_and_swap_12): Likewise.
15219 (sync_add<mode>): Likewise.
15220 (sync_<optab>_12): Likewise.
15221 (sync_old_<optab>_12): Likewise.
15222 (sync_new_<optab>_12): Likewise.
15223 (sync_nand_12): Likewise.
15224 (sync_old_nand_12): Likewise.
15225 (sync_new_nand_12): Likewise.
15226 (sync_sub<mode>): Likewise.
15227 (sync_old_add<mode>): Likewise.
15228 (sync_old_sub<mode>): Likewise.
15229 (sync_new_add<mode>): Likewise.
15230 (sync_new_sub<mode>): Likewise.
15231 (sync_<optab><mode>): Likewise.
15232 (sync_old_<optab><mode>): Likewise.
15233 (sync_new_<optab><mode>): Likewise.
15234 (sync_nand<mode>): Likewise.
15235 (sync_old_nand<mode>): Likewise.
15236 (sync_new_nand<mode>): Likewise.
15237 (sync_lock_test_and_set<mode>): Likewise.
15238 (test_and_set_12): Likewise.
15239 (atomic_compare_and_swap<mode>): Likewise.
15240 (atomic_exchange<mode>_llsc): Likewise.
15241 (atomic_fetch_add<mode>_llsc): Likewise.
15242 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15243 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15244 (umips_save_restore_pattern_p): Likewise.
15245 (umips_load_store_pair_p): Likewise.
15246 (umips_output_load_store_pair): Likewise.
15247 (umips_movep_target_p): Likewise.
15248 (umips_12bit_offset_address_p): Likewise.
15249 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15250 (mips_base_mips16): Rename this...
15251 (mips_base_compression_flags): ...to this. Update all uses.
15252 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15253 (mips_mips16_decl_p): Delete.
15254 (mips_nomips16_decl_p): Delete.
15255 (mips_get_compress_on_flags): New function.
15256 (mips_get_compress_off_flags): New function.
15257 (mips_get_compress_mode): New function.
15258 (mips_get_compress_on_name): New function.
15259 (mips_get_compress_off_name): New function.
15260 (mips_insert_attributes): Support multiple compression types.
15261 (mips_merge_decl_attributes): Likewise.
15262 (umips_12bit_offset_address_p): New function.
15263 (mips_start_function_definition): Emit .set micromips directive.
15264 (mips_call_may_need_jalx_p): New function.
15265 (mips_function_ok_for_sibcall): Add microMIPS support.
15266 (mips_print_operand_punctuation): Support short delay slots and
15268 (umips_swm_mask, umips_swm_encoding): New.
15269 (umips_build_save_restore): New function.
15270 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15271 (was_mips16_p): Remove.
15272 (old_compression_mode): New.
15273 (mips_set_compression_mode): New function.
15274 (mips_set_current_function): Add microMIPS support.
15275 (mips_option_override): Likewise.
15276 (umips_save_restore_pattern_p): New function.
15277 (umips_output_save_restore): New function.
15278 (umips_load_store_pair_p_1): New function.
15279 (umips_load_store_pair_p): New function.
15280 (umips_output_load_store_pair_1): New function.
15281 (umips_output_load_store_pair): New function.
15282 (umips_movep_target_p) New function.
15283 (mips_prepare_pch_save): Add microMIPS support.
15284 * config/mips/mips.h (TARGET_COMPRESSION): New.
15285 (TARGET_CPU_CPP_BUILTINS): Update macro
15286 to use new compression flags and to support microMIPS.
15287 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15288 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15289 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15290 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15291 (ASM_SPEC): Support mmicromips and mno-micromips.
15292 (M16STORE_REG_P): New macro.
15293 (MIPS_CALL): Support TARGET_MICROMIPS.
15294 (MICROMIPS_J): New macro.
15295 (mips_base_mips16): Rename this...
15296 (mips_base_compression_flags): ...to this.
15297 (UMIPS_12BIT_OFFSET_P): New macro.
15298 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15299 (MULTILIB_DIRNAMES): Likewise.
15300 2013-03-20 Richard Biener <rguenther@suse.de>
15302 PR tree-optimization/56661
15303 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
15304 the result does not have to be distinct.
15306 2013-03-20 Richard Biener <rguenther@suse.de>
15308 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
15311 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15312 Steven Bosscher <steven@gcc.gnu.org>
15314 PR rtl-optimization/56605
15315 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
15317 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
15320 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
15321 that require movd instead of movq.
15323 2013-03-20 Richard Biener <rguenther@suse.de>
15325 * tree-ssa-structalias.c (struct variable_info): Add pointer
15326 to the first field of an aggregate with sub-vars. Make
15327 this and the pointer to the next subfield its ID.
15328 (vi_next): New function.
15329 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
15330 storedanything_id, integer_id): Increment by one.
15331 (new_var_info, get_call_vi, lookup_call_clobber_vi,
15332 get_call_clobber_vi): Adjust.
15333 (solution_set_expand): Simplify and speedup.
15334 (solution_set_add): Inline into ...
15335 (set_union_with_increment): ... this. Adjust accordingly.
15336 (do_sd_constraint): Likewise.
15337 (do_ds_constraint): Likewise.
15338 (do_complex_constraint): Simplify.
15339 (build_pred_graph): Adjust.
15340 (solve_graph): Likewise. Simplify and speedup.
15341 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
15342 get_constraint_for_component_ref, get_constraint_for_1,
15343 first_vi_for_offset, first_or_preceding_vi_for_offset,
15344 create_function_info_for, create_variable_info_for_1,
15345 create_variable_info_for, intra_create_variable_infos): Adjust.
15346 (init_base_vars): Push NULL for ID zero.
15347 (compute_points_to_sets): Adjust.
15349 2013-03-20 Richard Biener <rguenther@suse.de>
15351 * cfgloop.c (verify_loop_structure): Streamline and avoid
15352 ICEing on corrupt loop tree.
15353 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
15356 2013-03-20 Richard Biener <rguenther@suse.de>
15358 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
15359 check whether an SSA update is needed.
15361 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
15363 * config/mips/constraints.md (T): Rename to...
15367 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
15368 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
15370 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15372 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
15373 (*subsi3_carryin_uxtw): Likewise.
15375 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15377 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
15378 (*rorsi3_insn_uxtw): Likewise.
15380 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15382 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
15383 (*extrsi5_insn_uxtw): Likewise.
15385 2013-03-19 Richard Biener <rguenther@suse.de>
15387 PR tree-optimization/56273
15388 * passes.c (init_optimization_passes): Move second VRP after DOM.
15390 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
15392 * config/i386/i386.md (*movti_internal): Merge from
15393 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
15394 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
15395 nox64 isa attributes.
15397 2013-03-18 Richard Biener <rguenther@suse.de>
15399 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
15401 (merge_node_constraints): Likewise.
15402 (build_succ_graph): Likewise.
15403 (valid_graph_edge): Inline into single caller.
15404 (unify_nodes): Likewise. Use bitmap_set_bit return value
15406 (scc_visit): Fix formatting and variable use.
15407 (do_sd_constraint): Use gcc_checking_assert.
15408 (do_ds_constraint): Likewise.
15409 (do_complex_constraint): Likewise.
15410 (condense_visit): Likewise. Cleanup.
15411 (dump_pred_graph): New function.
15412 (perform_var_substitution): Dump the pred-graph before
15413 variable substitution.
15414 (find_equivalent_node): Use gcc_checking_assert.
15415 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15417 2013-03-18 Richard Biener <rguenther@suse.de>
15419 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15420 Remove cond_expr_stmt_list argument and do not gimplify the
15422 (vect_loop_versioning): Adjust.
15423 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15424 Cleanup to use less temporaries.
15425 (vect_create_data_ref_ptr): Cleanup.
15427 2013-03-18 Jakub Jelinek <jakub@redhat.com>
15429 PR tree-optimization/56635
15430 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15431 require types_compatible_p types.
15433 2013-03-18 Nick Clifton <nickc@redhat.com>
15435 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15436 spurious backslash.
15438 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15439 Add missing line to comment describing function.
15441 2013-03-18 Richard Biener <rguenther@suse.de>
15443 PR tree-optimization/56210
15444 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15445 Handle string / character search functions.
15446 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15448 2013-03-18 Richard Biener <rguenther@suse.de>
15450 PR middle-end/56483
15451 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15452 and implement properly.
15453 * gimple.h (gimple_cond_single_var_p): Remove.
15455 2013-03-18 Richard Biener <rguenther@suse.de>
15457 * tree-data-ref.h (find_data_references_in_loop): Declare.
15458 * tree-data-ref.c (get_references_in_stmt): Use a stack
15459 vector pre-allocated in the callers.
15460 (find_data_references_in_stmt): Adjust.
15461 (graphite_find_data_references_in_stmt): Likewise.
15462 (create_rdg_vertices): Likewise.
15463 (find_data_references_in_loop): Export.
15464 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15465 Compute dependences here...
15466 (vect_analyze_data_refs): ...not here. When we encounter
15467 a non-vectorizable data reference in basic-block vectorization
15468 truncate the data reference vector. Do not bother to
15469 fixup data-dependence information for gather loads.
15470 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15471 of data references, as reported.
15473 2013-03-18 Richard Biener <rguenther@suse.de>
15475 PR tree-optimization/3713
15476 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
15477 has_constants and expr.
15478 (stmt_has_constants): Properly valueize SSA names when deciding
15479 whether the stmt has constants.
15481 2013-03-18 Richard Biener <rguenther@suse.de>
15483 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15484 whole function when there is nothing to do.
15485 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15486 * tree-vectorizer.c (vectorize_loops): Update virtual and
15487 loop-closed SSA once.
15488 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15490 2013-03-18 Richard Biener <rguenther@suse.de>
15492 PR middle-end/56113
15493 * domwalk.c (bb_postorder): New global static.
15494 (cmp_bb_postorder): New function.
15495 (walk_dominator_tree): Replace scheme imposing an order for
15496 visiting dominator sons by one sorting them at the time they
15497 are pushed on the stack.
15499 2013-03-18 Richard Biener <rguenther@suse.de>
15501 PR tree-optimization/39326
15502 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15503 (struct mem_ref): Replace mem member with ao_ref typed member.
15504 (MEM_ANALYZABLE): Adjust.
15505 (memref_eq): Likewise.
15506 (mem_ref_alloc): Likewise.
15507 (gather_mem_refs_stmt): Likewise.
15508 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15509 (execute_sm_if_changed_flag_set): Adjust.
15510 (execute_sm): Likewise.
15511 (ref_always_accessed_p): Likewise.
15512 (refs_independent_p): Likewise.
15513 (can_sm_ref_p): Likewise.
15515 2013-03-18 Jakub Jelinek <jakub@redhat.com>
15518 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15519 return 1 even for !unsignedp.
15521 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15523 * config/i386/i386.md (isa): Add x64 and nox64.
15524 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15525 (*pushtf): Enable *roF alternative for x64 isa only.
15526 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
15527 mode attribute of integer alternatives to DImode for TARGET_64BIT.
15528 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
15529 (*movtf_internal): Merge from *movtf_internal_rex64 and
15530 *movtf_internal_sse. Use x64 and nox64 isa attributes.
15531 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
15532 nox64 isa attributes.
15533 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
15534 nox64 isa attributes.
15535 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15537 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15539 * config/alpha/alpha.c (TARGET_LRA_P): New define.
15541 2013-03-17 Jakub Jelinek <jakub@redhat.com>
15544 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15545 class names. Remove trailing comma after "ALL_REGS".
15547 2013-03-16 Jan Hubicka <jh@suse.cz>
15549 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15550 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15551 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15552 of cgraph_get_create_node.
15553 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15555 2013-03-16 Jason Merrill <jason@redhat.com>
15558 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15559 with DW_AT_default_value.
15561 2013-03-16 Jakub Jelinek <jakub@redhat.com>
15563 * BASE-VER: Set to 4.9.0.
15565 2013-03-14 Andi Kleen <ak@linux.intel.com>
15568 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15569 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15570 Document _x* TSX intrinsics.
15572 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
15573 David Holsgrove <david.holsgrove@xilinx.com>
15575 * configure.ac: Add MicroBlaze TLS support detection.
15576 * configure: Regenerate.
15577 * config/microblaze/microblaze-protos.h
15578 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15579 symbol_mentioned_p, label_mentioned_p): Add prototypes.
15580 * config/microblaze/microblaze.c (microblaze_address_type): Add
15581 ADDRESS_TLS and tls_reloc address types.
15582 (microblaze_address_info): Add tls_reloc.
15583 (TARGET_HAVE_TLS): Define.
15584 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15585 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15586 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15587 load_tls_operand, microblaze_call_tls_get_addr,
15588 microblaze_legitimize_tls_address): New functions.
15589 (microblaze_classify_unspec): Handle UNSPEC_TLS.
15590 (get_base_reg): Use microblaze_tls_symbol_p.
15591 (microblaze_classify_address): Handle TLS.
15592 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15593 label_mentioned_p and microblaze_tls_referenced_p.
15594 (microblaze_legitimize_address): Handle TLS.
15595 (microblaze_address_insns): Handle ADDRESS_TLS.
15596 (pic_address_needs_scratch): Handle TLS.
15597 (print_operand_address): Handle TLS.
15598 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15599 (microblaze_expand_move): Handle TLS.
15600 (microblaze_legitimate_constant_p): Check
15601 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15602 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15603 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15604 (PIC_OFFSET_TABLE_REGNUM): Set.
15605 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15606 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15607 (addsi3, movsi_internal2, movdf_internal): Update constraints
15608 * config/microblaze/predicates.md (arith_plus_operand): Define
15609 (move_operand): Redefine as move_src_operand,
15610 check microblaze_tls_referenced_p.
15612 2013-03-14 Ian Bolton <ian.bolton@arm.com>
15614 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15615 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15617 2013-03-14 Ian Bolton <ian.bolton@arm.com>
15619 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15622 2013-03-14 Jakub Jelinek <jakub@redhat.com>
15624 PR tree-optimization/53265
15625 * common.opt (Waggressive-loop-optimizations): New option.
15626 * tree-ssa-loop-niter.c: Include tree-pass.h.
15627 (do_warn_aggressive_loop_optimizations): New function.
15628 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
15629 if number_of_latch_executions returned constant.
15630 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15631 early. If number_of_latch_executions returned constant, set
15632 nb_iterations_upper_bound back to it.
15633 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15635 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15636 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15638 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15639 (MULTILIB_OSDIRNAMES): Set.
15640 * genmultilib: If defaultosdirname doesn't start with :: , set
15641 defaultosdirname2 instead, clear it and emit two . multilib_raw
15642 entries instead of just one.
15644 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
15646 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15647 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15648 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15649 (SUBTARGET_OVERRIDE_OPTIONS): New.
15651 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
15654 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15655 (musermode): Convert to Var(TARGET_USERMODE).
15656 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15657 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15658 * config/sh/sh.c (sh_option_override): Use
15659 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15660 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15662 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15664 (udivsi3_i4_single, divsi3_i4_single): Use
15665 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15667 2013-03-13 Dave Korn <dave.korn.cygwin@....>
15669 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15672 2013-03-13 Richard Biener <rguenther@suse.de>
15674 PR tree-optimization/56608
15675 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15676 calls when vectorizing basic-blocks.
15678 2013-03-13 Jakub Jelinek <jakub@redhat.com>
15681 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15684 2013-03-12 Jakub Jelinek <jakub@redhat.com>
15686 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15688 2013-03-11 Jan Hubicka <jh@suse.cz>
15691 * lto-streamer-out.c (output_symbol_p): Skip references from
15692 constructors of external variables.
15694 2013-03-11 Jan Hubicka <jh@suse.cz>
15696 PR middle-end/56571
15697 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15699 * emit-rtl.c (verify_rtx_sharing): Likewise.
15700 (copy_insn_1): Likewise.
15701 * rtl.c (copy_rtx): Likewise.
15703 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
15706 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15707 output_operand_lossage message.
15709 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
15712 * arm.c (shift_op): Validate RTL pattern on the fly.
15713 (arm_print_operand, case 'S'): Don't use shift_operator to validate
15716 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
15719 * config/pa/pa.md (call_value): Check for calls to powf and direct to
15720 new call patterns that clobber %fr12.
15721 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15722 split and postreload patterns.
15723 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15724 registers %fr12 and %fr12R as call used.
15726 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
15728 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15729 (canon_address, record_store, replace_read, check_mem_read_rtx,
15730 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15731 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15732 rest_of_handle_dse): Likewise.
