1 2013-09-03 Richard Biener <rguenther@suse.de>
3 * tree-affine.c (add_elt_to_tree): Fix association issue,
4 avoid useless converts and make sure to always return a
7 2013-09-03 Richard Biener <rguenther@suse.de>
10 * fold-const.c (negate_expr_p): Fix division case.
11 (negate_expr): Likewise.
13 2013-09-03 Richard Biener <rguenther@suse.de>
16 * tree-streamer-out.c: Include tm.h.
17 * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
19 2013-09-03 Jan Hubicka <jh@suse.cz>
21 * tree-profile.c (tree_profiling): Cleanup CFG when done.
23 2013-09-03 Alan Modra <amodra@gmail.com>
25 * config.gcc (powerpc*-*-linux*): Add support for little-endian
26 multilibs to big-endian target and vice versa.
27 * config/rs6000/t-linux64: Use := assignment on all vars.
28 (MULTILIB_EXTRA_OPTS): Remove fPIC.
29 (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
30 * config/rs6000/t-linux64le: New file.
31 * config/rs6000/t-linux64bele: New file.
32 * config/rs6000/t-linux64lebe: New file.
34 2013-09-02 Jan Hubicka <jh@suse.cz>
36 * ipa-inline-transform.c (inline_transform): Do not
37 optimize_inline_calls when not optimizing.
39 2013-09-02 Jan Hubicka <jh@suse.cz>
41 * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
42 duplicated nodes for assembler names.
43 * symtab.c (symtab_unregister_node): Do not attempt to unlink
44 hard registers from assembler name hash.
46 2013-09-02 Jan Hubicka <jh@suse.cz>
48 * ipa-split.c (execute_split_functions): Split externally visible
49 functions called once.
51 2013-09-02 Martin Jambor <mjambor@suse.cz>
54 * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
55 linked list. When finding the correct duplicate, also consider also
56 the caller in additon to its inlined_to node.
58 2013-09-02 James Greenhalgh <james.greenhalgh@arm.com>
60 * config/aarch64/aarch64-simd-builtins.def
61 (dup_lane_scalar): Remove.
62 * config/aarch64/aarch64-simd.md
63 (aarch64_simd_dup): Add 'w->w' alternative.
64 (aarch64_dup_lane<mode>): Allow for VALL.
65 (aarch64_dup_lane_scalar<mode>): Remove.
66 (aarch64_dup_lane_<vswap_width_name><mode>): New.
67 (aarch64_get_lane_signed<mode>): Add w->w altenative.
68 (aarch64_get_lane_unsigned<mode>): Likewise.
69 (aarch64_get_lane<mode>): Likewise.
70 * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
71 (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
72 * config/aarch64/iterators.md (VSWAP_WIDTH): New.
73 (VCON): Change container of V2SF.
74 (vswap_width_name): Likewise.
75 * config/aarch64/arm_neon.h
76 (__aarch64_vdup_lane_any): New.
77 (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
78 (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
79 (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
81 2013-09-02 Eric Botcazou <ebotcazou@adacore.com>
84 * expr.c (emit_move_complex): Do not move complex FP values as parts if
85 the source or the destination is a single hard register.
87 2013-09-02 Richard Biener <rguenther@suse.de>
90 * tree-scalar-evolution.c (instantiate_scev_name): Allow
93 2013-09-02 Richard Biener <rguenther@suse.de>
95 * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
96 arithmetic to sizetype.
98 2013-09-02 Bin Cheng <bin.cheng@arm.com>
100 * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
101 Find auto-increment use both before and after candidate.
103 2013-09-02 Marek Polacek <polacek@redhat.com>
105 * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
107 2013-09-01 Jan Hubicka <jh@suse.cz>
109 * Makefile.in: Add ipa-profile.o
110 (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
111 * cgraph.c (struct cgraph_propagate_frequency_data,
112 cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
113 ipa-profile.c; replace cgraph_ by ipa_ prefix.
114 * cgraph.h (cgraph_propagate_frequency): Remove.
115 * ipa-inline-analysis.c: Include ipa-utils.h; drop duplicated cfgloop.h.
116 (inline_update_callee_summaries): Update.
117 * ipa-profile.c: New file.
118 * ipa-utils.h (ipa_propagate_frequency): Declare.
119 * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
120 data-streamer.h, value-prof.h
121 (symtab_remove_unreachable_nodes): Update profile.
122 (struct histogram_entry, histogram, histogram_pool, histogram_hash,
123 account_time_size, cmp_counts, dump_histogram,
124 ipa_profile_generate_summary, ipa_profile_write_summary,
125 ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
126 pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
127 Move to ipa-profile.c
129 2013-09-01 John David Anglin <danglin@gcc.gnu.org>
131 * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
133 2013-09-01 Jan Hubicka <jh@suse.cz>
135 * common.opt (fdevirtualize-speculatively): New function.
136 * invoke.texi (fdevirtualize-speculatively): Document.
137 * ipa-devirt.c: Include ipa-inline.h
138 (likely_target_p): New function.
139 (ipa_devirt): New function.
140 (gate_ipa_devirt): New function.
141 (pass_data_ipa_devirt): New static var.
142 (pass_ipa_devirt): Likewise.
143 (make_pass_ipa_devirt): New function.
144 * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
145 (common_handle_option): Disable devirtualization when
146 value range profiling is available.
147 * passes.def (pass_ipa_devirt): Add.
148 * timever.def (TV_IPA_DEVIRT): New timevar.
149 * tree-pass.h (make_pass_ipa_devirt):
151 2013-09-01 Iain Sandoe <iain@codesourcery.com>
153 * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
154 include sanitize(undefined).
156 2013-08-31 Diego Novillo <dnovillo@google.com>
158 * Makefile.in (TREE_CORE_H): Define.
160 (GTFILES): Add tree-core.h.
161 * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
163 * tree-core.h: New file.
164 Move all data structures, enum, typedefs, global
165 declarations and constants from ...
168 2013-08-31 Jan Hubicka <jh@suse.cz>
170 * bulitins.c (expand_builtin): Do not early exit for gcov
171 instrumented functions.
173 2013-08-31 Marek Polacek <polacek@redhat.com>
175 * ubsan.c: Include tm_p.h.
177 2013-08-31 Jan Hubicka <jh@suse.cz>
179 * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
182 * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
183 * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
184 * tree-cfg.c (verify_gimple_label): ... here.
185 * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
186 (ipa_merge_profiles): New function.
187 * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
188 (lto_input_function_body): Likewise.
189 * ipa-utils.h (ipa_merge_profiles): Declare.
190 * lto-streamer.h (lto_input_function_body): Update prototype.
191 (emit_label_in_global_context_p): Remove.
192 * lto-symtab.c: Include ipa-utils.h
193 (lto_cgraph_replace_node): Use ipa_merge_profiles.
195 2013-08-31 Jan Hubicka <jh@suse.cz>
197 * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
199 2013-08-31 Jan Hubicka <jh@suse.cz>
201 * basic-block.h (apply_scale): Make scale parmeter gcov_type.
203 2013-08-31 Uros Bizjak <ubizjak@gmail.com>
205 * config/alpha/alpha.c (alpha_emit_conditional_move): Update
206 "cmp" RTX before signed_comparison_operator check to account
209 2013-08-30 Jan Hubicka <jh@suse.cz>
211 * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
212 (detect_type_change_1): Rename to ...
213 (detect_type_change): ... this one; early return on non-polymorphic
215 (detect_type_change_ssa): Add comp_type parameter; update
216 use of detect_type_change.
217 (compute_complex_assign_jump_func): Add param_type parameter;
218 update use of detect_type_change_ssa.
219 (compute_complex_ancestor_jump_func): Likewise.
220 (ipa_get_callee_param_type): New function.
221 (ipa_compute_jump_functions_for_edge): Compute parameter type;
222 update calls to the jump function computation functions.
224 2013-08-30 Teresa Johnson <tejohnson@google.com>
225 Steven Bosscher <steven@gcc.gnu.org>
227 * cfgrtl.c (fixup_new_cold_bb): New routine.
228 (commit_edge_insertions): Invoke fixup_partitions.
229 (find_partition_fixes): New routine.
230 (fixup_partitions): Ditto.
231 (verify_hot_cold_block_grouping): Update comments.
232 (rtl_verify_edges): Invoke find_partition_fixes.
233 (rtl_verify_bb_pointers): Update comments.
234 (rtl_verify_bb_layout): Ditto.
235 * basic-block.h (probably_never_executed_edge_p): Declare.
236 (fixup_partitions): Ditto.
237 * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
238 * bb-reorder.c (sanitize_hot_paths): New function.
239 (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
241 * predict.c (probably_never_executed_edge_p): New routine.
242 * cfg.c (check_bb_profile): Add partition insanity warnings.
244 2013-08-30 Meador Inge <meadori@codesourcery.com>
246 * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
248 2013-08-30 Marek Polacek <polacek@redhat.com>
250 * Makefile.in (ubsan.o): Add.
251 (c-family/c-ubsan.o): Add.
252 (builtins.o): Add ubsan.h dependency.
255 * common.opt: Add -fsanitize=undefined option.
256 (flag_sanitize): Add variable.
257 (fsanitize=): Add option. Add Driver.
258 (fsanitize=thread): Remove option.
259 (fsanitize=address): Likewise.
260 (static-libubsan): New option.
261 * doc/invoke.texi: Document the new flag and -static-libubsan.
262 * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
263 (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
264 * builtin-attrs.def (ATTR_COLD): Define.
265 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
266 * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
267 BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
268 * flag-types.h (sanitize_code): New enum.
269 * opts.c (common_handle_option): Parse command line arguments
270 of -fsanitize=. Add -fsanitize=unreachable option.
271 * varasm.c (get_variable_section): Adjust.
272 (assemble_noswitch_variable): Likewise.
273 (assemble_variable): Likewise.
274 (output_constant_def_contents): Likewise.
275 (categorize_decl_for_section): Likewise.
276 (place_block_symbol): Likewise.
277 (output_object_block): Likewise.
278 * builtins.def: Likewise.
279 * toplev.c (compile_file): Likewise.
280 (process_options): Likewise.
281 * cppbuiltin.c: Likewise.
282 * tsan.c (tsan_pass): Likewise.
283 (tsan_gate): Likewise.
284 (tsan_gate_O0): Likewise.
285 * cfgexpand.c (partition_stack_vars): Likewise.
286 (expand_stack_vars): Likewise.
287 (defer_stack_allocation): Likewise.
288 (expand_used_vars): Likewise.
289 * cfgcleanup.c (old_insns_match_p): Likewise.
290 * asan.c (asan_finish_file): Likewise.
291 (asan_instrument): Likewise.
292 (gate_asan): Likewise.
293 (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
294 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
295 (asan_global_struct): Use pointer_sized_int_node instead
296 calling build_nonstandard_integer_type.
297 (initialize_sanitizer_builtins): Likewise.
298 (asan_finish_file): Likewise.
299 * gcc.c: Document %{%:function(args):X}.
300 (static_spec_functions): Add sanitize.
301 (handle_spec_function): Add retval_nonnull argument and if non-NULL,
302 store funcval != NULL there.
303 (do_spec_1): Adjust handle_spec_function caller.
304 (handle_braces): Allow %:function(args) as condition.
305 (sanitize_spec_function): New function.
306 (ADD_STATIC_LIBUBSAN_LIBS): Define.
307 (LIBUBSAN_SPEC): Likewise.
308 (LIBUBSAN_EARLY_SPEC): Likewise.
309 (SANITIZER_SPEC): Handle libubsan.
310 (SANITIZER_EARLY_SPEC): Likewise.
311 * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
312 instead of fsanitize=address.
313 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
314 instead of fsanitize=address*.
315 * builtins.c: Include ubsan.h.
316 (fold_builtin_0): Instrument __builtin_unreachable.
317 * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
318 instead of flag_asan.
319 * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
320 (pointer_sized_int_node): Define.
321 * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
323 2013-08-30 Mike Stump <mikestump@comcast.net>
325 * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
328 2013-08-29 Jan Hubicka <jh@suse.cz>
330 * cgraph.c (cgraph_function_body_availability): Handle weakref
332 * passes.def: Remove pass_fixup_cfg.
333 * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
334 track when we need to remove functions.
335 (gate_ipa_inline): Execute inlining always; add comment why.
336 (pass_data_ipa_inline): Remove TODO_remove_functions.
337 * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
338 do not produce summaries.
339 * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
340 (symtab_nonoverwritable_alias): Assert we are not called on weakref.
341 * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
342 constant pool and vtable.
344 2013-08-30 Tejas Belagod <tejas.belagod@arm.com>
346 * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
347 New arm_neon.h's internal macros to specify 64-bit constants.
348 Avoid using stdint.h's macros.
350 2013-08-30 Joern Rennecke <joern.rennecke@embecosm.com>
352 * recog.c (verify_changes): Verify that changes[i].old is non-zero
353 before applying REG_P.
355 2013-08-30 Jakub Jelinek <jakub@redhat.com>
357 PR tree-optimization/58277
358 * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
359 after seeing too many stmts with vdef in between dombb and current
360 bb, invalidate everything.
362 2013-08-30 Richard Biener <rguenther@suse.de>
364 * fold-const.c (fold_single_bit_test): Fix overflow test.
366 2013-08-30 Eric Botcazou <ebotcazou@adacore.com>
368 * function.c (assign_parm_setup_reg): For a parameter passed by pointer
369 and which can live in a register, always retrieve the value on entry.
370 * var-tracking.c (add_stores): Treat the copy on entry for a parameter
371 passed by invisible reference specially.
372 (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
373 (vt_add_function_parameter): Correctly deal with a parameter passed by
376 2013-08-30 Jan Hubicka <jh@suse.cz>
378 * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
380 2013-08-30 Richard Biener <rguenther@suse.de>
382 PR tree-optimization/58228
383 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
384 allow invariant loads in nested loop vectorization.
386 2013-08-30 Richard Biener <rguenther@suse.de>
388 PR tree-optimization/58223
389 * tree-loop-distribution.c (has_anti_dependence): Rename to ...
390 (has_anti_or_output_dependence): ... this and adjust to also
391 look for output dependences.
392 (mark_nodes_having_upstream_mem_writes): Adjust.
393 (rdg_flag_uses): Likewise.
395 2013-08-30 Richard Biener <rguenther@suse.de>
397 PR tree-optimization/58010
398 * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
399 assert that we have a loop-closed PHI.
401 2013-08-29 Jan Hubicka <jh@suse.cz>
403 * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
404 * cgraph.c (cgraph_release_function_body): Free decl_in_state.
405 * lto-section-in.c (lto_free_function_in_decl_state): New function.
406 (lto_free_function_in_decl_state_for_node): New function.
408 2013-08-29 Xinliang David Li <davidxl@google.com>
410 * loop-unroll.c (report_unroll_peel): Minor message change.
411 * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
412 Emit alignment peeling message with default -fopt-info.
413 (vect_loop_versioning): Emit loop version info message.
414 * tree-vectorizer.c (vectorize_loops): Minor message change.
415 (execute_vect_slp): Ditto.
417 2013-08-29 Eric Botcazou <ebotcazou@adacore.com>
419 * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
420 of the clone from the DECL_NAME of the original function.
422 2013-08-29 Oleg Endo <olegendo@gcc.gnu.org>
424 * passes.c (register_pass): Add overload.
425 * tree-pass.h (register_pass): Forward declare it. Add comment.
427 2013-08-29 Jan Hubicka <jh@suse.cz>
429 * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
430 DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
431 * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
432 DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
433 (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
434 * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
435 Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
436 (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
438 2013-08-29 Teresa Johnson <tejohnson@google.com>
440 * dumpfile.c (dump_loc): Output column number.
441 * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
442 * doc/invoke.texi: Document optall -fopt-info flag.
443 * profile.c (read_profile_edge_counts): Use new dump framework.
444 (compute_branch_probabilities): Ditto.
445 * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
446 when pass not in any opt group.
447 * pass_manager.h (pass_manager::get_pass_profile): New method.
448 * value-prof.c (check_counter): Use new dump framework.
449 (check_ic_target): Ditto.
450 * coverage.c (get_coverage_counts): Ditto.
451 (coverage_init): Setup new dump framework.
453 2013-08-29 Richard Biener <rguenther@suse.de>
455 PR tree-optimization/58246
456 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
457 handle the dominance check inside a basic-block.
459 2013-08-29 Richard Biener <rguenther@suse.de>
462 * tree-ssa-copy.c (may_propagate_copy): Allow propagating
463 of default defs that appear in abnormal PHI nodes.
465 2013-08-29 Richard Biener <rguenther@suse.de>
467 PR tree-optimization/57685
468 * tree-vrp.c (register_edge_assert_for_1): Recurse only for
469 single-use operands to avoid exponential complexity.
471 2013-08-28 Dehao Chen <dehao@google.com>
473 * ipa-inline.c (edge_badness): Fix integer underflow.
475 2013-08-28 Uros Bizjak <ubizjak@gmail.com>
477 * gtm-builtins.def (_ITM_free): Declare leaf.
479 2013-08-28 Jakub Jelinek <jakub@redhat.com>
482 * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
483 (*tls_local_dynamic_base_64_largepic): Likewise.
484 (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
485 Remove predicate from call operand.
486 * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
487 return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
489 2013-08-28 Jeff Law <law@redhat.com>
491 * tree-ssa-threadedge.c (thread_around_empty_block): Remove
492 checks for the number of predecessors and successors allowed.
493 * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
494 which require copying a joiner block if there is a request which
495 is a subpath that requires no joiner block copying.
497 2013-08-28 Jan Hubicka <jh@suse.cz>
499 * lto-streamer-out.c (DFS_write_tree_body): Drop
500 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
501 (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
502 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
503 * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
504 Do not read DECL_ERROR_ISSUED.
505 (unpack_ts_decl_with_vis_value_fields): Do not read
507 (lto_input_ts_binfo_tree_pointers): Do not read
508 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
509 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
510 write DECL_ERROR_ISSUED..
511 (pack_ts_decl_with_vis_value_fields): Do not write
513 (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
514 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
515 * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
516 * tree.h (tree_decl_common): Update comment.
517 (DECL_ERROR_ISSUED): Remove.
519 2013-08-28 Jakub Jelinek <jakub@redhat.com>
522 * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
524 2013-08-28 Jan Hubicka <jh@suse.cz>
526 * builtins.def (free): Declare leaf.
528 2013-08-27 David Malcolm <dmalcolm@redhat.com>
531 * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
532 * configure: Regenerate.
534 2013-08-27 Martin Jambor <mjambor@suse.cz>
536 * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
537 (ipa_ancestor_jf_data): Likewise.
538 (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
539 (ipa_get_jf_pass_through_type_preserved): New function.
540 (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
541 (ipa_get_jf_ancestor_type_preserved): New function.
542 * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
543 (ipa_get_jf_ancestor_result): Likewise.
544 (propagate_vals_accross_pass_through): Use
545 ipa_get_jf_pass_through_result to do all the value mappings.
546 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
548 (ipa_set_jf_cst_copy): New function.
549 (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
550 (ipa_set_jf_arith_pass_through): Likewise.
551 (ipa_set_ancestor_jf): Likewise.
552 (compute_complex_assign_jump_func): Set type_preserved instead of
554 (ipa_compute_jump_functions_for_edge): Likewise.
555 (combine_known_type_and_ancestor_jfs): Honor type_preserved.
556 (update_jump_functions_after_inlining): Update type_preserved.
557 Explicitely create jump functions when combining one with pass_through.
558 (ipa_write_jump_function): Stream the type_preserved flags.
559 (ipa_read_jump_function): Likewise.
561 2013-08-27 Jakub Jelinek <jakub@redhat.com>
562 Aldy Hernandez <aldyh@redhat.com>
564 * Makefile.in (omp-low.o): Depend on $(TARGET_H).
565 * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
566 * function.h (struct function): Add has_force_vect_loops and
568 * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
569 * gimple.c (gimple_build_omp_critical): Add KIND argument and
571 * gimple.def: Update CLAUSES comments.
572 * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
573 (gimple_build_omp_for): Add argument to prototype.
574 (gimple_omp_for_kind): New.
575 (gimple_omp_for_set_kind): New.
576 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
577 GOVD_DATA_SHARE_CLASS.
578 (enum omp_region_type): Add ORT_SIMD.
579 (gimple_add_tmp_var): Handle ORT_SIMD.
580 (gimplify_var_or_parm_decl): Same.
581 (is_gimple_stmt): Same.
582 (omp_firstprivatize_variable): Same.
583 (omp_add_variable): Only use splay_tree_insert if lookup failed.
584 (omp_notice_variable): Handle ORT_SIMD.
585 (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
586 (omp_check_private): Handle ORT_SIMD.
587 (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
589 (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
590 Handle OMP_CLAUSE_LASTPRIVATE.
591 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
593 (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
594 (gimplify_expr): Handle OMP_SIMD.
595 * internal-fn.c (expand_GOMP_SIMD_LANE): New.
596 (expand_GOMP_SIMD_VF): New.
597 (expand_GOMP_SIMD_LAST_LANE): New.
598 * internal-fn.def (GOMP_SIMD_LANE): New.
600 (GOMP_SIMD_LAST_LANE): New.
601 * omp-low.c: Include target.h.
602 (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
604 (check_omp_nesting_restrictions): Same.
606 (lower_rec_simd_input_clauses): New.
607 (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
609 (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
610 GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
611 (expand_omp_build_assign): New.
612 (expand_omp_for_init_counts): New.
613 (expand_omp_for_init_vars): New.
614 (extract_omp_for_update_vars): New.
615 (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
616 and rewrite accordingly.
617 (expand_omp_simd): New.
618 (expand_omp_for): Use expand_omp_simd.
619 (lower_omp_for_lastprivate): Unshare vinit when appropriate.
620 (lower_omp_for): Do not lower the body.
621 * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
623 * tree-flow.h (find_omp_clause): Remove prototype.
624 * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
625 forcing vectorization of the loop, or if flag_tree_vectorize.
626 (gate_tree_if_conversion): Similarly.
627 * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
628 gimple_build_omp_for.
629 (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
630 * tree-parloops (create_parallel_loop): Pass kind argument to
631 gimple_build_omp_for.
632 * tree-pretty-print.c (dump_omp_clause): Add cases for
633 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
634 OMP_CLAUSE__SIMDUID_.
635 (dump_generic_node): Handle OMP_SIMD.
636 * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
637 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
638 unroll OMP_SIMD loops here.
639 * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
640 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
642 (vect_analyze_data_refs): Handle simd loops.
643 * tree-vect-loop.c (vectorizable_live_operation): Handle
645 * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
646 (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
647 (vectorizable_load): Same.
648 * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
649 (struct simduid_to_vf): New.
650 (simduid_to_vf::hash): New.
651 (simduid_to-vf::equal): New.
652 (struct simd_array_to_simduid): New.
653 (simd_array_to_simduid::hash): New.
654 (simd_array_to_simduid::equal): New.
655 (adjust_simduid_builtins): New.
656 (struct note_simd_array_uses_struct): New.
657 (note_simd_array_uses_cb): New.
658 (note_simd_array_uses): New.
659 (vectorize_loops): Handle simd hints and adjust simd builtins
661 * tree-vectorizer.h (struct _stmt_vec_info): Add
662 simd_lane_access_p field.
663 (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
664 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
665 OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
666 (omp_clause_code_name): Same.
667 (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
668 OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
669 * tree.def (OMP_SIMD): New entry.
670 * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
671 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
672 (OMP_CLAUSE_DECL): Adjust range for new clauses.
673 (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
674 (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
675 (OMP_CLAUSE_LINEAR_STEP): New.
676 (OMP_CLAUSE_SAFELEN_EXPR): New.
677 (OMP_CLAUSE__SIMDUID__DECL): New.
678 (find_omp_clause): New prototype.
680 2013-08-27 H.J. Lu <hongjiu.lu@intel.com>
682 * config/i386/driver-i386.c (host_detect_local_cpu): Update
683 Haswell processor detection.
685 2013-08-27 Christian Widmer <shadow@umbrox.de>
688 * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
689 of Ivy Bridge and Haswell processors. Assume core-avx2 for unknown
690 AVX2 capable processors.
692 2013-08-27 Tejas Belagod <tejas.belagod@arm.com>
694 * config/aarch64/arm_neon.h: Replace all inline asm implementations
695 of vget_low_* with implementations in terms of other intrinsics.
697 2013-08-27 Marc Glisse <marc.glisse@inria.fr>
700 * doc/extend.texi (__builtin_isinf_sign): Restrict the return
701 values to -1, 0 and 1.
703 2013-08-27 Vidya Praveen <vidyapraveen@arm.com>
705 * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
706 UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
707 (<optab><mode>3_insn): Remove.
708 (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
709 (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
710 (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
711 (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
712 (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
713 (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
714 (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
715 (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
716 (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
717 (ror<mode>3_insn): Likewise.
718 * config/aarch64/predicates.md (aarch64_simd_register): New.
720 2013-08-27 Richard Biener <rguenther@suse.de>
722 PR tree-optimization/57521
723 * tree-if-conv.c (if_convertible_bb_p): Verify that at least
724 one edge is non-critical.
725 (find_phi_replacement_condition): Make sure to use a non-critical
726 edge. Cleanup and remove old bug workarounds.
727 (bb_postdominates_preds): Remove.
728 (if_convertible_loop_p_1): Do not compute post-dominators.
729 (combine_blocks): Do not free post-dominators.
730 (main_tree_if_conversion): Likewise.
731 (pass_data_if_conversion): Add TODO_verify_ssa.
733 2013-08-27 DJ Delorie <dj@redhat.com>
735 * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
737 2013-08-27 Yufeng Zhang <yufeng.zhang@arm.com>
739 * function.c (assign_parm_find_data_types): Set passed_mode and
740 nominal_mode to the TYPE_MODE of nominal_type for the built
741 pointer type in case of the struct-pass-by-reference.
743 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
745 * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
746 (UINT16_TYPE): Change default to "unsigned int".
748 * config/avr/avr.opt (mfract-convert-truncate): New option.
749 * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
750 is set, round negative fractional integers according to n1169
751 when converting to integer types.
753 2013-08-26 Jan Hubicka <jh@suse.cz>
755 * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
756 methods can not be called indirectly when their address is not taken.
758 2013-08-26 Jan Hubicka <jh@suse.cz>
760 * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
763 2013-08-26 Jan Hubicka <jh@suse.cz>
765 * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
768 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
770 * reload.c (find_valid_class): Allow classes that do not include
771 FIRST_PSEUDO_REGISTER - 1.
773 2013-08-26 Jan Hubicka <jh@suse.cz>
775 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
776 fix edge count/frequency when speculation failed; fix type check
779 2013-08-26 Jan Hubicka <jh@suse.cz>
781 * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
783 2013-08-26 Jan Hubicka <jh@suse.cz>
785 * ipa-inline-transform.c (inline_transform): Be ready for basic block
786 to be changed by edge redirection.
788 2013-08-26 Jan Hubicka <jh@suse.cz>
790 * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
791 formating; add sanity check.
792 (cgraph_resolve_speculation): Add FIXME about scaling profiles.
793 (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
794 * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
795 (resolve_noninline_speculation): Update callee keys, too.
797 2013-08-26 Jan Hubicka <jh@suse.cz>
799 * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
800 (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
802 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
804 * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
807 2013-08-26 Uros Bizjak <ubizjak@gmail.com>
809 * config/i386/i386.c (ix86_debug_options): Remove prototype.
810 (x86_64_elf_select_section): Ditto.