15734 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
15736 PR middle-end/56524
15737 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15739 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15740 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15741 (save_optabs_if_changed): Replace with...
15742 (init_tree_optimization_optabs): ...this.
15743 * optabs.c (save_optabs_if_changed): Rename to...
15744 (init_tree_optimization_optabs): ...this. Take the optimization node
15745 as argument. Do nothing if the base optabs are already correct.
15746 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15747 to recompute optabs.
15748 * function.h (function): Remove optabs field.
15749 * function.c (invoke_set_current_function_hook): Call
15750 init_tree_optimization_optabs. Use the result to initialize
15753 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
15755 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15756 if GTMA_HAS_NO_INSTRUMENTATION.
15757 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15758 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15759 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15760 * gimple-pretty-print.c (dump_gimple_transaction): Handle
15761 GTMA_HAS_NO_INSTRUMENTATION.
15763 2013-03-08 Jakub Jelinek <jakub@redhat.com>
15765 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15768 2013-03-08 Marek Polacek <polacek@redhat.com>
15769 Jakub Jelinek <jakub@redhat.com>
15771 PR tree-optimization/56478
15772 * predict.c (is_comparison_with_loop_invariant_p): Change the
15773 type of loop_step to tree.
15774 (predict_loops): Adjust.
15775 (predict_iv_comparison): Perform the computations on double_ints.
15777 2013-03-08 Richard Biener <rguenther@suse.de>
15779 PR tree-optimization/56570
15780 * tree-cfg.c (verify_expr_location_1): Verify locations for
15782 * tree-sra.c (create_access_replacement): Strip locations
15783 from DECL_DEBUG_EXPRs.
15785 2013-03-08 Richard Biener <rguenther@suse.de>
15787 * tree-inline.c (expand_call_inline): Do not associate
15788 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15789 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15791 2013-03-08 Richard Biener <rguenther@suse.de>
15793 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15794 or block changes with -Og. Fix for location / block encoding
15795 changes and PHI arguments with locations.
15797 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
15799 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15801 (struct output_info): Likewise.
15802 (register_overhead): Remove bad gcc_assert.
15803 (bitmap_find_bit): If there is only a single bitmap element, do not
15804 count a miss as a search.
15805 (print_statistics): Update for counter type changes.
15806 (dump_bitmap_statistics): Likewise. Print headers such that they
15807 are properly lined up with the printed counters.
15809 2013-03-07 Jakub Jelinek <jakub@redhat.com>
15811 PR tree-optimization/56559
15812 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15813 check that it has only a single use.
15815 2013-03-07 Richard Biener <rguenther@suse.de>
15817 * doc/invoke.texi (fwhole-program): Discourage use in combination
15820 2013-03-06 Jakub Jelinek <jakub@redhat.com>
15822 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15824 PR tree-optimization/56539
15825 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15826 instead of GSI_CONTINUE_LINKING as last argument to
15827 force_gimple_operand_gsi. Adjust function comment.
15829 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15832 PR middle-end/56548
15833 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15834 promoted mode, convert the result back to the original mode.
15836 2013-03-06 Richard Biener <rguenther@suse.de>
15838 PR middle-end/56294
15839 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15840 (insert_updated_phi_nodes_compare_uids): New function.
15841 (update_ssa): Sort symbols_to_rename after UID before
15842 traversing it to insert PHI nodes.
15844 2013-03-06 Richard Biener <rguenther@suse.de>
15846 PR middle-end/50494
15847 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15848 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15851 2013-02-13 Richard Biener <rguenther@suse.de>
15854 * varasm.c (output_constant_def_1): Get the decl representing
15855 the constant as argument.
15856 (output_constant_def): Wrap output_constant_def_1.
15857 (make_decl_rtl): Use output_constant_def_1 with the decl
15858 representing the constant.
15859 (build_constant_desc): Optionally re-use a decl already
15860 representing the constant.
15861 (tree_output_constant_def): Adjust.
15863 2013-03-06 Joey Ye <joey.ye@arm.com>
15866 * gcc.c (convert_white_space): New function.
15867 (main): Handles white space in function name.
15869 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
15872 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15873 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
15874 to SH_DIV_CALL_TABLE for TARGET_SH2.
15875 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15877 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15878 call-table options.
15880 2013-03-05 Sterling Augustine <saugustine@google.com>
15881 Cary Coutant <ccoutant@google.com>
15884 * dwarf2out.c (resolve_addr): Don't call
15885 remove_loc_list_addr_table_entries a second time for the same
15888 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15891 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15892 (avoid_complex_debug_insns): New function.
15893 (expand_debug_locations): Call it.
15895 PR rtl-optimization/56484
15896 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15897 lifetimes of hard registers on small register class machines.
15899 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
15901 * config/microblaze/microblaze-protos.h: Rename
15902 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15903 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15905 (microblaze_fast_interrupt_function_p): New function.
15906 (microblaze_is_interrupt_handler): Rename to
15907 microblaze_is_interrupt_variant and add fast_interrupt check.
15908 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15909 (save_restore_insns): Likewise.
15910 (compute_frame_size): Likewise.
15911 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15912 (microblaze_globalize_label): Likewise.
15913 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15914 * config/microblaze/microblaze.md: Use wrapper
15915 microblaze_is_interrupt_variant.
15917 2013-03-05 Kai Tietz <ktietz@redhat.com>
15919 * sdbout.c (sdbout_one_type): Switch to current function's section
15920 supporting cold/hot.
15922 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
15924 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15927 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15929 PR middle-end/56461
15930 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15931 if VALGRIND_GET_VBITS is defined, temporarily make object
15932 memory all defined, and restore previous valgrind addressability
15933 and definability afterwards. Free this_object at the end.
15935 PR middle-end/56461
15936 * lra.c (lra): Call lra_clear_live_ranges if live_p,
15937 right before calling lra_create_live_ranges, also call it
15938 when clearing live_p. Only call lra_clear_live_ranges
15939 at the end if live_p.
15941 PR middle-end/56461
15942 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15944 2013-03-05 Richard Biener <rguenther@suse.de>
15946 PR tree-optimization/56521
15947 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15950 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
15953 * except.h (remove_unreachable_eh_regions): New prototype.
15954 * except.c (remove_eh_handler_splicer): New function, split out
15955 of remove_eh_handler.
15956 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
15957 warning about running it on many EH regions one at a time.
15958 (remove_unreachable_eh_regions_worker): New function, walk the
15959 EH tree in depth-first order and remove non-marked regions.
15960 (remove_unreachable_eh_regions): New function.
15961 * tree-eh.c (mark_reachable_handlers): New function, split out
15962 from remove_unreachable_handlers.
15963 (remove_unreachable_handlers): Use mark_reachable_handlers and
15964 remove_unreachable_eh_regions.
15965 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15966 and remove_unreachable_eh_regions.
15968 2013-03-05 Richard Biener <rguenther@suse.de>
15970 PR middle-end/56525
15971 * loop-init.c (fix_loop_structure): Remove loops in two stages,
15972 not freeing them until the end.
15974 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
15976 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15978 2013-03-05 Richard Biener <rguenther@suse.de>
15980 PR tree-optimization/56270
15981 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15982 of loads after scheduling an SLP instance.
15984 2013-03-05 Jakub Jelinek <jakub@redhat.com>
15986 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15988 (check_gcc_parallelize): Run guality.exp as a separate job from
15989 vect.exp with unsorted.exp and $(dg_target_exps) separately from
15990 struct-layout-1.exp with stackalign.exp.
15992 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15994 PR middle-end/56461
15995 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15996 load_index sbitmap even if some bit in it isn't set.
15998 PR middle-end/56461
15999 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
16000 (discover_iteration_bound_by_body_walk): Change queues to
16001 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
16002 spelling in comment. Call safe_push on queues[bound_index] directly.
16003 Release queues[queue_index] in every iteration unconditionally.
16004 Release bounds vector.
16006 PR middle-end/56461
16007 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
16008 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
16009 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
16012 2013-03-05 Richard Biener <rguenther@suse.de>
16015 * tree-inline.c (remap_blocks_to_null): New function.
16016 (expand_call_inline): When expanding a call stmt without
16017 an associated block inline remap all callee blocks to NULL.
16019 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16021 PR rtl-optimization/56494
16022 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
16023 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
16024 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
16026 PR middle-end/56461
16027 * sel-sched-ir.c (free_sched_pools): Release
16028 succs_info_pool.stack[succs_info_pool.max_top] vectors too
16029 if succs_info_pool.max_top isn't -1.
16032 * opts.c (opts_obstack, opts_concat): Moved to...
16033 * opts-common.c (opts_obstack, opts_concat): ... here.
16035 2013-03-04 Jakub Jelinek <jakub@redhat.com>
16037 PR middle-end/56461
16038 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
16040 2013-03-04 Martin Jambor <mjambor@suse.cz>
16042 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
16043 all appropriate places.
16045 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
16047 PR tree-optimization/56424
16048 * ipa-split.c (split_function): Do not set the RSO flag if result is
16049 not by reference and its type is a register type.
16051 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
16053 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
16054 (microblaze_legitimate_pic_operand): Likewise
16055 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
16056 new function microblaze_legitimate_pic_operand
16057 * config/microblaze/microblaze-protos.h
16058 (microblaze_legitimate_pic_operand): Declare.
16060 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16062 * config/microblaze/predicates.md (call_insn_simple_operand):
16063 New predicate for supported rtx code types.
16064 * config/microblaze/microblaze.md (call_internal1): Use
16065 call_insn_simple_operand predicate.
16067 2013-03-04 Jakub Jelinek <jakub@redhat.com>
16069 PR middle-end/56461
16070 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
16071 partitions.ordered_remove.
16073 PR middle-end/56461
16074 * tree-vect-stmts.c (vectorizable_conversion): Don't call
16075 vec_oprnds0.create (1) for modifier == NONE.
16077 PR middle-end/56461
16078 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
16079 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
16080 vec_oprnds1 right before pushing anything to it for
16083 PR middle-end/56461
16084 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
16085 set nbbs to 0 instead of having separate code path.
16086 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
16087 instead of false as last argument if returning NULL.
16089 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
16091 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
16092 the attribute is now called "target" instead of "option".
16093 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
16094 * doc/tm.texi.in (Target Attributes): Likewise document the correct
16095 attribute/pragma name for TARGET_OPTION_VALID_P and
16096 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
16097 * doc/tm.texi: Regenerated.
16099 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
16101 * config/microblaze/microblaze.c:
16102 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
16103 * config/microblaze/microblaze.h: Add -mxl-reorder to
16105 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
16106 instructions emitted if TARGET_REORDER.
16107 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
16108 or 0 for -m/-mno case, but initialises as 2 to detect default use case
16111 2013-03-01 Xinliang David Li <davidxl@google.com>
16113 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
16116 2013-03-01 Jakub Jelinek <jakub@redhat.com>
16118 PR middle-end/56461
16119 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
16120 vector even when returning true. Fix up function comment formatting.
16122 PR middle-end/56461
16123 * ira-build.c (ira_loop_nodes_count): New variable.
16124 (create_loop_tree_nodes): Initialize it.
16125 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
16127 PR middle-end/56461
16128 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
16129 method on dr_chain and result_chain.
16130 * tree-vect-stmts.c (vectorizable_store): Only call
16131 result_chain.create if j == 0.
16133 PR middle-end/56461
16134 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
16135 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
16136 before overwriting it.
16138 2013-03-01 Tobias Burnus <burnus@net-b.de>
16140 * doc/extended.texi (C Extensions): Change order in @menu
16142 (Other MIPS Built-in Functions): Move last MIPS entry before
16143 "picoChip Built-in Functions".
16144 (SH Built-in Functions): Move after RX Built-in Functions.
16145 * doc/gcc.texi (Introduction): Change order in @menu
16147 * doc/md.texi (Constraints): Ditto.
16148 * gty.texi (Type Information): Ditto.
16149 (User-provided marking routines for template types): Make
16151 * doc/invoke.texi (AArch64 Options): Move before
16152 "Adapteva Epiphany Options".
16154 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
16155 Jakub Jelinek <jakub@redhat.com>
16158 * asan.c (gate_asan): Lookup no_sanitize_address instead of
16159 no_address_safety_analysis attribute.
16160 * doc/extend.texi (no_address_safety_attribute): Rename to
16161 no_sanitize_address attribute, mention no_address_safety_analysis
16162 attribute as deprecated alias.
16164 2013-02-28 Jakub Jelinek <jakub@redhat.com>
16166 PR middle-end/56461
16167 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
16168 type to vec<vec<tree> > *.
16169 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
16170 to be vec<tree> instead of vec<tree> *, set vec_defs
16171 to vNULL and call vec_defs.create (number_of_vects), adjust other
16173 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
16174 vectorizable_condition): Adjust vect_get_slp_defs callers.
16176 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16178 * config/aarch64/aarch64.c
16179 (aarch64_float_const_representable): Remove unused variable.
16181 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16183 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
16185 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16187 * config/aarch64/aarch64-builtins.c
16188 (aarch64_init_simd_builtins): Make static.
16190 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16192 * config/aarch64/aarch64.c
16193 (aarch64_simd_make_constant): Make static.
16195 2013-02-28 Martin Jambor <mjambor@suse.cz>
16197 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
16198 with no initialization to the RHS of debug statements.
16200 2013-02-28 Martin Jambor <mjambor@suse.cz>
16202 PR tree-optimization/56294
16203 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
16205 (get_access_replacement): Do not call create_access_replacement.
16206 Assert a replacement exists.
16207 (get_repl_default_def_ssa_name): Create the replacement declaration
16210 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16212 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
16213 final_end_function.
16215 2013-02-28 Marek Polacek <polacek@redhat.com>
16217 PR rtl-optimization/56466
16218 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
16219 if we're changing a loop.
16220 (peel_loops_completely): Likewise.
16222 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
16225 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
16227 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16230 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
16231 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
16232 INTX_FTYPE_FX, FX_FTYPE_INTX.
16233 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
16235 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16237 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
16238 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
16239 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
16240 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
16241 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
16242 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
16243 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
16244 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
16245 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
16246 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
16247 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
16248 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
16249 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
16250 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
16251 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
16252 (avrxmega6): Increase max flash segments from 5 to 6.
16253 * config/avr/t-multilib: Regenerate.
16254 * config/avr/avr-tables.opt: Regenerate.
16255 * doc/avr-mmcu.texi: Regenerate.
16257 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16259 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
16260 (avr_device_to_arch): Rename to avr_device_to_ld.
16261 (avr_device_to_as): New prototype.
16262 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
16263 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
16264 * config/avr/driver-avr.c (avr_device_to_as): New.
16265 (avr_device_to_arch): Rename to avr_device_to_ld.
16267 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16269 PR middle-end/56461
16270 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
16271 method on dr_chain and result_chain.
16273 PR middle-end/56461
16274 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
16275 pointer_set_destroy on not_executed_last_iteration.
16277 PR middle-end/56461
16278 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
16280 PR middle-end/56461
16281 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
16282 FOR_EACH_DEFINED_FUNCTION when freeing state.
16284 PR middle-end/56461
16285 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
16287 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
16290 PR middle-end/56461
16291 * ipa-cp.c (decide_whether_version_node): Call vec_free on
16292 known_aggs[i].items and release known_aggs vector.
16294 PR middle-end/56461
16295 * ipa-reference.c (propagate): Free node_info even for alias nodes.
16297 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16299 * config/microblaze/microblaze.c (microblaze_emit_compare):
16300 Use xor for EQ/NE comparisions.
16301 * config/microblaze/microblaze.md (cstoresf4): Add constraints
16302 (cbranchsf4): Adjust operator to comparison_operator.
16304 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16306 PR middle-end/56461
16307 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
16309 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
16310 vec_safe_push, always update *slot.
16311 (redirect_edge_var_map_clear): Use vec_free.
16312 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
16313 (free_var_map_entry): Use vec_free.
16314 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
16315 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
16317 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
16319 PR middle-end/45472
16320 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
16321 when the may_trap_p bit of the exprs being merged differs.
16322 Reorder tests for speculativeness in the logical and operator.
16324 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16326 * incpath.c (add_standard_paths): Use reconcat instead of concat
16327 where appropriate and avoid leaking memory.
16329 * opts.h: Include obstack.h.
16330 (opts_concat): New prototype.
16331 (opts_obstack): New declaration.
16332 * opts.c (opts_concat): New function.
16333 (opts_obstack): New variable.
16334 (init_options_struct): Call gcc_init_obstack on opts_obstack.
16335 (finish_options): Use opts_concat instead of concat
16336 and XOBNEWVEC instead of XNEWVEC.
16337 * opts-common.c (generate_canonical_option, decode_cmdline_option,
16338 generate_option): Likewise.
16339 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
16340 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
16343 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
16344 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
16346 2013-02-26 Jakub Jelinek <jakub@redhat.com>
16348 PR middle-end/56461
16349 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
16351 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
16353 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
16354 (arm_block_move_unaligned_straight): Likewise.
16355 (arm_adjust_block_mem): Likewise.
16357 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
16360 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
16361 temp, cond and label.
16362 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
16365 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16366 * config/c6x/c6x.h (dbx_register_map): Update declaration.
16369 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16370 of prologue/epilogue functions.
16373 * config/tilegx/tilegx.c (tilegx_expand_prologue):
16374 Remove unused variable cfa_offset.
16375 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16378 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16379 type promotion to unsigned.
16382 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
16383 for HOST_WIDE_INT of 32 bit / same size as int.
16384 (arm_block_move_unaligned_straight): Likewise.
16385 (arm_adjust_block_mem): Likewise.
16388 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
16391 2013-02-26 Marek Polacek <polacek@redhat.com>
16393 PR tree-optimization/56426
16394 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
16396 2013-02-26 Richard Biener <rguenther@suse.de>
16399 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
16400 unused variable loops.
16402 2013-02-26 Jakub Jelinek <jakub@redhat.com>
16404 PR tree-optimization/56448
16405 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16406 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16407 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16408 later operands of the references, or even first operand for
16409 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16411 PR tree-optimization/56443
16412 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16413 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16414 to type_for_mode langhook.
16416 2013-02-25 Matt Turner <mattst88@gmail.com>
16418 * doc/invoke.texi: Document r4700.
16420 2013-02-25 Richard Biener <rguenther@suse.de>
16422 PR tree-optimization/56175
16423 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16425 (simplify_bitwise_binary): ... here. Also guard the conversion
16426 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16428 2013-02-25 Catherine Moore <clm@codesourcery.com>
16431 2013-02-24 Catherine Moore <clm@codesourcery.com>
16432 Maciej W. Rozycki <macro@codesourcery.com>
16433 Tom de Vries <tom@codesourcery.com>
16434 Nathan Sidwell <nathan@codesourcery.com>
16435 Iain Sandoe <iain@codesourcery.com>
16436 Nathan Froyd <froydnj@codesourcery.com>
16437 Chao-ying Fu <fu@mips.com>
16439 * doc/extend.texi: (micromips, nomicromips, nocompression):
16440 Document new function attributes.
16441 * doc/invoke.texi (minterlink-compressed, mmicromips,
16442 m14k, m14ke, m14kec): Document new options.
16443 (minterlink-mips16): Update documentation.
16444 * doc/md.texi (ZC, ZD): Document new constraints.
16445 * configure.ac (gcc_cv_as_micromips): Check if linker
16446 supports the .set micromips directive.
16447 * configure: Regenerate.
16448 * config.in: Regenerate.
16449 * config/mips/mips-tables.opt: Regenerate.
16450 * config/mips/micromips.md: New file.
16451 * constraints.md (ZC, AD): New constraints.
16452 * config/mips/predicates.md (movep_src_register): New predicate.
16453 (movep_src_operand): New predicate.
16454 (non_volatile_mem_operand): New predicate.
16455 * config/mips/mips.md (multimem): New type.
16456 (length): Differentiate between 17-bit and 18-bit branch offsets.
16457 (MOVEP1, MOVEP2): New mode iterator.
16458 (mov_<load>l): Use ZC constraint.
16459 (mov_<load>r): Likewise.
16460 (mov_<store>l): Likewise.
16461 (mov_<store>r): Likewise.
16462 (*branch_equality<mode>_inverted): Add microMIPS support.
16463 (*branch_equality<mode>): Likewise.
16464 (*jump_absolute): Likewise.
16465 (indirect_jump_<mode>): Likewise.
16466 (tablejump_<mode>): Likewise.
16467 (<optab>_internal): Likewise.
16468 (sibcall_internal): Likewise.
16469 (sibcall_value_internal): Likewise.
16470 (prefetch): Use constraint ZD.
16471 * config/mips/mips.opt (minterlink-compressed): New option.
16472 (minterlink-mips16): Now an alias for minterlink-compressed.
16473 (mmicromips): New option.
16474 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16475 (compare_and_swap_12): Likewise.
16476 (sync_add<mode>): Likewise.
16477 (sync_<optab>_12): Likewise.
16478 (sync_old_<optab>_12): Likewise.
16479 (sync_new_<optab>_12): Likewise.
16480 (sync_nand_12): Likewise.
16481 (sync_old_nand_12): Likewise.
16482 (sync_new_nand_12): Likewise.
16483 (sync_sub<mode>): Likewise.
16484 (sync_old_add<mode>): Likewise.
16485 (sync_old_sub<mode>): Likewise.
16486 (sync_new_add<mode>): Likewise.
16487 (sync_new_sub<mode>): Likewise.
16488 (sync_<optab><mode>): Likewise.
16489 (sync_old_<optab><mode>): Likewise.
16490 (sync_new_<optab><mode>): Likewise.
16491 (sync_nand<mode>): Likewise.
16492 (sync_old_nand<mode>): Likewise.
16493 (sync_new_nand<mode>): Likewise.
16494 (sync_lock_test_and_set<mode>): Likewise.
16495 (test_and_set_12): Likewise.
16496 (atomic_compare_and_swap<mode>): Likewise.
16497 (atomic_exchange<mode>_llsc): Likewise.
16498 (atomic_fetch_add<mode>_llsc): Likewise.
16499 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16500 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16501 (umips_save_restore_pattern_p): Likewise.
16502 (umips_load_store_pair_p): Likewise.
16503 (umips_output_load_store_pair): Likewise.
16504 (umips_movep_target_p): Likewise.
16505 (umips_12bit_offset_address_p): Likewise.
16506 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16507 (mips_base_mips16): Rename this...
16508 (mips_base_compression_flags): ...to this. Update all uses.
16509 (mips_attribute_table): Add micromips, nomicromips and nocompression.
16510 (mips_mips16_decl_p): Delete.
16511 (mips_nomips16_decl_p): Delete.
16512 (mips_get_compress_on_flags): New function.
16513 (mips_get_compress_off_flags): New function.
16514 (mips_get_compress_mode): New function.
16515 (mips_get_compress_on_name): New function.
16516 (mips_get_compress_off_name): New function.
16517 (mips_insert_attributes): Support multiple compression types.
16518 (mips_merge_decl_attributes): Likewise.
16519 (umips_12bit_offset_address_p): New function.
16520 (mips_start_function_definition): Emit .set micromips directive.
16521 (mips_call_may_need_jalx_p): New function.
16522 (mips_function_ok_for_sibcall): Add microMIPS support.
16523 (mips_print_operand_punctuation): Support short delay slots and
16525 (umips_swm_mask, umips_swm_encoding): New.
16526 (umips_build_save_restore): New function.
16527 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16528 (was_mips16_p): Remove.
16529 (old_compression_mode): New.
16530 (mips_set_compression_mode): New function.
16531 (mips_set_current_function): Add microMIPS support.
16532 (mips_option_override): Likewise.
16533 (umips_save_restore_pattern_p): New function.
16534 (umips_output_save_restore): New function.
16535 (umips_load_store_pair_p_1): New function.
16536 (umips_load_store_pair_p): New function.
16537 (umips_output_load_store_pair_1): New function.
16538 (umips_output_load_store_pair): New function.
16539 (umips_movep_target_p) New function.
16540 (mips_prepare_pch_save): Add microMIPS support.
16541 * config/mips/mips.h (TARGET_COMPRESSION): New.
16542 (TARGET_CPU_CPP_BUILTINS): Update macro
16543 to use new compression flags and to support microMIPS.
16544 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16545 (MIPS_ARCH_FLOAT_SPEC): Likewise.
16546 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16547 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16548 (ASM_SPEC): Support mmicromips and mno-micromips.
16549 (M16STORE_REG_P): New macro.
16550 (MIPS_CALL): Support TARGET_MICROMIPS.
16551 (MICROMIPS_J): New macro.
16552 (mips_base_mips16): Rename this...
16553 (mips_base_compression_flags): ...to this.
16554 (UMIPS_12BIT_OFFSET_P): New macro.
16555 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16556 (MULTILIB_DIRNAMES): Likewise.
16558 2013-02-25 Tom de Vries <tom@codesourcery.com>
16560 PR rtl-optimization/56131
16561 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16562 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16563 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16565 2013-02-25 Tobias Burnus <burnus@net-b.de>
16567 * doc/invoke.texi (-fsanitize=): Move from optimization
16568 to debugging options.
16570 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
16572 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16574 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
16575 Alexander Monakov <amonakov@ispras.ru>
16577 PR middle-end/56077
16578 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16579 flush pending lists also on non-jumps. Adjust comment.
16581 2013-02-24 Catherine Moore <clm@codesourcery.com>
16582 Maciej W. Rozycki <macro@codesourcery.com>
16583 Tom de Vries <tom@codesourcery.com>
16584 Nathan Sidwell <nathan@codesourcery.com>
16585 Iain Sandoe <iain@codesourcery.com>
16586 Nathan Froyd <froydnj@codesourcery.com>
16587 Chao-ying Fu <fu@mips.com>
16589 * doc/extend.texi: (micromips, nomicromips, nocompression):
16590 Document new function attributes.
16591 * doc/invoke.texi (minterlink-compressed, mmicromips,
16592 m14k, m14ke, m14kec): Document new options.
16593 (minterlink-mips16): Update documentation.
16594 * doc/md.texi (ZC, ZD): Document new constraints.
16595 * configure.ac (gcc_cv_as_micromips): Check if linker
16596 supports the .set micromips directive.
16597 * configure: Regenerate.
16598 * config.in: Regenerate.
16599 * config/mips/mips-tables.opt: Regenerate.
16600 * config/mips/micromips.md: New file.
16601 * constraints.md (ZC, AD): New constraints.
16602 * config/mips/predicates.md (movep_src_register): New predicate.
16603 (movep_src_operand): New predicate.
16604 (non_volatile_mem_operand): New predicate.
16605 * config/mips/mips.md (multimem): New type.
16606 (length): Differentiate between 17-bit and 18-bit branch offsets.
16607 (MOVEP1, MOVEP2): New mode iterator.
16608 (mov_<load>l): Use ZC constraint.
16609 (mov_<load>r): Likewise.
16610 (mov_<store>l): Likewise.
16611 (mov_<store>r): Likewise.
16612 (*branch_equality<mode>_inverted): Add microMIPS support.
16613 (*branch_equality<mode>): Likewise.
16614 (*jump_absolute): Likewise.
16615 (indirect_jump_<mode>): Likewise.
16616 (tablejump_<mode>): Likewise.
16617 (<optab>_internal): Likewise.
16618 (sibcall_internal): Likewise.
16619 (sibcall_value_internal): Likewise.
16620 (prefetch): Use constraint ZD.
16621 * config/mips/mips.opt (minterlink-compressed): New option.
16622 (minterlink-mips16): Now an alias for minterlink-compressed.
16623 (mmicromips): New option.
16624 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16625 (compare_and_swap_12): Likewise.
16626 (sync_add<mode>): Likewise.
16627 (sync_<optab>_12): Likewise.
16628 (sync_old_<optab>_12): Likewise.
16629 (sync_new_<optab>_12): Likewise.
16630 (sync_nand_12): Likewise.
16631 (sync_old_nand_12): Likewise.
16632 (sync_new_nand_12): Likewise.
16633 (sync_sub<mode>): Likewise.
16634 (sync_old_add<mode>): Likewise.
16635 (sync_old_sub<mode>): Likewise.
16636 (sync_new_add<mode>): Likewise.
16637 (sync_new_sub<mode>): Likewise.
16638 (sync_<optab><mode>): Likewise.
16639 (sync_old_<optab><mode>): Likewise.
16640 (sync_new_<optab><mode>): Likewise.
16641 (sync_nand<mode>): Likewise.
16642 (sync_old_nand<mode>): Likewise.
16643 (sync_new_nand<mode>): Likewise.
16644 (sync_lock_test_and_set<mode>): Likewise.
16645 (test_and_set_12): Likewise.
16646 (atomic_compare_and_swap<mode>): Likewise.
16647 (atomic_exchange<mode>_llsc): Likewise.
16648 (atomic_fetch_add<mode>_llsc): Likewise.
16649 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16650 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16651 (umips_save_restore_pattern_p): Likewise.
16652 (umips_load_store_pair_p): Likewise.
16653 (umips_output_load_store_pair): Likewise.
16654 (umips_movep_target_p): Likewise.
16655 (umips_12bit_offset_address_p): Likewise.
16656 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16657 (mips_base_mips16): Rename this...
16658 (mips_base_compression_flags): ...to this. Update all uses.
16659 (mips_attribute_table): Add micromips, nomicromips and nocompression.
16660 (mips_mips16_decl_p): Delete.
16661 (mips_nomips16_decl_p): Delete.
16662 (mips_get_compress_on_flags): New function.
16663 (mips_get_compress_off_flags): New function.
16664 (mips_get_compress_mode): New function.
16665 (mips_get_compress_on_name): New function.
16666 (mips_get_compress_off_name): New function.
16667 (mips_insert_attributes): Support multiple compression types.
16668 (mips_merge_decl_attributes): Likewise.
16669 (umips_12bit_offset_address_p): New function.
16670 (mips_start_function_definition): Emit .set micromips directive.
16671 (mips_call_may_need_jalx_p): New function.
16672 (mips_function_ok_for_sibcall): Add microMIPS support.
16673 (mips_print_operand_punctuation): Support short delay slots and
16675 (umips_swm_mask, umips_swm_encoding): New.
16676 (umips_build_save_restore): New function.
16677 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16678 (was_mips16_p): Remove.
16679 (old_compression_mode): New.
16680 (mips_set_compression_mode): New function.
16681 (mips_set_current_function): Add microMIPS support.
16682 (mips_option_override): Likewise.
16683 (umips_save_restore_pattern_p): New function.
16684 (umips_output_save_restore): New function.
16685 (umips_load_store_pair_p_1): New function.
16686 (umips_load_store_pair_p): New function.
16687 (umips_output_load_store_pair_1): New function.
16688 (umips_output_load_store_pair): New function.
16689 (umips_movep_target_p) New function.
16690 (mips_prepare_pch_save): Add microMIPS support.
16691 * config/mips/mips.h (TARGET_COMPRESSION): New.
16692 (TARGET_CPU_CPP_BUILTINS): Update macro
16693 to use new compression flags and to support microMIPS.
16694 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16695 (MIPS_ARCH_FLOAT_SPEC): Likewise.
16696 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16697 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16698 (ASM_SPEC): Support mmicromips and mno-micromips.
16699 (M16STORE_REG_P): New macro.
16700 (MIPS_CALL): Support TARGET_MICROMIPS.
16701 (MICROMIPS_J): New macro.
16702 (mips_base_mips16): Rename this...
16703 (mips_base_compression_flags): ...to this.
16704 (UMIPS_12BIT_OFFSET_P): New macro.
16705 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16706 (MULTILIB_DIRNAMES): Likewise.
16708 2013-02-24 Jakub Jelinek <jakub@redhat.com>
16711 * target-globals.c (save_target_globals): For init_reg_sets and
16712 target_reinit remporarily set this_fn_optabs to this_target_optabs.
16714 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
16716 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16717 * config/aarch64/t-aarch64
16718 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16720 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
16722 PR inline-asm/56148
16723 * lra-constraints.c (process_alt_operands): Reload operand
16724 conflicting with earlier clobber only if no more other conflicting
16727 2013-02-22 Jakub Jelinek <jakub@redhat.com>
16730 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16731 if not linking a shared library.
16733 2013-02-22 Seth LaForge <sethml@google.com>
16735 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16737 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
16739 * config/arm/arm.md (split for extendsidi): Update condition.
16740 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16741 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16742 (qhs_zextenddi_cstr): Likewise.
16744 2013-02-21 Jakub Jelinek <jakub@redhat.com>
16746 PR middle-end/56420
16747 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16748 avoid signed wrapping.
16749 (expand_mult): Handle properly multiplication by
16750 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
16751 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
16752 in the compiler if coeff is HOST_WIDE_INT_MIN.