811 (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
813 (ix86_pass_by_reference): Ditto.
814 (output_set_got): Ditto.
815 (ix86_unary_operator_ok): Ditto.
816 (ix86_expand_builtin): Ditto.
818 2013-08-23 Jan Hubicka <jh@suse.cz>
820 * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
822 2013-08-23 Jan Hubicka <jh@suse.cz>
824 * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
825 (tree_decl_with_vis): Add FINAL field.
827 2013-08-23 Jeff Law <law@redhat.com>
829 * tree-ssa-pre.c (do_regular_insertion): Include the expression in
830 the debugging dump when the expression is fully redundant.
832 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
834 * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
835 * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
836 * pretty-print.c (pp_formatted_text_data): Likewise.
837 (pp_write_text_to_stream): Likewise.
838 (pp_write_text_as_dot_label_to_stream): Likewise.
839 (pp_append_r): Likewise.
840 (pp_format): Likewise.
841 (pp_flush): Likewise.
842 (pp_clear_output_area): Likewise.
843 (pp_append_text): Likewise.
844 (pp_formatted_text): Likewise.
845 (pp_remaining_character_count_for_line): Likewise.
846 (pp_newline): Likewise.
847 (pp_character): Likewise.
848 (output_buffer::~output_buffer): Define.
849 (pretty_printer::~pretty_printer): Destruct output buffer.
850 * pretty-print.h (output_buffer::~output_buffer): Declare.
851 (pretty_printer::~pretty_printer): Declare virtual.
853 2013-08-24 Marc Glisse <marc.glisse@inria.fr>
856 * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
857 HOST_WIDE_INT_M1U): New macros.
858 * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
859 fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
860 unsigned -1 for lshift.
861 * cse.c (cse_insn): Likewise.
862 * double-int.c (rshift_double, lshift_double): Likewise.
863 * builtins.c (fold_builtin_bitop): Likewise.
864 * combine.c (force_to_mode): Likewise.
865 * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
866 * simplify-rtx.c (simplify_const_unary_operation,
867 simplify_const_binary_operation): Likewise.
868 * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
869 check_va_list_escapes): Likewise.
870 * rtlanal.c (nonzero_bits1): Likewise.
871 * expmed.c (expand_smod_pow2): Likewise.
872 * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
874 2013-08-23 Jan Hubicka <jh@suse.cz>
876 * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
877 as having address taken.
879 2013-08-23 Jan Hubicka <jh@suse.cz>
881 * ipa-utils.h (method_class_type): Declare.
882 * ipa-devirt.c (method_class_type): Export.
884 * cgraphunit.c (analyze_functions): Do basic devirtualization;
885 do not walk base classes of anonymous types.
887 2013-08-23 Kaz Kojima <kkojima@gcc.gnu.org>
889 PR rtl-optimization/58220
891 * final.c (reemit_insn_block_notes): Use NEXT_INSN to
892 handle SEQUENCE insns properly.
894 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
896 * pretty-print.h (pp_newline_and_flush): Declare. Remove macro
898 (pp_newline_and_indent): Likewise.
899 (pp_separate_with): Likewise.
900 * pretty-print.c (pp_newline_and_flush): Define.
901 (pp_newline_and_indent): Likewise.
902 (pp_separate_with): Likewise.
904 2013-08-23 Jakub Jelinek <jakub@redhat.com>
907 * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
908 * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
910 2013-08-23 Kirill Yukhin <kirill.yukhin@intel.com>
912 * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
913 * gcc/config/i386/i386.md (*movti_internal): Use
914 predicate to determine if EVEX is needed.
915 (*movsi_internal): Ditto.
916 (*movdf_internal): Ditto.
917 (*movsf_internal): Ditto.
918 * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
920 2013-08-23 Jakub Jelinek <jakub@redhat.com>
922 PR tree-optimization/58209
923 * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
924 (find_tail_calls): Give up for pointer result types if m is non-NULL.
925 (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
926 emit POINTER_PLUS_EXPR.
927 (create_tailcall_accumulator): For pointer result type accumulate in
930 2013-08-22 Paolo Carlini <paolo.carlini@oracle.com>
932 * configure.ac: Add backslashes missing from the last change.
933 * configure: Regenerate.
935 2013-08-22 Jan Hubicka <jh@susue.cz>
937 * ipa.c (function_and_variable_visibility): First remember function
938 was global and then make it local.
940 2013-08-22 Julian Brown <julian@codesourcery.com>
942 * configure.ac: Add aarch64 to list of arches which use "nop" in
944 * configure: Regenerate.
946 2013-08-22 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
948 * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
949 gnu_libc_has_function.
950 * config/s390/tpf.h: Likewise.
952 2013-08-22 Jan Hubicka <jh@susue.cz>
954 * timevar.c (validate_phases): Add cast.
956 2013-08-22 Jan Hubicka <jh@susue.cz>
958 * timevar.c (validate_phases): Use size_t for memory.
959 * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
961 2013-08-22 Gabriel Dos Reis <gdr@integrable-solutions.net>
963 * pretty-print.h (output_buffer::output_buffer): Declare.
964 (pretty_printer::pretty_printer): Likewise.
965 (pp_construct): Remove.
966 * pretty-print.c (output_buffer::output_buffer): Define.
967 (pretty_printer::pretty_printer): Rename from pp_construct. Simplify.
968 * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
969 (print_gimple_expr): Likewise.
970 (print_gimple_seq): Likewise.
971 (gimple_dump_bb): Likewise.
972 * sched-vis.c (dump_value_slim): Likewise.
973 (dump_insn_slim): Likewise.
974 (dump_rtl_slim): Likewise.
975 (str_pattern_slim): Likewise.
976 * tree-mudflap.c (mf_varname_tree): Likewise.
977 * graph.c (print_graph_cfg): Likewise.
978 (start_graph_dump): Likewise.
979 * tree-pretty-print.c (maybe_init_pretty_print): Likewise. Use
981 * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
982 pretty printer initialization.
983 * coretypes.h (diagnostic_context): Remove superflous type alias
985 (pretty_printer): Likewise. Declare directly as a class.
986 (pretty_print_info): Remove declaration as class.
987 * asan.c (asan_emit_stack_protection): Remove call to pp_construct
988 and pp_clear_output_area.
989 (asan_add_global): Likewise.
991 2013-08-22 Jan Hubicka <jh@suse.cz>
993 * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
994 * ipa-utils.h (update_type_inheritance_graph): Declare.
995 (possible_polymorphic_call_target_p): Declare.
996 (possible_polymorphic_call_target_p): New.
997 * ipa-devirt.c: Update toplevel comments.
998 (cached_polymorphic_call_targets): Move up.
999 (odr_type_d): Move ID down.
1000 (polymorphic_type_binfo_p): Update comment.
1001 (odr_hasher::remove): Likewise;
1002 (get_odr_type): Set anonymous_namespace.
1003 (dump_odr_type): Dump it.
1004 (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
1005 (maybe_record_node): Record node in cached_polymorphic_call_targets.
1006 (record_binfo): Add comment.
1007 (free_polymorphic_call_targets_hash): Do not ICE when cache is not
1009 (devirt_node_removal_hook): Do not iCE when cache is freed.
1010 (possible_polymorphic_call_target_p): New predicate.
1011 (update_type_inheritance_graph): New function.
1013 2013-08-22 Alexander Ivchenko <alexander.ivchenko@intel.com>
1014 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
1015 Sergey Lega <sergey.s.lega@intel.com>
1016 Anna Tikhonova <anna.tikhonova@intel.com>
1017 Ilya Tocar <ilya.tocar@intel.com>
1018 Andrey Turetskiy <andrey.turetskiy@intel.com>
1019 Ilya Verbin <ilya.verbin@intel.com>
1020 Kirill Yukhin <kirill.yukhin@intel.com>
1021 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
1023 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
1024 (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
1025 (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
1026 (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
1027 (OPTION_MASK_ISA_AVX2_UNSET): Update.
1028 (OPTION_MASK_ISA_AVX512F_UNSET): New.
1029 (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
1030 (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
1031 (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
1032 (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
1033 OPT_mavx512pf, OPT_mavx512er cases.
1034 * config/i386/constraints.md (v): New constraint.
1035 (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
1036 * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
1037 (bit_AVX512CD): New.
1038 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1039 AVX512F, AVX512ER, AVX512PF, AVX512CD features.
1040 * config/i386/i386-c.c (ix86_target_macros_internal):
1041 Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
1043 * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
1044 (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
1045 * config/i386/i386.c (regclass_map, dbx_register_map)
1046 (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
1047 (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
1048 (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
1050 (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
1051 PTA_AVX512PF, PTA_AVX512CD. Handle -mavx512f, -mavx512er, -mavx512cd,
1052 -mavx512pf options. Fix formatting.
1053 (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
1054 targets. Squash EVEX_SSE_REGS if AVX512F is disabled.
1055 (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
1056 -mavx512cd, -mavx512pf options.
1057 (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
1058 (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
1059 (ix86_preferred_output_reload_class): Replace SSE_REGS with
1061 (ix86_hard_regno_mode_ok): Support 512-bit registers.
1062 (ix86_set_reg_reg_cost): Ditto.
1063 (x86_order_regs_for_local_alloc): Ditto.
1064 (MAX_VECT_LEN): Extend to 64-byte.
1065 (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
1066 * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
1067 (TARGET_AVX512ER, TARGET_AVX512CD): New.
1068 (BIGGEST_ALIGNMENT): Extend to 512-bits.
1069 (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
1070 (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
1071 (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
1072 (SSE_REG_MODE_P): Support new modes.
1073 (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
1074 (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
1075 (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
1076 (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
1077 (REG_CLASS_CONTENTS): Add new registers.
1078 (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
1079 (EXT_REX_SSE_REGNO_P): New.
1080 (HI_REGISTER_NAMES): Add new registers.
1081 * config/i386/i386.md: Define constants for new registers.
1082 (mode): Add new 512-bit modes.
1083 (prefix): Support evex prefix.
1084 (isa): Support avx512f, noavx512f, fma_avx512f.
1085 (ssemodesuffix): Add new 512-bit modes.
1087 (*movxi_internal_avx512f): Ditto.
1088 (*movdi_internal): Replace constraint "x" with the new constraint "v".
1090 (*movsi_internal): Likewise.
1091 (*movdf_internal): Likewise.
1092 (*movsf_internal): Likewise.
1093 (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
1094 (<code><mode>3): Likewise.
1095 * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
1097 * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
1098 with the new constraint "v".
1099 * config/i386/sse.md (*mov<mode>_internal): Support new registers and
1101 (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
1102 with the new constraint "v".
1103 (<sse2>_loaddqu<avxsizesuffix>): Likewise.
1104 (<sse2>_storedqu<avxsizesuffix>): Likewise.
1105 (*<plusminus_insn><mode>3): Likewise.
1106 (<sse>_vm<plusminus_insn><mode>3): Likewise.
1107 (*mul<mode>3): Likewise.
1108 (<sse>_vmmul<mode>3): Likewise.
1109 (<sse>_div<mode>3): Likewise.
1110 (<sse>_vmdiv<mode>3): Likewise.
1111 (<sse>_sqrt<mode>2): Likewise.
1112 (<sse>_vmsqrt<mode>2): Likewise.
1113 (*<code><mode>3_finite): Likewise.
1114 (*<code><mode>3) <smaxmin>: Likewise.
1115 (<sse>_vm<code><mode>3): Likewise.
1116 (*<code><mode>3) <any_logic>: Likewise.
1117 (*fma_fmadd_<mode>): Likewise.
1118 (*fma_fmsub_<mode>): Likewise.
1119 (*fma_fnmadd_<mode>): Likewise.
1120 (*fma_fnmsub_<mode>): Likewise.
1121 (*fma_fmaddsub_<mode>): Likewise.
1122 (*fma_fmsubadd_<mode>): Likewise.
1123 (*fmai_fmadd_<mode>): Likewise.
1124 (*fmai_fmsub_<mode>): Likewise.
1125 (*fmai_fnmadd_<mode>): Likewise.
1126 (*fmai_fnmsub_<mode>): Likewise.
1127 (sse_cvtsi2ss): Likewise.
1128 (sse_cvtsi2ssq): Likewise.
1129 (sse_cvtss2si): Likewise.
1130 (sse_cvtss2si_2): Likewise.
1131 (sse_cvtss2siq): Likewise.
1132 (sse_cvtss2siq_2): Likewise.
1133 (sse_cvttss2si): Likewise.
1134 (sse_cvtss2siq_2): Likewise.
1135 (float<sseintvecmodelower><mode>2): Likewise.
1136 (sse2_cvtsd2si_2): Likewise.
1137 (sse2_cvtsd2siq_2): Likewise.
1138 (*<plusminus_insn><mode>3): Likewise.
1139 (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
1140 (*<sse4_1_avx2>_mul<mode>3): Likewise.
1141 (ashr<mode>3): Likewise.
1142 (<shift_insn><mode>3): Likewise.
1143 (avx2_<code><mode>3): Likewise.
1144 (*avx2_<code><mode>3): Likewise.
1145 (*andnot<mode>3): Likewise.
1146 (*<code><mode>3) <any_logic>: Likewise.
1147 (abs<mode>2): Likewise.
1148 (avx2_permvar<mode>): Likewise.
1149 (avx2_perm<mode>_1): Likewise.
1150 (*avx_vpermilp<mode>): Likewise.
1151 (avx_vpermilvar<mode>3): Likewise.
1152 (avx2_ashrv<mode>): Likewise.
1153 (avx2_<shift_insn>v<mode>): Likewise.
1154 * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
1156 * doc/rtl.texi: Document XImode.
1158 2013-08-21 Jeff Law <law@redhat.com>
1160 * tree-flow.h (register_jump_thread): Pass vector of edges
1161 instead of each important edge.
1162 * tree-ssa-threadedge.c (thread_across_edge): Build the jump
1163 thread path into a vector and pass that to register_jump_thread.
1164 * tree-ssa-threadupdate.c (register_jump_thread): Conver the
1165 passed in edge vector to the current 3-edge form.
1168 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
1170 * dce.c (fini_dce): Call df_analyze again just in case
1171 delete_unmarked_insns removed anything.
1173 2013-08-21 Joern Rennecke <joern.rennecke@embecosm.com>
1175 * reload.h (struct reg_equivs): Rename to ..
1176 (struct reg_equivs_s): .. this.
1178 2013-08-20 Martin Liska <marxin.liska@gmail.com>
1180 * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
1182 2013-08-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1184 * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
1186 2013-08-21 Jeff Law <law@redhat.com>
1188 * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
1189 simplify assignments too. If the RHS collapses to a singleton
1190 range, then return the value for the range.
1192 2013-08-21 Kirill Yukhin <kirill.yukhin@intel.com>
1194 * config/i386/sse.md (V16): Rename to...
1196 (mov<mode>): Update iterator name.
1197 (*mov<mode>_internal): Ditto.
1198 (push<mode>1): Ditto.
1199 (movmisalign<mode>): Ditto.
1201 2013-08-20 Jan Hubicka <jh@suse.cz>
1204 * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
1205 entry for direct edges.
1206 (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
1208 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1210 Revert my last two changes, r201865 and r201864:
1213 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1215 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1216 instances can own GC refs.
1218 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1219 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1220 (gcc::context::gt_pch_nx): Likewise.
1221 (gcc::context::gt_pch_nx): Likewise.
1222 * ggc.h (gt_ggc_mx <T>): New.
1223 (gt_pch_nx_with_op <T>): New.
1224 (gt_pch_nx <T>): New.
1225 * passes.c (opt_pass::gt_ggc_mx): New.
1226 (opt_pass::gt_pch_nx): New.
1227 (opt_pass::gt_pch_nx_with_op): New.
1228 (pass_manager::gt_ggc_mx): New.
1229 (pass_manager::gt_pch_nx): New.
1230 (pass_manager::gt_pch_nx_with_op): New.
1231 (pass_manager::operator new): Use
1232 ggc_internal_cleared_alloc_stat rather than xcalloc.
1233 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1234 (pass_manager::gt_ggc_mx): New.
1235 (pass_manager::gt_pch_nx): New.
1236 (pass_manager::gt_pch_nx_with_op): New.
1237 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1238 (opt_pass::operator new): New.
1239 (opt_pass::gt_ggc_mx): New.
1240 (opt_pass::gt_pch_nx): New.
1241 (opt_pass::gt_pch_nx_with_op): New.
1244 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1246 * Makefile.in (GTFILES): Add context.h.
1247 * context.c (gcc::context::operator new): New.
1248 (gcc::context::gt_ggc_mx): New.
1249 (gcc::context::gt_pch_nx): New.
1250 (gcc::context::gt_pch_nx): New.
1251 * context.h (gcc::context): Add GTY((user)) marking.
1252 (gcc::context::operator new): New.
1253 (gcc::context::gt_ggc_mx): New.
1254 (gcc::context::gt_pch_nx): New.
1255 (gcc::context::gt_pch_nx): New.
1256 (g): Add GTY marking.
1257 (gt_ggc_mx (gcc::context *)): New.
1258 (gt_pch_nx (gcc::context *)): New.
1259 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1260 void *cookie)): New.
1261 * gengtype.c (open_base_files) <ifiles>: Add context.h.
1263 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
1265 * dce.c (fini_dce): Call df_analyze again just in case
1266 delete_unmarked_insns removed anything.
1268 2013-08-20 Teresa Johnson <tejohnson@google.com>
1270 PR rtl-optimizations/57451
1271 * final.c (reemit_insn_block_notes): Prevent lexical blocks
1272 from crossing split section boundaries.
1274 2013-08-20 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1276 * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
1277 * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
1278 with MULTLIB_DEFAULTS.
1280 2013-08-20 Nick Clifton <nickc@redhat.com>
1282 * target.def (narrow_volatile_bitfield): Note that the default
1283 value is false, not !TARGET_STRICT_ALIGN.
1284 * doc/tm.texi: Regenerate.
1286 2013-08-20 Pavel Chupin <pavel.v.chupin@intel.com>
1288 Fix LIB_SPEC for systems without libpthread.
1290 * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
1291 * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
1293 * config/i386/linux-common.h: Likewise.
1294 * config/mips/linux-common.h: Likewise.
1296 2013-08-20 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
1298 * tree-ssa-ccp.c (get_default_value): Remove redundant condition
1301 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1303 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1304 instances can own GC refs.
1306 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1307 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1308 (gcc::context::gt_pch_nx): Likewise.
1309 (gcc::context::gt_pch_nx): Likewise.
1310 * ggc.h (gt_ggc_mx <T>): New.
1311 (gt_pch_nx_with_op <T>): New.
1312 (gt_pch_nx <T>): New.
1313 * passes.c (opt_pass::gt_ggc_mx): New.
1314 (opt_pass::gt_pch_nx): New.
1315 (opt_pass::gt_pch_nx_with_op): New.
1316 (pass_manager::gt_ggc_mx): New.
1317 (pass_manager::gt_pch_nx): New.
1318 (pass_manager::gt_pch_nx_with_op): New.
1319 (pass_manager::operator new): Use
1320 ggc_internal_cleared_alloc_stat rather than xcalloc.
1321 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1322 (pass_manager::gt_ggc_mx): New.
1323 (pass_manager::gt_pch_nx): New.
1324 (pass_manager::gt_pch_nx_with_op): New.
1325 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1326 (opt_pass::operator new): New.
1327 (opt_pass::gt_ggc_mx): New.
1328 (opt_pass::gt_pch_nx): New.
1329 (opt_pass::gt_pch_nx_with_op): New.
1331 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1333 * Makefile.in (GTFILES): Add context.h.
1334 * context.c (gcc::context::operator new): New.
1335 (gcc::context::gt_ggc_mx): New.
1336 (gcc::context::gt_pch_nx): New.
1337 (gcc::context::gt_pch_nx): New.
1338 * context.h (gcc::context): Add GTY((user)) marking.
1339 (gcc::context::operator new): New.
1340 (gcc::context::gt_ggc_mx): New.
1341 (gcc::context::gt_pch_nx): New.
1342 (gcc::context::gt_pch_nx): New.
1343 (g): Add GTY marking.
1344 (gt_ggc_mx (gcc::context *)): New.
1345 (gt_pch_nx (gcc::context *)): New.
1346 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1347 void *cookie)): New.
1348 * gengtype.c (open_base_files) <ifiles>: Add context.h.
1350 2013-08-20 Alan Modra <amodra@gmail.com>
1353 * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
1354 (rs6000_emit_epilogue): Likewise.
1356 2013-08-19 Dehao Chen <dehao@google.com>
1358 * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
1360 2013-08-19 Peter Bergner <bergner@vnet.ibm.com>
1361 Jakub Jelinek <jakub@redhat.com>
1363 * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
1364 (BUILT_IN_FABSD64): Likewise.
1365 (BUILT_IN_FABSD128): Likewise.
1366 * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
1367 (fold_builtin_1): Likewise.
1368 * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
1369 destination and source operands.
1370 (*abstd2_fpr): Likewise.
1371 (*nabstd2_fpr): Likewise.
1373 2013-08-19 Richard Sandiford <rdsandiford@googlemail.com>
1375 * config/mips/mips.c (mips_adjust_insn_length): Add checks for
1378 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
1380 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
1383 2013-08-19 Alexander Ivchenko <alexander.ivchenko@intel.com>
1385 * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
1386 * builtins.c (default_libc_has_function): New.
1387 (gnu_libc_has_function): Ditto.
1388 (no_c99_libc_has_function): Ditto.
1389 (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
1390 instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
1391 (fold_builtin_sincos): Likewise.
1392 (fold_builtin_cexp): Likewise.
1393 * builtins.def (DEF_C94_BUILTIN): Likewise.
1394 (DEF_C99_BUILTIN): Likewise.
1395 (DEF_C99_C90RES_BUILTIN): Likewise.
1396 (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
1397 definitions to using this define.
1398 * config/darwin-protos.h (darwin_libc_has_function): New.
1399 * config/darwin.c (darwin_libc_has_function): Ditto.
1400 * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
1401 TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
1402 * config/darwin.h: Ditto.
1403 * config/elfos.h: Ditto.
1404 * config/freebsd.h: Ditto.
1405 * config/i386/cygming.h: Ditto.
1406 * config/i386/djgpp.h: Ditto.
1407 * config/i386/i386-interix.h: Ditto.
1408 * config/microblaze/microblaze.h: Ditto.
1409 * config/mmix/mmix.h: Ditto.
1410 * config/gnu-user.h: Ditto.
1411 * config/ia64/hpux.h: Ditto.
1412 * config/pa/pa-hpux.h: Ditto.
1413 * config/pdp11/pdp11.h: Ditto.
1414 * config/picochip/picochip.h: Ditto.
1415 * config/linux.h: Ditto.
1416 * config/netbsd.h: Ditto.
1417 * config/openbsd.h: Ditto.
1418 * config/rs6000/aix43.h: Ditto.
1419 * config/rs6000/aix51.h: Ditto.
1420 * config/rs6000/aix52.h: Ditto.
1421 * config/rs6000/aix53.h: Ditto.
1422 * config/rs6000/aix61.h: Ditto.
1423 * config/rs6000/darwin.h: Ditto.
1424 * config/rs6000/linux.h: Ditto.
1425 * config/rs6000/linux64.h: Ditto.
1426 * config/s390/tpf.h: Ditto.
1427 * config/sol2-10.h: Ditto.
1428 * config/sol2.h: Ditto.
1429 * config/vms/vms.h: Ditto.
1430 * config/vxworks.h: Ditto.
1431 * config/linux-android.c (linux_android_libc_has_function):
1432 New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
1433 * config/linux-protos.h (linux_android_libc_has_function):
1435 * config/i386/i386.c (ix86_libc_has_function): New.
1436 * config/i386/i386-protos.h
1437 (ix86_libc_has_function): New declaration.
1438 * config/i386/i386.md
1439 ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
1440 ("isinf<mode>2): Likewise.
1441 * convert.c (convert_to_integer): Using new target hook
1442 TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
1443 TARGET_C99_FUNCTIONS.
1444 * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
1445 * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
1446 * coretypes.h (function_class): New enum for different
1447 classes of functions.
1448 * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
1449 * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
1450 (TARGET_HAS_SINCOS): Likewise.
1451 (TARGET_LIBC_HAS_FUNCTION): New.
1452 * doc/tm.texi: Regenerated.
1453 * targhooks.h (default_libc_has_function): New declaration.
1454 (no_c99_libc_has_function): Ditto.
1455 (gnu_libc_has_function): Ditto.
1456 * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
1457 and TARGET_HAS_SINCOS.
1459 2013-08-18 Jan Hubicka <jh@suse.cz>
1461 * Makeifle-in (ipa-devirt.o): New.
1462 (GTFILES): Add ipa-utils.h and ipa-devirt.c
1463 * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
1464 (analyze_functions): Look into possible targets of polymorphic call.
1465 * dumpfile.c (dump_files): Add type-inheritance dump.
1466 * dumpfile.h (TDI_inheritance): New.
1467 * ipa-devirt.c: New file.
1468 * ipa-utils.h (odr_type_d): Forward declare.
1469 (odr_type): New type.
1470 (build_type_inheritance_graph): Declare.
1471 (possible_polymorphic_call_targets): Declare and introduce inline
1472 variant when only edge is pased.
1473 (dump_possible_polymorphic_call_targets): Likewise.
1474 * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
1475 * tree.c (type_in_anonymous_namespace_p): Break out from ...
1476 (types_same_for_odr): ... here.
1477 * tree.h (type_in_anonymous_namespace_p): Declare.
1479 2013-08-18 Jakub Jelinek <jakub@redhat.com>
1481 PR tree-optimization/58006
1482 * tree-parloops.c (take_address_of): Don't ICE if get_name
1484 (eliminate_local_variables_stmt): Remove clobber stmts.
1486 2013-08-18 Eric Botcazou <ebotcazou@adacore.com>
1488 * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
1489 error message is issued for an alias to undefined symbol.
1491 2013-08-18 Jan Hubicka <jh@suse.cz>
1493 * cgraph.c (cgraph_create_indirect_edge): Discover
1494 polymorphic calls and record basic info into indirect_info.
1495 * gimple-fold.c (gimple_fold_call): When doing BINFO based
1496 devirtualization, ignore objc function calls.
1497 * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
1498 call with no parm index info.
1499 * ipa-prop.c (ipa_analyze_call_uses): Likewise.
1500 * tree.c (virtual_method_call_p): New function.
1501 * tree.h (virtual_method_call_p): Declare.
1503 2013-08-16 Jan Hubicka <jh@suse.cz>
1506 * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
1508 2013-08-16 David Edelsohn <dje.gcc@gmail.com>
1510 * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
1513 2013-08-16 David Malcolm <dmalcolm@redhat.com>
1515 * gengtype.c (type_for_name): Add special-case support for
1516 locating types within the "gcc::" namespace.
1517 (open_base_files): Emit a "using namespace gcc" directive.
1519 2013-08-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1522 * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1523 memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1525 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1526 array instead of each individual operand as a separate argument.
1527 (emit_fusion_gpr_load): Likewise.
1528 (expand_fusion_gpr_load): Add new function declaration.
1530 * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1531 signature to have the operands passed as an array, instead of as
1532 separate arguments. Allow ZERO_EXTEND to be in the memory
1533 address, and also SIGN_EXTEND if -mpower8-fusion-sign. Do not
1534 depend on the register live/dead flags when peepholes are run.
1535 (expand_fusion_gpr_load): New function to be called from the
1536 peephole2 pass, to change the register that addis sets to be the
1538 (emit_fusion_gpr_load): Change the calling signature to have the
1539 operands passed as an array, instead of as separate arguments.