16753 (expand_divmod): Don't make ext_op1 static, change it's type to
16754 uhwi. Avoid undefined behavior in -INTVAL (op1).
16756 PR rtl-optimization/50339
16757 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16759 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16760 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16761 into splitting_ashiftrt field.
16762 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16764 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16767 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
16769 PR middle-end/56108
16770 * trans-mem.c (execute_tm_mark): Do not expand transactions that
16771 are sure to go irrevocable.
16773 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
16775 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16776 scalars are valid operands.
16778 2013-02-21 Martin Jambor <mjambor@suse.cz>
16780 PR tree-optimization/56310
16781 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16782 only matching indices and non-negative final offsets.
16783 (intersect_aggregates_with_edge): Pass src_idx to
16784 agg_replacements_to_vector. Pass src_idx insstead of index to
16785 intersect_with_agg_replacements.
16787 2013-02-21 Martin Jambor <mjambor@suse.cz>
16789 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16790 instead of hard-wired defaults.
16792 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
16794 * doc/invoke.texi (MIPS Options): Update documentation of the
16795 floating-point multiply-accumulate instruction restrictions.
16797 2013-02-21 Kostya Serebryany <kcc@google.com>
16799 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16800 asan_shadow_offset on x86_64 linux.
16802 2013-02-21 Richard Biener <rguenther@suse.de>
16804 PR tree-optimization/56415
16806 2013-02-11 Richard Biener <rguenther@suse.de>
16808 PR tree-optimization/56273
16809 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16812 2013-02-21 Jakub Jelinek <jakub@redhat.com>
16815 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16818 PR inline-asm/56405
16819 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16820 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16822 2013-02-20 Jan Hubicka <jh@suse.cz>
16824 PR tree-optimization/56265
16825 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16826 when target is referenced for first time.
16828 2013-02-20 Richard Biener <rguenther@suse.de>
16830 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16831 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16832 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16833 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16834 not return anything.
16835 (rename_ssa_copies): Do not remove unused locals.
16836 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16837 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16838 * passes.c (execute_function_todo): Do not schedule unused locals
16839 removal if cleanup_tree_cfg did something.
16840 * tree-ssa-live.c (remove_unused_locals): Dump statistics
16841 about the number of removed locals.
16843 2013-02-20 Richard Biener <rguenther@suse.de>
16845 PR tree-optimization/56398
16846 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16848 2013-02-20 Martin Jambor <mjambor@suse.cz>
16850 PR tree-optimization/55334
16851 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16852 restricted pointers to arrays.
16854 2013-02-20 Richard Biener <rguenther@suse.de>
16855 Jakub Jelinek <jakub@redhat.com>
16857 PR tree-optimization/56396
16858 * tree-ssa-ccp.c (n_const_val): New static variable.
16859 (get_value): Return NULL for SSA names we don't have a lattice
16861 (ccp_initialize): Initialize n_const_val.
16862 * tree-ssa-copy.c (n_copy_of): New static variable.
16863 (init_copy_prop): Initialize n_copy_of.
16864 (get_value): Return NULL_TREE for SSA names we don't have a
16867 2013-02-20 Martin Jambor <mjambor@suse.cz>
16869 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16871 2013-02-20 Richard Biener <rguenther@suse.de>
16873 * genpreds.c (write_lookup_constraint): Do not compare first
16874 letter of the constraint again.
16876 2013-02-20 Richard Biener <rguenther@suse.de>
16878 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16880 (get_use_iv_cost): Terminate hashtable walk when coming across
16883 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
16885 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16886 reassociation for avx2 targets.
16888 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16890 * config/microblaze/microblaze.c: microblaze_has_clz = 0
16891 Add version check for v8.10.a to enable microblaze_has_clz
16892 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16893 version and TARGET_PATTERN_COMPARE check
16894 * config/microblaze/microblaze.md: New clzsi2 instruction
16896 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16898 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16899 function before branching.
16901 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
16903 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16905 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16907 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
16909 PR middle-end/55889
16910 * sel-sched.c: Include ira.h.
16911 (implicit_clobber_conflict_p): New function.
16912 (moveup_expr): Use it.
16913 * Makefile.in (sel-sched.o): Depend on ira.h.
16915 2013-02-19 Richard Biener <rguenther@suse.de>
16917 PR tree-optimization/56384
16918 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16919 (vn_hash_type): Split out from ...
16920 (vn_hash_constant_with_type): ... here.
16921 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16922 (vn_phi_eq): Compare types from vn_phi_s structure.
16923 (vn_phi_lookup): Populate vn_phi_s type.
16924 (vn_phi_insert): Likewise.
16926 2013-02-19 Jakub Jelinek <jakub@redhat.com>
16928 PR tree-optimization/56350
16929 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16930 if haven't found reduction or nested cycle operand, rather than
16931 asserting we must find it.
16933 PR tree-optimization/56381
16934 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16937 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
16938 Jakub Jelinek <jakub@redhat.com>
16941 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16942 (swap_optab_enable): Same.
16943 (init_all_optabs): Use argument instead of global.
16944 * tree.h (struct tree_optimization_option): New field target_optabs.
16945 * expr.h (init_all_optabs): Add argument to prototype.
16946 (TREE_OPTIMIZATION_OPTABS): New.
16947 (save_optabs_if_changed): Protoize.
16948 * optabs.h: Declare this_fn_optabs.
16949 * optabs.c (save_optabs_if_changed): New.
16950 Declare this_fn_optabs.
16951 (init_optabs): Add argument to init_all_optabs() call.
16952 * function.c (invoke_set_current_function_hook): Handle per
16954 * function.h (struct function): New field optabs.
16955 * config/mips/mips.c (mips_set_mips16_mode): Handle when
16956 optimization_current_node has changed.
16957 * target-globals.h (save_target_globals_default_opts): Protoize.
16958 * target-globals.c (save_target_globals_default_opts): New.
16960 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
16963 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16964 registers %fr12 and %fr12R as call used.
16967 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16968 and HImode, require all displacements to be an integer multiple of
16970 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16971 only allow QImode and HImode when reload is in progress and strict is
16972 true. Likewise for symbolic addresses. Use base14_operand to check
16973 displacements in REG+BASE addresses.
16975 2013-02-18 Richard Biener <rguenther@suse.de>
16977 PR tree-optimization/56366
16978 * tree-vect-loop.c (get_initial_def_for_induction): Properly
16979 handle sign-conversion of outer-loop initial induction value.
16981 2013-02-18 Richard Biener <rguenther@suse.de>
16983 PR middle-end/56349
16984 * cfghooks.c (merge_blocks): If we merge a latch into another
16985 block adjust references to it.
16986 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16987 (verify_loop_structure): Verify that a recorded latch is in fact
16990 2013-02-18 Richard Biener <rguenther@suse.de>
16992 PR tree-optimization/56321
16993 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16994 order SSA name release and virtual operand unlinking.
16996 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16998 * config/microblaze/microblaze.md (save_stack_block): Define.
16999 (restore_stack_block): Likewise.
17001 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17003 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
17004 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
17005 * config/microblaze/microblaze.c (microblaze_option_override):
17006 Bail out early for PIC modes when target does not support PIC.
17008 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17010 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
17011 Replace with a microblaze version.
17012 (microblaze_trampoline_init): Adapt for microblaze.
17013 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
17016 2013-02-16 Jakub Jelinek <jakub@redhat.com>
17017 Dodji Seketeli <dodji@redhat.com>
17020 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
17021 (instrument_mem_region_access): Do not forget to always put
17022 instrumentation of the of 'base' and 'base + len' in a "if (len !=
17023 0) statement, even for cases where either 'base' or 'base + len'
17024 are not instrumented -- because they have been previously
17025 instrumented. Simplify the logic by putting all the statements
17026 instrument 'base + len' inside a sequence, and then insert that
17027 sequence right before the current insertion point. Then, to
17028 instrument 'base + len', just get an iterator on that statement.
17029 And do not forget to update the pointer to iterator the function
17030 received as argument.
17032 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
17034 PR rtl-optimization/56348
17035 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
17037 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
17039 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
17040 (clean_graph_dump_file): Pass base to start_graph_dump.
17042 2013-02-14 Richard Henderson <rth@redhat.com>
17045 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
17047 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
17049 * collect2-aix.h: Define F_LOADONLY.
17051 2013-02-14 Richard Biener <rguenther@suse.de>
17054 * varasm.c (output_constant_def_1): Get the decl representing
17055 the constant as argument.
17056 (output_constant_def): Wrap output_constant_def_1.
17057 (make_decl_rtl): Use output_constant_def_1 with the decl
17058 representing the constant.
17059 (build_constant_desc): Optionally re-use a decl already
17060 representing the constant.
17061 (tree_output_constant_def): Adjust.
17063 2013-02-14 Dodji Seketeli <dodji@redhat.com>
17066 * asan.c (instrument_builtin_call): Really put the length of the
17067 second source argument into src1_len.
17069 2013-02-13 Jakub Jelinek <jakub@redhat.com>
17071 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
17072 argument. If it is false, don't create edge from then_bb to
17074 (insert_if_then_before_iter): Pass true to it.
17075 (build_check_stmt): Pass false to it.
17076 (transform_statements): Flush hash table only on extended basic
17077 block boundaries, rather than at the beginning of every bb.
17078 Don't flush hash table on nonfreeing_call_p calls.
17079 * tree-flow.h (nonfreeing_call_p): New prototype.
17080 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
17082 2013-02-13 David S. Miller <davem@davemloft.net>
17084 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
17086 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
17089 * ira.c (max_regno_before_ira): Move from ...
17091 (fix_reg_equiv_init): Use max_regno_before_ira instead of
17094 2013-02-13 Jakub Jelinek <jakub@redhat.com>
17096 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
17098 2013-02-13 Richard Biener <rguenther@suse.de>
17101 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
17102 globals in MEM_REFs.
17104 2013-02-13 Richard Biener <rguenther@suse.de>
17106 * loop-init.c (loop_optimizer_init): Clear loop state when
17107 re-initializing preserved loops.
17108 * loop-unswitch.c (unswitch_single_loop): Return whether
17109 we unswitched the loop. Do not verify loop state here.
17110 (unswitch_loops): When we unswitched a loop discover new loops.
17112 2013-02-13 Kostya Serebryany <kcc@google.com>
17114 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
17116 * sanitizer.def: Rename __asan_init to __asan_init_v1.
17118 2013-02-12 Dodji Seketeli <dodji@redhat.com>
17120 Avoid instrumenting duplicated memory access in the same basic block
17121 * Makefile.in (asan.o): Add new dependency on hash-table.h
17122 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
17123 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
17124 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
17125 (free_mem_ref_resources, has_mem_ref_been_instrumented)
17126 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
17127 (get_mem_ref_of_assignment): New functions.
17128 (get_mem_refs_of_builtin_call): Extract from
17129 instrument_builtin_call and tweak a little bit to make it fit with
17131 (instrument_builtin_call): Use the new
17132 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
17133 of is_gimple_builtin_call.
17134 (instrument_derefs, instrument_mem_region_access): Insert the
17135 instrumented memory reference into the hash table.
17136 (maybe_instrument_assignment): Renamed instrument_assignment into
17137 this, and change it to advance the iterator when instrumentation
17138 actually happened and return true in that case. This makes it
17139 homogeneous with maybe_instrument_assignment, and thus give a
17140 chance to callers to be more 'regular'.
17141 (transform_statements): Clear the memory reference hash table
17142 whenever we enter a new BB, when we cross a function call, or when
17143 we are done transforming statements. Use
17144 maybe_instrument_assignment instead of instrumentation. No more
17145 need to special case maybe_instrument_assignment and advance the
17146 iterator after calling it; it's now handled just like
17147 maybe_instrument_call. Update comment.
17149 2013-02-13 Richard Biener <rguenther@suse.de>
17151 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
17152 Fix loop discovery code.
17154 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
17156 PR inline-asm/56148
17157 * lra-constraints.c (process_alt_operands): Match early clobber
17158 operand with itself. Check conflicts with earlyclobber only if
17159 the operand is not reloaded. Prefer to reload conflicting operand
17160 if earlyclobber and matching operands are the same.
17162 2013-02-12 Richard Biener <rguenther@suse.de>
17165 * lto-streamer-out.c (write_symbol): Do not output symbols
17166 for hard register variables.
17168 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
17171 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
17172 (umulsidi3_insn, mulsidi3_insn): New insns.
17174 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
17176 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
17177 (struct tune_params): Add vec_costs field.
17178 * config/arm/arm.c (arm_builtin_vectorization_cost)
17179 (arm_add_stmt_cost): New functions.
17180 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
17181 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
17182 (arm_default_vec_cost): New struct of type cpu_vec_costs.
17183 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
17184 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
17185 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
17186 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
17188 2013-02-12 Richard Biener <rguenther@suse.de>
17191 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
17192 decls again if possible.
17194 2013-02-12 Richard Biener <rguenther@suse.de>
17196 PR middle-end/56288
17197 * tree-ssa.c (verify_ssa_name): Fix check, move
17198 SSA_NAME_IN_FREE_LIST check up.
17200 2013-02-12 Jakub Jelinek <jakub@redhat.com>
17201 Steven Bosscher <steven@gcc.gnu.org>
17203 PR rtl-optimization/56151
17204 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
17205 equal to op0 or op1, and last_insn pattern is CODE operation
17206 with MEM dest and one of the operands matches that MEM.
17208 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
17210 * doc/extend.texi: Document Function Multiversioning and "default"
17211 parameter string to target attribute.
17212 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
17213 target attribute parameter is "default".
17214 (ix86_compare_version_priority): Remove checks for target attribute.
17215 (ix86_mangle_function_version_assembler_name): Change error to sorry.
17216 Remove check for target attribute equal to NULL. Add assert.
17217 (ix86_generate_version_dispatcher_body): Change error to sorry.
17219 2013-02-11 Iain Sandoe <iain@codesourcery.com>
17220 Jack Howarth <howarth@bromo.med.uc.edu>
17221 Patrick Marlier <patrick.marlier@gmail.com>
17224 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
17225 define ENDFILE_SPEC as TM_DESTRUCTOR.
17226 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
17228 2013-02-11 Alexander Potapenko <glider@google.com>
17229 Jack Howarth <howarth@bromo.med.uc.edu>
17230 Jakub Jelinek <jakub@redhat.com>
17233 * config/darwin.c (cdtor_record): Rename ctor_record.
17234 (sort_cdtor_records): Rename sort_ctor_records.
17235 (finalize_dtors): New routine to sort destructors by
17236 priority before use in assemble_integer.
17237 (machopic_asm_out_destructor): Use finalize_dtors if needed.
17239 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
17241 PR rtl-optimization/56275
17242 * simplify-rtx.c (avoid_constant_pool_reference): Check that
17243 offset is non-negative and less than cmode size before
17244 calling simplify_subreg.
17246 2013-02-11 Richard Biener <rguenther@suse.de>
17248 PR tree-optimization/56264
17249 * cfgloop.h (fix_loop_structure): Adjust prototype.
17250 * loop-init.c (fix_loop_structure): Return the number of
17251 newly discovered loops.
17252 * tree-cfgcleanup.c (repair_loop_structures): When new loops
17253 are discovered, do a full loop-closed SSA rewrite.
17255 2013-02-11 Richard Biener <rguenther@suse.de>
17257 PR tree-optimization/56273
17258 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17260 (check_array_ref): Fix missing newline in dumps.
17261 (search_for_addr_array): Likewise.
17263 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
17265 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
17267 2013-02-09 Jakub Jelinek <jakub@redhat.com>
17270 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
17272 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
17274 PR rtl-optimization/56246
17275 * lra-constraints.c (simplify_operand_subreg): Try to reuse
17277 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
17278 constraints are satisfied.
17280 2013-02-08 Jeff Law <law@redhat.com>
17283 * emit-rtl.c (reg_is_parm_p): New function.
17284 * regs.h (reg_is_parm_p): New prototype.
17285 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
17286 callee-clobbered registers.
17288 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
17291 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
17292 If there is no implicit builtin declaration, just return NULL.
17294 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17296 * config/i386/sse.md (FMAMODEM): New mode iterator.
17297 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
17298 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
17300 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17302 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
17303 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
17304 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
17306 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17308 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
17309 (microblaze*-*-elf): Likewise.
17310 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
17312 * config/microblaze/microblaze-c.c: Add builtin defines for
17313 _LITTLE_ENDIAN and _BIG_ENDIAN.