1540 Allow ZERO_EXTEND to be in the memory address, and also
1541 SIGN_EXTEND if -mpower8-fusion-sign.
1543 * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1545 (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1546 adjust the register addis loads up in the peephole2 pass. Do not
1547 depend on the register live/dead state when the peephole pass is done.
1549 2013-08-16 David Malcolm <dmalcolm@redhat.com>
1551 * gengtype.c (create_user_defined_type): Ensure that the kind
1552 is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1553 declaration is seen before the GTY((user)) marking.
1555 2013-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
1558 * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1560 2013-08-16 Jan Hubicka <jh@suse.cz>
1562 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1563 arugment expected_type.
1564 (gimple_fold_call): Use it.
1565 * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1566 * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1567 * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1568 (try_make_edge_direct_virtual_call): Likewise.
1569 * tree.c (obj_type_ref_class): New.
1570 * tree.h (obj_type_ref_class): Use it.
1572 2013-08-16 Gabriel Dos Reis <gdr@integrable-solutions.net>
1574 * sched-vis.c (rtl_slim_pp_initialized): Remove.
1575 (rtl_slim_pp): Likewise.
1576 (init_rtl_slim_pretty_print): Likewise.
1577 (dump_value_slim): Don't call it. Use local pretty printer.
1578 (dump_insn_slim): Likewise.
1579 (dump_rtl_slim): Likewise.
1580 (str_pattern_slim): Likewise.
1581 * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1584 2013-08-16 Jakub Jelinek <jakub@redhat.com>
1586 PR tree-optimization/58164
1587 * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1588 walk gimple_goto_dest of GIMPLE_GOTO.
1590 PR tree-optimization/58165
1591 * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1592 bi_call must be the last stmt in a bb, don't split_block, instead
1593 use fallthru edge from it and give up if there is none.
1594 Release conds vector when returning early.
1596 2013-08-14 Xinliang David Li <davidxl@google.com>
1598 * config/i386/i386.c (ix86_option_override_internal):
1599 Remove unused variable and field.
1601 2013-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1604 * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1605 * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1606 * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1607 and Linux, correct BLKmode alignment when 128-bit alignment is
1608 required and compatibility flag is not set.
1609 (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1610 for zero-size arguments when compatibility flag is not set.
1612 2013-08-14 Jakub Jelinek <jakub@redhat.com>
1614 PR tree-optimization/58145
1615 * tree-sra.c (build_ref_for_offset): If prev_base has
1616 TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1618 2013-08-14 Xinliang David Li <davidxl@google.com>
1620 * config/i386/i386.c (ix86_option_override_internal):
1621 Fix uninitialized variable error.
1623 2013-08-14 Xinliang David Li <davidxl@google.com>
1625 * config/i386/i386.opt: Define two new options.
1626 * config/i386/x86-tune.def: Add arch selector field in macros.
1627 * config/i386/i386.h: Adjust macro definition.
1628 * config/i386/i386.c (ix86_option_override_internal):
1630 (parse_mtune_ctrl_str): New function.
1631 (set_ix86_tune_features): New function.
1632 (ix86_function_specific_restore): Call the new helper function.
1634 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
1636 PR rtl-optimization/57662
1637 * sel-sched.c (code_motion_process_successors): When the current insn
1638 is removed after the recursive traversal, break from the loop.
1639 Add comments and debug printouts.
1641 2013-08-14 Jakub Jelinek <jakub@redhat.com>
1642 Alexandre Oliva <aoliva@redhat.com>
1645 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1646 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1647 there also UNSPEC_PLTOFF.
1649 2013-08-14 Marek Polacek <polacek@redhat.com>
1651 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1652 NUM_CONDITIONS bit positions.
1654 2013-08-13 Cary Coutant <ccoutant@google.com>
1656 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1657 (attr_checksum): Hash vector contents instead of pointer.
1658 (attr_checksum_ordered): Likewise.
1660 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
1662 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1663 when Pmode != word_mode. Add length_address attribute.
1664 (sse3_monitor_<mode>): Merge from sse3_monitor and
1665 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
1666 Pmode != word_mode. Update insn length attribute.
1667 * config/i386/i386.c (ix86_option_override_internal): Update
1668 ix86_gen_monitor selection for merged sse3_monitor insn.
1670 2013-08-13 Julian Brown <julian@codesourcery.com>
1672 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1673 perform invalid legitimization on greater-than-word-size modes for
1676 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
1678 * ira.c (setup_class_translate_array): Use aclass instead of cl
1679 for classes not fully covered by allocno classes.
1681 2013-08-13 Jakub Jelinek <jakub@redhat.com>
1683 PR tree-optimization/57661
1684 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1685 * tree-inline.c (tree_function_versioning): Initialize it.
1686 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1687 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1688 that is not being copied.
1691 * asan.c (instrument_strlen_call): Fix typo in comment.
1692 Use char * type even for the lhs of POINTER_PLUS_EXPR.
1694 2013-08-13 Steve Ellcey <sellcey@mips.com>
1696 * config/mips/mips.md (prefetch): Use lw instead of ld on
1697 loongson in 32bit mode.
1699 2013-08-13 Nick Clifton <nickc@redhat.com>
1701 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1703 2013-08-13 Jan Hubicka <jh@suse.cz>
1705 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1706 introduced edge; fix typo in sanity check.
1707 (cgraph_resolve_speculation): Export; improve diagnostic.
1708 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1709 speculation at type mismatch.
1710 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1711 (cgraph_resolve_speculation): Declare.
1712 (symtab_can_be_discarded): New function.
1713 * value-prof.c (gimple_ic_transform): Remove actual transform code.
1714 * ipa-inline-transform.c (speculation_removed): New global var.
1715 (clone_inlined_nodes): See if speculation can be removed.
1716 (inline_call): If speculations was removed, we growths may not match.
1717 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1718 (speculation_useful_p): New function.
1719 (resolve_noninline_speculation): New function.
1720 (inline_small_functions): Resolve useless speculations.
1721 * ipa-inline.h (speculation_useful_p): Declare
1722 * ipa.c (can_replace_by_local_alias): Simplify.
1723 (ipa_profile): Produce speculative calls in non-lto, too;
1724 add simple cost model; produce local aliases.
1726 2013-08-13 David Malcolm <dmalcolm@redhat.com>
1728 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1731 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
1733 * config/i386/i386.c (ix86_function_versions): Use error + inform.
1735 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
1737 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1738 iterator instead of X87MODEF.
1740 2013-08-12 Perez Read <netfirewall@gmail.com>
1743 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
1744 operand 0 for intel asm alternative.
1745 (*movabs<mode>_2): Ditto for operand 1.
1747 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
1749 * config/aarch64/arm_none.h
1750 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1752 2013-08-12 Nick Clifton <nickc@redhat.com>
1754 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1756 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
1758 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1759 expand for QI/HImode operand to produce more effictive code for
1760 unsigned char(short) --> float(double) conversion.
1762 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
1764 * doc/invoke.texi: Mention that -ftls-model does not force the final
1767 2013-08-12 Marek Polacek <polacek@redhat.com>
1768 Marc Glisse <marc.glisse@inria.fr>
1770 PR tree-optimization/57980
1771 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1772 when creating -1 constant.
1774 2013-08-10 Jan Hubicka <jh@suse.cz>
1776 Workaround binutils PR14342.
1777 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1778 (gimple_init_edge_profiler): Likewise.
1779 (gimple_gen_ic_func_profiler): Likewise.
1781 2013-08-09 Jan Hubicka <jh@suse.cz>
1783 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1785 2013-08-09 Xinliang David Li <davidxl@google.com>
1787 * config/i386/stringop.def: New file.
1788 * config/i386/stringop.opt: New file.
1789 * config/i386/i386-opts.h: Include stringopt.def.
1790 * config/i386/i386.opt: Include stringopt.opt.
1791 * config/i386/i386.c (ix86_option_override_internal):
1792 Override default size based stringop inline strategies with options.
1793 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1796 2013-08-09 Jan Hubicka <jh@suse.cz>
1798 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1800 2013-08-09 Jan Hubicka <jh@suse.cz>
1802 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1804 (dump_cgraph_node): Dump profile-id.
1805 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1806 and common_target_probability.
1807 * lto-cgraph.c (lto_output_edge): Stream common targets.
1808 (lto_output_node): Stream profile ids.
1809 (input_node): Stream profile ids.
1810 (input_edge): Stream common targets.
1811 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1812 * ipa.c: Include value-prof.h
1813 (ipa_profile_generate_summary): Turn indirect call statement histograms
1814 into common targets.
1815 (ipa_profile): Turn common targets into speculative edges.
1817 2013-08-09 Jan Hubicka <jh@suse.cz>
1819 * cgraph.h (cgraph_node): Add profile_id.
1820 * value-prof.c (cgraph_node_map): Turn into pointer_map.
1821 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1822 (del_node_map): Update.
1823 (find_func_by_funcdef_no): Replace by ...
1824 (find_func_by_profile_id): ... this one.
1825 (gimple_ic_transform): Do not remove useful histograms when
1826 speculation is not done; dump info when indirect call removal
1828 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1829 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1830 (__gcov_indirect_call_profiler_v2): .. this one.
1831 * profile.h (init_node_map): Update.
1832 * coverage.c (coverage_compute_profile_id): New function.
1833 * coverage.h (coverage_compute_profile_id): Declare.
1834 * tree-profile.c (init_ic_make_global_vars): Make
1835 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
1836 (gimple_init_edge_profiler): Update prototype of
1837 __gcov_indirect_call_profiler.
1838 (gimple_gen_ic_func_profiler): Simplify.
1839 (tree_profiling): Use init_node_map
1841 2013-08-09 Jan Hubicka <jh@suse.cz>
1843 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1844 non-speculative refs.
1845 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1846 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1847 (cgraph_set_call_stmt): Likewise.
1848 (cgraph_create_edge_1): Fix release checking compilatoin;
1850 (cgraph_free_edge): Free indirect info.
1851 (cgraph_turn_edge_to_speculative): New function.
1852 (cgraph_speculative_call_info): New function.
1853 (cgraph_make_edge_direct): Return direct edge; handle speculation.
1854 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1855 (dump_cgraph_node): Dump speculation.
1856 (verify_edge_count_and_frequency): Accept speculative edges.
1857 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1858 (verify_cgraph_node): Handle speculation.
1859 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1860 (cgraph_set_call_stmt): Update prototype.
1861 (cgraph_make_edge_direct): Update prototype.
1862 (cgraph_speculative_call_info): Declare.
1863 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1864 to change; update call of ipa_find_references.
1865 * ipa-ref.c (ipa_record_reference): Fix return value; clear
1866 lto_stmt_uid and speculative flags.
1867 (ipa_dump_references): Dump speculation.
1868 (ipa_clone_references): Clone speculative flag.
1869 (ipa_clone_referring): Likewise.
1870 (ipa_clone_ref): New function.
1871 (ipa_find_reference): Look into lto_stmt_uids
1872 (ipa_clear_stmts_in_references): Do not clear speculative calls.
1873 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1874 (ipa_find_reference): Update declaration.
1875 (ipa_clone_ref): Declare.
1876 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1877 stream speculative flag.
1878 (lto_output_ref): Stream statements uids and speculation.
1879 (input_ref): Likewise.
1880 (input_edge): Stream speuclation.
1881 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1882 (cgraph_set_call_stmt_including_clones): Handle speculation.
1883 * ipa-inline.c (heap_edge_removal_hook): New function.
1884 (inline_small_functions): Register it.
1885 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1886 also initialize refs.
1887 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1889 (try_make_edge_direct_simple_call): Likewise.
1890 (try_make_edge_direct_simple_call): Likewise.
1891 (update_indirect_edges_after_inlining): Likewise.
1892 (remove_described_reference): Look proper lto_stmt_uid.
1893 (propagate_controlled_uses): Likewise.
1894 (propagate_controlled_uses): Liekwise.
1895 * tree-inline.c (copy_bb): Copy speculative edges.
1896 (redirect_all_calls): New function.
1897 (copy_cfg_body): Do redirection after loop info is updated.
1898 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1900 2013-08-09 Jan Hubicka <jh@suse.cz>
1902 * lto-streamer-out.c (output_function): Renumber PHIs.
1903 * lto-streamer-in.c (input_function): Likewise.
1905 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
1907 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1908 (get_lane_unsigned): Likewise.
1909 (dup_lane_scalar): Likewise.
1910 (get_lane): enable for VALL.
1911 * config/aarch64/aarch64-simd.md
1912 (aarch64_dup_lane_scalar<mode>): Remove.
1913 (aarch64_get_lane_signed<mode>): Likewise.
1914 (aarch64_get_lane_unsigned<mode>): Likewise.
1915 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1916 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1917 (aarch64_get_lane<mode>): Enable for all vector modes.
1918 (aarch64_get_lanedi): Remove misleading constraints.
1919 * config/aarch64/arm_neon.h
1920 (__aarch64_vget_lane_any): Define.
1921 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1922 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1923 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1924 * config/aarch64/iterators.md (VDQQH): New.
1928 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
1930 * configure.ac: Add GAS check for LEON instructions on SPARC.
1931 * configure: Regenerate.
1932 * config.in: Likewise.
1933 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1935 * config/sparc/sparc.opt (LEON, LEON3): New masks.
1936 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1938 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1939 (AS_LEON_FLAG): New macro.
1940 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1941 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1942 Deal with LEON and LEON3 for the memory model.
1943 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1944 (atomic_compare_and_swap<mode>_1): Likewise.
1945 (*atomic_compare_and_swap<mode>_1): Likewise.
1947 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1949 * config/arm/neon.md (vcond): Fix floating-point vector
1950 comparisons against 0.
1952 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
1954 * lra-constraints.c (emit_spill_move): Remove assert.
1955 (process_alt_operands): Add more debugging
1956 output. Increase reject for spilling into memory. Decrease
1957 reject for reloading scratch.
1958 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1960 2013-08-08 Steve Ellcey <sellcey@mips.com>
1962 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1963 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1964 micromips incompatible. Add nan2008.
1965 (MULTILIB_DIRNAMES): Add nan2008.
1966 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1967 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1968 and micromips incompatible. Add nan2008.
1969 (MULTILIB_DIRNAMES): Add nan2008.
1970 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1972 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
1974 PR rtl-optimization/58079
1975 * combine.c (combine_simplify_rtx): Avoid using SUBST if
1976 simplify_comparison has widened a comparison with an integer.
1978 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1980 * config/arm/neon.md (movmisalign<mode>): Disable when we
1981 don't allow unaligned accesses.
1982 (*movmisalign<mode>_neon_store): Likewise.
1983 (*movmisalign<mode>_neon_load): Likewise.
1984 (*movmisalign<mode>_neon_store): Likewise.
1985 (*movmisalign<mode>_neon_load): Likewise.
1987 2013-08-08 Jan Hubicka <jh@suse.cz>
1989 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1990 (make_pass_rebuild_cgraph_edges): Also clear references.
1991 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1992 * ipa-inline-transform.c (inline_transform): Remove all references
1994 * cgraphunit.c (expand_function): Remove all references after
1996 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1997 (ipa_find_reference): Rewrite to iterator.
1998 (remove_stmt_references): Likewise.
1999 (ipa_clear_stmts_in_references): New function.
2000 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
2001 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
2003 * ipa-split.c (split_function): Remove references in split function.
2005 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
2008 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
2009 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
2011 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
2014 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
2015 suggested mode for the assignment isn't compatible with the
2018 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2021 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
2023 2013-08-07 Xinliang David Li <davidxl@google.com>
2025 * config/i386/i386.opt: New option -mtune-ctrl=.
2026 * config/i386/x86-tune.def: New file.
2027 * config/i386/i386.h: include x86-tune.def.
2028 * config/i386/i386.c (ix86_option_override_internal):
2029 Parsing -mtune-ctrl= option and set tune features.
2031 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
2034 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
2035 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
2038 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
2040 * rtl.h (update_alignments): Declare.
2041 * final.c (grow_label_align): New function extracted from...
2042 (shorten_branches): ...here. Call it.
2043 (update_alignments): New function.
2044 * reorg.c (sibling_labels): New variable.
2045 (get_label_before): Add SIBLING parameter. If it is non-zero, push
2046 the new label along with it onto the sibling_labels vector.
2047 (fill_simple_delay_slots): Adjust call to get_label_before.
2048 (fill_slots_from_thread): Likewise.
2049 (relax_delay_slots): Likewise.
2050 (make_return_insns): Likewise.
2051 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
2053 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
2055 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
2056 document its semantics.
2057 (diagnostic_report_diagnostic): Adjust accordingly.
2059 2013-08-07 David Malcolm <dmalcolm@redhat.com>
2061 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
2062 (sparc_option_override): ...and port to new C++ pass API.
2063 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
2065 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
2067 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
2069 2013-08-06 Caroline Tice <cmtice@google.com>
2071 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
2072 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
2073 * tree-pass.h: Add pass_vtable_verify.
2074 * varasm.c (assemble_variable): Add code to properly set the comdat
2075 section and name for the .vtable_map_vars section.
2076 (assemble_vtyv_preinit_initializer): New function.
2077 (default_sectin_type_flags): Make sure .vtable_map_vars section has
2079 * output.h: Add function decl for assemble_vtv_preinit_initializer.
2080 * vtable-verify.c: New file.
2081 * vtable-verify.h: New file.
2082 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
2083 initialiation levels.
2084 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
2085 * passes.def: Insert pass_vtable_verify.
2086 * aclocal.m4: Reorder includes.
2087 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
2088 -fvtv-counts options.
2089 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
2090 as appropriate, if -fvtable-verify=... is used.
2091 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
2092 -fvtable-verify=... is used.
2093 * Makefile.in (OBJS): Add vtable-verify.o to list.
2094 (vtable-verify.o): Add new build rule.
2095 (GTFILES): Add vtable-verify.c to list.
2096 * common.opt (fvtable-verify=): New flag.
2097 (vtv_priority): Values for fvtable-verify= flag.
2098 (fvtv-counts): New flag.
2099 (fvtv-debug): New flag.
2100 * tree.h (save_vtable_map_decl): New extern function decl.
2102 2013-08-07 David Malcolm <dmalcolm@redhat.com>
2104 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
2105 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
2106 (pass_data_rl78_devirt): ...new pass_data instance and...
2107 (make_pass_rl78_devirt): ...new function.
2108 (rl78_asm_file_start): Port pass registration to new C++ API.
2110 2013-08-07 David Malcolm <dmalcolm@redhat.com>
2112 * coretypes.h (rtl_opt_pass): Add.
2113 (gcc::context): Add.
2114 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
2115 (epiphany_init): Port to new C++ pass API.
2116 (epiphany_optimize_mode_switching): Likewise.
2117 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
2118 (pass_manager::get_pass_mode_switching): New.
2119 (pass_manager::get_pass_peephole2): New.
2120 * mode-switching.c (pass_mode_switching): Add clone method.
2121 * recog.c (pass_peephole2): Add clone method.
2122 (pass_split_all_insns): Add clone method.
2124 2013-08-06 David Malcolm <dmalcolm@redhat.com>
2126 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
2127 (mips_option_override): ...here, porting to new C++ API for passes.
2129 2013-08-06 Jan Hubicka <jh@suse.cz>
2131 * cgraph.c (cgraph_get_body): New function based on lto.c
2133 * cgraph.h (cgraph_get_body): Declare.
2134 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
2136 * cgraphunit.c (expand_function): Get body prior expanding.
2137 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
2138 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
2140 * passes.c (do_per_function_toporder): Get body.
2141 * tree-inline.c (expand_call_inline): Get body prior inlining it.
2142 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
2144 2013-08-06 Martin Jambor <mjambor@suse.cz>
2147 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
2148 is not re-finalized. Rename second parameter to no_collect.
2150 2013-08-06 Martin Jambor <mjambor@suse.cz>
2153 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
2154 MEM_REF has proper alignment information.
2156 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
2159 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
2161 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
2162 rtx (*) (rtx, ...) with insn_gen_fn.
2163 * genoutput.c (output_insn_data): Cast gen_? function pointers to
2164 insn_gen_fn::stored_funcptr. Add initializer braces.
2166 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2168 Rewrite how instances of passes are cloned to remove assumptions
2169 about their sizes (thus allowing pass subclasses to have
2170 additional data fields, albeit non-GC-managed ones at this point).
2172 * passes.c (make_pass_instance): Now that passes have clone
2173 methods, rewrite this function to eliminate XNEW and memcpy
2174 calls that used hardcoded sizes. Since this function no longer
2175 creates pass instances, rename it to...
2176 (add_pass_instance): ...this. Document the old way that passes were
2177 numbered and flagged, and rework this function to continue using it.
2178 (next_pass_1): Add an initial_pass argument for use by
2180 (position_pass): When adding multiple instances of a pass, use
2181 the pass's clone method, rather than relying on the XNEW/memcpy
2182 within the former make_pass_instance (now add_pass_instance).
2183 (pass_manager::pass_manager): When invoking next_pass_1, also supply
2184 the initial instance of the current pass within the pass manager.
2186 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2188 This is the automated part of the conversion of passes from C
2189 structs to C++ classes.
2191 Patch autogenerated by refactor_passes.py from
2192 https://github.com/davidmalcolm/gcc-refactoring-scripts
2193 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
2195 * asan.c (pass_asan): Convert from a global struct to a subclass of
2196 gimple_opt_pass along with...
2197 (pass_data_asan): ...new pass_data instance and...
2198 (make_pass_asan): ...new function.
2199 (pass_asan_O0): Convert from a global struct to a subclass of
2200 gimple_opt_pass along with...
2201 (pass_data_asan_O0): ...new pass_data instance and...
2202 (make_pass_asan_O0): ...new function.
2203 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
2204 subclass of rtl_opt_pass along with...
2205 (pass_data_inc_dec): ...new pass_data instance and...
2206 (make_pass_inc_dec): ...new function.
2207 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
2208 a subclass of rtl_opt_pass along with...
2209 (pass_data_reorder_blocks): ...new pass_data instance and...
2210 (make_pass_reorder_blocks): ...new function.
2211 (pass_duplicate_computed_gotos): Convert from a global struct to a
2212 subclass of rtl_opt_pass along with...
2213 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
2214 (make_pass_duplicate_computed_gotos): ...new function.
2215 (pass_partition_blocks): Convert from a global struct to a subclass of
2216 rtl_opt_pass along with...
2217 (pass_data_partition_blocks): ...new pass_data instance and...
2218 (make_pass_partition_blocks): ...new function.
2219 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
2220 struct to a subclass of rtl_opt_pass along with...
2221 (pass_data_branch_target_load_optimize1): ...new pass_data instance
2223 (make_pass_branch_target_load_optimize1): ...new function.
2224 (pass_branch_target_load_optimize2): Convert from a global struct to a
2225 subclass of rtl_opt_pass along with...
2226 (pass_data_branch_target_load_optimize2): ...new pass_data instance
2228 (make_pass_branch_target_load_optimize2): ...new function.
2229 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
2230 of rtl_opt_pass along with...
2231 (pass_data_jump): ...new pass_data instance and...
2232 (make_pass_jump): ...new function.
2233 (pass_jump2): Convert from a global struct to a subclass of
2234 rtl_opt_pass along with...
2235 (pass_data_jump2): ...new pass_data instance and...
2236 (make_pass_jump2): ...new function.
2237 * cfgexpand.c (pass_expand): Convert from a global struct to a
2238 subclass of rtl_opt_pass along with...
2239 (pass_data_expand): ...new pass_data instance and...
2240 (make_pass_expand): ...new function.
2241 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
2242 of rtl_opt_pass along with...
2243 (pass_data_free_cfg): ...new pass_data instance and...
2244 (make_pass_free_cfg): ...new function.
2245 (pass_into_cfg_layout_mode): Convert from a global struct to a
2246 subclass of rtl_opt_pass along with...
2247 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
2248 (make_pass_into_cfg_layout_mode): ...new function.
2249 (pass_outof_cfg_layout_mode): Convert from a global struct to a
2250 subclass of rtl_opt_pass along with...
2251 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
2252 (make_pass_outof_cfg_layout_mode): ...new function.
2253 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
2254 struct to a subclass of gimple_opt_pass along with...
2255 (pass_data_build_cgraph_edges): ...new pass_data instance and...
2256 (make_pass_build_cgraph_edges): ...new function.
2257 (pass_rebuild_cgraph_edges): Convert from a global struct to a
2258 subclass of gimple_opt_pass along with...
2259 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
2260 (make_pass_rebuild_cgraph_edges): ...new function.
2261 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
2262 subclass of gimple_opt_pass along with...
2263 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
2265 (make_pass_remove_cgraph_callee_edges): ...new function.
2266 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
2267 struct to a subclass of rtl_opt_pass along with...
2268 (pass_data_stack_adjustments): ...new pass_data instance and...
2269 (make_pass_stack_adjustments): ...new function.
2270 * combine.c (pass_combine): Convert from a global struct to a subclass
2271 of rtl_opt_pass along with...
2272 (pass_data_combine): ...new pass_data instance and...
2273 (make_pass_combine): ...new function.
2274 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
2275 global struct to a subclass of rtl_opt_pass along with...
2276 (pass_data_compare_elim_after_reload): ...new pass_data instance
2278 (make_pass_compare_elim_after_reload): ...new function.
2279 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
2280 of rtl_opt_pass along with...
2281 (pass_data_rtl_cprop): ...new pass_data instance and...
2282 (make_pass_rtl_cprop): ...new function.
2283 * cse.c (pass_cse): Convert from a global struct to a subclass of
2284 rtl_opt_pass along with...
2285 (pass_data_cse): ...new pass_data instance and...
2286 (make_pass_cse): ...new function.
2287 (pass_cse2): Convert from a global struct to a subclass of
2288 rtl_opt_pass along with...
2289 (pass_data_cse2): ...new pass_data instance and...
2290 (make_pass_cse2): ...new function.
2291 (pass_cse_after_global_opts): Convert from a global struct to a
2292 subclass of rtl_opt_pass along with...
2293 (pass_data_cse_after_global_opts): ...new pass_data instance and...
2294 (make_pass_cse_after_global_opts): ...new function.
2295 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
2296 of rtl_opt_pass along with...
2297 (pass_data_ud_rtl_dce): ...new pass_data instance and...
2298 (make_pass_ud_rtl_dce): ...new function.
2299 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
2300 rtl_opt_pass along with...
2301 (pass_data_fast_rtl_dce): ...new pass_data instance and...
2302 (make_pass_fast_rtl_dce): ...new function.
2303 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
2304 a subclass of rtl_opt_pass along with...
2305 (pass_data_df_initialize_opt): ...new pass_data instance and...
2306 (make_pass_df_initialize_opt): ...new function.
2307 (pass_df_initialize_no_opt): Convert from a global struct to a
2308 subclass of rtl_opt_pass along with...
2309 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
2310 (make_pass_df_initialize_no_opt): ...new function.
2311 (pass_df_finish): Convert from a global struct to a subclass of
2312 rtl_opt_pass along with...
2313 (pass_data_df_finish): ...new pass_data instance and...
2314 (make_pass_df_finish): ...new function.
2315 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
2316 rtl_opt_pass along with...
2317 (pass_data_rtl_dse1): ...new pass_data instance and...
2318 (make_pass_rtl_dse1): ...new function.
2319 (pass_rtl_dse2): Convert from a global struct to a subclass of
2320 rtl_opt_pass along with...
2321 (pass_data_rtl_dse2): ...new pass_data instance and...