17314 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
17315 add to TARGET_DEFAULT flags.
17316 Expand ASM_SPEC and LINK_SPEC.
17317 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
17318 * config/microblaze/microblaze.md: Update extendsidi2 and
17319 movdi_internal instructions to use low-order / high-order reg
17321 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
17322 options and inversemask / mask of LITTLE_ENDIAN.
17323 * config/microblaze/t-microblaze: Expand multilib options to
17324 include mlittle-endian (le) and update exceptions patterns.
17326 2013-02-08 Jakub Jelinek <jakub@redhat.com>
17328 PR rtl-optimization/56195
17329 * lra-constraints.c (get_reload_reg): Don't reuse regs
17330 if they have smaller mode than requested, if they have
17331 wider mode than requested, try to return a SUBREG.
17333 PR tree-optimization/56250
17334 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
17335 if type is unsigned and code isn't MULT_EXPR.
17337 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
17339 PR tree-optimization/56064
17340 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
17341 bits according to mode.
17342 * fixed-value.h (fixed_from_double_int)
17343 (const_fixed_from_double_int): Adjust comments.
17345 2013-02-08 Richard Biener <rguenther@suse.de>
17348 * lto-streamer.h (struct data_in): Remove current_file, current_line
17349 and current_col members.
17350 * lto-streamer-out.c (lto_output_location): Stream changed bits
17351 en-block for efficiency.
17352 * lto-streamer-in.c (clear_line_info): Remove.
17353 (lto_input_location): Cache current file, line and column
17354 globally via local statics. Read changed bits en-block.
17355 (input_function): Do not call clear_line_info.
17356 (lto_read_body): Likewise.
17357 (lto_input_toplevel_asms): Likewise.
17359 2013-02-08 Michael Matz <matz@suse.de>
17361 PR tree-optimization/52448
17362 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
17363 (nt_call_phase): New static.
17364 (add_or_mark_expr): Only mark accesses with newer phase than any
17366 (nonfreeing_call_p): New.
17367 (nt_init_block): Update nt_call_phase, mark blocks as visited.
17368 (nt_fini_block): Keep blocks marked as visited.
17369 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
17371 2013-02-08 Richard Biener <rguenther@suse.de>
17373 * ira.c (ira): Free broken dominator information.
17375 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17377 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
17379 2013-02-08 Marek Polacek <polacek@redhat.com>
17381 * cfgloop.c (verify_loop_structure): Add more checking of headers.
17383 2013-02-08 Richard Biener <rguenther@suse.de>
17385 PR middle-end/56181
17386 * cfgloop.h (flow_loops_find): Adjust.
17387 (bb_loop_header_p): Declare.
17388 * cfgloop.c (bb_loop_header_p): New function split out from ...
17389 (flow_loops_find): ... here. Adjust function signature,
17390 support incremental loop structure update.
17391 (verify_loop_structure): Cleanup. Verify a loop is a loop.
17392 * cfgloopmanip.c (fix_loop_structure): Move ...
17393 * loop-init.c (fix_loop_structure): ... here.
17394 (apply_loop_flags): Split out from ...
17395 (loop_optimizer_init): ... here.
17396 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
17397 in incremental mode, only remove dead loops here.
17399 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
17402 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
17403 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
17404 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17405 (*round<mode>3.libgcc): New insns for fixed-modes.
17406 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17407 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17408 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17409 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17410 implementations. Define to __builtin_avr_absFX,
17411 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17412 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17413 __builtin_avr_countlsFX, respectively.
17414 * config/avr/avr-c.c (target.h): Include it.
17415 (enum avr_builtin_id): New enum.
17416 (avr_resolve_overloaded_builtin): New static function.
17417 (avr_register_target_pragmas): Use it to set
17418 targetm.resolve_overloaded_builtin.
17419 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17420 tree nodes used by DEF_BUILTIN.
17421 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17422 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17423 <AVR_BUILTIN_xxBITS>: Same.
17425 2013-02-08 Richard Biener <rguenther@suse.de>
17427 * cfgloop.c (verify_loop_structure): Properly handle
17428 a loop exiting to another loop header.
17429 * ira-int.h (ira_loops): Remove.
17430 * ira.c (ira_loops): Remove.
17431 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17432 (do_reload): Use loop_optimizer_finalize.
17433 * ira-build.c (create_loop_tree_nodes): Use get_loops and
17434 number_of_loops to access the loop tree.
17435 (more_one_region_p): Likewise.
17436 (finish_loop_tree_nodes): Likewise.
17437 (rebuild_regno_allocno_maps): Likewise.
17438 (mark_loops_for_removal): Likewise.
17439 (mark_all_loops_for_removal): Likewise.
17440 (remove_unnecessary_regions): Likewise.
17441 (ira_build): Likewise.
17442 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17444 2013-02-08 Richard Biener <rguenther@suse.de>
17446 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17447 * ipa-pure-const.c (analyze_function): Avoid calling
17448 mark_irreducible_loops twice.
17449 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17451 2013-02-07 David S. Miller <davem@davemloft.net>
17453 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17455 * var-tracking.c (vt_add_function_parameter): Test the presence of
17456 HAVE_window_save properly and do not remap argument registers when
17457 we have a leaf function.
17459 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
17462 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17464 * config/i386/i386.c (ix86_print_operand): Ditto.
17466 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
17468 * lra-constraints.c (process_alt_operands): Fix recently added comment.
17470 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
17472 PR rtl-optimization/56225
17473 * lra-constraints.c (process_alt_operands): Check that reload hard
17474 reg can hold value for strict_low_part.
17476 2013-02-07 Jakub Jelinek <jakub@redhat.com>
17479 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17480 dwarf2out_end_function.
17481 (in_first_function_p, maybe_at_text_label_p,
17482 first_loclabel_num_not_at_text_label): New variables.
17483 (dwarf2out_var_location): In the first function find out
17484 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17485 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17488 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
17490 PR rtl-optimization/56178
17491 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17492 SUBREG of a register. Tidy up related block of code.
17493 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17494 note if the source is a register or a SUBREG of a register.
17496 2013-02-07 Jakub Jelinek <jakub@redhat.com>
17499 * config/rs6000/rs6000.md (ptrm): New mode attr.
17500 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17501 call_value_indirect_aix<pttrsize>,
17502 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17505 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
17507 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17508 if -bnortl. Convert to strcmp and strncmp.
17510 2013-02-07 Alan Modra <amodra@gmail.com>
17513 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17514 addresses won't wrap when offsetting.
17515 (rs6000_secondary_reload): Provide secondary reloads needed for
17516 wrapping LO_SUM addresses.
17518 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
17520 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17521 MACH, just __MACH__.
17523 2013-02-06 Richard Biener <rguenther@suse.de>
17525 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17526 instead of calling fix_loop_structure.
17528 2013-02-06 Jakub Jelinek <jakub@redhat.com>
17530 PR middle-end/56217
17531 * omp-low.c (use_pointer_for_field): Return false if
17532 lower_send_shared_vars doesn't generate any copy-out code.
17534 2013-02-06 Tom de Vries <tom@codesourcery.com>
17536 PR rtl-optimization/56131
17537 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17538 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
17539 of the label is NULL. Add comment.
17541 2013-02-05 Jakub Jelinek <jakub@redhat.com>
17543 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17546 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17547 (STATIC_LIBTSAN_LIBS): Likewise.
17548 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17549 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17550 is defined, don't add anything else beyond that.
17551 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17552 (LINK_COMMAND_SPEC): Use them.
17554 PR tree-optimization/56205
17555 * tree-stdarg.c (check_all_va_list_escapes): Return true if
17556 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17557 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17559 2013-02-05 Richard Biener <rguenther@suse.de>
17561 PR tree-optimization/53342
17562 PR tree-optimization/53185
17563 * tree-vectorizer.h (vect_check_strided_load): Remove.
17564 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17565 not disallow peeling for vectorized strided loads.
17566 (vect_check_strided_load): Make static and simplify.
17567 (vect_analyze_data_refs): Adjust.
17568 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17569 correctly when vectorizing strided loads.
17571 2013-02-05 Richard Biener <rguenther@suse.de>
17573 * doc/install.texi: Refer to ISL, not PPL.
17575 2013-02-05 Jan Hubicka <jh@suse.cz>
17577 PR tree-optimization/55789
17578 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17580 2013-02-05 Jan Hubicka <jh@suse.cz>
17582 PR tree-optimization/55789
17583 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17584 the dead call anyway.
17586 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
17589 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17591 2013-02-04 Alexander Potapenko <glider@google.com>
17592 Jack Howarth <howarth@bromo.med.uc.edu>
17593 Jakub Jelinek <jakub@redhat.com>
17596 * config/darwin.c (sort_ctor_records): Stabilized qsort
17597 on constructor priority by using original position.
17598 (finalize_ctors): New routine to sort constructors by
17599 priority before use in assemble_integer.
17600 (machopic_asm_out_constructor): Use finalize_ctors if needed.
17602 2013-02-04 Jakub Jelinek <jakub@redhat.com>
17605 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17606 about visibility on artificial decls.
17607 * config/sol2.c (solaris_assemble_visibility): Likewise.
17609 2013-02-04 Kai Tietz <ktietz@redhat.com>
17612 * config/i386/i386.c (function_value_ms_64): Add additional valtype
17613 argument and improve checking of return-argument types for 16-byte
17615 (ix86_function_value_1): Add additional valtype argument on call
17616 of function_value_64.
17617 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17618 handling infunction_value_64 function.
17620 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
17622 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17624 2013-02-04 Richard Biener <rguenther@suse.de>
17626 PR tree-optimization/56188
17627 * tree-ssa-structalias.c (label_visit): Consider case with
17628 initially non-empty points-to set.
17629 (perform_var_substitution): Dump node mapping and clean up.
17631 2013-02-04 Richard Guenther <rguenther@suse.de>
17634 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17635 node prevail as last resort.
17636 (lto_symtab_merge_decls): Remove guard on LTRANS here.
17637 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17639 2013-02-04 Richard Biener <rguenther@suse.de>
17641 PR tree-optimization/56113
17642 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17644 (equiv_class_lookup_or_add): ... this.
17645 (label_visit): Adjust and fix error in previous patch.
17646 (perform_var_substitution): Adjust.
17648 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
17650 * config/sh/divtab.c: Fix formatting and comments throughout the file.
17651 * config/sh/sh4-300.md: Likewise.
17652 * config/sh/sh4a.md: Likewise.
17653 * config/sh/constraints.md: Likewise.
17654 * config/sh/sh.md: Likewise.
17655 * config/sh/netbsd-elf.h: Likewise.
17656 * config/sh/predicates.md: Likewise.
17657 * config/sh/sh-protos.h: Likewise.
17658 * config/sh/ushmedia.h: Likewise.
17659 * config/sh/linux.h: Likewise.
17660 * config/sh/sh.c: Likewise.
17661 * config/sh/superh.h: Likewise.
17662 * config/sh/elf.h: Likewise.
17663 * config/sh/sh4.md: Likewise.
17664 * config/sh/sh.h: Likewise.
17666 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17668 * config/pa/constraints.md: Adjust unused letters. Change "T"
17669 constraint to match_test floating_point_store_memory_operand().
17670 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17671 (base14_operand): New.
17672 (floating_point_store_memory_operand): New.
17673 (integer_store_memory_operand): Revise to use base14_operand and
17674 reg_plus_base_memory_operand.
17675 (move_dest_operand): Allow symbolic_memory_operands.
17676 (symbolic_memory_operand): Check for LO_SOM.
17677 (symbolic_operand): Change default case to break.
17678 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17679 CONST_DOUBLE values to be reloaded by putting them into memory when
17680 the destination is a floating point register.
17681 (movdf): Remove code to handle CONST_DOUBLE.
17683 (reload_indf_r1): New.
17684 (reload_insf_r1): New.
17685 Consistently use "Q" and "T" constraints with integer and floating
17686 point move instructions, respectively.
17687 (movdi): Remove FAIL.
17688 Change predicate for source operand unamed DImode move from
17689 general_operand to move_src_operand.
17690 (umulsidi3): Change predicate for destination operand to
17692 Likewise for similar unamed patterns.
17693 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17694 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17695 (hppa_legitimize_address): Simplify mask calculation.
17696 (pa_emit_move_sequence): Revised handling of secondary reloads from
17697 REG+D addresses for floating point loads and stores. Directly handle
17698 loading CONST0_RTX (mode) to a floating point register.
17699 (pa_secondary_reload): Handle reloading DF and SFmode constant values
17700 to floating point registers. Don't restrict secondary reloads to
17701 floating point registers to integer modes. Revise some comments and
17703 (TARGET_LEGITIMATE_ADDRESS_P): Define.
17704 (pa_legitimate_address_p): New.
17705 (pa_legitimize_reload_address): New.
17706 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17707 (STRICT_REG_OK_FOR_BASE_P): New.
17708 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
17709 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17711 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
17712 Andrew Dixie <andrewd@gentrack.com>
17714 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17717 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
17719 * expmed.c (extract_bit_field_1): Pass the full width of the
17720 structure to get_best_reg_extraction_insn.
17722 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
17725 * configure.ac (use_cxa_atexit): Add AIX.
17726 * configure: Regenerate.
17728 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17730 2013-02-01 Jakub Jelinek <jakub@redhat.com>
17733 * final.c (need_profile_function): New variable.
17734 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17735 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17736 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17737 notes, targetm.asm_out.function_prologue doesn't emit anything,
17738 HAVE_prologue and profiler should be emitted before prologue,
17739 set need_profile_function instead of emitting it.
17740 (final_scan_insn): If need_profile_function, emit
17741 profile_function on the first NOTE_INSN_BASIC_BLOCK or
17742 NOTE_INSN_FUNCTION_BEG note.
17744 2013-02-01 Richard Henderson <rth@redhat.com>
17746 * config/rs6000/rs6000.md (smulditi3): New.
17749 * config/alpha/alpha.md (umulditi3): New.
17751 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
17753 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17754 (ASM_OUTPUT_ALIGNED_LOCAL): New.
17756 2013-02-01 Richard Biener <rguenther@suse.de>
17758 PR tree-optimization/56113
17759 * tree-ssa-structalias.c (label_visit): Reduce work for
17760 single-predecessor nodes.
17762 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
17764 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17765 range isn't testing for zero.
17767 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
17769 PR middle-end/56113
17770 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17772 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
17773 Nick Clifton <nickc@redhat.com>
17775 * config/v850/constraints.md (Q): Define as a memory constraint.
17776 * config/v850/predicates.md (label_ref_operand): New predicate.
17777 (e3v5_shift_operand): New predicate.
17778 (ior_operator): New predicate.
17779 * config/v850/t-v850: Add e3v5 multilib.
17780 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17781 (v850_gen_movdi): Prototype.
17782 * config/v850/v850.c: Add support for e3v5 architecture.
17783 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17785 (construct_save_jarl): Add e3v5 long JARL support.
17786 (v850_adjust_insn_length): New function. Adjust length of call
17787 insns when using e3v5 instructions.
17788 (v850_gen_movdi): New function: Generate instructions to move a
17790 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17791 (CPP_SPEC): Define __v850e3v5__ as appropriate.
17792 (TARGET_USE_FPU): Enable for e3v5.
17793 (CONST_OK_FOR_W): New macro.
17794 (ADJUST_INSN_LENGTH): Define.
17795 * config/v850/v850.md (UNSPEC_LOOP): Define.
17796 (attr cpu): Add v850e3v5.
17797 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17798 (movdi): New pattern.
17799 (movdi_internal): New pattern.
17800 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17801 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17802 (cstoresf4): Likewise.
17803 (cstoredf4): Likewise.
17804 (insv): New pattern.
17805 (rotlso3_a): New pattern.
17806 (rotlsi3_b): New pattern
17807 (rotlsi3_v850e3v5): New pattern.
17808 (doloop_begin): New pattern.
17809 (fix_loop_counter): New pattern.
17810 (doloop_end): New pattern.
17811 (branch_normal): Add e3v5 long branch support.
17812 (branch_invert): Likewise.
17813 (branch_z_normal): Likewise.
17814 (branch_z_invert): Likewise.
17815 (branch_nz_normal): Likewise.
17816 (branch_nz_invert): Likewise.
17817 (call_internal_short): Add e3v5 register-indirect JARL support.
17818 (call_internal_long): Likewise.
17819 (call_value_internal_short): Likewise.
17820 (call_value_internal_long): Likewise.