2322 (make_pass_rtl_dse2): ...new function.
2323 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
2324 subclass of rtl_opt_pass along with...
2325 (pass_data_dwarf2_frame): ...new pass_data instance and...
2326 (make_pass_dwarf2_frame): ...new function.
2327 * except.c (pass_set_nothrow_function_flags): Convert from a global
2328 struct to a subclass of rtl_opt_pass along with...
2329 (pass_data_set_nothrow_function_flags): ...new pass_data instance
2331 (make_pass_set_nothrow_function_flags): ...new function.
2332 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
2333 subclass of rtl_opt_pass along with...
2334 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
2336 (make_pass_convert_to_eh_region_ranges): ...new function.
2337 * final.c (pass_compute_alignments): Convert from a global struct to a
2338 subclass of rtl_opt_pass along with...
2339 (pass_data_compute_alignments): ...new pass_data instance and...
2340 (make_pass_compute_alignments): ...new function.
2341 (pass_final): Convert from a global struct to a subclass of
2342 rtl_opt_pass along with...
2343 (pass_data_final): ...new pass_data instance and...
2344 (make_pass_final): ...new function.
2345 (pass_shorten_branches): Convert from a global struct to a subclass of
2346 rtl_opt_pass along with...
2347 (pass_data_shorten_branches): ...new pass_data instance and...
2348 (make_pass_shorten_branches): ...new function.
2349 (pass_clean_state): Convert from a global struct to a subclass of
2350 rtl_opt_pass along with...
2351 (pass_data_clean_state): ...new pass_data instance and...
2352 (make_pass_clean_state): ...new function.
2353 * function.c (pass_instantiate_virtual_regs): Convert from a global
2354 struct to a subclass of rtl_opt_pass along with...
2355 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
2356 (make_pass_instantiate_virtual_regs): ...new function.
2357 (pass_leaf_regs): Convert from a global struct to a subclass of
2358 rtl_opt_pass along with...
2359 (pass_data_leaf_regs): ...new pass_data instance and...
2360 (make_pass_leaf_regs): ...new function.
2361 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
2362 subclass of rtl_opt_pass along with...
2363 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
2365 (make_pass_thread_prologue_and_epilogue): ...new function.
2366 (pass_match_asm_constraints): Convert from a global struct to a
2367 subclass of rtl_opt_pass along with...
2368 (pass_data_match_asm_constraints): ...new pass_data instance and...
2369 (make_pass_match_asm_constraints): ...new function.
2370 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
2371 subclass of rtl_opt_pass along with...
2372 (pass_data_rtl_fwprop): ...new pass_data instance and...
2373 (make_pass_rtl_fwprop): ...new function.
2374 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
2375 rtl_opt_pass along with...
2376 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
2377 (make_pass_rtl_fwprop_addr): ...new function.
2378 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
2379 rtl_opt_pass along with...
2380 (pass_data_rtl_pre): ...new pass_data instance and...
2381 (make_pass_rtl_pre): ...new function.
2382 (pass_rtl_hoist): Convert from a global struct to a subclass of
2383 rtl_opt_pass along with...
2384 (pass_data_rtl_hoist): ...new pass_data instance and...
2385 (make_pass_rtl_hoist): ...new function.
2386 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
2387 subclass of gimple_opt_pass along with...
2388 (pass_data_lower_cf): ...new pass_data instance and...
2389 (make_pass_lower_cf): ...new function.
2390 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
2391 from a global struct to a subclass of gimple_opt_pass along with...
2392 (pass_data_strength_reduction): ...new pass_data instance and...
2393 (make_pass_strength_reduction): ...new function.
2394 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
2395 of rtl_opt_pass along with...
2396 (pass_data_rtl_ifcvt): ...new pass_data instance and...
2397 (make_pass_rtl_ifcvt): ...new function.
2398 (pass_if_after_combine): Convert from a global struct to a subclass of
2399 rtl_opt_pass along with...
2400 (pass_data_if_after_combine): ...new pass_data instance and...
2401 (make_pass_if_after_combine): ...new function.
2402 (pass_if_after_reload): Convert from a global struct to a subclass of
2403 rtl_opt_pass along with...
2404 (pass_data_if_after_reload): ...new pass_data instance and...
2405 (make_pass_if_after_reload): ...new function.
2406 * init-regs.c (pass_initialize_regs): Convert from a global struct to
2407 a subclass of rtl_opt_pass along with...
2408 (pass_data_initialize_regs): ...new pass_data instance and...
2409 (make_pass_initialize_regs): ...new function.
2410 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
2411 of ipa_opt_pass_d along with...
2412 (pass_data_ipa_cp): ...new pass_data instance and...
2413 (make_pass_ipa_cp): ...new function.
2414 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
2415 global struct to a subclass of gimple_opt_pass along with...
2416 (pass_data_inline_parameters): ...new pass_data instance and...
2417 (make_pass_inline_parameters): ...new function.
2418 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
2419 subclass of gimple_opt_pass along with...
2420 (pass_data_early_inline): ...new pass_data instance and...
2421 (make_pass_early_inline): ...new function.
2422 (pass_ipa_inline): Convert from a global struct to a subclass of
2423 ipa_opt_pass_d along with...
2424 (pass_data_ipa_inline): ...new pass_data instance and...
2425 (make_pass_ipa_inline): ...new function.
2426 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
2427 struct to a subclass of gimple_opt_pass along with...
2428 (pass_data_local_pure_const): ...new pass_data instance and...
2429 (make_pass_local_pure_const): ...new function.
2430 (pass_ipa_pure_const): Convert from a global struct to a subclass of
2431 ipa_opt_pass_d along with...
2432 (pass_data_ipa_pure_const): ...new pass_data instance and...
2433 (make_pass_ipa_pure_const): ...new function.
2434 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
2435 to a subclass of ipa_opt_pass_d along with...
2436 (pass_data_ipa_reference): ...new pass_data instance and...
2437 (make_pass_ipa_reference): ...new function.
2438 * ipa-split.c (pass_split_functions): Convert from a global struct to
2439 a subclass of gimple_opt_pass along with...
2440 (pass_data_split_functions): ...new pass_data instance and...
2441 (make_pass_split_functions): ...new function.
2442 (pass_feedback_split_functions): Convert from a global struct to a
2443 subclass of gimple_opt_pass along with...
2444 (pass_data_feedback_split_functions): ...new pass_data instance and...
2445 (make_pass_feedback_split_functions): ...new function.
2446 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
2447 global struct to a subclass of simple_ipa_opt_pass along with...
2448 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
2450 (make_pass_ipa_function_and_variable_visibility): ...new function.
2451 (pass_ipa_free_inline_summary): Convert from a global struct to a
2452 subclass of simple_ipa_opt_pass along with...
2453 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
2454 (make_pass_ipa_free_inline_summary): ...new function.
2455 (pass_ipa_whole_program_visibility): Convert from a global struct to a
2456 subclass of ipa_opt_pass_d along with...
2457 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
2459 (make_pass_ipa_whole_program_visibility): ...new function.
2460 (pass_ipa_profile): Convert from a global struct to a subclass of
2461 ipa_opt_pass_d along with...
2462 (pass_data_ipa_profile): ...new pass_data instance and...
2463 (make_pass_ipa_profile): ...new function.
2464 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
2465 ipa_opt_pass_d along with...
2466 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
2467 (make_pass_ipa_cdtor_merge): ...new function.
2468 * ira.c (pass_ira): Convert from a global struct to a subclass of
2469 rtl_opt_pass along with...
2470 (pass_data_ira): ...new pass_data instance and...
2471 (make_pass_ira): ...new function.
2472 (pass_reload): Convert from a global struct to a subclass of
2473 rtl_opt_pass along with...
2474 (pass_data_reload): ...new pass_data instance and...
2475 (make_pass_reload): ...new function.
2476 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
2477 subclass of rtl_opt_pass along with...
2478 (pass_data_cleanup_barriers): ...new pass_data instance and...
2479 (make_pass_cleanup_barriers): ...new function.
2480 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
2481 of rtl_opt_pass along with...
2482 (pass_data_loop2): ...new pass_data instance and...
2483 (make_pass_loop2): ...new function.
2484 (pass_rtl_loop_init): Convert from a global struct to a subclass of
2485 rtl_opt_pass along with...
2486 (pass_data_rtl_loop_init): ...new pass_data instance and...
2487 (make_pass_rtl_loop_init): ...new function.
2488 (pass_rtl_loop_done): Convert from a global struct to a subclass of
2489 rtl_opt_pass along with...
2490 (pass_data_rtl_loop_done): ...new pass_data instance and...
2491 (make_pass_rtl_loop_done): ...new function.
2492 (pass_rtl_move_loop_invariants): Convert from a global struct to a
2493 subclass of rtl_opt_pass along with...
2494 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
2495 (make_pass_rtl_move_loop_invariants): ...new function.
2496 (pass_rtl_unswitch): Convert from a global struct to a subclass of
2497 rtl_opt_pass along with...
2498 (pass_data_rtl_unswitch): ...new pass_data instance and...
2499 (make_pass_rtl_unswitch): ...new function.
2500 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
2501 subclass of rtl_opt_pass along with...
2502 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
2504 (make_pass_rtl_unroll_and_peel_loops): ...new function.
2505 (pass_rtl_doloop): Convert from a global struct to a subclass of
2506 rtl_opt_pass along with...
2507 (pass_data_rtl_doloop): ...new pass_data instance and...
2508 (make_pass_rtl_doloop): ...new function.
2509 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
2510 a subclass of rtl_opt_pass along with...
2511 (pass_data_lower_subreg): ...new pass_data instance and...
2512 (make_pass_lower_subreg): ...new function.
2513 (pass_lower_subreg2): Convert from a global struct to a subclass of
2514 rtl_opt_pass along with...
2515 (pass_data_lower_subreg2): ...new pass_data instance and...
2516 (make_pass_lower_subreg2): ...new function.
2517 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
2518 struct to a subclass of ipa_opt_pass_d along with...
2519 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
2520 (make_pass_ipa_lto_gimple_out): ...new function.
2521 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
2522 of ipa_opt_pass_d along with...
2523 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2524 (make_pass_ipa_lto_finish_out): ...new function.
2525 * mode-switching.c (pass_mode_switching): Convert from a global struct
2526 to a subclass of rtl_opt_pass along with...
2527 (pass_data_mode_switching): ...new pass_data instance and...
2528 (make_pass_mode_switching): ...new function.
2529 * modulo-sched.c (pass_sms): Convert from a global struct to a
2530 subclass of rtl_opt_pass along with...
2531 (pass_data_sms): ...new pass_data instance and...
2532 (make_pass_sms): ...new function.
2533 * omp-low.c (pass_expand_omp): Convert from a global struct to a
2534 subclass of gimple_opt_pass along with...
2535 (pass_data_expand_omp): ...new pass_data instance and...
2536 (make_pass_expand_omp): ...new function.
2537 (pass_lower_omp): Convert from a global struct to a subclass of
2538 gimple_opt_pass along with...
2539 (pass_data_lower_omp): ...new pass_data instance and...
2540 (make_pass_lower_omp): ...new function.
2541 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2542 of gimple_opt_pass along with...
2543 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2544 (make_pass_diagnose_omp_blocks): ...new function.
2545 * passes.c (pass_early_local_passes): Convert from a global struct to
2546 a subclass of simple_ipa_opt_pass along with...
2547 (pass_data_early_local_passes): ...new pass_data instance and...
2548 (make_pass_early_local_passes): ...new function.
2549 (pass_all_early_optimizations): Convert from a global struct to a
2550 subclass of gimple_opt_pass along with...
2551 (pass_data_all_early_optimizations): ...new pass_data instance and...
2552 (make_pass_all_early_optimizations): ...new function.
2553 (pass_all_optimizations): Convert from a global struct to a subclass
2554 of gimple_opt_pass along with...
2555 (pass_data_all_optimizations): ...new pass_data instance and...
2556 (make_pass_all_optimizations): ...new function.
2557 (pass_all_optimizations_g): Convert from a global struct to a subclass
2558 of gimple_opt_pass along with...
2559 (pass_data_all_optimizations_g): ...new pass_data instance and...
2560 (make_pass_all_optimizations_g): ...new function.
2561 (pass_rest_of_compilation): Convert from a global struct to a subclass
2562 of rtl_opt_pass along with...
2563 (pass_data_rest_of_compilation): ...new pass_data instance and...
2564 (make_pass_rest_of_compilation): ...new function.
2565 (pass_postreload): Convert from a global struct to a subclass of
2566 rtl_opt_pass along with...
2567 (pass_data_postreload): ...new pass_data instance and...
2568 (make_pass_postreload): ...new function.
2569 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2570 subclass of rtl_opt_pass along with...
2571 (pass_data_gcse2): ...new pass_data instance and...
2572 (make_pass_gcse2): ...new function.
2573 * postreload.c (pass_postreload_cse): Convert from a global struct to
2574 a subclass of rtl_opt_pass along with...
2575 (pass_data_postreload_cse): ...new pass_data instance and...
2576 (make_pass_postreload_cse): ...new function.
2577 * predict.c (pass_profile): Convert from a global struct to a subclass
2578 of gimple_opt_pass along with...
2579 (pass_data_profile): ...new pass_data instance and...
2580 (make_pass_profile): ...new function.
2581 (pass_strip_predict_hints): Convert from a global struct to a subclass
2582 of gimple_opt_pass along with...
2583 (pass_data_strip_predict_hints): ...new pass_data instance and...
2584 (make_pass_strip_predict_hints): ...new function.
2585 * recog.c (pass_peephole2): Convert from a global struct to a subclass
2586 of rtl_opt_pass along with...
2587 (pass_data_peephole2): ...new pass_data instance and...
2588 (make_pass_peephole2): ...new function.
2589 (pass_split_all_insns): Convert from a global struct to a subclass of
2590 rtl_opt_pass along with...
2591 (pass_data_split_all_insns): ...new pass_data instance and...
2592 (make_pass_split_all_insns): ...new function.
2593 (pass_split_after_reload): Convert from a global struct to a subclass
2594 of rtl_opt_pass along with...
2595 (pass_data_split_after_reload): ...new pass_data instance and...
2596 (make_pass_split_after_reload): ...new function.
2597 (pass_split_before_regstack): Convert from a global struct to a
2598 subclass of rtl_opt_pass along with...
2599 (pass_data_split_before_regstack): ...new pass_data instance and...
2600 (make_pass_split_before_regstack): ...new function.
2601 (pass_split_before_sched2): Convert from a global struct to a subclass
2602 of rtl_opt_pass along with...
2603 (pass_data_split_before_sched2): ...new pass_data instance and...
2604 (make_pass_split_before_sched2): ...new function.
2605 (pass_split_for_shorten_branches): Convert from a global struct to a
2606 subclass of rtl_opt_pass along with...
2607 (pass_data_split_for_shorten_branches): ...new pass_data instance
2609 (make_pass_split_for_shorten_branches): ...new function.
2610 * ree.c (pass_ree): Convert from a global struct to a subclass of
2611 rtl_opt_pass along with...
2612 (pass_data_ree): ...new pass_data instance and...
2613 (make_pass_ree): ...new function.
2614 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2615 subclass of rtl_opt_pass along with...
2616 (pass_data_stack_regs): ...new pass_data instance and...
2617 (make_pass_stack_regs): ...new function.
2618 (pass_stack_regs_run): Convert from a global struct to a subclass of
2619 rtl_opt_pass along with...
2620 (pass_data_stack_regs_run): ...new pass_data instance and...
2621 (make_pass_stack_regs_run): ...new function.
2622 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2623 subclass of rtl_opt_pass along with...
2624 (pass_data_cprop_hardreg): ...new pass_data instance and...
2625 (make_pass_cprop_hardreg): ...new function.
2626 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2627 subclass of rtl_opt_pass along with...
2628 (pass_data_reginfo_init): ...new pass_data instance and...
2629 (make_pass_reginfo_init): ...new function.
2630 * regmove.c (pass_regmove): Convert from a global struct to a subclass
2631 of rtl_opt_pass along with...
2632 (pass_data_regmove): ...new pass_data instance and...
2633 (make_pass_regmove): ...new function.
2634 * regrename.c (pass_regrename): Convert from a global struct to a
2635 subclass of rtl_opt_pass along with...
2636 (pass_data_regrename): ...new pass_data instance and...
2637 (make_pass_regrename): ...new function.
2638 * reorg.c (pass_delay_slots): Convert from a global struct to a
2639 subclass of rtl_opt_pass along with...
2640 (pass_data_delay_slots): ...new pass_data instance and...
2641 (make_pass_delay_slots): ...new function.
2642 (pass_machine_reorg): Convert from a global struct to a subclass of
2643 rtl_opt_pass along with...
2644 (pass_data_machine_reorg): ...new pass_data instance and...
2645 (make_pass_machine_reorg): ...new function.
2646 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2647 of rtl_opt_pass along with...
2648 (pass_data_sched): ...new pass_data instance and...
2649 (make_pass_sched): ...new function.
2650 (pass_sched2): Convert from a global struct to a subclass of
2651 rtl_opt_pass along with...
2652 (pass_data_sched2): ...new pass_data instance and...
2653 (make_pass_sched2): ...new function.
2654 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2655 to a subclass of rtl_opt_pass along with...
2656 (pass_data_stack_ptr_mod): ...new pass_data instance and...
2657 (make_pass_stack_ptr_mod): ...new function.
2658 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2659 to a subclass of rtl_opt_pass along with...
2660 (pass_data_rtl_store_motion): ...new pass_data instance and...
2661 (make_pass_rtl_store_motion): ...new function.
2662 * tracer.c (pass_tracer): Convert from a global struct to a subclass
2663 of gimple_opt_pass along with...
2664 (pass_data_tracer): ...new pass_data instance and...
2665 (make_pass_tracer): ...new function.
2666 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2667 to a subclass of gimple_opt_pass along with...
2668 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2669 (make_pass_diagnose_tm_blocks): ...new function.
2670 (pass_lower_tm): Convert from a global struct to a subclass of
2671 gimple_opt_pass along with...
2672 (pass_data_lower_tm): ...new pass_data instance and...
2673 (make_pass_lower_tm): ...new function.
2674 (pass_tm_init): Convert from a global struct to a subclass of
2675 gimple_opt_pass along with...
2676 (pass_data_tm_init): ...new pass_data instance and...
2677 (make_pass_tm_init): ...new function.
2678 (pass_tm_mark): Convert from a global struct to a subclass of
2679 gimple_opt_pass along with...
2680 (pass_data_tm_mark): ...new pass_data instance and...
2681 (make_pass_tm_mark): ...new function.
2682 (pass_tm_edges): Convert from a global struct to a subclass of
2683 gimple_opt_pass along with...
2684 (pass_data_tm_edges): ...new pass_data instance and...
2685 (make_pass_tm_edges): ...new function.
2686 (pass_tm_memopt): Convert from a global struct to a subclass of
2687 gimple_opt_pass along with...
2688 (pass_data_tm_memopt): ...new pass_data instance and...
2689 (make_pass_tm_memopt): ...new function.
2690 (pass_ipa_tm): Convert from a global struct to a subclass of
2691 simple_ipa_opt_pass along with...
2692 (pass_data_ipa_tm): ...new pass_data instance and...
2693 (make_pass_ipa_tm): ...new function.
2694 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2695 subclass of gimple_opt_pass along with...
2696 (pass_data_call_cdce): ...new pass_data instance and...
2697 (make_pass_call_cdce): ...new function.
2698 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2699 subclass of gimple_opt_pass along with...
2700 (pass_data_build_cfg): ...new pass_data instance and...
2701 (make_pass_build_cfg): ...new function.
2702 (pass_split_crit_edges): Convert from a global struct to a subclass of
2703 gimple_opt_pass along with...
2704 (pass_data_split_crit_edges): ...new pass_data instance and...
2705 (make_pass_split_crit_edges): ...new function.
2706 (pass_warn_function_return): Convert from a global struct to a
2707 subclass of gimple_opt_pass along with...
2708 (pass_data_warn_function_return): ...new pass_data instance and...
2709 (make_pass_warn_function_return): ...new function.
2710 (pass_warn_function_noreturn): Convert from a global struct to a
2711 subclass of gimple_opt_pass along with...
2712 (pass_data_warn_function_noreturn): ...new pass_data instance and...
2713 (make_pass_warn_function_noreturn): ...new function.
2714 (pass_warn_unused_result): Convert from a global struct to a subclass
2715 of gimple_opt_pass along with...
2716 (pass_data_warn_unused_result): ...new pass_data instance and...
2717 (make_pass_warn_unused_result): ...new function.
2718 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2719 a subclass of gimple_opt_pass along with...
2720 (pass_data_merge_phi): ...new pass_data instance and...
2721 (make_pass_merge_phi): ...new function.
2722 * tree-complex.c (pass_lower_complex): Convert from a global struct to
2723 a subclass of gimple_opt_pass along with...
2724 (pass_data_lower_complex): ...new pass_data instance and...
2725 (make_pass_lower_complex): ...new function.
2726 (pass_lower_complex_O0): Convert from a global struct to a subclass of
2727 gimple_opt_pass along with...
2728 (pass_data_lower_complex_O0): ...new pass_data instance and...
2729 (make_pass_lower_complex_O0): ...new function.
2730 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2731 subclass of gimple_opt_pass along with...
2732 (pass_data_lower_eh): ...new pass_data instance and...
2733 (make_pass_lower_eh): ...new function.
2734 (pass_refactor_eh): Convert from a global struct to a subclass of
2735 gimple_opt_pass along with...
2736 (pass_data_refactor_eh): ...new pass_data instance and...
2737 (make_pass_refactor_eh): ...new function.
2738 (pass_lower_resx): Convert from a global struct to a subclass of
2739 gimple_opt_pass along with...
2740 (pass_data_lower_resx): ...new pass_data instance and...
2741 (make_pass_lower_resx): ...new function.
2742 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2743 of gimple_opt_pass along with...
2744 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2745 (make_pass_lower_eh_dispatch): ...new function.
2746 (pass_cleanup_eh): Convert from a global struct to a subclass of
2747 gimple_opt_pass along with...
2748 (pass_data_cleanup_eh): ...new pass_data instance and...
2749 (make_pass_cleanup_eh): ...new function.
2750 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2751 to a subclass of simple_ipa_opt_pass along with...
2752 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2753 (make_pass_ipa_lower_emutls): ...new function.
2754 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2755 a subclass of gimple_opt_pass along with...
2756 (pass_data_if_conversion): ...new pass_data instance and...
2757 (make_pass_if_conversion): ...new function.
2758 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2759 subclass of gimple_opt_pass along with...
2760 (pass_data_build_ssa): ...new pass_data instance and...
2761 (make_pass_build_ssa): ...new function.
2762 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2763 global struct to a subclass of gimple_opt_pass along with...
2764 (pass_data_loop_distribution): ...new pass_data instance and...
2765 (make_pass_loop_distribution): ...new function.
2766 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2767 subclass of gimple_opt_pass along with...
2768 (pass_data_mudflap_1): ...new pass_data instance and...
2769 (make_pass_mudflap_1): ...new function.
2770 (pass_mudflap_2): Convert from a global struct to a subclass of
2771 gimple_opt_pass along with...
2772 (pass_data_mudflap_2): ...new pass_data instance and...
2773 (make_pass_mudflap_2): ...new function.
2774 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2775 subclass of gimple_opt_pass along with...
2776 (pass_data_mudflap_1): ...new pass_data instance and...
2777 (make_pass_mudflap_1): ...new function.
2778 (pass_mudflap_2): Convert from a global struct to a subclass of
2779 gimple_opt_pass along with...
2780 (pass_data_mudflap_2): ...new pass_data instance and...
2781 (make_pass_mudflap_2): ...new function.
2782 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2783 gimple_opt_pass along with...
2784 (pass_data_nrv): ...new pass_data instance and...
2785 (make_pass_nrv): ...new function.
2786 (pass_return_slot): Convert from a global struct to a subclass of
2787 gimple_opt_pass along with...
2788 (pass_data_return_slot): ...new pass_data instance and...
2789 (make_pass_return_slot): ...new function.
2790 * tree-object-size.c (pass_object_sizes): Convert from a global struct
2791 to a subclass of gimple_opt_pass along with...
2792 (pass_data_object_sizes): ...new pass_data instance and...
2793 (make_pass_object_sizes): ...new function.
2794 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2795 global struct to a subclass of gimple_opt_pass along with...
2796 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2798 (make_pass_cleanup_cfg_post_optimizing): ...new function.
2799 (pass_fixup_cfg): Convert from a global struct to a subclass of
2800 gimple_opt_pass along with...
2801 (pass_data_fixup_cfg): ...new pass_data instance and...
2802 (make_pass_fixup_cfg): ...new function.
2803 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2804 (make_pass_mudflap_1): ...new function.
2805 (pass_mudflap_2): Replace declaration with that of...
2806 (make_pass_mudflap_2): ...new function.
2807 (pass_asan): Replace declaration with that of...
2808 (make_pass_asan): ...new function.
2809 (pass_asan_O0): Replace declaration with that of...
2810 (make_pass_asan_O0): ...new function.
2811 (pass_tsan): Replace declaration with that of...
2812 (make_pass_tsan): ...new function.
2813 (pass_tsan_O0): Replace declaration with that of...
2814 (make_pass_tsan_O0): ...new function.
2815 (pass_lower_cf): Replace declaration with that of...
2816 (make_pass_lower_cf): ...new function.
2817 (pass_refactor_eh): Replace declaration with that of...
2818 (make_pass_refactor_eh): ...new function.
2819 (pass_lower_eh): Replace declaration with that of...
2820 (make_pass_lower_eh): ...new function.
2821 (pass_lower_eh_dispatch): Replace declaration with that of...
2822 (make_pass_lower_eh_dispatch): ...new function.
2823 (pass_lower_resx): Replace declaration with that of...
2824 (make_pass_lower_resx): ...new function.
2825 (pass_build_cfg): Replace declaration with that of...
2826 (make_pass_build_cfg): ...new function.
2827 (pass_early_tree_profile): Replace declaration with that of...
2828 (make_pass_early_tree_profile): ...new function.
2829 (pass_cleanup_eh): Replace declaration with that of...
2830 (make_pass_cleanup_eh): ...new function.
2831 (pass_sra): Replace declaration with that of...
2832 (make_pass_sra): ...new function.
2833 (pass_sra_early): Replace declaration with that of...
2834 (make_pass_sra_early): ...new function.
2835 (pass_early_ipa_sra): Replace declaration with that of...
2836 (make_pass_early_ipa_sra): ...new function.
2837 (pass_tail_recursion): Replace declaration with that of...
2838 (make_pass_tail_recursion): ...new function.
2839 (pass_tail_calls): Replace declaration with that of...
2840 (make_pass_tail_calls): ...new function.
2841 (pass_tree_loop): Replace declaration with that of...
2842 (make_pass_tree_loop): ...new function.
2843 (pass_tree_loop_init): Replace declaration with that of...
2844 (make_pass_tree_loop_init): ...new function.
2845 (pass_lim): Replace declaration with that of...
2846 (make_pass_lim): ...new function.
2847 (pass_tree_unswitch): Replace declaration with that of...
2848 (make_pass_tree_unswitch): ...new function.
2849 (pass_predcom): Replace declaration with that of...
2850 (make_pass_predcom): ...new function.
2851 (pass_iv_canon): Replace declaration with that of...
2852 (make_pass_iv_canon): ...new function.
2853 (pass_scev_cprop): Replace declaration with that of...
2854 (make_pass_scev_cprop): ...new function.