17821 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17822 (mloop): New option.
17823 * config.gcc: Add support for configuring v840e3v5 target.
17824 * doc/invoke.texi: Document new v850 specific command line options.
17826 2013-01-31 Paul Koning <ni1d@arrl.net>
17830 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17831 children if parent is a class.
17832 (prune_unused_types_prune): Don't add DW_AT_declaration.
17834 2013-01-31 Richard Biener <rguenther@suse.de>
17836 PR tree-optimization/56157
17837 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17838 match up operand with SLP child.
17840 2013-01-31 Jason Merrill <jason@redhat.com>
17843 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17844 parameters the first time.
17845 (gen_scheduled_generic_parms_dies): Check completeness here.
17847 2013-01-31 Richard Biener <rguenther@suse.de>
17849 PR middle-end/53073
17850 * common.opt (faggressive-loop-optimizations): New flag,
17851 enabled by default.
17852 * doc/invoke.texi (faggressive-loop-optimizations): Document.
17853 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17854 infer_loop_bounds_from_undefined by it.
17856 2013-01-31 Richard Biener <rguenther@suse.de>
17858 PR tree-optimization/56150
17859 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17860 visit virtual operands.
17861 (find_uses_to_rename_bb): Likewise.
17863 2013-01-31 Richard Biener <rguenther@suse.de>
17865 PR tree-optimization/56150
17866 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17867 mixed store non-store stmts.
17869 2013-01-30 Jakub Jelinek <jakub@redhat.com>
17872 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17873 LIBASAN_EARLY_SPEC is defined.
17874 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17875 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17877 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17880 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17881 invalid args instead of ICEing on it.
17882 (ix86_valid_target_attribute_tree): Return error_mark_node if
17883 ix86_valid_target_attribute_inner_p failed.
17884 (ix86_valid_target_attribute_p): Return false only if
17885 ix86_valid_target_attribute_tree returned error_mark_node. Allow
17886 target("default") attribute.
17887 (sorted_attr_string): Change argument from const char * to tree,
17888 merge in all target attribute arguments rather than just one.
17889 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
17890 instead of free. Avoid using strcat.
17891 (ix86_mangle_function_version_assembler_name): Mangle
17892 target("default") as if no target attribute is present. Adjust
17893 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
17894 instead of xmalloc and XDELETEVEC instead of free.
17895 (ix86_function_versions): Don't return true if one of the decls
17896 doesn't have target attribute. If they don't and one of the decls
17897 is DECL_FUNCTION_VERSIONED, report an error. Adjust
17898 sorted_attr_string caller. Use XDELETEVEC instead of free.
17899 (ix86_supports_function_versions): Remove.
17900 (make_name): Fix up formatting.
17901 (make_dispatcher_decl): Remove resolver_name and its initialization.
17902 Avoid leaking memory.
17903 (is_function_default_version): Return true if there is
17904 target("default") attribute rather than no target attribute at all.
17905 (make_resolver_func): Avoid leaking memory.
17906 (ix86_generate_version_dispatcher_body): Likewise.
17907 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17908 * target.def (supports_function_versions): Remove.
17909 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17910 * doc/tm.texi: Regenerated.
17912 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
17914 PR rtl-optimization/56144
17915 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17916 for values with side effects.
17918 2013-01-30 Richard Biener <rguenther@suse.de>
17920 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17921 (sparseset_pop): Likewise.
17922 * cfganal.c (compute_idf): Likewise. Increase work-stack size
17923 to be able to use quick_push in the worker loop.
17925 2013-01-30 Marek Polacek <polacek@redhat.com>
17927 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17929 2013-01-30 Richard Biener <rguenther@suse.de>
17932 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17934 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
17936 PR tree-optimization/56064
17937 * fixed-value.c (fixed_from_double_int): New function.
17938 * fixed-value.h (fixed_from_double_int): New prototype.
17939 (const_fixed_from_double_int): New static inline function.
17940 * fold-const.c (native_interpret_fixed): New static function.
17941 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17942 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17943 (native_encode_fixed): New static function.
17944 (native_encode_expr) <FIXED_CST>: Use it.
17945 (native_interpret_int): Move double_int worker code to...
17946 * double-int.c (double_int::from_buffer): ...this new static method.
17947 * double-int.h (double_int::from_buffer): Prototype it.
17949 2013-01-30 Richard Biener <rguenther@suse.de>
17951 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17952 New pointer-map and obstack.
17953 (init_alias_vars): Allocate pointer-map and obstack.
17954 (delete_points_to_sets): Free them.
17955 (find_what_var_points_to): Cache result.
17956 (find_what_p_points_to): Adjust for changed interface of
17957 find_what_var_points_to.
17958 (compute_points_to_sets): Likewise.
17959 (ipa_pta_execute): Likewise.
17961 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17963 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17964 * configure: Regenerate.
17965 * config.in: Regenerate.
17966 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17967 #nobits/#progbits if supported.
17969 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
17972 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17973 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17974 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17976 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17978 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17979 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17981 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17983 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17984 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17986 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17988 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17990 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17991 * config/arm/cortex-a7.md: New bypasses using
17992 arm_mac_accumulator_is_result.
17994 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
17996 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
17997 (cortex_a7_neon_mla): Likewise.
17998 (cortex_a7_fpfmad): New reservation.
17999 (cortex_a7_fpmacs): Use ffmas and update required units.
18000 (cortex_a7_fpmuld): Update required units and latency.
18001 (cortex_a7_fpmacd): Likewise.
18002 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
18003 (cortex_a7_neon). Likewise.
18004 (bypass) Update participating units.
18006 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18008 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
18009 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
18011 * config/arm/vfp11.md (vfp_farith): Use ffmas.
18012 (vfp_fmul): Use ffmad.
18013 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
18014 (cortex_r4_fmacd): Use ffmad.
18015 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
18016 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
18017 (cortex_a9_fmacd): Use ffmad.
18018 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
18019 (cortex_a8_vfp_macd): Use ffmad.
18020 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
18021 (cortex_a5_fpmacd): Use ffmad.
18022 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
18023 (cortex_a15_vfp_macd): Use ffmad.
18024 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
18026 2013-01-29 Jason Merrill <jason@redhat.com>
18029 * varasm.c (default_assemble_visibility): Don't warn about
18030 visibility on artificial decls.
18032 2013-01-29 Richard Biener <rguenther@suse.de>
18034 PR tree-optimization/56113
18035 * tree-ssa-structalias.c (equiv_class_lookup): Also return
18037 (label_visit): Free duplicate bitmaps and record the leader instead.
18038 (perform_var_substitution): Adjust.
18040 2013-01-29 Richard Biener <rguenther@suse.de>
18042 PR tree-optimization/55270
18043 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
18044 the CFG, schedule loops for fixup.
18046 2013-01-29 Nick Clifton <nickc@redhat.com>
18048 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
18051 2013-01-28 Leif Ekblad <leif@rdos.net>
18053 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
18054 * config/i386/i386.h (TARGET_RDOS): New macro.
18055 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
18056 * config/i386/i386.c (ix86_option_override_internal): For 64bit
18057 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
18058 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
18059 DEFAULT_LARGE_SECTION_THRESHOLD.
18060 * config/i386/i386.md (R14_REG, R15_REG): New constants.
18061 * config/i386/rdos.h: New file.
18062 * config/i386/rdos64.h: New file.
18064 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
18067 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
18070 2013-01-28 Jakub Jelinek <jakub@redhat.com>
18072 PR rtl-optimization/56117
18073 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
18074 call cselib_lookup_from_insn on the MEM before calling
18075 add_insn_mem_dependence.
18077 2013-01-28 Richard Biener <rguenther@suse.de>
18079 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
18080 to a stmt that didn't have one.
18081 (copy_phis_for_bb): Likewise for PHI arguments.
18082 (copy_debug_stmt): Likewise for debug stmts.
18084 2013-01-28 Richard Biener <rguenther@suse.de>
18086 PR tree-optimization/56034
18087 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
18088 (partition_builtin_p): Adjust.
18089 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
18090 it is the last partition.
18091 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
18092 up the vertex for the definition.
18093 (classify_partition): Classify whether a partition is a
18094 PKIND_REDUCTION, thus has uses outside of the loop.
18095 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
18096 Merge all PKIND_REDUCTION partitions into the last partition.
18097 (tree_loop_distribution): Seed partitions from reductions as well.
18099 2013-01-28 Jakub Jelinek <jakub@redhat.com>
18101 PR tree-optimization/56125
18102 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
18103 pow(x,c) into sqrt(x) * powi(x, n/2) or
18104 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
18105 optimizing for size.
18106 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
18107 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
18110 PR tree-optimization/56094
18111 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
18112 to UNKNOWN_LOCATION while gimplifying expr.
18114 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
18117 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
18118 operand 0 in movabs insn template for -masm=intel asm alternative.
18119 (*movabs<mode>_2): Ditto for operand 1.
18121 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
18124 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
18127 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
18129 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
18132 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
18134 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
18135 Undef to avoid warning.
18137 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
18139 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
18140 * configure: Regenerate.
18142 2013-01-25 Jakub Jelinek <jakub@redhat.com>
18144 PR tree-optimization/56098
18145 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
18146 for stmts with volatile ops.
18147 (cond_store_replacement): Don't optimize if assign has volatile ops.
18148 (cond_if_else_store_replacement_1): Don't optimize if either
18149 then_assign or else_assign have volatile ops.
18150 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
18153 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18155 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
18157 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18159 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
18160 missing ':' in asm example.
18162 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
18164 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
18165 entries into lane and laneq entries.
18166 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
18167 Remove AdvSIMD scalar modes.
18168 (aarch64_sq<r>dmulh_laneq<mode>): New.
18169 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
18171 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
18172 builtin implementations to relfect changes in RTL in aarch64-simd.md.
18173 * config/aarch64/iterators.md (VCOND): New.
18176 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18179 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
18180 Add NULL LIBNAME argument to existing definitions.
18181 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
18182 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
18183 * config/avr/avr.c (DEF_BUILTIN): Same.
18184 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
18185 (avr_expand_builtin): Expand to a vanilla call if a libgcc
18186 implementation is available (DECL_ASSEMBLER_NAME is set).
18187 (avr_fold_absfx): New static function.
18188 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
18189 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
18190 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
18191 AVR_BUILTIN_ABSLLK.
18192 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
18193 (abshk, absk, abslk, absllk): Provide as static inline functions.
18195 2013-01-25 Marek Polacek <polacek@redhat.com>
18197 PR tree-optimization/56035
18198 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
18200 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
18202 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
18203 (*movtf_internal_rex64): Add (!o,C) alternative
18204 (*movxf_internal_rex64): Ditto.
18205 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
18207 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
18209 * doc/invoke.texi: fix typo.
18210 * doc/objc.texi: fix typo.
18212 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
18214 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
18215 for the first two alternatives.
18217 2013-01-24 Diego Novillo <dnovillo@google.com>
18219 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
18220 (ggc-zone.o): Remove.
18221 * configure.ac: Remove option --with-gc.
18222 * configure: Re-generate.
18223 * doc/install.texi: Remove documentation for --with-gc.
18224 * gengtype.c (write_enum_defn): Remove. Update all users.
18225 (write_Types_process_field): Remove generation of gt_e_* argument.
18226 (output_type_enum): Remove. Update all users.
18227 (write_enum_defn): Remove. Update all users.
18228 (enum alloc_zone): Remove. Update all users.
18229 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
18230 * ggc-common.c (ggc_splay_alloc): Remove first argument.
18231 Update all callers.
18232 (struct ptr_data): Remove field TYPE. Update all users.
18233 (gt_pch_note_object): Remove argument TYPE. Update all users.
18234 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
18236 * ggc-none.c (ggc_alloc_typed_stat): Remove.
18237 (struct alloc_zone): Remove.
18238 (ggc_internal_alloc_zone_stat): Remove.
18239 (ggc_internal_cleared_alloc_zone_stat): Remove.
18240 * ggc-page.c (ggc_alloc_typed_stat): Remove.
18241 (ggc_pch_count_object): Remove last argument. Update all users.
18242 (ggc_pch_alloc_object): Remove last argument. Update all users.
18243 (struct alloc_zone): Remove.
18244 * ggc-zone.c: Remove.
18245 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
18246 (struct alloc_zone): Remove.
18247 (ggc_alloc_typed_stat): Remove.
18248 (ggc_alloc_typed): Remove.
18249 (ggc_splay_alloc): Remove first argument.
18250 (rtl_zone): Remove. Update all users.
18251 (tree_zone): Remove. Update all users.
18252 (tree_id_zone): Remove. Update all users.
18253 (ggc_internal_zone_alloc_stat): Remove. Update all users.
18254 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
18255 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
18256 * tree-ssanames.c: Remove references to zone allocator in comments.
18258 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
18260 * config/avr/avr.c (avr_out_fract): Make register numbers that
18261 might be outside of source operand signed.
18263 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
18265 * config/i386/constraints.md (Yf): New constraint.
18266 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
18267 of f constraint to conditionaly disable x87 register preferences.
18268 (*movdf_internal): Ditto.
18269 (*movsf_internal): Ditto.
18271 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
18273 PR inline-asm/55934
18274 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
18275 that have operands with impossible constraints.
18276 Add a FIXME for a speed-up opportunity.
18277 * lra-constraints.c (process_alt_operands): Verify that a class
18278 selected from constraints on asms is valid for the operand mode.
18279 (curr_insn_transform): Remove incorrect comment.
18281 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
18283 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
18284 TOC operand is a valid symbol ref in the constant pool.
18286 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
18288 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
18290 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
18293 * config/avr/stdfix.h: New file.
18294 * t-avr (stdfix-gcc.h): New rule to build it.
18295 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
18297 2013-01-23 Kostya Serebryany <kcc@google.com>
18299 * config/darwin.h: remove dependency on
18300 CoreFoundation (asan on Mac OS).
18302 2013-01-23 Jakub Jelinek <jakub@redhat.com>
18305 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
18306 instead of cmpdi_operand for first comparison operand.
18307 Don't assert that comparison operands aren't both constants.
18309 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
18311 * doc/install.texi (Downloading the Source): Update references to
18312 downloading separate components.
18314 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
18316 * doc/extend.texi (__int128): Improve grammar.
18318 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
18321 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
18322 alternative to (o,r).
18323 (*movdi_internal_rex64): Remove (!o,n) alternative.
18324 (DImode immediate->memory splitter): Remove.
18325 (DImode immediate->memory peephole2): Remove.
18326 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
18327 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
18328 alternative to (!o,*r).
18329 (*movtf_internal_sse): New pattern.
18330 (*movxf_internal_rex64): New pattern.
18331 (*movxf_internal): Disable for TARGET_64BIT.
18332 (*movdf_internal_rex64): Remove (!o,F) alternative.
18334 2013-01-22 Jakub Jelinek <jakub@redhat.com>
18336 PR middle-end/56074
18337 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
18338 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
18339 * tree-vect-loop-manip.c (find_loop_location): Also ignore
18340 stmt locations where LOCATION_LOCUS of the stmt location is
18341 UNKNOWN_LOCATION or BUILTINS_LOCATION.
18344 * config/i386/i386.md (UNSPEC_STOS): New.
18345 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
18346 *strsetqi_1): Add UNSPEC_STOS.
18348 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
18351 * doc/invoke.texi: Remove left over -Wsynth example.
18353 2013-01-21 Jakub Jelinek <jakub@redhat.com>
18355 PR tree-optimization/56051
18356 * fold-const.c (fold_binary_loc): Don't fold
18357 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
18358 a narrowing conversion, or widening conversion from signed
18361 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
18363 PR rtl-optimization/56023
18364 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
18365 dependent on debug instruction.
18367 2013-01-21 Martin Jambor <mjambor@suse.cz>
18369 PR middle-end/56022
18370 * function.c (allocate_struct_function): Call
18371 invoke_set_current_function_hook earlier.
18373 2013-01-21 Jakub Jelinek <jakub@redhat.com>
18375 * reload1.c (init_reload): Only initialize reload_obstack
18376 during the first call.
18378 2013-01-21 Marek Polacek <polacek@redhat.com>
18380 * cfgloop.c (verify_loop_structure): Fix up grammar.
18382 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
18384 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
18385 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
18387 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18390 * config/arm/marvell-pj4.md: Update copyright year.
18391 Fix up use of alu to alu_reg and simple_alu_imm.
18393 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
18395 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
18397 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
18400 * lra-constraints.c (curr_insn_transform): Don't reuse original
18401 insn for secondary memory move when memory mode should be different.