2855 (pass_empty_loop): Replace declaration with that of...
2856 (make_pass_empty_loop): ...new function.
2857 (pass_record_bounds): Replace declaration with that of...
2858 (make_pass_record_bounds): ...new function.
2859 (pass_graphite): Replace declaration with that of...
2860 (make_pass_graphite): ...new function.
2861 (pass_graphite_transforms): Replace declaration with that of...
2862 (make_pass_graphite_transforms): ...new function.
2863 (pass_if_conversion): Replace declaration with that of...
2864 (make_pass_if_conversion): ...new function.
2865 (pass_loop_distribution): Replace declaration with that of...
2866 (make_pass_loop_distribution): ...new function.
2867 (pass_vectorize): Replace declaration with that of...
2868 (make_pass_vectorize): ...new function.
2869 (pass_slp_vectorize): Replace declaration with that of...
2870 (make_pass_slp_vectorize): ...new function.
2871 (pass_complete_unroll): Replace declaration with that of...
2872 (make_pass_complete_unroll): ...new function.
2873 (pass_complete_unrolli): Replace declaration with that of...
2874 (make_pass_complete_unrolli): ...new function.
2875 (pass_parallelize_loops): Replace declaration with that of...
2876 (make_pass_parallelize_loops): ...new function.
2877 (pass_loop_prefetch): Replace declaration with that of...
2878 (make_pass_loop_prefetch): ...new function.
2879 (pass_iv_optimize): Replace declaration with that of...
2880 (make_pass_iv_optimize): ...new function.
2881 (pass_tree_loop_done): Replace declaration with that of...
2882 (make_pass_tree_loop_done): ...new function.
2883 (pass_ch): Replace declaration with that of...
2884 (make_pass_ch): ...new function.
2885 (pass_ccp): Replace declaration with that of...
2886 (make_pass_ccp): ...new function.
2887 (pass_phi_only_cprop): Replace declaration with that of...
2888 (make_pass_phi_only_cprop): ...new function.
2889 (pass_build_ssa): Replace declaration with that of...
2890 (make_pass_build_ssa): ...new function.
2891 (pass_build_alias): Replace declaration with that of...
2892 (make_pass_build_alias): ...new function.
2893 (pass_build_ealias): Replace declaration with that of...
2894 (make_pass_build_ealias): ...new function.
2895 (pass_dominator): Replace declaration with that of...
2896 (make_pass_dominator): ...new function.
2897 (pass_dce): Replace declaration with that of...
2898 (make_pass_dce): ...new function.
2899 (pass_dce_loop): Replace declaration with that of...
2900 (make_pass_dce_loop): ...new function.
2901 (pass_cd_dce): Replace declaration with that of...
2902 (make_pass_cd_dce): ...new function.
2903 (pass_call_cdce): Replace declaration with that of...
2904 (make_pass_call_cdce): ...new function.
2905 (pass_merge_phi): Replace declaration with that of...
2906 (make_pass_merge_phi): ...new function.
2907 (pass_split_crit_edges): Replace declaration with that of...
2908 (make_pass_split_crit_edges): ...new function.
2909 (pass_pre): Replace declaration with that of...
2910 (make_pass_pre): ...new function.
2911 (pass_profile): Replace declaration with that of...
2912 (make_pass_profile): ...new function.
2913 (pass_strip_predict_hints): Replace declaration with that of...
2914 (make_pass_strip_predict_hints): ...new function.
2915 (pass_lower_complex_O0): Replace declaration with that of...
2916 (make_pass_lower_complex_O0): ...new function.
2917 (pass_lower_complex): Replace declaration with that of...
2918 (make_pass_lower_complex): ...new function.
2919 (pass_lower_vector): Replace declaration with that of...
2920 (make_pass_lower_vector): ...new function.
2921 (pass_lower_vector_ssa): Replace declaration with that of...
2922 (make_pass_lower_vector_ssa): ...new function.
2923 (pass_lower_omp): Replace declaration with that of...
2924 (make_pass_lower_omp): ...new function.
2925 (pass_diagnose_omp_blocks): Replace declaration with that of...
2926 (make_pass_diagnose_omp_blocks): ...new function.
2927 (pass_expand_omp): Replace declaration with that of...
2928 (make_pass_expand_omp): ...new function.
2929 (pass_expand_omp_ssa): Replace declaration with that of...
2930 (make_pass_expand_omp_ssa): ...new function.
2931 (pass_object_sizes): Replace declaration with that of...
2932 (make_pass_object_sizes): ...new function.
2933 (pass_strlen): Replace declaration with that of...
2934 (make_pass_strlen): ...new function.
2935 (pass_fold_builtins): Replace declaration with that of...
2936 (make_pass_fold_builtins): ...new function.
2937 (pass_stdarg): Replace declaration with that of...
2938 (make_pass_stdarg): ...new function.
2939 (pass_early_warn_uninitialized): Replace declaration with that of...
2940 (make_pass_early_warn_uninitialized): ...new function.
2941 (pass_late_warn_uninitialized): Replace declaration with that of...
2942 (make_pass_late_warn_uninitialized): ...new function.
2943 (pass_cse_reciprocals): Replace declaration with that of...
2944 (make_pass_cse_reciprocals): ...new function.
2945 (pass_cse_sincos): Replace declaration with that of...
2946 (make_pass_cse_sincos): ...new function.
2947 (pass_optimize_bswap): Replace declaration with that of...
2948 (make_pass_optimize_bswap): ...new function.
2949 (pass_optimize_widening_mul): Replace declaration with that of...
2950 (make_pass_optimize_widening_mul): ...new function.
2951 (pass_warn_function_return): Replace declaration with that of...
2952 (make_pass_warn_function_return): ...new function.
2953 (pass_warn_function_noreturn): Replace declaration with that of...
2954 (make_pass_warn_function_noreturn): ...new function.
2955 (pass_cselim): Replace declaration with that of...
2956 (make_pass_cselim): ...new function.
2957 (pass_phiopt): Replace declaration with that of...
2958 (make_pass_phiopt): ...new function.
2959 (pass_forwprop): Replace declaration with that of...
2960 (make_pass_forwprop): ...new function.
2961 (pass_phiprop): Replace declaration with that of...
2962 (make_pass_phiprop): ...new function.
2963 (pass_tree_ifcombine): Replace declaration with that of...
2964 (make_pass_tree_ifcombine): ...new function.
2965 (pass_dse): Replace declaration with that of...
2966 (make_pass_dse): ...new function.
2967 (pass_nrv): Replace declaration with that of...
2968 (make_pass_nrv): ...new function.
2969 (pass_rename_ssa_copies): Replace declaration with that of...
2970 (make_pass_rename_ssa_copies): ...new function.
2971 (pass_sink_code): Replace declaration with that of...
2972 (make_pass_sink_code): ...new function.
2973 (pass_fre): Replace declaration with that of...
2974 (make_pass_fre): ...new function.
2975 (pass_check_data_deps): Replace declaration with that of...
2976 (make_pass_check_data_deps): ...new function.
2977 (pass_copy_prop): Replace declaration with that of...
2978 (make_pass_copy_prop): ...new function.
2979 (pass_vrp): Replace declaration with that of...
2980 (make_pass_vrp): ...new function.
2981 (pass_uncprop): Replace declaration with that of...
2982 (make_pass_uncprop): ...new function.
2983 (pass_return_slot): Replace declaration with that of...
2984 (make_pass_return_slot): ...new function.
2985 (pass_reassoc): Replace declaration with that of...
2986 (make_pass_reassoc): ...new function.
2987 (pass_rebuild_cgraph_edges): Replace declaration with that of...
2988 (make_pass_rebuild_cgraph_edges): ...new function.
2989 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2990 (make_pass_remove_cgraph_callee_edges): ...new function.
2991 (pass_build_cgraph_edges): Replace declaration with that of...
2992 (make_pass_build_cgraph_edges): ...new function.
2993 (pass_local_pure_const): Replace declaration with that of...
2994 (make_pass_local_pure_const): ...new function.
2995 (pass_tracer): Replace declaration with that of...
2996 (make_pass_tracer): ...new function.
2997 (pass_warn_unused_result): Replace declaration with that of...
2998 (make_pass_warn_unused_result): ...new function.
2999 (pass_diagnose_tm_blocks): Replace declaration with that of...
3000 (make_pass_diagnose_tm_blocks): ...new function.
3001 (pass_lower_tm): Replace declaration with that of...
3002 (make_pass_lower_tm): ...new function.
3003 (pass_tm_init): Replace declaration with that of...
3004 (make_pass_tm_init): ...new function.
3005 (pass_tm_mark): Replace declaration with that of...
3006 (make_pass_tm_mark): ...new function.
3007 (pass_tm_memopt): Replace declaration with that of...
3008 (make_pass_tm_memopt): ...new function.
3009 (pass_tm_edges): Replace declaration with that of...
3010 (make_pass_tm_edges): ...new function.
3011 (pass_split_functions): Replace declaration with that of...
3012 (make_pass_split_functions): ...new function.
3013 (pass_feedback_split_functions): Replace declaration with that of...
3014 (make_pass_feedback_split_functions): ...new function.
3015 (pass_strength_reduction): Replace declaration with that of...
3016 (make_pass_strength_reduction): ...new function.
3017 (pass_ipa_lower_emutls): Replace declaration with that of...
3018 (make_pass_ipa_lower_emutls): ...new function.
3019 (pass_ipa_function_and_variable_visibility): Replace declaration with
3021 (make_pass_ipa_function_and_variable_visibility): ...new function.
3022 (pass_ipa_tree_profile): Replace declaration with that of...
3023 (make_pass_ipa_tree_profile): ...new function.
3024 (pass_early_local_passes): Replace declaration with that of...
3025 (make_pass_early_local_passes): ...new function.
3026 (pass_ipa_whole_program_visibility): Replace declaration with that
3028 (make_pass_ipa_whole_program_visibility): ...new function.
3029 (pass_ipa_lto_gimple_out): Replace declaration with that of...
3030 (make_pass_ipa_lto_gimple_out): ...new function.
3031 (pass_ipa_increase_alignment): Replace declaration with that of...
3032 (make_pass_ipa_increase_alignment): ...new function.
3033 (pass_ipa_inline): Replace declaration with that of...
3034 (make_pass_ipa_inline): ...new function.
3035 (pass_ipa_free_lang_data): Replace declaration with that of...
3036 (make_pass_ipa_free_lang_data): ...new function.
3037 (pass_ipa_free_inline_summary): Replace declaration with that of...
3038 (make_pass_ipa_free_inline_summary): ...new function.
3039 (pass_ipa_cp): Replace declaration with that of...
3040 (make_pass_ipa_cp): ...new function.
3041 (pass_ipa_reference): Replace declaration with that of...
3042 (make_pass_ipa_reference): ...new function.
3043 (pass_ipa_pure_const): Replace declaration with that of...
3044 (make_pass_ipa_pure_const): ...new function.
3045 (pass_ipa_pta): Replace declaration with that of...
3046 (make_pass_ipa_pta): ...new function.
3047 (pass_ipa_lto_finish_out): Replace declaration with that of...
3048 (make_pass_ipa_lto_finish_out): ...new function.
3049 (pass_ipa_tm): Replace declaration with that of...
3050 (make_pass_ipa_tm): ...new function.
3051 (pass_ipa_profile): Replace declaration with that of...
3052 (make_pass_ipa_profile): ...new function.
3053 (pass_ipa_cdtor_merge): Replace declaration with that of...
3054 (make_pass_ipa_cdtor_merge): ...new function.
3055 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
3057 (make_pass_cleanup_cfg_post_optimizing): ...new function.
3058 (pass_init_datastructures): Replace declaration with that of...
3059 (make_pass_init_datastructures): ...new function.
3060 (pass_fixup_cfg): Replace declaration with that of...
3061 (make_pass_fixup_cfg): ...new function.
3062 (pass_expand): Replace declaration with that of...
3063 (make_pass_expand): ...new function.
3064 (pass_instantiate_virtual_regs): Replace declaration with that of...
3065 (make_pass_instantiate_virtual_regs): ...new function.
3066 (pass_rtl_fwprop): Replace declaration with that of...
3067 (make_pass_rtl_fwprop): ...new function.
3068 (pass_rtl_fwprop_addr): Replace declaration with that of...
3069 (make_pass_rtl_fwprop_addr): ...new function.
3070 (pass_jump): Replace declaration with that of...
3071 (make_pass_jump): ...new function.
3072 (pass_jump2): Replace declaration with that of...
3073 (make_pass_jump2): ...new function.
3074 (pass_lower_subreg): Replace declaration with that of...
3075 (make_pass_lower_subreg): ...new function.
3076 (pass_cse): Replace declaration with that of...
3077 (make_pass_cse): ...new function.
3078 (pass_fast_rtl_dce): Replace declaration with that of...
3079 (make_pass_fast_rtl_dce): ...new function.
3080 (pass_ud_rtl_dce): Replace declaration with that of...
3081 (make_pass_ud_rtl_dce): ...new function.
3082 (pass_rtl_dce): Replace declaration with that of...
3083 (make_pass_rtl_dce): ...new function.
3084 (pass_rtl_dse1): Replace declaration with that of...
3085 (make_pass_rtl_dse1): ...new function.
3086 (pass_rtl_dse2): Replace declaration with that of...
3087 (make_pass_rtl_dse2): ...new function.
3088 (pass_rtl_dse3): Replace declaration with that of...
3089 (make_pass_rtl_dse3): ...new function.
3090 (pass_rtl_cprop): Replace declaration with that of...
3091 (make_pass_rtl_cprop): ...new function.
3092 (pass_rtl_pre): Replace declaration with that of...
3093 (make_pass_rtl_pre): ...new function.
3094 (pass_rtl_hoist): Replace declaration with that of...
3095 (make_pass_rtl_hoist): ...new function.
3096 (pass_rtl_store_motion): Replace declaration with that of...
3097 (make_pass_rtl_store_motion): ...new function.
3098 (pass_cse_after_global_opts): Replace declaration with that of...
3099 (make_pass_cse_after_global_opts): ...new function.
3100 (pass_rtl_ifcvt): Replace declaration with that of...
3101 (make_pass_rtl_ifcvt): ...new function.
3102 (pass_into_cfg_layout_mode): Replace declaration with that of...
3103 (make_pass_into_cfg_layout_mode): ...new function.
3104 (pass_outof_cfg_layout_mode): Replace declaration with that of...
3105 (make_pass_outof_cfg_layout_mode): ...new function.
3106 (pass_loop2): Replace declaration with that of...
3107 (make_pass_loop2): ...new function.
3108 (pass_rtl_loop_init): Replace declaration with that of...
3109 (make_pass_rtl_loop_init): ...new function.
3110 (pass_rtl_move_loop_invariants): Replace declaration with that of...
3111 (make_pass_rtl_move_loop_invariants): ...new function.
3112 (pass_rtl_unswitch): Replace declaration with that of...
3113 (make_pass_rtl_unswitch): ...new function.
3114 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
3115 (make_pass_rtl_unroll_and_peel_loops): ...new function.
3116 (pass_rtl_doloop): Replace declaration with that of...
3117 (make_pass_rtl_doloop): ...new function.
3118 (pass_rtl_loop_done): Replace declaration with that of...
3119 (make_pass_rtl_loop_done): ...new function.
3120 (pass_web): Replace declaration with that of...
3121 (make_pass_web): ...new function.
3122 (pass_cse2): Replace declaration with that of...
3123 (make_pass_cse2): ...new function.
3124 (pass_df_initialize_opt): Replace declaration with that of...
3125 (make_pass_df_initialize_opt): ...new function.
3126 (pass_df_initialize_no_opt): Replace declaration with that of...
3127 (make_pass_df_initialize_no_opt): ...new function.
3128 (pass_reginfo_init): Replace declaration with that of...
3129 (make_pass_reginfo_init): ...new function.
3130 (pass_inc_dec): Replace declaration with that of...
3131 (make_pass_inc_dec): ...new function.
3132 (pass_stack_ptr_mod): Replace declaration with that of...
3133 (make_pass_stack_ptr_mod): ...new function.
3134 (pass_initialize_regs): Replace declaration with that of...
3135 (make_pass_initialize_regs): ...new function.
3136 (pass_combine): Replace declaration with that of...
3137 (make_pass_combine): ...new function.
3138 (pass_if_after_combine): Replace declaration with that of...
3139 (make_pass_if_after_combine): ...new function.
3140 (pass_ree): Replace declaration with that of...
3141 (make_pass_ree): ...new function.
3142 (pass_partition_blocks): Replace declaration with that of...
3143 (make_pass_partition_blocks): ...new function.
3144 (pass_match_asm_constraints): Replace declaration with that of...
3145 (make_pass_match_asm_constraints): ...new function.
3146 (pass_regmove): Replace declaration with that of...
3147 (make_pass_regmove): ...new function.
3148 (pass_split_all_insns): Replace declaration with that of...
3149 (make_pass_split_all_insns): ...new function.
3150 (pass_fast_rtl_byte_dce): Replace declaration with that of...
3151 (make_pass_fast_rtl_byte_dce): ...new function.
3152 (pass_lower_subreg2): Replace declaration with that of...
3153 (make_pass_lower_subreg2): ...new function.
3154 (pass_mode_switching): Replace declaration with that of...
3155 (make_pass_mode_switching): ...new function.
3156 (pass_sms): Replace declaration with that of...
3157 (make_pass_sms): ...new function.
3158 (pass_sched): Replace declaration with that of...
3159 (make_pass_sched): ...new function.
3160 (pass_ira): Replace declaration with that of...
3161 (make_pass_ira): ...new function.
3162 (pass_reload): Replace declaration with that of...
3163 (make_pass_reload): ...new function.
3164 (pass_clean_state): Replace declaration with that of...
3165 (make_pass_clean_state): ...new function.
3166 (pass_branch_prob): Replace declaration with that of...
3167 (make_pass_branch_prob): ...new function.
3168 (pass_value_profile_transformations): Replace declaration with that
3170 (make_pass_value_profile_transformations): ...new function.
3171 (pass_postreload_cse): Replace declaration with that of...
3172 (make_pass_postreload_cse): ...new function.
3173 (pass_gcse2): Replace declaration with that of...
3174 (make_pass_gcse2): ...new function.
3175 (pass_split_after_reload): Replace declaration with that of...
3176 (make_pass_split_after_reload): ...new function.
3177 (pass_branch_target_load_optimize1): Replace declaration with that
3179 (make_pass_branch_target_load_optimize1): ...new function.
3180 (pass_thread_prologue_and_epilogue): Replace declaration with that
3182 (make_pass_thread_prologue_and_epilogue): ...new function.
3183 (pass_stack_adjustments): Replace declaration with that of...
3184 (make_pass_stack_adjustments): ...new function.
3185 (pass_peephole2): Replace declaration with that of...
3186 (make_pass_peephole2): ...new function.
3187 (pass_if_after_reload): Replace declaration with that of...
3188 (make_pass_if_after_reload): ...new function.
3189 (pass_regrename): Replace declaration with that of...
3190 (make_pass_regrename): ...new function.
3191 (pass_cprop_hardreg): Replace declaration with that of...
3192 (make_pass_cprop_hardreg): ...new function.
3193 (pass_reorder_blocks): Replace declaration with that of...
3194 (make_pass_reorder_blocks): ...new function.
3195 (pass_branch_target_load_optimize2): Replace declaration with that
3197 (make_pass_branch_target_load_optimize2): ...new function.
3198 (pass_leaf_regs): Replace declaration with that of...
3199 (make_pass_leaf_regs): ...new function.
3200 (pass_split_before_sched2): Replace declaration with that of...
3201 (make_pass_split_before_sched2): ...new function.
3202 (pass_compare_elim_after_reload): Replace declaration with that of...
3203 (make_pass_compare_elim_after_reload): ...new function.
3204 (pass_sched2): Replace declaration with that of...
3205 (make_pass_sched2): ...new function.
3206 (pass_stack_regs): Replace declaration with that of...
3207 (make_pass_stack_regs): ...new function.
3208 (pass_stack_regs_run): Replace declaration with that of...
3209 (make_pass_stack_regs_run): ...new function.
3210 (pass_df_finish): Replace declaration with that of...
3211 (make_pass_df_finish): ...new function.
3212 (pass_compute_alignments): Replace declaration with that of...
3213 (make_pass_compute_alignments): ...new function.
3214 (pass_duplicate_computed_gotos): Replace declaration with that of...
3215 (make_pass_duplicate_computed_gotos): ...new function.
3216 (pass_variable_tracking): Replace declaration with that of...
3217 (make_pass_variable_tracking): ...new function.
3218 (pass_free_cfg): Replace declaration with that of...
3219 (make_pass_free_cfg): ...new function.
3220 (pass_machine_reorg): Replace declaration with that of...
3221 (make_pass_machine_reorg): ...new function.
3222 (pass_cleanup_barriers): Replace declaration with that of...
3223 (make_pass_cleanup_barriers): ...new function.
3224 (pass_delay_slots): Replace declaration with that of...
3225 (make_pass_delay_slots): ...new function.
3226 (pass_split_for_shorten_branches): Replace declaration with that of...
3227 (make_pass_split_for_shorten_branches): ...new function.
3228 (pass_split_before_regstack): Replace declaration with that of...
3229 (make_pass_split_before_regstack): ...new function.
3230 (pass_convert_to_eh_region_ranges): Replace declaration with that
3232 (make_pass_convert_to_eh_region_ranges): ...new function.
3233 (pass_shorten_branches): Replace declaration with that of...
3234 (make_pass_shorten_branches): ...new function.
3235 (pass_set_nothrow_function_flags): Replace declaration with that of...
3236 (make_pass_set_nothrow_function_flags): ...new function.
3237 (pass_dwarf2_frame): Replace declaration with that of...
3238 (make_pass_dwarf2_frame): ...new function.
3239 (pass_final): Replace declaration with that of...
3240 (make_pass_final): ...new function.
3241 (pass_rtl_seqabstr): Replace declaration with that of...
3242 (make_pass_rtl_seqabstr): ...new function.
3243 (pass_release_ssa_names): Replace declaration with that of...
3244 (make_pass_release_ssa_names): ...new function.
3245 (pass_early_inline): Replace declaration with that of...
3246 (make_pass_early_inline): ...new function.
3247 (pass_inline_parameters): Replace declaration with that of...
3248 (make_pass_inline_parameters): ...new function.
3249 (pass_update_address_taken): Replace declaration with that of...
3250 (make_pass_update_address_taken): ...new function.
3251 (pass_convert_switch): Replace declaration with that of...
3252 (make_pass_convert_switch): ...new function.
3253 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
3254 to a subclass of simple_ipa_opt_pass along with...
3255 (pass_data_ipa_tree_profile): ...new pass_data instance and...
3256 (make_pass_ipa_tree_profile): ...new function.
3257 * tree-sra.c (pass_sra_early): Convert from a global struct to a
3258 subclass of gimple_opt_pass along with...
3259 (pass_data_sra_early): ...new pass_data instance and...
3260 (make_pass_sra_early): ...new function.
3261 (pass_sra): Convert from a global struct to a subclass of
3262 gimple_opt_pass along with...
3263 (pass_data_sra): ...new pass_data instance and...
3264 (make_pass_sra): ...new function.
3265 (pass_early_ipa_sra): Convert from a global struct to a subclass of
3266 gimple_opt_pass along with...
3267 (pass_data_early_ipa_sra): ...new pass_data instance and...
3268 (make_pass_early_ipa_sra): ...new function.
3269 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
3270 subclass of gimple_opt_pass along with...
3271 (pass_data_ccp): ...new pass_data instance and...
3272 (make_pass_ccp): ...new function.
3273 (pass_fold_builtins): Convert from a global struct to a subclass of
3274 gimple_opt_pass along with...
3275 (pass_data_fold_builtins): ...new pass_data instance and...
3276 (make_pass_fold_builtins): ...new function.
3277 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
3278 subclass of gimple_opt_pass along with...
3279 (pass_data_copy_prop): ...new pass_data instance and...
3280 (make_pass_copy_prop): ...new function.
3281 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
3282 global struct to a subclass of gimple_opt_pass along with...
3283 (pass_data_rename_ssa_copies): ...new pass_data instance and...
3284 (make_pass_rename_ssa_copies): ...new function.
3285 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
3286 subclass of gimple_opt_pass along with...
3287 (pass_data_dce): ...new pass_data instance and...
3288 (make_pass_dce): ...new function.
3289 (pass_dce_loop): Convert from a global struct to a subclass of
3290 gimple_opt_pass along with...
3291 (pass_data_dce_loop): ...new pass_data instance and...
3292 (make_pass_dce_loop): ...new function.
3293 (pass_cd_dce): Convert from a global struct to a subclass of
3294 gimple_opt_pass along with...
3295 (pass_data_cd_dce): ...new pass_data instance and...
3296 (make_pass_cd_dce): ...new function.
3297 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
3298 subclass of gimple_opt_pass along with...
3299 (pass_data_dominator): ...new pass_data instance and...
3300 (make_pass_dominator): ...new function.
3301 (pass_phi_only_cprop): Convert from a global struct to a subclass of
3302 gimple_opt_pass along with...
3303 (pass_data_phi_only_cprop): ...new pass_data instance and...
3304 (make_pass_phi_only_cprop): ...new function.
3305 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
3306 subclass of gimple_opt_pass along with...
3307 (pass_data_dse): ...new pass_data instance and...
3308 (make_pass_dse): ...new function.
3309 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
3310 a subclass of gimple_opt_pass along with...
3311 (pass_data_forwprop): ...new pass_data instance and...
3312 (make_pass_forwprop): ...new function.
3313 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
3314 struct to a subclass of gimple_opt_pass along with...
3315 (pass_data_tree_ifcombine): ...new pass_data instance and...
3316 (make_pass_tree_ifcombine): ...new function.
3317 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
3318 subclass of gimple_opt_pass along with...
3319 (pass_data_ch): ...new pass_data instance and...
3320 (make_pass_ch): ...new function.
3321 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
3322 subclass of gimple_opt_pass along with...
3323 (pass_data_tree_loop): ...new pass_data instance and...
3324 (make_pass_tree_loop): ...new function.
3325 (pass_tree_loop_init): Convert from a global struct to a subclass of
3326 gimple_opt_pass along with...
3327 (pass_data_tree_loop_init): ...new pass_data instance and...
3328 (make_pass_tree_loop_init): ...new function.
3329 (pass_lim): Convert from a global struct to a subclass of
3330 gimple_opt_pass along with...
3331 (pass_data_lim): ...new pass_data instance and...
3332 (make_pass_lim): ...new function.
3333 (pass_tree_unswitch): Convert from a global struct to a subclass of
3334 gimple_opt_pass along with...
3335 (pass_data_tree_unswitch): ...new pass_data instance and...
3336 (make_pass_tree_unswitch): ...new function.
3337 (pass_predcom): Convert from a global struct to a subclass of
3338 gimple_opt_pass along with...
3339 (pass_data_predcom): ...new pass_data instance and...
3340 (make_pass_predcom): ...new function.
3341 (pass_vectorize): Convert from a global struct to a subclass of
3342 gimple_opt_pass along with...
3343 (pass_data_vectorize): ...new pass_data instance and...
3344 (make_pass_vectorize): ...new function.
3345 (pass_graphite): Convert from a global struct to a subclass of
3346 gimple_opt_pass along with...
3347 (pass_data_graphite): ...new pass_data instance and...
3348 (make_pass_graphite): ...new function.
3349 (pass_graphite_transforms): Convert from a global struct to a subclass
3350 of gimple_opt_pass along with...
3351 (pass_data_graphite_transforms): ...new pass_data instance and...
3352 (make_pass_graphite_transforms): ...new function.
3353 (pass_check_data_deps): Convert from a global struct to a subclass of
3354 gimple_opt_pass along with...
3355 (pass_data_check_data_deps): ...new pass_data instance and...
3356 (make_pass_check_data_deps): ...new function.
3357 (pass_iv_canon): Convert from a global struct to a subclass of
3358 gimple_opt_pass along with...
3359 (pass_data_iv_canon): ...new pass_data instance and...
3360 (make_pass_iv_canon): ...new function.
3361 (pass_scev_cprop): Convert from a global struct to a subclass of
3362 gimple_opt_pass along with...
3363 (pass_data_scev_cprop): ...new pass_data instance and...
3364 (make_pass_scev_cprop): ...new function.
3365 (pass_record_bounds): Convert from a global struct to a subclass of
3366 gimple_opt_pass along with...
3367 (pass_data_record_bounds): ...new pass_data instance and...
3368 (make_pass_record_bounds): ...new function.
3369 (pass_complete_unroll): Convert from a global struct to a subclass of
3370 gimple_opt_pass along with...
3371 (pass_data_complete_unroll): ...new pass_data instance and...
3372 (make_pass_complete_unroll): ...new function.
3373 (pass_complete_unrolli): Convert from a global struct to a subclass of
3374 gimple_opt_pass along with...
3375 (pass_data_complete_unrolli): ...new pass_data instance and...
3376 (make_pass_complete_unrolli): ...new function.
3377 (pass_parallelize_loops): Convert from a global struct to a subclass
3378 of gimple_opt_pass along with...
3379 (pass_data_parallelize_loops): ...new pass_data instance and...
3380 (make_pass_parallelize_loops): ...new function.
3381 (pass_loop_prefetch): Convert from a global struct to a subclass of
3382 gimple_opt_pass along with...
3383 (pass_data_loop_prefetch): ...new pass_data instance and...
3384 (make_pass_loop_prefetch): ...new function.
3385 (pass_iv_optimize): Convert from a global struct to a subclass of
3386 gimple_opt_pass along with...
3387 (pass_data_iv_optimize): ...new pass_data instance and...
3388 (make_pass_iv_optimize): ...new function.
3389 (pass_tree_loop_done): Convert from a global struct to a subclass of
3390 gimple_opt_pass along with...
3391 (pass_data_tree_loop_done): ...new pass_data instance and...
3392 (make_pass_tree_loop_done): ...new function.
3393 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
3394 struct to a subclass of gimple_opt_pass along with...
3395 (pass_data_cse_reciprocals): ...new pass_data instance and...
3396 (make_pass_cse_reciprocals): ...new function.
3397 (pass_cse_sincos): Convert from a global struct to a subclass of
3398 gimple_opt_pass along with...
3399 (pass_data_cse_sincos): ...new pass_data instance and...
3400 (make_pass_cse_sincos): ...new function.
3401 (pass_optimize_bswap): Convert from a global struct to a subclass of
3402 gimple_opt_pass along with...
3403 (pass_data_optimize_bswap): ...new pass_data instance and...
3404 (make_pass_optimize_bswap): ...new function.
3405 (pass_optimize_widening_mul): Convert from a global struct to a
3406 subclass of gimple_opt_pass along with...
3407 (pass_data_optimize_widening_mul): ...new pass_data instance and...
3408 (make_pass_optimize_widening_mul): ...new function.
3409 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
3410 subclass of gimple_opt_pass along with...
3411 (pass_data_phiopt): ...new pass_data instance and...
3412 (make_pass_phiopt): ...new function.
3413 (pass_cselim): Convert from a global struct to a subclass of
3414 gimple_opt_pass along with...
3415 (pass_data_cselim): ...new pass_data instance and...
3416 (make_pass_cselim): ...new function.
3417 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
3418 subclass of gimple_opt_pass along with...
3419 (pass_data_phiprop): ...new pass_data instance and...
3420 (make_pass_phiprop): ...new function.
3421 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
3422 subclass of gimple_opt_pass along with...
3423 (pass_data_pre): ...new pass_data instance and...
3424 (make_pass_pre): ...new function.
3425 (pass_fre): Convert from a global struct to a subclass of
3426 gimple_opt_pass along with...
3427 (pass_data_fre): ...new pass_data instance and...
3428 (make_pass_fre): ...new function.
3429 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
3430 subclass of gimple_opt_pass along with...
3431 (pass_data_reassoc): ...new pass_data instance and...
3432 (make_pass_reassoc): ...new function.
3433 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
3434 subclass of gimple_opt_pass along with...
3435 (pass_data_sink_code): ...new pass_data instance and...
3436 (make_pass_sink_code): ...new function.
3437 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
3438 subclass of gimple_opt_pass along with...
3439 (pass_data_strlen): ...new pass_data instance and...
3440 (make_pass_strlen): ...new function.
3441 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
3442 struct to a subclass of gimple_opt_pass along with...
3443 (pass_data_build_alias): ...new pass_data instance and...
3444 (make_pass_build_alias): ...new function.
3445 (pass_build_ealias): Convert from a global struct to a subclass of
3446 gimple_opt_pass along with...
3447 (pass_data_build_ealias): ...new pass_data instance and...
3448 (make_pass_build_ealias): ...new function.
3449 (pass_ipa_pta): Convert from a global struct to a subclass of
3450 simple_ipa_opt_pass along with...
3451 (pass_data_ipa_pta): ...new pass_data instance and...
3452 (make_pass_ipa_pta): ...new function.
3453 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
3454 subclass of gimple_opt_pass along with...
3455 (pass_data_uncprop): ...new pass_data instance and...
3456 (make_pass_uncprop): ...new function.
3457 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
3458 global struct to a subclass of gimple_opt_pass along with...
3459 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
3460 (make_pass_late_warn_uninitialized): ...new function.
3461 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
3462 to a subclass of gimple_opt_pass along with...
3463 (pass_data_init_datastructures): ...new pass_data instance and...
3464 (make_pass_init_datastructures): ...new function.
3465 (pass_early_warn_uninitialized): Convert from a global struct to a
3466 subclass of gimple_opt_pass along with...
3467 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
3468 (make_pass_early_warn_uninitialized): ...new function.
3469 (pass_update_address_taken): Convert from a global struct to a
3470 subclass of gimple_opt_pass along with...
3471 (pass_data_update_address_taken): ...new pass_data instance and...
3472 (make_pass_update_address_taken): ...new function.
3473 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
3474 struct to a subclass of gimple_opt_pass along with...
3475 (pass_data_release_ssa_names): ...new pass_data instance and...
3476 (make_pass_release_ssa_names): ...new function.
3477 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
3478 subclass of gimple_opt_pass along with...
3479 (pass_data_stdarg): ...new pass_data instance and...
3480 (make_pass_stdarg): ...new function.
3481 * tree-switch-conversion.c (pass_convert_switch): Convert from a
3482 global struct to a subclass of gimple_opt_pass along with...
3483 (pass_data_convert_switch): ...new pass_data instance and...
3484 (make_pass_convert_switch): ...new function.
3485 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
3486 to a subclass of gimple_opt_pass along with...
3487 (pass_data_tail_recursion): ...new pass_data instance and...
3488 (make_pass_tail_recursion): ...new function.
3489 (pass_tail_calls): Convert from a global struct to a subclass of
3490 gimple_opt_pass along with...
3491 (pass_data_tail_calls): ...new pass_data instance and...
3492 (make_pass_tail_calls): ...new function.
3493 * tree-vect-generic.c (pass_lower_vector): Convert from a global
3494 struct to a subclass of gimple_opt_pass along with...
3495 (pass_data_lower_vector): ...new pass_data instance and...
3496 (make_pass_lower_vector): ...new function.
3497 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
3498 gimple_opt_pass along with...
3499 (pass_data_lower_vector_ssa): ...new pass_data instance and...
3500 (make_pass_lower_vector_ssa): ...new function.
3501 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
3502 to a subclass of gimple_opt_pass along with...
3503 (pass_data_slp_vectorize): ...new pass_data instance and...
3504 (make_pass_slp_vectorize): ...new function.
3505 (pass_ipa_increase_alignment): Convert from a global struct to a
3506 subclass of simple_ipa_opt_pass along with...
3507 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
3508 (make_pass_ipa_increase_alignment): ...new function.
3509 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
3510 gimple_opt_pass along with...
3511 (pass_data_vrp): ...new pass_data instance and...
3512 (make_pass_vrp): ...new function.
3513 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
3514 subclass of simple_ipa_opt_pass along with...
3515 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
3516 (make_pass_ipa_free_lang_data): ...new function.
3517 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
3518 gimple_opt_pass along with...
3519 (pass_data_tsan): ...new pass_data instance and...
3520 (make_pass_tsan): ...new function.
3521 (pass_tsan_O0): Convert from a global struct to a subclass of
3522 gimple_opt_pass along with...
3523 (pass_data_tsan_O0): ...new pass_data instance and...
3524 (make_pass_tsan_O0): ...new function.
3525 * var-tracking.c (pass_variable_tracking): Convert from a global
3526 struct to a subclass of rtl_opt_pass along with...
3527 (pass_data_variable_tracking): ...new pass_data instance and...
3528 (make_pass_variable_tracking): ...new function.
3529 * web.c (pass_web): Convert from a global struct to a subclass of
3530 rtl_opt_pass along with...
3531 (pass_data_web): ...new pass_data instance and...
3532 (make_pass_web): ...new function.
3533 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3534 declaration with that of...
3535 (make_pass_mode_switch_use): ...new function.
3536 (pass_resolve_sw_modes): Replace declaration with that of...
3537 (make_pass_resolve_sw_modes): ...new function.
3538 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3539 from a global struct to a subclass of rtl_opt_pass along with...
3540 (pass_data_mode_switch_use): ...new pass_data instance and...
3541 (make_pass_mode_switch_use): ...new function.
3542 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3543 from a global struct to a subclass of rtl_opt_pass along with...
3544 (pass_data_resolve_sw_modes): ...new pass_data instance and...
3545 (make_pass_resolve_sw_modes): ...new function.
3546 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3547 struct to a subclass of rtl_opt_pass along with...
3548 (pass_data_insert_vzeroupper): ...new pass_data instance and...
3549 (make_pass_insert_vzeroupper): ...new function.
3550 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3551 global struct to a subclass of rtl_opt_pass along with...
3552 (pass_data_work_around_errata): ...new pass_data instance and...
3553 (make_pass_work_around_errata): ...new function.
3554 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3555 struct to a subclass of rtl_opt_pass along with...
3556 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3557 (make_pass_mips_machine_reorg2): ...new function.
3559 2013-08-05 David Malcolm <dmalcolm@redhat.com>
3561 * passes.c (pass_manager::operator new): New.
3563 2013-08-05 David Malcolm <dmalcolm@redhat.com>
3565 Handwritten part of conversion of passes to C++ classes.
3567 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3568 (toplev.o): Add dep on PASS_MANAGER_H.
3569 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3570 of early local pases to reflect this moving from a global to a
3571 member of gcc::pass_manager.
3572 (cgraph_add_new_function): Likewise.
3573 * lto-cgraph.c (lto_output_node): Update for conversion of
3574 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3575 * passes.c (opt_pass::clone): New.
3576 (opt_pass::gate): New.
3577 (opt_pass::execute): New.
3578 (opt_pass::opt_pass): New.
3579 (pass_manager::execute_early_local_passes): New.
3580 (pass_manager::execute_pass_mode_switching): new.
3581 (finish_optimization_passes): Convert to...
3582 (pass_manager::finish_optimization_passes): ...this.
3583 (finish_optimization_passes): Update for conversion of passes to
3585 (register_dump_files_1): Use has_gate since we cannot portably
3586 check a vtable entry against NULL.
3587 (dump_one_pass): Likewise.
3588 (ipa_write_summaries_2): Likewise.
3589 (ipa_write_optimization_summaries_1): Likewise.
3590 (ipa_read_summaries_1): Likewise.
3591 (ipa_read_optimization_summaries_1): Likewise.
3592 (execute_ipa_stmt_fixups): Likewise.
3593 (pass_manager::pass_manager): Rewrite pass-creation, invoking
3594 pass-creation functions rather than wiring up globals, and
3595 storing the results in fields of pass_manager generated using
3597 (pass_manager::dump_profile_report): Update for conversion of
3598 passes to C++ classes.
3599 (pass_manager::execute_ipa_summary_passes): Likewise.
3600 (execute_one_ipa_transform_pass): Likewise.
3601 (execute_one_pass): Use has_gate and has_execute since we cannot
3602 portably check a vtable entry against NULL.
3603 * pass_manager.h (pass_manager::finish_optimization_passes): New.
3604 (pass_manager): Use pass-instances.def to add fields for the
3605 various pass instances.
3606 * toplev.c (finalize): Update for move of
3607 finish_optimization_passes to a method of gcc::pass_manager.
3608 * toplev.h (finish_optimization_passes): Move to method of class
3610 * tree-pass.h (struct pass_data): New.
3611 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3612 (opt_pass::gate): Convert to virtual function.
3613 (opt_pass::~opt_pass): New.
3614 (opt_pass::clone): New.
3615 (opt_pass::execute): Convert to virtual function.
3616 (opt_pass::opt_pass): New.
3617 (opt_pass::ctxt_): new.
3618 (gimple_opt_pass): Convert to subclass of opt_pass.
3619 (gimple_opt_pass::gimple_opt_pass): New.
3620 (rtl_opt_pass): Convert to subclass of opt_pass.
3621 (rtl_opt_pass::rtl_opt_pass): New.
3622 (ipa_opt_pass_d): Convert to subclass of opt_pass.
3623 (ipa_opt_pass_d::ipa_opt_pass_d): New.
3624 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3625 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3626 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3627 invocation of pass_mode_switching to reflect this moving from a
3628 global to a member of gcc::pass_manager.
3629 (ix86_option_override): Rework how pass_insert_vzeroupper is
3630 added to the pass_manager to reflect autogenerated changes.
3631 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3633 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
3635 PR rtl-optimization/57708
3636 * recog.c (peep2_find_free_register): Validate all regs in a
3639 2013-08-05 Jan Hubicka <jh@suse.cz>
3642 * cgraph.c (verify_cgraph_node): Accept local flags from other
3644 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3645 (function_and_variable_visibility): Likewise.
3646 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3648 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3650 * graph.c (init_graph_slim_pretty_print): Remove.
3651 (print_graph_cfg): Do not call it. Use local pretty printer.
3652 (start_graph_dump): Likewise.
3654 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3656 * gimple-pretty-print.c (buffer): Remove.
3657 (initialized): Likewise.
3658 (maybe_init_pretty_print): Likewise.
3659 (print_gimple_stmt): Do not call it. Use non-static local
3660 pretty_printer variable.
3661 (print_gimple_expr): Likewise.
3662 (print_gimple_seq): Likewise.
3663 (gimple_dump_bb): Likewise.
3665 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3667 * asan.c (asan_pp): Remove.
3668 (asan_pp_initialized): Likewise.
3669 (asan_pp_initialize): Likewise.
3670 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
3671 (asan_emit_stack_protection): Tidy. Use local pretty printer.
3672 (asan_add_global): Likewise.
3674 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
3676 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
3677 * diagnostic.h (diagnostic_flush_buffer): Adjust.
3678 * pretty-print.c (pp_formatted_text_data): Likewise.
3679 (pp_indent): Rename from pp_base_indent.
3680 (pp_format): Rename from pp_base_format.
3681 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3682 (pp_format_verbatim): Rename from pp_base_format_verbatim.
3683 (pp_flush): Rename from pp_base_flush.
3684 (pp_set_line_maximum_length): Rename from
3685 pp_base_set_line_maximum_length.
3686 (pp_clear_output_area): Rename from pp_base_clear_output_area.
3687 (pp_set_prefix): Rename from pp_base_set_prefix.
3688 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3689 (pp_emit_prefix): Rename from pp_base_emit_prefix.
3690 (pp_append_text): Rename from pp_base_append_text.
3691 (pp_formatted_text): Rename from pp_base_formatted_text.
3692 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3693 (pp_remaining_character_count_for_line): Rename from
3694 pp_base_remaining_character_count_for_line.
3695 (pp_newline): Rename from pp_base_newline.
3696 (pp_character): Rename from pp_base_character.
3697 (pp_string): Rename from pp_base_string.
3698 (pp_maybe_space): Rename from pp_base_maybe_space.
3699 * asan.c (asan_pp_string): Adjust.
3700 (asan_emit_stack_protection): Likewise.
3701 (asan_add_global): Likewise.
3702 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3703 * tree-mudflap.c (mf_varname_tree): Likewise.
3704 * tree-pretty-print.c (pp_tree_identifier): Rename from
3705 pp_base_tree_identifier.
3706 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3707 Declare as function.
3709 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3711 * pretty-print.h (pp_bar_bar): New.
3712 (pp_ampersand_ampersand): Likewise.
3713 (pp_less_equal): Likewise.
3714 (pp_greater_equal): Likewise.
3715 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3716 printer functions instead of pp_string or operators and punctuators.
3717 (dump_gimple_call): Likewise.
3718 (dump_gimple_omp_for): Likewise.
3719 (dump_gimple_transaction): Likewise.
3720 (dump_gimple_phi): Likewise.
3721 (pp_gimple_stmt_1): Likewise.
3722 * sched-vis.c (print_insn): Likewise.
3723 * tree-mudflap.c (mf_varname_tree): Likewise.
3724 * tree-pretty-print.c (dump_block_node): Likewise.
3725 (dump_generic_node): Likewise.
3727 2013-08-02 Jan Hubicka <jh@suse.cz>
3729 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3731 * lto-streamer-out.c (tree_is_indexable): Results decls and
3732 parm decls are not indexable.
3733 (DFS_write_tree_body): Do not follow args and results.
3734 (hash_tree): Likewise.
3735 (output_functions): Rearrange so struct function is needed
3736 only when real body is output; be able to also ouptut abstract
3737 functions; output DECL_ARGUMENTS and DECL_RESULT.
3738 (lto_output): When not in WPA, ale store abstract functions.
3739 (write_symbol): Do not care about RESULT_DECL.
3740 (output_symbol_p): Handle correctly sbtract decls.
3741 * lto-streamer-in.c (input_function): Rearrange so struct
3742 function can be NULL at entry; allow streaming of
3743 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3744 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3745 sanity check during LTO.
3746 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3747 RESULT_DECl and DECL_ARGUMENTS.
3748 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3751 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3753 * pretty-print.h (pp_underscore): New.
3755 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3756 printer functions instead of pp_character.
3757 (dump_binary_rhs): Likewise.
3758 (dump_ternary_rhs): Likewise.
3759 (dump_gimple_call_args): Likewise.
3760 (pp_points_to_solution): Likewise.
3761 (dump_gimple_call): Likewise.
3762 (dump_gimple_switch): Likewise.
3763 (dump_gimple_cond): Likewise.
3764 (dump_gimple_bind): Likewise.
3765 (dump_gimple_try): Likewise.
3766 (dump_gimple_omp_for): Likewise.
3767 (dump_gimple_omp_continue): Likewise.
3768 (dump_gimple_omp_single): Likewise.
3769 (dump_gimple_omp_sections): Likewise.
3770 (dump_gimple_omp_block): Likewise.
3771 (dump_gimple_omp_critical): Likewise.
3772 (dump_gimple_transaction): Likewise.
3773 (dump_gimple_asm): Likewise.
3774 (dump_gimple_phi): Likewise.
3775 (dump_gimple_omp_parallel): Likewise.
3776 (dump_gimple_omp_task): Likewise.
3777 (dump_gimple_omp_atomic_load): Likewise.
3778 (dump_gimple_omp_atomic_store): Likewise.
3779 (dump_gimple_mem_ops): Likewise.
3780 (pp_gimple_stmt_1): Likewise.
3781 (pp_cfg_jump): Likewise.
3782 (dump_implicit_edges): Likewise.
3783 (gimple_dump_bb_for_graph): Likewise.
3784 * graph.c (draw_cfg_node): Likewise.
3785 * langhooks.c (lhd_print_error_function): Likewise.
3786 * sched-vis.c (print_exp): Likewise.
3787 (print_value): Likewise.
3788 (print_pattern): Likewise.
3789 (print_insn): Likewise.
3790 (rtl_dump_bb_for_graph): Likewise.
3791 * tree-pretty-print.c (dump_function_declaration): Likewise.
3792 (dump_array_domain): Likewise.
3793 (dump_omp_clause): Likewise.
3794 (dump_location): Likewise.
3795 (dump_generic_node): Likewise.
3796 (print_struct_decl): Likewise.
3797 * diagnostic.c (diagnostic_show_locus): Use pp_space.
3799 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3801 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3802 candidate table when replacing a candidate statement.
3803 (replace_rhs_if_not_dup): Likewise.
3804 (replace_one_candidate): Likewise.
3806 2013-08-02 Jan Hubicka <jh@suse.cz>
3807 Martin Liska <marxin.liska@gmail.com>
3809 * cgraphunit.c (add_new_function): Fix logic when adding from
3811 (assemble_thunk): Rename to ...
3812 (expand_thunk); .. this one; export; get it working with
3813 general functions; make produced gimple valid.
3814 * cgraph.h (expand_thunk): Declare.
3816 2013-08-02 Jan Hubicka <jh@suse.cz>
3818 * ipa-cp.c (gather_context_independent_values): Use
3819 ipa_get_param_move_cost.
3820 (get_replacement_map): Remove PARAM; move parameter folding
3822 (create_specialized_node): Update.
3823 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3824 assert that we have gimple body; update move_cost.
3825 (count_formal_params): Assert that we have gimple body.
3826 (ipa_dump_param): New function.
3827 (ipa_alloc_node_params): Break out from ...
3828 (ipa_initialize_node_params): ... here.
3829 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3830 (ipa_write_node_info): Stream move costs.
3831 (ipa_read_node_info): Read move costs.
3832 (ipa_update_after_lto_read): Do not recompute node params.
3833 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3834 (ipa_get_param): Check we are not in WPA.
3835 (ipa_get_param_move_cost): New.
3836 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3837 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3838 parm numbers to be present.
3840 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
3842 PR rtl-optimization/58048
3843 * lra-constraints.c (process_alt_operands): Don't check asm
3844 operand on register.
3846 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
3848 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3849 the implied StoreLoad barrier for atomic operations if before.
3851 2013-08-02 Jan Hubicka <jh@suse.cz>
3852 Martin Liska <marxin.liska@gmail.com>
3854 * cgraph.c (cgraph_function_body_availability): Do not check
3856 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3857 symtab_node_availability): Declare.
3858 * ipa.c (can_replace_by_local_alias): New.
3859 (function_and_variable_visibility): Use it.
3860 * symtab.c (symtab_for_node_and_aliases,
3861 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3863 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
3865 PR rtl-optimization/57963
3866 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3867 (lra_constraints): Use them.
3869 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
3871 * config/arm/types.md (define_attr "type"): Add "load_acq"
3873 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3875 (cortex_a53_store1): Likewise.
3877 2013-08-01 Jan Hubicka <jh@suse.cz>
3879 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3880 partitions are not needed.
3882 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
3884 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3885 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3886 MAYBE_NON_Q_CLASS_P where appropriate.
3888 2013-08-01 Jan Hubicka <jh@suse.cz>
3890 * cgraph.h (release_function_body): Declare.
3891 * tree.c (free_lang_data_in_decl): Free, parameters and return values
3892 of unused delcarations.
3894 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3896 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3897 RTL form when subtracting a constant.
3899 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3901 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3902 Generate canonical plus rtx with negated immediate instead of minus
3904 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3906 2013-08-01 Jan Hubicka <jh@suse.cz>
3908 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3909 (cgraph_release_function_body): Likewise.
3910 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3911 * cgraph.h (cgrpah_node): Rename abstract_and_needed
3912 to used_as_abstract_origin.
3913 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3914 symbols used as abstract origins.
3915 * cgraphunit.c (analyze_functions): Update.
3916 * ipa.c (symtab_remove_unreachable_nodes): Recompute
3917 used_as_abstract_origin.
3918 * tree-inline.c (tree_function_versioning): Update
3919 used_as_abstract_origin; be ready for DECL_RESULT and
3920 DECL_ARGUMENTS to be NULL.
3922 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3923 for abstract functions.
3924 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3927 2013-08-01 Jan Hubicka <jh@suse.cz>
3929 * profile.c (compute_value_histograms): Fix thinko.
3931 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
3933 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
3934 aarch-common-protos.h to extra_headers.
3935 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3936 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3937 * config/aarch64/t-aarch64 (aarch-common.o): Define.
3939 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
3941 * config/aarch64/aarch64.md (define_attr "type"): Delete.
3942 Include "../arm/types.md". Define "type" attribute for all patterns.
3943 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3946 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3948 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3949 to support power8 load fusion.
3950 (fusion_gpr_mem_load): Likewise.
3952 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3954 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3955 declarations for power8 load fusion.
3956 (emit_fusion_gpr_load): Likewise.
3958 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3959 tuning for power8, turn on fusion mode by default. Turn on sign
3960 extending fusion mode if normal fusion mode is on, and we are at
3962 (fusion_gpr_load_p): New function, return true if we can fuse an
3963 addis instruction with a dependent load to a GPR.
3964 (emit_fusion_gpr_load): Emit the instructions for power8 load
3967 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3968 (VSX load fusion peepholes): New peepholes to fuse together an
3969 addi instruction with a VSX load instruction.
3971 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3972 peepholes to fuse an addis instruction with a load to a GPR base
3973 register. If we are supporting sign extending fusions, convert
3974 sign extending loads to zero extending loads and add an explicit
3977 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
3979 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
3980 aarch-common-protos.h to extra_headers.
3981 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3982 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3983 (arm_early_store_addr_dep): Likewise.
3984 (arm_no_early_alu_shift_dep): Likewise.
3985 (arm_no_early_alu_shift_value_dep): Likewise.
3986 (arm_no_early_mul_dep): Likewise.
3987 (arm_no_early_store_addr_dep): Likewise.
3988 (arm_mac_accumulator_is_mul_result): Likewise.
3989 (arm_mac_accumulator_is_result): Likewise.
3990 * config/arm/aarch-common.c: ... here. New file.
3991 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3993 (arm_early_store_addr_dep): Likewise.
3994 (arm_no_early_alu_shift_dep): Likewise.
3995 (arm_no_early_alu_shift_value_dep): Likewise.
3996 (arm_no_early_mul_dep): Likewise.
3997 (arm_no_early_store_addr_dep): Likewise.
3998 (arm_mac_accumulator_is_mul_result): Likewise.
3999 (arm_mac_accumulator_is_result): Likewise.
4000 * config/arm/aarch-common-protos.h: ... here. New file.
4001 * config/arm/t-arm (aarch-common.o): Define.
4003 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
4005 * config/arm/arm.md: Include new file "types.md".
4006 (define_attr "type"): Move from here to ...
4007 (define_attr "mul32"): Likewise.
4008 (define_attr "mul64"): Likewise.
4009 * config/arm/types.md: ... here. New file.
4011 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
4013 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
4014 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
4016 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4018 * gen-pass-instances.awk: Fix offset of substr().
4020 2013-07-31 David Malcolm <dmalcolm@redhat.com>
4022 * Makefile.in (pass-instances.def): New.
4023 (passes.o): Replace dependency on passes.def with one on
4026 * gen-pass-instances.awk: New.