18403 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18405 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18406 atomic_storedi_1): New patterns.
18408 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
18410 btver2 pipeline descriptions.
18411 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18413 * config/i386/i386.md (btver2_decode): New type attributes.
18414 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18416 * config/i386/btver2.md: New file describing btver2 pipelines.
18418 2013-01-19 Andrew Pinski <apinski@cavium.com>
18420 PR tree-optimization/52631
18421 * tree-ssa-sccvn (visit_use): Before looking up the original
18422 statement, try looking up the simplified expression.
18424 2013-01-19 Anthony Green <green@moxielogic.com>
18426 * config/moxie/moxie.c (moxie_expand_prologue): Set
18427 current_function_static_stack_size.
18429 2013-01-18 Jakub Jelinek <jakub@redhat.com>
18431 PR tree-optimization/56029
18432 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18433 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18435 2013-01-18 Sharad Singhai <singhai@google.com>
18437 PR tree-optimization/55995
18438 * dumpfile.c (dump_loc): Print location only if available.
18439 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18441 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
18444 * lra-constraints.c (curr_insn_transform): Reuse original insn for
18445 secondary memory move.
18446 (inherit_reload_reg): Use rclass instead of cl for
18447 check_secondary_memory_needed_p.
18449 2013-01-18 Jakub Jelinek <jakub@redhat.com>
18451 PR middle-end/56015
18452 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18453 the case where writing real complex part of target modifies op1.
18455 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
18457 * config/aarch64/aarch64-simd.md
18458 (aarch64_vcond_internal<mode>): Handle unordered cases.
18459 * config/aarch64/iterators.md (v_cmp_result): New.
18461 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
18462 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18464 * config/arm/marvell-pj4.md: New file.
18465 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18466 * config/arm/arm.md (generic_sched): Add marvell_pj4.
18467 (generic_vfp): Likewise.
18468 * config/arm/arm-cores.def: Add marvell-pj4.
18469 * config/arm/arm-tune.md: Regenerate.
18470 * config/arm/arm-tables.opt: Regenerate.
18471 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18472 * doc/invoke.texi: Document marvell-pj4.
18474 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
18476 * config/aarch64/arm_neon.h: Map scalar types to standard types.
18478 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18483 * var-tracking.c (negative_power_of_two_p): New.
18484 (global_get_addr_cache, local_get_addr_cache): New.
18485 (get_addr_from_global_cache, get_addr_from_local_cache): New.
18486 (vt_canonicalize_addr): Rewrite using the above. Adjust the
18488 (vt_stack_offset_p): Remove.
18489 (vt_canon_true_dep): Always canonicalize loc's address.
18490 (clobber_overlapping_mems): Make sure we have a MEM.
18491 (local_get_addr_clear_given_value): New.
18492 (val_reset): Clear local cached entries.
18493 (compute_bb_dataflow): Create and release the local cache.
18494 Disable duplicate MEMs clobbering.
18495 (emit_notes_in_bb): Clobber MEMs likewise.
18496 (vt_emit_notes): Create and release the local cache.
18497 (vt_initialize, vt_finalize): Create and release the global
18498 cache, respectively.
18499 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18501 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18503 PR libmudflap/53359
18504 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18505 not found in the symtab.
18507 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18510 PR rtl-optimization/55547
18511 PR rtl-optimization/53827
18514 * alias.c (offset_overlap_p): New, factored out of...
18515 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
18516 the conservative special case for symbolic constants. Don't
18517 adjust zero sizes on alignment.
18519 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
18521 PR rtl-optimization/52573
18522 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18523 REG_UNUSED for the same register.
18525 2013-01-17 Richard Biener <rguenther@suse.de>
18526 Marek Polacek <polacek@redhat.com>
18528 PR rtl-optimization/55833
18529 * loop-unswitch.c (unswitch_loops): Move loop verification...
18530 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
18531 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18532 Set it to true when we're removing a loop from hierarchy tree in
18533 an irreducible region.
18534 (fix_bb_placements): Adjust caller.
18535 (fix_loop_placements): Likewise.
18537 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
18539 * config/avr/builtins.def (DEF_BUILTIN): Factor out
18540 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18541 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18542 Remove ID. Adjust comments.
18543 * config/avr/avr-c.c (avr_builtin_name): Remove.
18544 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18545 * config/avr/avr.c (avr_tolower): New static function.
18546 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
18547 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18548 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18551 2013-01-17 Jan Hubicka <jh@suse.cz>
18553 PR tree-optimization/55273
18554 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18556 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
18559 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18560 store through atomic_store<mode>_1.
18561 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18563 2013-01-17 Martin Jambor <mjambor@suse.cz>
18565 PR tree-optimizations/55264
18566 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18567 for virtual methods.
18568 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18569 virtual methods before inlining is over.
18570 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18572 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18575 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
18577 PR rtl-optimization/56005
18578 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18579 pending reads for prefetch.
18581 2013-01-16 Ian Bolton <ian.bolton@arm.com>
18583 * config/aarch64/aarch64.md
18584 (*cstoresi_neg_uxtw): New pattern.
18585 (*cmovsi_insn_uxtw): New pattern.
18586 (*<optab>si3_uxtw): New pattern.
18587 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18588 (*<optab>si3_insn_uxtw): New pattern.
18589 (*bswapsi2_uxtw): New pattern.
18591 2013-01-16 Richard Biener <rguenther@suse.de>
18593 * tree-inline.c (tree_function_versioning): Remove set but
18594 never used variable.
18596 2013-01-16 Richard Biener <rguenther@suse.de>
18598 PR tree-optimization/55964
18599 * tree-flow.h (rename_variables_in_loop): Remove.
18600 (rename_variables_in_bb): Likewise.
18601 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18602 (copy_loop_before): Adjust and delete update-ssa status.
18603 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18604 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
18605 (rename_variables_in_loop): Remove.
18606 (slpeel_update_phis_for_duplicate_loop): Likewise.
18607 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18608 use available cfg machinery instead of duplicating it.
18609 Update PHI nodes and perform poor-mans SSA update here.
18610 (slpeel_tree_peel_loop_to_edge): Adjust.
18612 2013-01-16 Richard Biener <rguenther@suse.de>
18614 PR tree-optimization/54767
18615 PR tree-optimization/53465
18616 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18617 (vrp_visit_phi_node): For PHI arguments coming via backedges
18618 drop all symbolical range information.
18619 (execute_vrp): Compute backedges.
18621 2013-01-16 Richard Biener <rguenther@suse.de>
18623 * doc/install.texi: Update CLooG and ISL requirements to
18626 2013-01-16 Christian Bruel <christian.bruel@st.com>
18629 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18630 (broken_move): Handle UNSPECV_SP_SWITCH_B.
18631 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18633 2013-01-16 DJ Delorie <dj@redhat.com>
18635 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18636 (UNSPECV_SP_SWITCH_E): New.
18637 (sp_switch_1): Change to an unspec.
18638 (sp_switch_2): Change to an unspec. Don't use post-inc when we
18641 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
18643 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18644 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18645 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18646 (expand_mem_thread_fence): Ditto.
18647 (expand_mem_signal_fence): Ditto.
18648 (expand_atomic_load): Ditto.
18649 (expand_atomic_store): Ditto.
18651 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
18653 PR rtl-optimization/55547
18654 PR rtl-optimization/53827
18657 * alias.c (memrefs_conflict_p): Set sizes to negative after
18660 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18663 * function.c (thread_prologue_and_epilogue_insns): Always
18664 add crtl->drap_reg to set_up_by_prologue.set, even if
18665 stack_realign_drap is false.
18667 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
18669 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18670 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18671 *call): Fix indention.
18673 2013-01-15 Tom de Vries <tom@codesourcery.com>
18676 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18679 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
18681 PR rtl-optimization/55153
18682 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18684 2013-01-15 Martin Jambor <mjambor@suse.cz>
18686 PR tree-optimization/55920
18687 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18688 accesses as grp_to_be_debug_replaced.
18690 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18692 PR tree-optimization/55920
18693 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18694 there is non-useless type conversion needed from debug rhs to lhs,
18695 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18697 2013-01-15 Joseph Myers <joseph@codesourcery.com>
18698 Mikael Pettersson <mikpe@it.uu.se>
18701 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18703 (ASM_OUTPUT_CASE_LABEL): Remove.
18704 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18705 * final.c (shorten_branches): Update alignment of labels before
18706 jump tables if CASE_VECTOR_SHORTEN_MODE.
18708 2013-01-15 Richard Biener <rguenther@suse.de>
18711 * system.h: Do not include gmp.h for building host tools.
18713 2013-01-15 Richard Biener <rguenther@suse.de>
18715 PR middle-end/55882
18716 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18717 account for bitpos when computing alignment.
18719 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
18721 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18722 (ix86_target_macros_internal): Likewise.
18724 * config/i386/i386.c (m_CORE2I7): Removed.
18725 (m_CORE_HASWELL): New macro.
18726 (m_CORE_ALL): Likewise.
18727 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18728 (initial_ix86_arch_features): Likewise.
18729 (processor_target_table): Initializations for Core avx2.
18730 (cpu_names): New names "core-avx2".
18731 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18732 PROCESSOR_CORE_HASWELL.
18733 (ix86_issue_rate): New case.
18734 (ia32_multipass_dfa_lookahead): Likewise.
18735 (ix86_sched_init_global): Likewise.
18737 * config/i386/i386.h (TARGET_HASWELL): New macro.
18738 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18739 (processor_type): New PROCESSOR_HASWELL.
18741 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18743 PR tree-optimization/55955
18744 * tree-vect-loop.c (vectorizable_reduction): Give up early on
18745 *SHIFT_EXPR and *ROTATE_EXPR codes.
18747 PR tree-optimization/48766
18748 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18749 -ftrapv disable -fwrapv.
18751 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18754 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18755 etc. to 1 and not to __flash.
18756 Use LL suffix for __INT24_MAX__ with -mint8.
18757 Use ULL suffix for __UINT24_MAX__ with -mint8.
18759 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18761 * config/avr/avr-arch.h
18762 (struct base_arch_s): Use typedef avr_arch_t instead.
18763 (struct arch_info_s): Use typedef avr_arch_info_t instead.
18764 (struct mcu_type_s): Use typedef avr_mcu_t instead.
18765 * config/avr/avr.c: Same.
18766 * config/avr/avr-devices.c: Same.
18767 * config/avr/driver-avr.c: Same.
18768 * config/avr/gen-avr-mmcu-texi.c: Same.
18769 * config/avr/avr-mcus.def: Adjust comment.
18771 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
18773 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18774 * config/aarch64/iterators.md (VALLDI): New.
18776 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
18777 Andi Kleen <ak@linux.intel.com>
18780 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18781 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18784 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18786 * config/avr/avr-stdint.h: Remove trailing blanks.
18787 * config/avr/avr-log.h: Same.
18788 * config/avr/avr-arch.h: Same.
18789 * config/avr/avr-devices.c: Same.
18790 * config/avr/avr-dimode.md: Same.
18791 * config/avr/predicates.md: Same.
18792 * config/avr/avr-c.c: Same. And fix typo.
18794 * config/avr/avr-protos.h: Same. And:
18795 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18796 (init_cumulative_args): Rename to avr_init_cumulative_args.
18797 (expand_prologue): Rename to avr_expand_prologue.
18798 (expand_epilogue): Rename to avr_expand_epilogue.
18799 (adjust_insn_length): Rename to avr_adjust_insn_length.
18800 (notice_update_cc): Rename to avr_notice_update_cc.
18801 (final_prescan_insn): Rename to avr_final_prescan_insn.
18802 * config/avr/avr.c: Same.
18803 * config/avr/avr.h: Same.
18804 * config/avr/avr.md: Remove trailing blanks.
18805 (prologue): Use avr_expand_prologue.
18806 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18808 2013-01-14 Richard Biener <rguenther@suse.de>
18810 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18811 verify_location, collect_subblocks): New functions.
18812 (verify_gimple_in_cfg): Verify that locations only reference
18813 BLOCKs in the functions BLOCK tree.
18815 2013-01-14 Richard Biener <rguenther@suse.de>
18817 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18819 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18821 (insert_out_of_ssa_copy_on_edge): Likewise.
18822 (rewrite_close_phi_out_of_ssa): Likewise.
18823 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18825 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18826 propagated constants.
18827 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18830 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18832 * config/avr/avr-modes.def: Add GPL copyright notice.
18834 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
18836 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18837 MEMMODEL_MASK to determine memory model.
18838 (atomic_store<mode>): Ditto from operands[2].
18839 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18841 2013-01-13 Jakub Jelinek <jakub@redhat.com>
18844 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18845 (fold_gimple_assign): Don't call unshare_expr here.
18846 (fold_ctor_reference): Call unshare_expr.
18848 2013-01-13 Terry Guo <terry.guo@arm.com>
18850 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18851 * doc/fragments.texi: Document MULTILIB_REUSE.
18852 * gcc.c (multilib_reuse): New internal spec.
18853 (set_multilib_dir): Also search multilib from multilib_reuse.
18854 * genmultilib (tmpmultilib3): Refactor code.
18855 (tmpmultilib4): Ditto.
18856 (multilib_reuse): New multilib argument.
18858 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
18860 * Makefile.in: Update copyright.
18862 2013-01-12 Tom de Vries <tom@codesourcery.com>
18864 PR middle-end/55890
18865 * calls.c (expand_call): Check if arg_nr is valid.
18867 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18869 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18870 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18871 documentation. Add missing '__' in front of
18872 __builtin_ia32_packssdw256.
18874 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
18877 * config/s390/s390.c (s390_preferred_reload_class): Do not return
18878 NO_REGS for larl operands.
18879 (s390_reload_larl_operand): Use s390_load_address instead of
18882 2013-01-11 Richard Biener <rguenther@suse.de>
18884 * tree-cfg.c (verify_node_sharing_1): Split out from ...
18885 (verify_node_sharing): ... here.
18886 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18888 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
18890 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18891 Substitute TREECHECKING.
18892 * configure: Regenerate.
18893 * Makefile.in (TREECHECKING): New.
18895 2013-01-11 Richard Guenther <rguenther@suse.de>
18897 PR tree-optimization/44061
18898 * tree-vrp.c (extract_range_basic): Compute zero as
18899 value-range for __builtin_constant_p of function parameters.
18901 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
18903 Update copyright years.
18905 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
18907 PR rtl-optimization/55672
18908 * lra-eliminations.c (mark_not_eliminable): Permit addition with
18909 const to be eliminable.
18911 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
18913 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18914 * configure: Regenerate.
18916 2013-01-10 Richard Biener <rguenther@suse.de>
18918 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18920 2013-01-10 Richard Biener <rguenther@suse.de>
18923 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18924 locations for virtual PHI arguments.
18925 (rewrite_update_phi_arguments): Likewise.
18927 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
18929 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18932 2013-01-10 Jakub Jelinek <jakub@redhat.com>
18934 PR tree-optimization/55921
18935 * tree-complex.c (expand_complex_asm): New function.
18936 (expand_complex_operations_1): Call it for GIMPLE_ASM.
18938 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
18941 * config/s390/s390.c (s390_symref_operand_p)
18942 (s390_loadrelative_operand_p): Merge the two functions.
18943 (s390_check_qrst_address, print_operand_address): Add parameters
18944 to s390_loadrelative_operand_p invokation.
18945 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18946 (s390_reload_larl_operand, s390_secondary_reload): Use
18947 s390_loadrelative_operand_p instead of s390_symref_operand_p.
18948 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18950 2013-01-09 Mike Stump <mikestump@comcast.net>
18952 * dse.c (record_store): Remove unnecessary assert.
18954 2013-01-09 Jan Hubicka <jh@suse.cz>
18956 PR tree-optimization/55569
18957 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18958 * cfgloop.h (scale_loop_profile): Likewise.
18960 2013-01-09 Jan Hubicka <jh@suse.cz>
18963 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18965 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18967 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
18969 PR middle-end/55114
18970 * expr.h (maybe_emit_group_store): Declare.
18971 * expr.c (maybe_emit_group_store): New function.
18972 * builtins.c (expand_builtin_int_roundingfn): Call it.
18973 (expand_builtin_int_roundingfn_2): Likewise.
18975 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
18977 PR rtl-optimization/55829
18978 * lra-constraints.c (match_reload): Add code for absent output.
18979 (curr_insn_transform): Add code for reloads of matched inputs
18982 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
18984 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18985 attribute of movddup insn to DF.