4028 * passes.c (pass_manager::pass_manager): Use pass-instances.def
4029 rather than passes.def, updating local definition of NEXT_PASS
4030 macro to add an extra NUM parameter (currently unused).
4032 2013-07-30 David Malcolm <dmalcolm@redhat.com>
4034 * Makefile.in (PASS_MANAGER_H): New.
4035 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
4036 (passes.o): Likewise.
4037 (statistics.o): Likewise.
4038 (cgraphunit.o): Likewise.
4039 (context.o): Depend on PASS_MANAGER_H.
4041 * pass_manager.h: New.
4043 * cgraphunit.c (cgraph_add_new_function): Update for moves
4044 of globals to fields of pass_manager.
4045 (analyze_function): Likewise.
4046 (expand_function): Likewise.
4047 (ipa_passes): Likewise.
4048 (compile): Likewise.
4050 * context.c (context::context): New.
4051 * context.h (context::context): New.
4052 (context::get_passes): New.
4053 (context::passes_): New.
4055 * lto-cgraph.c (input_node): Update for moves of globals to
4056 fields of pass_manager.
4058 * passes.c (all_passes): Remove, in favor of a field of the
4059 same name within the new class pass_manager.
4060 (all_small_ipa_passes): Likewise.
4061 (all_lowering_passes): Likewise.
4062 (all_regular_ipa_passes): Likewise.
4063 (all_late_ipa_passes): Likewise.
4064 (all_lto_gen_passes): Likewise.
4065 (passes_by_id): Likewise.
4066 (passes_by_id_size): Likewise.
4067 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4068 the new class pass_manager.
4069 (set_pass_for_id): Convert to...
4070 (pass_manager::set_pass_for_id): ...method.
4071 (get_pass_for_id): Convert to...
4072 (pass_manager::get_pass_for_id): ...method.
4073 (register_one_dump_file): Move body of implementation into...
4074 (pass_manager::register_one_dump_file): ...here.
4075 (register_dump_files_1): Convert to...
4076 (pass_manager::register_dump_files_1): ...method.
4077 (register_dump_files): Convert to...
4078 (pass_manager::register_dump_files): ...method.
4079 (create_pass_tab): Update for moves of globals to fields of
4081 (dump_passes): Move body of implementation into...
4082 (pass_manager::dump_passes): ...here.
4083 (register_pass): Move body of implementation into...
4084 (pass_manager::register_pass): ...here.
4085 (init_optimization_passes): Convert into...
4086 (pass_manager::pass_manager): ...constructor for new
4087 pass_manager class, and initialize the pass_lists array.
4088 (check_profile_consistency): Update for moves of globals to
4089 fields of pass_manager.
4090 (dump_profile_report): Move body of implementation into...
4091 (pass_manager::dump_profile_report): ...here.
4092 (ipa_write_summaries_1): Update for moves of pass lists from
4093 being globals to fields of pass_manager.
4094 (ipa_write_optimization_summaries): Likewise.
4095 (ipa_read_summaries): Likewise.
4096 (ipa_read_optimization_summaries): Likewise.
4097 (execute_all_ipa_stmt_fixups): Likewise.
4099 * statistics.c (statistics_fini): Update for moves of globals to
4100 fields of pass_manager.
4102 * toplev.c (general_init): Replace call to
4103 init_optimization_passes with construction of the pass_manager
4106 * tree-pass.h (all_passes): Remove, in favor of a field of the
4107 same name within the new class pass_manager.
4108 (all_small_ipa_passes): Likewise.
4109 (all_lowering_passes): Likewise.
4110 (all_regular_ipa_passes): Likewise.
4111 (all_lto_gen_passes): Likewise.
4112 (all_late_ipa_passes): Likewise.
4113 (passes_by_id): Likewise.
4114 (passes_by_id_size): Likewise.
4115 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4116 the new class pass_manager.
4117 (get_pass_for_id): Remove.
4119 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
4121 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
4124 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
4126 * arm.md (mulhi3): New expand pattern.
4128 2013-07-30 Jan Hubicka <jh@suse.cz>
4129 Martin Liska <marxin.liska@gmail.com>
4131 * profile.c (compute_value_histograms): Do not ICE when
4132 there is mismatch only on some counters.
4134 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
4136 PR rtl-optimization/57637
4137 * function.c (move_insn_for_shrink_wrap): Also check the
4138 GEN set of the LIVE problem for the liveness analysis
4139 if it exists, otherwise give up.
4141 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4143 PR tree-optimization/57993
4144 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
4145 replaced statement in the candidate table.
4146 (phi_add_costs): Return infinite cost when the hidden basis does
4147 not dominate all phis on which the candidate is dependent.
4148 (replace_one_candidate): Record replaced statement in the
4151 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
4153 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
4154 (ashlv2si3): New expander.
4155 (*ashlv2si3_i): New define_insn_and_split.
4156 * predicates.md (float_operation): Allow patterns with three
4159 PR rtl-optimization/58021
4160 * mode-switching.c (create_pre_exit): Always split off preceding
4161 insns if we are not at the basic block head.
4163 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
4165 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
4166 (UCLIBC_DYNAMIC_LINKER): New macro.
4167 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
4169 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
4170 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
4172 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
4173 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
4174 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
4175 for SF and DF modes. Use ieee_quad_format for TF mode.
4176 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
4177 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
4178 (mips_option_override): Handle `-mnan=legacy'.
4179 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
4180 `-mabs=2008' and `-mnan=2008'.
4181 (OPTION_DEFAULT_SPECS): Add "nan" default.
4182 (ASM_SPEC): Handle `-mnan='.
4183 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
4184 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
4185 comment accordingly.
4186 (neg<mode>2): Likewise.
4187 * config/mips/mips.opt (mabs, mnan): New options.
4188 * doc/install.texi (Configuration): Document `--with-nan=' option.
4189 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
4191 (MIPS Options): Document them.
4192 * config.gcc <mips*-*-*>: Handle `--with-nan='.
4193 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
4194 * configure: Regenerate.
4195 * config.in: Regenerate.
4197 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
4199 * config/i386/i386.md (float post-reload splitters): Do not check
4200 for subregs of SSE registers.
4202 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
4203 H.J. Lu <hongjiu.lu@intel.com>
4207 * config/i386/i386.md (post-reload splitter
4208 to avoid partial SSE reg dependency stalls): New pattern.
4210 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
4212 * config/s390/s390.md ("movcc"): Swap load and store instructions.
4214 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
4216 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
4217 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
4219 2013-07-26 Cary Coutant <ccoutant@google.com>
4221 * dwarf2out.c (die_checksum_ordered): Don't include template
4222 instantiations in signature.
4223 (is_template_parameter): New function.
4224 (is_template_instantiation): New function.
4225 (generate_skeleton_bottom_up): Don't include template instantiations
4227 (generate_skeleton): Likewise.
4228 (break_out_comdat_types): Move recursive call to break out nested
4230 (prune_unused_types_mark_generic_parms_dies): Call
4231 is_template_parameter.
4233 2013-07-26 Ian Bolton <ian.bolton@arm.com>
4235 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
4236 uses vector registers.
4237 * config/aarch64/iterators.md: Add attributes rtn and vas.
4239 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4240 Richard Earnshaw <richard.earnshaw@arm.com>
4242 * combine.c (simplify_comparison): Re-canonicalize operands
4244 * config/arm/arm.md (movcond_addsi): New splitter.
4246 2013-07-25 Sterling Augustine <saugustine@google.com>
4248 * dwarf2out.c (size_of_pubnames): Move code to...
4249 (include_pubname_in_output): ...here. New.
4250 (want_pubnames): Rearrange.
4251 (output_pubnames): Call include_pubname_in_output. Move assertion.
4253 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
4255 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
4257 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
4260 * doc/extend.texi: Remove obsolete builtins. Fix
4261 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
4263 2013-07-25 Jan Hubicka <jh@suse.cz>
4265 * cgraph.c (release_function_body): Break out from ...
4266 (cgraph_release_function_body): ... this one; also release DECL_RESULT
4268 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
4269 old_tree in the map.
4270 (create_specialized_node): Update.
4271 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
4273 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
4274 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
4275 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
4276 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
4279 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4281 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
4282 addsi3_carryin_alt2_<optab>): Correct output template.
4284 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4286 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
4287 Adjust for arm_restrict_it.
4288 Remove trailing whitespace.
4290 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
4292 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
4293 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
4295 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
4297 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
4299 PR rtl-optimization/57960
4300 * lra-constraints.c (process_alt_operands): Use the right mode
4301 when checking strict_low.
4303 2013-07-25 Jan Hubicka <jh@suse.cz>
4305 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
4306 * cgraph.c (cgraph_remove_node): Do not release function body
4307 when in cgraph streaming.
4308 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
4309 in other partitions are not considered reachable; fix handling of
4312 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4314 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
4316 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4321 * config/arm/arm.md ("*sibcall_insn): Replace use of
4322 Ss with US. Adjust output for v5 and v4t.
4323 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
4325 * config/arm/constraints.md ("Ss"): Rename to US.
4327 2013-07-25 Terry Guo <terry.guo@arm.com>
4329 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
4330 shift_add/shift_sub0/shift_sub1 RTXs.
4332 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
4333 Anton Blanchard <anton@au1.ibm.com>
4335 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
4336 (altivec_vpks<VI_char>ss): Likewise.
4337 (altivec_vpks<VI_char>us): Likewise.
4338 (altivec_vpku<VI_char>us): Likewise.
4339 (altivec_vpku<VI_char>um): Likewise.
4341 2013-07-24 David Malcolm <dmalcolm@redhat.com>
4343 Introduce context class.
4345 * Makefile.in (CONTEXT_H): New.
4346 (OBJS): Add context.o.
4347 (toplev.o): Add CONTEXT_H to dependencies.
4350 * toplev.c (general_init): Create the singleton gcc::context instance.
4356 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
4358 PR rtl-optimization/57968
4359 * mode-switching.c (create_pre_exit): Allow instructions that
4360 don't set a return register to need a non-exit mode.
4362 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4363 Anton Blanchard <anton@au1.ibm.com>
4365 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
4366 operands to vperm for little endian.
4367 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
4368 of lvsl to create the control mask for a vperm for little endian.
4370 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4371 Anton Blanchard <anton@au1.ibm.com>
4373 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
4374 two operands for little-endian.
4376 2013-07-23 Steve Ellcey <sellcey@mips.com>
4378 * config/mips/mips.c (mips_case_values_threshold): New.
4379 (TARGET_CASE_VALUES_THRESHOLD): Define.
4381 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4382 Anton Blanchard <anton@au1.ibm.com>
4384 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
4385 selection of field for vector splat in little endian mode.
4387 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4389 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
4390 expanders to rs6000.md.
4391 (ior<mode>3): Likewise.
4392 (and<mode>3): Likewise.
4393 (one_cmpl<mode>2): Likewise.
4394 (nor<mode>3): Likewise.
4395 (andc<mode>3): Likewise.
4396 (eqv<mode>3): Likewise.
4397 (nand<mode>3): Likewise.
4398 (orc<mode>3): Likewise.
4400 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
4403 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
4404 to split multi-word logical operations.
4405 (rs6000_split_logical_di): Likewise.
4406 (rs6000_split_logical): Likewise.
4408 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
4409 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
4410 and allow TImode operations in 32-bit.
4411 (vsx_and<mode>3_64bit): Likewise.
4412 (vsx_ior<mode>3_32bit): Likewise.
4413 (vsx_ior<mode>3_64bit): Likewise.
4414 (vsx_xor<mode>3_32bit): Likewise.
4415 (vsx_xor<mode>3_64bit): Likewise.
4416 (vsx_one_cmpl<mode>2_32bit): Likewise.
4417 (vsx_one_cmpl<mode>2_64bit): Likewise.
4418 (vsx_nor<mode>3_32bit): Likewise.
4419 (vsx_nor<mode>3_64bit): Likewise.
4420 (vsx_andc<mode>3_32bit): Likewise.
4421 (vsx_andc<mode>3_64bit): Likewise.
4422 (vsx_eqv<mode>3_32bit): Likewise.
4423 (vsx_eqv<mode>3_64bit): Likewise.
4424 (vsx_nand<mode>3_32bit): Likewise.
4425 (vsx_nand<mode>3_64bit): Likewise.
4426 (vsx_orc<mode>3_32bit): Likewise.
4427 (vsx_orc<mode>3_64bit): Likewise.
4429 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
4430 logical types in GPRs.
4432 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
4433 logical insns to rs6000.md, and allow TImode operations in
4435 (altivec_ior<mode>3): Likewise.
4436 (altivec_xor<mode>3): Likewise.
4437 (altivec_one_cmpl<mode>2): Likewise.
4438 (altivec_nor<mode>3): Likewise.
4439 (altivec_andc<mode>3): Likewise.
4441 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
4442 attributes for moving the 128-bit logical operations into
4444 (BOOL_REGS_OUTPUT): Likewise.
4445 (BOOL_REGS_OP1): Likewise.
4446 (BOOL_REGS_OP2): Likewise.
4447 (BOOL_REGS_UNARY): Likewise.
4448 (BOOL_REGS_AND_CR0): Likewise.
4449 (one_cmpl<mode>2): Add support for DI logical operations on
4450 32-bit, splitting the operations to 32-bit.
4454 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
4455 changes to combine the 32/64-bit code, allow logical operations on
4456 TI mode in 32-bit, and to use similar match_operator patterns like
4457 scalar mode uses. Combine the Altivec and VSX code for logical
4458 operations, and move it here.
4459 (ior<mode>3, 128-bit types): Likewise.
4460 (xor<mode>3, 128-bit types): Likewise.
4461 (one_cmpl<mode>3, 128-bit types): Likewise.
4462 (nor<mode>3, 128-bit types): Likewise.
4463 (andc<mode>3, 128-bit types): Likewise.
4464 (eqv<mode>3, 128-bit types): Likewise.
4465 (nand<mode>3, 128-bit types): Likewise.
4466 (orc<mode>3, 128-bit types): Likewise.
4467 (and<mode>3_internal): Likewise.
4468 (bool<mode>3_internal): Likewise.
4469 (boolc<mode>3_internal1): Likewise.
4470 (boolc<mode>3_internal2): Likewise.
4471 (boolcc<mode>3_internal1): Likewise.
4472 (boolcc<mode>3_internal2): Likewise.
4473 (eqv<mode>3_internal1): Likewise.
4474 (eqv<mode>3_internal2): Likewise.
4475 (one_cmpl1<mode>3_internal): Likewise.
4477 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
4479 * config/microblaze/microblaze.c (microblaze_expand_prologue):
4480 Rename flag_stack_usage to flag_stack_usage_info.
4482 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
4484 * config/microblaze/sync.md: New file.
4485 * config/microblaze/microblaze.md: Include sync.md
4486 * config/microblaze/microblaze.c: Add print_operand 'y'.
4487 * config/microblaze/constraints.md: Add memory_contraint
4488 'Q' which is a single register.
4490 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
4492 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
4494 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
4496 * reload.c (find_reloads): Exit loop once we find this operand
4497 cannot be reloaded somehow for this alternative.
4499 * reload.c (find_reloads): Exit loop once we find a hard register.
4501 * rtlanal.c (computed_jump_p): Exit loop once we find label
4504 * i386.c (ix86_pad_returns): Exit loop after setting replace.
4506 * cfgloopmanip.c (remove_path): Exit loop after setting
4509 * gensupport.c (subst_dup): Avoid loop if code is not
4510 MATCH_DUP nor MATCH_OP_DUP.
4512 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
4514 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
4516 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4518 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
4519 true for SP_REGNUM if mode == ptr_mode.
4520 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
4521 with value R0_REGNUM + 31.
4523 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4525 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4526 pad pointer-typed argument downward.
4528 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4530 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4531 and __ILP32__ when the ILP32 model is in use.
4533 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4535 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4536 (aarch64_load_symref_appropriately): In the case of
4537 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4538 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4539 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4540 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4541 if 'mode' doesn't equal to 'ptr_mode'.
4542 (aarch64_output_mi_thunk): Add an assertion on the alignment of
4543 'vcall_offset'; change to call aarch64_emit_move differently depending
4544 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4545 to calculate the upper bound of 'vcall_offset'.
4546 (aarch64_cannot_force_const_mem): Change to also return true if
4548 (aarch64_legitimize_reload_address): In the case of large
4549 displacements, add new local variable 'xmode' and an assertion
4550 based on it; change to use 'xmode' to generate the new rtx and
4552 (aarch64_asm_trampoline_template): Change to generate the template
4553 differently depending on TARGET_ILP32 or not; change to use
4554 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4555 (aarch64_trampoline_size): Removed.
4556 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4557 and replace immediate literals with it. Change to use 'ptr_mode'
4558 instead of 'DImode' and call convert_memory_address if the mode
4559 of 'fnaddr' doesn't equal to 'ptr_mode'.
4560 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4562 (aarch64_elf_asm_destructor): Likewise.
4563 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4564 on TARGET_ILP32 instead of aarch64_trampoline_size.
4565 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4566 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4567 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4568 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
4569 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4570 (storewb_pair<GPI:mode>_<P:mode>): ... this.
4571 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4572 depending on the value of 'mode'.
4573 (add_losym_<mode>): New.
4574 (ldr_got_small_<mode>): New, based on ldr_got_small.
4575 (ldr_got_small): Remove.
4576 (ldr_got_small_sidi): New.
4577 * config/aarch64/iterators.md (P): New.
4578 (PTR): Change to 'ptr_mode' in the condition.
4580 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4582 * config.gcc (aarch64*-*-*): Support --with-abi.
4583 (aarch64*-*-elf): Support --with-multilib-list.
4584 (aarch64*-*-linux*): Likewise.
4585 (supported_defaults): Add abi to aarch64*-*-*.
4586 * configure.ac: Mention AArch64 for --with-multilib-list.
4587 * configure: Re-generated.
4588 * config/aarch64/biarchilp32.h: New file.
4589 * config/aarch64/biarchlp64.h: New file.
4590 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4592 (MULTILIB_DEFAULTS): Ditto.
4593 (DRIVER_SELF_SPECS): Ditto.
4594 (ASM_SPEC): Update to also substitute -mabi.
4595 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4596 file whose name depends on -mabi= and -mbig-endian.
4597 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4599 (POINTER_SIZE): New define.
4600 (POINTERS_EXTEND_UNSIGNED): Ditto.
4601 (enum aarch64_abi_type): New enumeration tag.
4602 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4603 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4604 (TARGET_ILP32): New define.
4605 * config/aarch64/aarch64.opt (mabi): New.
4607 (ilp32, lp64): New values for -mabi.
4608 * config/aarch64/t-aarch64 (comma): New define.
4609 (MULTILIB_OPTIONS): Ditto.
4610 (MULTILIB_DIRNAMES): Ditto.
4611 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4612 * doc/invoke.texi: Document -mabi for AArch64.
4614 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
4616 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4618 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4619 Anton Blanchard <anton@au1.ibm.com>
4621 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4622 endianness when selecting field to splat.
4624 2013-07-22 Eric Christopher <echristo@gmail.com>
4626 * dwarf2out.c (die_odr_checksum): New function to use
4627 CHECKSUM_ macros and ULEB128 for DIE tag.
4628 (generate_type_signature): Use.
4630 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
4632 * config.gcc (sparc*-*-*): Accept leon3 processor.
4633 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4634 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4635 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4636 * config/sparc/sparc.opt (enum processor_type): Add leon3.
4637 (mfix-ut699): Adjust comment.
4638 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4639 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4640 (CPP_CPU_SPEC): Likewise.
4641 (ASM_CPU_SPEC): Likewise.
4642 * config/sparc/sparc.c (leon3_cost): New constant.
4643 (sparc_option_override): Add leon3 support.
4644 (mem_ref): New function.
4645 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4646 (sparc_do_work_around_errata): Look into the instruction in the delay
4647 slot and adjust accordingly. Add fix for the data cache nullify issues
4648 of the UT699. Change insertion position for the NOP.
4649 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4650 (leon3_load): New reservation.
4651 (leon_store): Bump latency to 2.
4652 (grfpu): New automaton.
4653 (grfpu_alu): New unit.
4654 (grfpu_ds): Likewise.
4655 (leon_fp_alu): Adjust.
4656 (leon_fp_mult): Delete.
4657 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4658 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4659 * config/sparc/sparc.md (cpu): Add leon3.
4660 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4662 (atomic_test_and_set): Likewise.
4665 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
4667 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
4668 default for R5900 targets.
4669 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4670 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4671 * config/mips/mips.c (mips_option_override): Report an error for
4672 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
4673 for -march=r5900 -mhard-float.
4675 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
4677 * df-problems.c (can_move_insns_across): Exit loop once we
4678 find a non-fixed, non-global register.
4680 * ipa-pure-const.c (propagate_nothrow): Exit loop after
4683 * omega.c (omega_eliminate_red): Break after setting red_found.
4684 (omega_problem_has_red_equations): Similarly after setting found.
4685 (omega_query_variable): Similarly after setting coupled.
4687 2013-07-22 Marek Polacek <polacek@redhat.com>
4689 * gimplify.c: Don't include gimple.h twice.
4691 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4693 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4694 instead of TARGET_THUMB1.
4695 (Pz): New constraint.
4696 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4698 (compare_negsi_si): Likewise.
4699 (compare_addsi2_op0): Likewise.
4700 (compare_addsi2_op1): Likewise.
4701 (addsi3_carryin_<optab>): Likewise.
4702 (addsi3_carryin_alt2_<optab>): Likewise.
4703 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4704 for arm_restrict_it.
4705 (subsi3_carryin): Likewise.
4706 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4707 (minmax_arithsi): Disable for arm_restrict_it.
4708 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4709 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4710 (satsi_<SAT:code>_shift): Likewise.
4711 (arm_shiftsi3): Add alternative for 16-bit encoding.
4712 (arm32_movhf): Disable for arm_restrict_it.
4713 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4714 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4716 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
4718 * config/arm/arm.md (attribute "insn"): Delete.
4719 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4720 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4721 (not_shiftsi): Update for attribute change.
4722 (not_shiftsi_compare0): Likewise.
4723 (not_shiftsi_compare0_scratch): Likewise.
4724 (arm_one_cmplsi2): Likewise.
4725 (thumb1_one_cmplsi2): Likewise.
4726 (notsi_compare0): Likewise.
4727 (notsi_compare0_scratch): Likewise.
4728 (thumb1_movdi_insn): Likewise.
4729 (arm_movsi_insn): Likewise.
4730 (movhi_insn_arch4): Likewise.
4731 (movhi_bytes): Likewise.
4732 (arm_movqi_insn): Likewise.
4733 (thumb1_movqi_insn): Likewise.
4734 (arm32_movhf): Likewise.
4735 (thumb1_movhf): Likewise.
4736 (arm_movsf_soft_insn): Likewise.
4737 (thumb1_movsf_insn): Likewise.
4738 (thumb_movdf_insn): Likewise.
4739 (movsicc_insn): Likewise.
4740 (movsfcc_soft_insn): Likewise.
4741 (and_scc): Likewise.
4742 (cond_move): Likewise.
4743 (if_move_not): Likewise.
4744 (if_not_move): Likewise.
4745 (if_shift_move): Likewise.
4746 (if_move_shift): Likewise.
4747 (if_shift_shift): Likewise.
4748 (if_not_arith): Likewise.
4749 (if_arith_not): Likewise.
4750 (cond_move_not): Likewise.
4751 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4752 (neon_mov<mode>): Likewise.
4753 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4754 (thumb2_movsi_vfp): Likewise.
4755 (movsf_vfp): Likewise.
4756 (thumb2_movsf_vfp): Likewise.
4757 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4759 (cortexa7_older_only): Likewise.
4760 (cortexa7_younger): Likewise.
4761 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4762 (1020alu_shift_op): Likewise.
4763 (1020alu_shift_reg_op): Likewise.
4764 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4765 (alu_shift_op): Likewise.
4766 (alu_shift_reg_op): Likewise.
4767 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4768 (11_alu_shift_op): Likewise.
4769 (11_alu_shift_reg_op): Likewise.
4770 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4771 (9_alu_shift_reg_op): Likewise.
4772 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4774 (cortex_a15_alu_shift): Likewise.
4775 (cortex_a15_alu_shift_reg): Likewise.
4776 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4777 (cortex_a5_alu_shift): Likewise.
4778 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4780 (cortex_a53_alu_shift): Likewise.
4781 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4783 (cortex_a7_alu_reg): Likewise.
4784 (cortex_a7_alu_shift): Likewise.
4785 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4786 (cortex_a8_alu_shift): Likewise.
4787 (cortex_a8_alu_shift_reg): Likewise.
4788 (cortex_a8_mov): Likewise.
4789 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4790 (cortex_a9_dp_shift): Likewise.
4791 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4792 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4793 (cortex_r4_mov): Likewise.
4794 (cortex_r4_alu_shift): Likewise.
4795 (cortex_r4_alu_shift_reg): Likewise.
4796 * config/arm/fa526.md (526_alu_op): Update for attribute change.
4797 (526_alu_shift_op): Likewise.
4798 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4799 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4800 (626te_alu_shift_op): Likewise.
4801 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4802 (726te_alu_op): Likewise.
4803 (726te_alu_shift_op): Likewise.
4804 (726te_alu_shift_reg_op): Likewise.
4805 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4806 (mp626_alu_shift_op): Likewise.
4807 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4808 (pj4_alu_e1_conds): Likewise.
4809 (pj4_alu): Likewise.
4810 (pj4_alu_conds): Likewise.
4811 (pj4_shift): Likewise.
4812 (pj4_shift_conds): Likewise.
4813 (pj4_alu_shift): Likewise.
4814 (pj4_alu_shift_conds): Likewise.
4816 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4818 * config/arm/predicates.md (shiftable_operator_strict_it):
4820 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4821 Disable cond_exec version for arm_restrict_it.
4822 (thumb2_smaxsi3): Convert to generate cond_exec.
4823 (thumb2_sminsi3): Likewise.
4824 (thumb32_umaxsi3): Likewise.
4825 (thumb2_uminsi3): Likewise.
4826 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4827 (thumb2_neg_abssi2): Likewise.
4828 (thumb2_mov_scc): Add alternative for 16-bit encoding.
4829 (thumb2_movsicc_insn): Adjust alternatives.
4830 (thumb2_mov_negscc): Disable for arm_restrict_it.
4831 (thumb2_mov_negscc_strict_it): New pattern.
4832 (thumb2_mov_notscc_strict_it): New pattern.
4833 (thumb2_mov_notscc): Disable for arm_restrict_it.
4834 (thumb2_ior_scc): Likewise.
4835 (thumb2_ior_scc_strict_it): New pattern.
4836 (thumb2_cond_move): Adjust for arm_restrict_it.
4837 (thumb2_cond_arith): Disable for arm_restrict_it.
4838 (thumb2_cond_arith_strict_it): New pattern.
4839 (thumb2_cond_sub): Adjust for arm_restrict_it.
4840 (thumb2_movcond): Likewise.
4841 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4842 (thumb2_zero_extendhisi2_v6): Likewise.
4843 (thumb2_zero_extendqisi2_v6): Likewise.
4844 (orsi_notsi_si): Likewise.
4845 (orsi_not_shiftsi_si): Likewise.
4847 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
4849 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4850 instruction sequence is 1 byte shorter.
4852 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
4854 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4855 it is not needed after split.
4857 2013-07-20 Iain Sandoe <iain@codesourcery.com>
4860 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4861 second label for nonlocal goto receivers. Don't output pic base labels
4862 unless we're producing PIC; mark that action unreachable().