18986 (*vec_interleave_lowv2df): Ditto.
18987 (vec_dupv2df): Ditto.
18989 2013-01-09 Jan Hubicka <jh@suse.cz>
18991 PR tree-optimiation/55875
18992 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18993 EVERY_ITERATION parameter.
18994 (number_of_iterations_exit): Check if exit is executed every iteration.
18995 (idx_infer_loop_bounds): Similarly here.
18996 (n_of_executions_at_most): Simplify
18997 to only test for cases where statement is dominated by the
18998 particular bound; handle correctly the "postdominance" test.
18999 (scev_probably_wraps_p): Use max loop iterations info
19000 as a global bound first.
19002 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
19003 Nick Clifton <nickc@redhat.com>
19005 * config/v850/v850.md (cbranchsf4): New pattern.
19006 (cstoresf4): New pattern.
19007 (cbranchdf4): New pattern.
19008 (cstoredf4): New pattern.
19009 (movsicc): Disallow floating point comparisons.
19010 (cmpsf_le_insn): Fix order of operators.
19011 (cmpsf_lt_insn): Likewise.
19012 (cmpsf_eq_insn): Likewise.
19013 (cmpdf_le_insn): Likewise.
19014 (cmpdf_lt_insn): Likewise.
19015 (cmpdf_eq_insn): Likewise.
19016 (cmpsf_ge_insn): Use LE comparison.
19017 (cmpdf_ge_insn): Likewise.
19018 (cmpsf_gt_insn): Use LT comparison.
19019 (cmpdf_gt_insn): Likewise.
19020 (cmpsf_ne_insn): Delete pattern.
19021 (cmpdf_ne_insn): Delete pattern.
19022 * config/v850/v850.c (v850_gen_float_compare): Use
19023 gen_cmpdf_eq_insn for NE comparison.
19024 (v850_float_z_comparison_operator)
19025 (v850_float_nz_comparison_operator): Move from here ...
19026 * config/v850/predicates.md: ... to here. Move GT and GE
19027 comparisons into v850_float_z_comparison_operator.
19028 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
19030 (v850_float_nz_comparison_operator): Likewise.
19032 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19034 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
19035 with calls to gen_insvsi/gen_insvdi.
19037 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
19039 * config/i386/i386.c (initial_ix86_tune_features): Set up
19040 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
19042 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
19043 Jakub Jelinek <jakub@redhat.com>
19045 PR tree-optimization/48189
19046 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
19047 If nitercst is 0, don't predict the exit edge.
19049 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
19051 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
19052 in asm_fprintf with reg_names.
19053 (aarch64_print_operand_address): Likewise.
19054 (aarch64_return_addr): Likewise.
19055 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
19057 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19059 * config/pa/pa.h (VAL_U6_BITS_P): Define.
19060 (INT_U6_BITS): Likewise.
19061 * config/pa/predicates.md (uint6_operand): New predicate.
19062 (shift5_operand, shift6_operand): Likewise.
19063 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
19065 (lshrdi3): Use shift6_operand.
19066 (shrpsi4, shrpdi4): New insn patterns.
19067 (extzv): Delete expander.
19068 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
19069 predicates in unamed zero extract patterns. Tighten common constraint.
19070 (extv): Delete expander.
19071 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
19072 predicates in unamed sign extract patterns. Tighten common constraint.
19073 (insv): Delete expander.
19074 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
19075 predicates in unamed insert patterns. Tighten common constraint.
19076 Change uint32_operand predicate to uint6_operand predicate in unamed
19077 DImode pattern to insert constant values of type 1...1xxxx.
19079 2013-01-04 Jan Hubicka <jh@suse.cz>
19081 PR tree-optimization/55823
19082 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
19085 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19086 Uros Bizjak <ubizjak@gmail.com>
19088 PR rtl-optimization/55845
19089 * df-problems.c (can_move_insns_across): Stop scanning at
19090 volatile_insn_p source instruction or give up if
19091 across_from .. across_to range contains any volatile_insn_p
19094 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
19096 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
19097 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
19099 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
19100 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
19102 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19105 * asan.c (asan_clear_shadow): New function.
19106 (asan_emit_stack_protection): Use it.
19108 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
19110 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
19111 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
19112 with tab instead of space.
19114 2013-01-08 Nick Clifton <nickc@redhat.com>
19116 * config/rl78/rl78.c (rl78_expand_prologue): Always select
19117 register bank 0 at the start of an interrupt handler.
19118 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
19121 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
19123 * config/aarch64/aarch64-simd.md
19124 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
19125 (aarch64_simd_bsl): Likewise.
19126 (aarch64_vcond_internal<mode>): Likewise.
19127 (vcond<mode><mode>): Likewise.
19128 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
19129 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
19131 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
19133 * config/aarch64/aarch64-builtins.c
19134 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
19136 2013-01-08 Martin Jambor <mjambor@suse.cz>
19139 * tree-sra.c (analyze_access_subtree): Return true also after
19140 potentially creating a debug-only replacement.
19142 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19144 PR middle-end/55890
19145 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
19147 PR tree-optimization/54120
19148 * tree-vrp.c (range_fits_type_p): Don't allow
19149 src_precision < precision from signed vr to unsigned_p
19150 if vr->min or vr->max is negative.
19151 (simplify_float_conversion_using_ranges): Test can_float_p
19152 against CODE_FOR_nothing.
19154 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19155 Richard Biener <rguenther@suse.de>
19157 PR middle-end/55851
19158 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
19159 types instead of just INTEGER_TYPE types.
19161 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
19163 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
19166 2013-01-07 Steve Ellcey <sellcey@mips.com>
19169 * config/mips/mips.opt: Change mad to mmad to match documentation.
19171 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19174 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
19175 .progmemx.data now.
19177 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19180 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
19181 (avr_addrspace_t): Add .section_name field.
19182 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
19184 (avr_addrspace): Same. Initialize .section_name. Remove last
19185 NULL entry. Put __memx into .progmemx.data.
19186 (progmem_section_prefix): Remove.
19187 (avr_asm_init_sections): No need to initialize progmem_section.
19188 (avr_asm_named_section): Use avr_addrspace[].section_name to get
19189 section name prefix.
19190 (avr_asm_select_section): Ditto. And use get_unnamed_section to
19191 retrieve the progmem section.
19192 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
19193 boundary to run over avr_addrspace[].
19194 (avr_register_target_pragmas): Ditto.
19196 2013-01-06 Jakub Jelinek <jakub@redhat.com>
19198 * varasm.c (output_constant_def_contents): For asan_protect_global
19199 protected strings, adjust DECL_ALIGN if needed, before testing for
19201 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
19202 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
19204 (output_object_block): For asan protected decls, emit asan padding
19205 after their contents.
19206 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
19207 (asan_finish_file): Test it here instead.
19209 2013-01-07 Nick Clifton <nickc@redhat.com>
19210 Matthias Klose <doko@debian.org>
19211 Doug Kwan <dougkwan@google.com>
19212 H.J. Lu <hongjiu.lu@intel.com>
19215 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
19217 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
19219 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
19221 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
19223 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
19225 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19228 * doc/install.texi (Cross-Compiler-Specific Options): Document
19231 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
19233 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
19234 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
19235 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
19236 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
19237 vqmovun_high_s64): Fix source operand number and update copyright.
19239 2013-01-07 Richard Biener <rguenther@suse.de>
19241 PR middle-end/55890
19242 * gimple.h (gimple_call_builtin_p): New overload.
19243 * gimple.c (validate_call): New function.
19244 (gimple_call_builtin_p): Likewise.
19245 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19246 Use gimple_call_builtin_p.
19247 (find_func_clobbers): Likewise.
19248 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
19249 (strlen_optimize_stmt): Likewise.
19251 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
19253 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
19254 (vld1q_dup_*): Likewise.
19255 (vld1_*): Likewise.
19256 (vld1q_*): Likewise.
19257 (vld1_lane_*): Likewise.
19258 (vld1q_lane_*): Likewise.
19260 2013-01-07 Richard Biener <rguenther@suse.de>
19262 * lto-streamer.h (LTO_minor_version): Bump to 2.
19264 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
19266 * config/aarch64/aarch64-protos.h
19267 (aarch64_const_double_zero_rtx_p): Rename to...
19268 (aarch64_float_const_zero_rtx_p): ...this.
19269 (aarch64_float_const_representable_p): New.
19270 (aarch64_output_simd_mov_immediate): Likewise.
19271 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
19272 move immediate case.
19273 * config/aarch64/aarch64.c
19274 (aarch64_const_double_zero_rtx_p): Rename to...
19275 (aarch64_float_const_zero_rtx_p): ...this.
19276 (aarch64_print_operand): Allow printing of new constants.
19277 (aarch64_valid_floating_const): New.
19278 (aarch64_legitimate_constant_p): Check for valid floating-point
19280 (aarch64_simd_valid_immediate): Likewise.
19281 (aarch64_vect_float_const_representable_p): New.
19282 (aarch64_float_const_representable_p): Likewise.
19283 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
19284 (aarch64_output_simd_mov_immediate): New.
19285 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
19286 (*movdf_aarch64): Likewise.
19287 * config/aarch64/constraints.md (Ufc): New.
19288 (Y): call aarch64_float_const_zero_rtx.
19289 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
19291 2013-01-07 Richard Biener <rguenther@suse.de>
19293 PR tree-optimization/55888
19294 PR tree-optimization/55862
19295 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
19296 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
19297 not if it is contained therein.
19299 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19301 * config/avr/t-avr: Typo.
19303 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19306 * config/avr/t-avr: Don't automatically rebuild
19307 $(srcdir)/config/avr/t-multilib
19308 $(srcdir)/config/avr/avr-tables.opt
19309 $(srcdir)/doc/avr-mmcu.texi
19310 (avr-mcus): New phony target to build them on request.
19311 (s-avr-mlib, s-avr-mmcu-texi): Remove.
19312 * avr/avr-mcus.def: Adjust comments.
19314 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
19316 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
19318 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
19320 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
19322 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
19324 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
19326 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
19328 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
19329 to generate profiling.
19330 * config/rs6000/aix64.h (LIB_SPEC): Same.
19332 2013-01-04 Andrew Pinski <apinski@cavium.com>
19334 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
19336 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
19338 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
19340 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
19342 (ix86_expand_move): Ditto.
19343 (ix86_zero_extend_to_Pmode): Ditto.
19344 (ix86_expand_call): Ditto.
19345 (ix86_expand_special_args_builtin): Ditto.
19346 (ix86_expand_builtin): Ditto.
19348 2013-01-04 Richard Biener <rguenther@suse.de>
19350 PR tree-optimization/55862
19351 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
19352 translating them through PHI nodes.
19354 2013-01-04 Martin Jambor <mjambor@suse.cz>
19356 PR tree-optimization/55755
19357 * tree-sra.c (sra_modify_assign): Do not check that an access has no
19358 children when trying to avoid producing a VIEW_CONVERT_EXPR.
19360 2013-01-04 Marek Polacek <polacek@redhat.com>
19362 PR middle-end/55859
19363 * opts.c (default_options_optimization): Clarify error message.
19365 2013-01-04 Richard Biener <rguenther@suse.de>
19367 PR middle-end/55863
19368 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
19371 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19374 * config/pa/pa.md (movsi): Revert previous change.
19375 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
19378 2013-01-03 Richard Henderson <rth@redhat.com>
19380 * config/i386/i386.c (ix86_expand_move): Always assign to op1
19381 after eliminating TLS symbols.
19383 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
19386 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
19387 * graphite-poly.c (debug_gmp_value): Likewise.
19389 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
19392 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
19393 selected code model, define __code_mode_small__, __code_model_medium__,
19394 __code_model_large__, __code_model_32__ or __code_model_kernel__.
19395 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
19396 xchg temporary register with %k. Declare temporary register as
19398 [__x86_64__]: For medium and large code models, preserve %rbx register.
19400 2013-01-03 Richard Biener <rguenther@suse.de>
19402 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
19403 (dump_subscript): Adjust.
19404 (finalize_ddr_dependent): Do not dump redundant info.
19405 (analyze_siv_subscript): Adjust.
19406 (subscript_dependence_tester): Likewise.
19407 (compute_affine_dependence): Likewise.
19409 2013-01-03 Richard Biener <rguenther@suse.de>
19412 2013-01-03 Richard Biener <rguenther@suse.de>
19414 PR tree-optimization/55857
19415 * tree-vect-stmts.c (vectorizable_load): Do not setup
19416 re-alignment for invariant loads.
19418 2013-01-02 Richard Biener <rguenther@suse.de>
19420 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19421 invariant load do not generate a vector load from the scalar location.
19423 2013-01-03 Richard Biener <rguenther@suse.de>
19425 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19426 for not vectorizing.
19427 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19428 not build INDIRECT_REFs, call get_name once only.
19429 (vect_create_data_ref_ptr): Likewise. Dump base object kind
19430 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19432 2013-01-03 Richard Biener <rguenther@suse.de>
19434 PR tree-optimization/55857
19435 * tree-vect-stmts.c (vectorizable_load): Do not setup
19436 re-alignment for invariant loads.
19438 2013-01-03 Richard Biener <rguenther@suse.de>
19441 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19442 prefer a built-in decl.
19444 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19446 * gcc.c (process_command): Update copyright notice dates.
19447 * gcov.c (print_version): Likewise.
19448 * gcov-dump.c (print_version): Likewise.
19450 PR rtl-optimization/55838
19451 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19452 iv0.step, iv1.step and step.
19454 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19455 Marc Glisse <marc.glisse@inria.fr>
19457 PR tree-optimization/55832
19458 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19459 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19460 integer_{one,zero}_node.
19462 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19465 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19466 * var-tracking.c (reverse_op): Don't add reverse ops to
19467 VALUEs that have already
19468 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19470 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19472 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19474 2013-01-02 Teresa Johnson <tejohnson@google.com>
19476 * dumpfile.c (dump_loc): Print filename with location.
19477 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19478 new location_t parameter to emit complete unroll message with
19479 new dump framework.
19480 (canonicalize_loop_induction_variables): Compute loops location
19481 and pass to try_unroll_loop_completely.
19482 * loop-unroll.c (report_unroll_peel): New function.
19483 (peel_loops_completely): Use new dump format with location
19484 for main dumpfile message, and invoke report_unroll_peel on success.
19485 (decide_unrolling_and_peeling): Ditto.
19486 (decide_peel_once_rolling): Remove old dumpfile message subsumed
19487 by report_unroll_peel.
19488 (decide_peel_completely): Ditto.
19489 (decide_unroll_constant_iterations): Ditto.
19490 (decide_unroll_runtime_iterations): Ditto.
19491 (decide_peel_simple): Ditto.
19492 (decide_unroll_stupid): Ditto.
19493 * cfgloop.c (get_loop_location): New function.
19494 * cfgloop.h (get_loop_location): Declare.
19496 2013-01-02 Sriraman Tallam <tmsriram@google.com>
19498 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19501 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19503 PR middle-end/55198
19504 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19505 BLKmode objects when EXPAND_MEMORY is specified.
19507 2013-01-02 Sriraman Tallam <tmsriram@google.com>
19509 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19511 (fold_builtin_cpu): Do not share cpu model decls across statements.
19513 2013-01-02 Jason Merrill <jason@redhat.com>
19516 * tree.c (build_array_type_1): Revert earlier change.
19518 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
19520 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19522 * config/aarch64/aarch64-tune.md: Re-generate.
19524 2013-01-02 Richard Biener <rguenther@suse.de>
19526 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19527 invariant load do not generate a vector load from the scalar location.
19529 2013-01-02 Richard Biener <rguenther@suse.de>
19532 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19533 * configure: Regenerate.
19535 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
19537 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19538 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19539 (expand_builtin_int_roundingfn_2): Keep the original target around
19540 for the fallback case.
19542 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
19544 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19545 to be clear for sign changes.
19547 2013-01-01 Jan Hubicka <jh@suse.cz>
19549 * ipa-inline-analysis.c: Fix formatting.
19551 2013-01-01 Jakub Jelinek <jakub@redhat.com>
19553 PR tree-optimization/55831
19554 * tree-vect-loop.c (get_initial_def_for_induction): Use
19555 gsi_after_labels instead of gsi_start_bb.
19557 Copyright (C) 2013 Free Software Foundation, Inc.
19559 Copying and distribution of this file, with or without modification,
19560 are permitted in any medium without royalty provided the copyright
19561 notice and this notice are preserved.