4863 (ix86_save_reg): If the function contains a nonlocal label, save the
4865 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4866 * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4867 (update_pic_label_number_if_needed): New.
4868 (machopic_output_function_base_name): Adjust for nonlocal receiver
4870 (machopic_should_output_picbase_label): New.
4871 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4872 (nonlocal_goto_receiver): New insn and split.
4874 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
4876 * config/aarch64/aarch64-builtins.c
4877 (aarch64_fold_builtin): Fold abs in all modes.
4878 * config/aarch64/aarch64-simd-builtins.def
4879 (abs): Enable for all modes.
4880 * config/aarch64/arm_neon.h
4881 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4882 (vabs_f64): Add missing intrinsic.
4884 2013-07-19 Ian Bolton <ian.bolton@arm.com>
4886 * config/aarch64/arm_neon.h (vabs_s64): New function
4888 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
4891 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4892 * config/avr/avr.md (adjust_len): Add `round'.
4893 * config/avr/avr-protos.h (avr_out_round): New prototype.
4894 (avr_out_plus): Add `out_label' argument.
4895 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4896 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4897 Handle the case where `insn' is just a pattern.
4898 (avr_out_bitop): Handle the case where `insn' is just a pattern.
4899 (avr_out_round): New function.
4900 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4902 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
4904 * config/microblaze/microblaze.c (microblaze_expand_prologue):
4905 Add check for flag_stack_usage to handle -fstack-usage support
4907 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
4909 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4910 interaction for new Power8 flags and VSX.
4912 2013-07-18 Sriraman Tallam <tmsriram@google.com>
4915 * tree-inline.c (expand_call_inline): Emit errors during
4916 early_inlining only if optimization is not turned on.
4918 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4922 * passes.c (init_optimization_passes): Move the construction of
4923 the pass hierarchy into a new passes.def file.
4925 * Makefile.in (passes.o): Add dependency on passes.def.
4927 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4929 * passes.c (init_optimization_passes): Introduce macros for
4930 constructing the tree of passes (INSERT_PASSES_AFTER,
4931 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4932 TERMINATE_PASS_LIST).
4934 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
4935 Wei Mi <wmi@google.com>
4937 PR rtl-optimization/57878
4938 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4940 (reload_pseudo_compare_func): Check nregs first for reload
4943 2013-07-18 David Malcolm <dmalcolm@redhat.com>
4945 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4947 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
4949 * read-rtl.c (validate_const_int): Once an invalid character is
4950 seen, quit the loop.
4952 * gengtype.c (write_roots): Similarly once we find the "deletable"
4953 or "if_marked" option.
4955 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
4957 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4958 "xtab" and "sat". Move value "clz" from here to ...
4959 (attriubte "type"): ... here.
4960 (satsi_<SAT:code>): Delete "insn" attribute.
4961 (satsi_<SAT:code>_shift): Likewise.
4962 (arm_zero_extendqisi2addsi): Likewise.
4963 (arm_extendqisi2addsi): Likewise.
4964 (clzsi2): Update for attribute changes.
4965 (rbitsi2): Likewise.
4966 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4968 (arm_usatsihi): Likewise.
4969 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4971 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
4973 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4974 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
4975 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
4976 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
4977 in alphabetical order.
4978 (attribute "core_cycles"): Update for attribute changes.
4979 (arm_addsi3): Likewise.
4980 (addsi3_compare0): Likewise.
4981 (addsi3_compare0_scratch): Likewise.
4982 (addsi3_compare_op1): Likewise.
4983 (addsi3_compare_op2): Likewise.
4984 (compare_addsi2_op0): Likewise.
4985 (compare_addsi2_op1): Likewise.
4986 (addsi3_carryin_shift_<optab>): Likewise.
4987 (subsi3_carryin_shift): Likewise.
4988 (rsbsi3_carryin_shift): Likewise.
4989 (arm_subsi3_insn): Likewise.
4990 (subsi3_compare0): Likewise.
4991 (subsi3_compare): Likewise.
4992 (arm_andsi3_insn): Likewise.
4993 (thumb1_andsi3_insn): Likewise.
4994 (andsi3_compare0): Likewise.
4995 (andsi3_compare0_scratch): Likewise.
4996 (zeroextractsi_compare0_scratch
4997 (andsi_not_shiftsi_si): Likewise.
4998 (iorsi3_insn): Likewise.
4999 (iorsi3_compare0): Likewise.
5000 (iorsi3_compare0_scratch): Likewise.
5001 (arm_xorsi3): Likewise.
5002 (thumb1_xorsi3_insn): Likewise.
5003 (xorsi3_compare0): Likewise.
5004 (xorsi3_compare0_scratch): Likewise.
5005 (satsi_<SAT:code>_shift): Likewise.
5007 (arm_shiftsi3): Likewise.
5008 (shiftsi3_compare0): Likewise.
5009 (not_shiftsi): Likewise.
5010 (not_shiftsi_compare0): Likewise.
5011 (not_shiftsi_compare0_scratch): Likewise.
5012 (arm_one_cmplsi2): Likewise.
5013 (thumb_one_complsi2): Likewise.
5014 (notsi_compare0): Likewise.
5015 (notsi_compare0_scratch): Likewise.
5016 (thumb1_zero_extendhisi2): Likewise.
5017 (arm_zero_extendhisi2): Likewise.
5018 (arm_zero_extendhisi2_v6): Likewise.
5019 (arm_zero_extendhisi2addsi): Likewise.
5020 (thumb1_zero_extendqisi2): Likewise.
5021 (thumb1_zero_extendqisi2_v6): Likewise.
5022 (arm_zero_extendqisi2): Likewise.
5023 (arm_zero_extendqisi2_v6): Likewise.
5024 (arm_zero_extendqisi2addsi): Likewise.
5025 (thumb1_extendhisi2): Likewise.
5026 (arm_extendhisi2): Likewise.
5027 (arm_extendhisi2_v6): Likewise.
5028 (arm_extendqisi): Likewise.
5029 (arm_extendqisi_v6): Likewise.
5030 (arm_extendqisi2addsi): Likewise.
5031 (thumb1_extendqisi2): Likewise.
5032 (thumb1_movdi_insn): Likewise.
5033 (arm_movsi_insn): Likewise.
5034 (movsi_compare0): Likewise.
5035 (movhi_insn_arch4): Likewise.
5036 (movhi_bytes): Likewise.
5037 (arm_movqi_insn): Likewise.
5038 (thumb1_movqi_insn): Likewise.
5039 (arm32_movhf): Likewise.
5040 (thumb1_movhf): Likewise.
5041 (arm_movsf_soft_insn): Likewise.
5042 (thumb1_movsf_insn): Likewise.
5043 (movdf_soft_insn): Likewise.
5044 (thumb_movdf_insn): Likewise.
5045 (arm_cmpsi_insn): Likewise.
5046 (cmpsi_shiftsi): Likewise.
5047 (cmpsi_shiftsi_swp): Likewise.
5048 (arm_cmpsi_negshiftsi_si): Likewise.
5049 (movsicc_insn): Likewise.
5050 (movsfcc_soft_insn): Likewise.
5051 (arith_shiftsi): Likewise.
5052 (arith_shiftsi_compare0
5053 (arith_shiftsi_compare0_scratch
5054 (sub_shiftsi): Likewise.
5055 (sub_shiftsi_compare0
5056 (sub_shiftsi_compare0_scratch
5057 (and_scc): Likewise.
5058 (cond_move): Likewise.
5059 (if_plus_move): Likewise.
5060 (if_move_plus): Likewise.
5061 (if_move_not): Likewise.
5062 (if_not_move): Likewise.
5063 (if_shift_move): Likewise.
5064 (if_move_shift): Likewise.
5065 (if_shift_shift): Likewise.
5066 (if_not_arith): Likewise.
5067 (if_arith_not): Likewise.
5068 (cond_move_not): Likewise.
5069 (thumb1_ashlsi3): Set type attribute.
5070 (thumb1_ashrsi3): Likewise.
5071 (thumb1_lshrsi3): Likewise.
5072 (thumb1_rotrsi3): Likewise.
5073 (shiftsi3_compare0_scratch): Likewise.
5074 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
5075 (neon_mov<mode>): Likewise.
5076 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
5078 (thumb2_movsi_insn): Likewise.
5079 (thumb2_cmpsi_neg_shiftsi): Likewise.
5080 (thumb2_extendqisi_v6): Likewise.
5081 (thumb2_zero_extendhisi2_v6): Likewise.
5082 (thumb2_zero_extendqisi2_v6): Likewise.
5083 (thumb2_shiftsi3_short): Likewise.
5084 (thumb2_addsi3_compare0_scratch): Likewise.
5085 (orsi_not_shiftsi_si): Likewise.
5086 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
5087 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
5089 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
5090 (1020alu_shift_op): Likewise.
5091 (1020alu_shift_reg_op): Likewise.
5092 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
5093 (alu_shift_op): Likewise.
5094 (alu_shift_reg_op): Likewise.
5095 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
5096 (11_alu_shift_op): Likewise.
5097 (11_alu_shift_reg_op): Likewise.
5098 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
5099 (9_alu_shift_reg_op): Likewise.
5100 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
5102 (cortex_a15_alu_shift): Likewise.
5103 (cortex_a15_alu_shift_reg): Likewise.
5104 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
5106 (cortex_a5_alu_shift): Likewise.
5107 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
5109 (cortex_a53_alu_shift): Likewise.
5110 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
5112 (cortex_a7_alu_reg): Likewise.
5113 (cortex_a7_alu_shift): Likewise.
5114 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
5116 (cortex_a8_alu_shift): Likewise.
5117 (cortex_a8_alu_shift_reg): Likewise.
5118 (cortex_a8_mov): Likewise.
5119 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
5120 (cortex_a9_dp_shift): Likewise.
5121 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
5123 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
5125 (cortex_r4_mov): Likewise.
5126 (cortex_r4_alu_shift): Likewise.
5127 (cortex_r4_alu_shift_reg): Likewise.
5128 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
5129 (526_alu_shift_op): Likewise.
5130 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
5131 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
5132 (626te_alu_shift_op): Likewise.
5133 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
5134 (726te_alu_op): Likewise.
5135 (726te_alu_shift_op): Likewise.
5136 (726te_alu_shift_reg_op): Likewise.
5137 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
5138 (mp626_alu_shift_op): Likewise.
5139 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
5140 (pj4_alu_e1_conds): Likewise.
5141 (pj4_alu): Likewise.
5142 (pj4_alu_conds): Likewise.
5143 (pj4_shift): Likewise.
5144 (pj4_shift_conds): Likewise.
5145 (pj4_alu_shift): Likewise.
5146 (pj4_alu_shift_conds): Likewise.
5147 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5149 (cortexa7_older_only): Likewise.
5150 (cortexa7_younger): Likewise.
5152 2013-07-18 David Malcolm <dmalcolm@redhat.com>
5154 * ipa-pure-const.c (generate_summary): Rename to...
5155 (pure_const_generate_summary): ... this.
5157 2013-07-17 Iain Sandoe <iain@codesourcery.com>
5159 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
5161 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
5164 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
5167 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5169 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
5170 enabled without -march=zEC12.
5171 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
5174 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5176 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
5177 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
5178 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
5179 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
5180 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
5181 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
5182 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
5184 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
5185 and ISA_HAS_NMADD3_NMSUB3.
5186 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
5187 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
5188 (nmsub4<mode>, nmsub3<mode>): Likewise.
5189 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
5191 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5193 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
5194 TARGET_MIPS5400 checking.
5196 2013-07-16 Jakub Jelinek <jakub@redhat.com>
5197 Peter Bergner <bergner@vnet.ibm.com>
5199 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
5200 registers in the comment.
5201 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
5202 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
5203 rather than FIRST_PSEUDO_REGISTERS.
5205 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
5207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
5208 enable extra ISA flags with TARGET_HTM.
5210 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5212 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
5215 2013-07-15 Cong Hou <congh@google.com>
5217 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
5218 in compare function for sorting.
5220 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
5222 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
5223 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
5224 * config/rs6000/rs6000.opt: Add -mhtm option.
5225 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
5226 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
5227 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
5228 __HTM__ if the HTM instructions are available.
5229 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
5230 htm_spr_reg_operand): New define_predicates.
5231 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
5232 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
5234 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
5235 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
5236 HTM builtin functions.
5237 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
5238 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
5239 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
5240 (rs6000_builtin_mask_calculate): Likewise.
5241 (rs6000_option_override_internal): Likewise.
5242 (bdesc_htm): Add new HTM builtin support.
5243 (htm_spr_num): New function.
5244 (htm_spr_regno): Likewise.
5245 (rs6000_htm_spr_icode): Likewise.
5246 (htm_expand_builtin): Likewise.
5247 (htm_init_builtins): Likewise.
5248 (rs6000_expand_builtin): Add support for HTM builtin functions.
5249 (rs6000_init_builtins): Likewise.
5250 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
5252 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
5253 (TARGET_HTM, MASK_HTM): Define macros.
5254 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
5255 (FIXED_REGISTERS): Likewise.
5256 (CALL_USED_REGISTERS): Likewise.
5257 (CALL_REALLY_USED_REGISTERS): Likewise.
5258 (REG_ALLOC_ORDER): Likewise.
5259 (enum reg_class): Likewise.
5260 (REG_CLASS_NAMES): Likewise.
5261 (REG_CLASS_CONTENTS): Likewise.
5262 (REGISTER_NAMES): Likewise.
5263 (ADDITIONAL_REGISTER_NAMES): Likewise.
5264 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
5265 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
5266 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
5267 * config/rs6000/htm.md: New file.
5268 * config/rs6000/htmintrin.h: New file.
5269 * config/rs6000/htmxlintrin.h: New file.
5271 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
5273 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5274 Define SYMBOL_TINY_GOT, update comment.
5275 * config/aarch64/aarch64.c
5276 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
5277 (aarch64_expand_mov_immediate): Likewise.
5278 (aarch64_print_operand): Likewise.
5279 (aarch64_classify_symbol): Likewise.
5280 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
5281 (ldr_got_tiny): Define.
5283 2013-07-13 Tobias Grosser <tobias@grosser.es>
5285 PR tree-optimization/54094
5286 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
5287 scheduling dimension for the parallelism check from the polyhedral
5288 information in the AST.
5289 * graphite-dependences.c (carries_deps): Do not assume the schedule is
5292 2013-07-13 Jason Merrill <jason@redhat.com>
5294 * print-tree.c (debug_vec_tree): Use debug_raw.
5295 (debug_raw (vec<tree, va_gc> &)): New.
5296 (debug_raw (vec<tree, va_gc> *)): New.
5297 * tree.h: Declare them.
5299 2013-07-13 Bin Cheng <bin.cheng@arm.com>
5301 * ifcvt.c (ifcvt_after_combine): New static variable.
5302 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
5304 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
5305 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
5306 rest_of_handle_if_after_reload): Pass new argument for if_convert.
5308 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
5310 * config/mips/mips.c (mips_expand_call): Remove empty statement.
5312 2013-07-12 Michael Matz <matz@suse.de>
5315 * convert.c (convert_to_real): Reject non-float inner types.
5317 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
5319 * config/aarch64/aarch64-protos.h
5320 (aarch64_simd_immediate_valid_for_move): Remove.
5321 * config/aarch64/aarch64.c (simd_immediate_info): New member.
5322 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
5324 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
5326 2013-07-11 Steve Ellcey <sellcey@mips.com>
5328 * config/mips/mips.c (mips_conditional_register_usage): Do not
5329 use t[0-7] registers in MIPS16 mode when optimizing for size.
5331 2013-07-11 Sriraman Tallam <tmsriram@google.com>
5333 * config/i386/i386.c (dispatch_function_versions): Fix array
5334 indexing of function_version_info to match actual_versions.
5336 2013-07-11 Teresa Johnson <tejohnson@google.com>
5338 * vec.h (struct va_gc): Move release out-of-line.
5339 (va_gc::release): Call ggc_free on released vec.
5341 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
5343 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
5344 Require GOT register as additional operand in UNSPEC.
5345 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
5346 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5347 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5348 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
5349 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
5350 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
5351 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
5353 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
5356 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
5357 name seen by assembler/linker rather if available.
5359 2013-07-11 Andreas Schwab <schwab@suse.de>
5361 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
5363 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
5365 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
5367 2013-07-10 Joseph Myers <joseph@codesourcery.com>
5369 * doc/tm.texi.in: Move hook documentation to ....
5370 * target.def: ... here.
5372 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
5374 * doc/tm.texi: Regenerate.
5376 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
5379 * doc/invoke.texi: Document -Wconditionally-supported.
5381 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5384 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5387 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5390 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
5391 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
5392 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
5393 Remove duplicate devices.
5394 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
5395 * config/avr/t-multilib: Regenerate.
5396 * config/avr/avr-tables.opt: Regenerate.
5397 * doc/avr-mmcu.texi: Regenerate.
5399 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5402 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
5404 2013-07-10 Graham Stott <graham.stott@btinternet.com>
5406 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
5407 the cost of MULT when optimizing for size.
5409 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5411 * config/cr16/cr16-protos.h: Don't include target.h.
5413 2013-07-09 Joseph Myers <joseph@codesourcery.com>
5415 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
5416 adjust register size for TDmode and TFmode for VSX registers.
5418 2013-07-08 Kai Tietz <ktietz@redhat.com>
5421 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
5422 hook_bool_const_tree_true.
5424 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5426 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
5427 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
5428 * config/s390/s390.md: Define FPR*_REGNUM constants.
5429 Fix FPR2_REGNUM constant (18 -> 17).
5430 ("*trunc<BFP:mode><DFP_ALL:mode>2")
5431 ("*trunc<DFP_ALL:mode><BFP:mode>2")
5432 ("trunc<BFP:mode><DFP_ALL:mode>2")
5433 ("trunc<DFP_ALL:mode><BFP:mode>2")
5434 ("*extend<BFP:mode><DFP_ALL:mode>2")
5435 ("*extend<DFP_ALL:mode><BFP:mode>2")
5436 ("extend<BFP:mode><DFP_ALL:mode>2")
5437 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
5440 2013-07-08 Graham Stott <graham.stott@btinternet.com>
5442 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
5444 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5446 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
5447 and cfun_fpr_bit_p to cfun_fpr_save_p.
5448 (s390_frame_area, s390_register_info, s390_frame_info)
5449 (s390_emit_prologue, s390_emit_epilogue)
5450 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
5452 * config/s390/s390.h: Define *_REGNUM macros for floating point
5455 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
5457 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
5459 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
5461 PR rtl-optimization/57786
5462 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
5463 and break out of the loop when it is set to false.
5465 2013-07-08 Jakub Jelinek <jakub@redhat.com>
5468 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
5469 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
5470 (const_int 63)) 0)).
5471 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
5472 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
5473 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
5475 PR rtl-optimization/57829
5476 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
5477 mask bits outside of mode are just sign-extension from mode to HWI.
5479 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
5481 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
5482 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
5483 adjust_address instead of change_address to keep info about alignment.
5484 (emit_strmov): Remove.
5485 (emit_memmov): New function.
5486 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
5487 (expand_movmem_epilogue): Likewise and return updated rtx for
5489 (expand_constant_movmem_prologue): Likewise and return updated rtx for
5490 destination and source.
5491 (decide_alignment): Refactor, handle vector_loop.
5492 (ix86_expand_movmem): Likewise.
5493 (ix86_expand_setmem): Likewise.
5494 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
5496 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
5498 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
5499 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
5501 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
5503 * config/i386/sse.md (sse_movlhps): Change alternative 3
5504 of operand 2 to "m".
5506 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
5509 * config/i386/sse.md (iptr): New mode attribute.
5510 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
5511 (<sse>_vm<plusminus_insn><mode>3): Ditto.
5512 (<sse>_vmmul<mode>3): Ditto.
5513 (<sse>_vmdiv<mode>3): Ditto.
5514 (sse_vmrcpv4sf2): Ditto.
5515 (<sse>_vmsqrt<mode>2): Ditto.
5516 (sse_vmrsqrtv4sf2): Ditto.
5517 (<sse>_vm<code><mode>3): Ditto.
5518 (avx_vmcmp<mode>3): Ditto.
5519 (<sse>_vmmaskcmp<mode>3): Ditto.
5520 (<sse>_comi): Ditto.
5521 (<sse>_ucomi): Ditto.
5522 (*xop_vmfrcz_<mode>): Ditto.
5523 (*fmai_fmadd_<mode>): Ditto.
5524 (*fmai_fmsub_<mode>): Ditto.
5525 (*fmai_fnmadd_<mode>): Ditto.
5526 (*fmai_fnmsub_<mode>): Ditto.
5527 (*fma4i_vmfmadd_<mode>): Ditto.
5528 (*fma4i_vmfmsub_<mode>): Ditto.
5529 (*fma4i_vmfnmadd_<mode>): Ditto.
5530 (*fma4i_vmfnmsub_<mode>): Ditto.
5531 (*xop_vmfrcz_<mode>): Ditto.
5532 (sse_cvtps2pi): Ditto.
5533 (sse_cvttps2pi): Ditto.
5534 (sse_cvtss2si): Ditto.
5535 (sse_cvtss2si_2): Ditto.
5536 (sse_cvtss2siq_2): Ditto.
5537 (sse_cvttss2si): Ditto.
5538 (sse_cvttss2siq): Ditto.
5539 (sse_cvtsd2si): Ditto.
5540 (sse_cvtsd2si_2): Ditto.
5541 (sse_cvtsd2siq_2): Ditto.
5542 (sse_cvttsd2si): Ditto.
5543 (sse_cvttsd2siq): Ditto.
5544 (sse_cvtsd2ss): Ditto.
5545 (sse_cvtss2sd): Ditto.
5546 (avx2_pbroadcast<mode>): Ditto.
5547 (avx2_pbroadcast<mode>_1): Ditto.
5548 (*avx_vperm_broadcast_v4sf): Ditto.
5550 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5551 (sse_movlhps): Ditto.
5552 (sse_storehps): Ditto.
5553 (sse_loadhps): Ditto.
5554 (sse_storelps): Ditto.
5555 (sse_loadlps): Ditto.
5556 (*vec_concatv4sf): Ditto.
5557 (*vec_interleave_highv2df): Ditto.
5558 (*vec_interleave_lowv2df): Ditto.
5559 (*vec_extractv2df_1_sse): Ditto.
5560 (*vec_extractv2df_0_sse): Ditto.
5561 (sse2_storelpd): Ditto.
5562 (sse2_loadlpd): Ditto.
5563 (sse2_movsd): Ditto.
5564 (*vec_concatv4si): Ditto.
5565 (vec_concatv2di): Ditto.
5567 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5568 for Intel asm dialect.
5569 (mmx_punpcklwd): Ditto.
5570 (mmx_punpckldq): Ditto.
5572 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5573 for intel assembler dialect.
5575 2013-07-06 Jakub Jelinek <jakub@redhat.com>
5578 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5579 for BUILT_IN_C{LZ,LRSB}*.
5580 * tree.h (CASE_INT_FN): Add FN##IMAX case.
5581 * tree-vrp.c (extract_range_basic): Handle
5582 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
5583 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5584 fall thru to code calling set_value*.
5585 * builtins.c (expand_builtin): Remove *IMAX cases.
5586 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5587 if width is bigger than 2*HWI.
5589 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
5591 PR rtl-optimization/55342
5592 * lra-int.h (lra_subreg_reload_pseudos): New.
5593 * lra.c: Add undoing optional reloads to the block diagram.
5594 (lra_subreg_reload_pseudos): New.
5595 (lra_optional_reload_pseudos): Change comments.
5596 (lra): Init and clear lra_subreg_reload_pseudos. Clear
5597 lra_optional_reload_pseudos after undo transformations.
5598 * lra-assigns.c (pseudo_prefix_title): New.
5599 (lra_setup_reg_renumber): Use it.
5600 (spill_for): Ditto. Check subreg reload pseudos too.
5601 (assign_by_spills): Consider subreg reload pseudos too.
5602 * lra-constraints.c (simplify_operand_subreg): Use
5603 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5604 (curr_insn_transform): Recognize and do optional reloads.
5605 (undo_optional_reloads): New.
5606 (lra_undo_inheritance): Call undo_optional_reloads.
5608 2013-07-05 Thomas Quinot <quinot@adacore.com>
5610 * tree-complex.c (expand_complex_operations_1): Fix typo.
5612 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
5614 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5615 (tune_params): New member 'const vec_costs'.
5616 * config/aarch64/aarch64.c (generic_vector_cost): New.
5617 (generic_tunings): New member 'generic_vector_cost'.
5618 (aarch64_builtin_vectorization_cost): New.
5619 (aarch64_add_stmt_cost): New.
5620 (TARGET_VECTORIZE_ADD_STMT_COST): New.
5621 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5623 2013-07-03 Jakub Jelinek <jakub@redhat.com>
5626 * config/i386/predicates.md (vsib_address_operand): Disallow
5627 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5629 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
5632 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5633 expand_builtin_setjmp_receiver.
5634 (expand_label): Adjust, call expand_builtin_setjmp_receiver
5635 with NULL for the label parameter.
5636 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5637 the frame-pointer. Adjust comments.
5638 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5639 only if LABEL is non-NULL.
5641 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
5643 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5644 (ARM_ABI_AAPCS64): Ditto.
5646 (ARM_DEFAULT_ABI): Ditto.
5648 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
5650 * config/aarch64/aarch64-builtins.c
5651 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5652 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5654 * config/aarch64/aarch64-simd.md
5655 (aarch64_ld1<VALL:mode>): New.
5656 (aarch64_st1<VALL:mode>): Likewise.
5657 * config/aarch64/arm_neon.h
5658 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5660 2013-07-02 Sriraman Tallam <tmsriram@google.com>
5662 * config/i386/i386.c (gate_insert_vzeroupper): Check if
5664 (ix86_option_override_internal):Turn on all -mavx target flags by
5665 default as they are dependent on AVX anyway.
5667 2013-07-02 Cary Coutant <ccoutant@google.com>
5669 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5671 (loc_checksum_ordered): Likewise.
5672 (hash_loc_operands): Remove inline keyword.
5674 2013-07-02 Jakub Jelinek <jakub@redhat.com>
5676 PR tree-optimization/57741
5677 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5678 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5679 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5680 Allow REAL_CST step_exprs if flag_associative_math.
5681 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5683 2013-07-02 Ian Bolton <ian.bolton@arm.com>
5685 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5687 2013-07-02 Ian Bolton <ian.bolton@arm.com>
5689 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5691 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5693 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5695 (iorsi3_insn): Likewise.
5696 (arm_xorsi3): Likewise.
5698 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
5700 * arm.md (attribute "wtype"): Delete. Move attribute values from here
5702 (attribute "type"): ... here, and prefix with "wmmx_".
5703 (attribute "core_cycles"): Update for attribute changes.
5704 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5705 (tbcstv4hi): Likewise.
5706 (tbcstv2si): Likewise.
5707 (iwmmxt_iordi3): Likewise.
5708 (iwmmxt_xordi3): Likewise.
5709 (iwmmxt_anddi3): Likewise.
5710 (iwmmxt_nanddi3): Likewise.
5711 (iwmmxt_arm_movdi): Likewise.
5712 (iwmmxt_movsi_insn): Likewise.
5713 (mov<mode>_internal): Likewise.
5714 (and<mode>3_iwmmxt): Likewise.
5715 (ior<mode>3_iwmmxt): Likewise.
5716 (xor<mode>3_iwmmxt): Likewise.
5717 (add<mode>3_iwmmxt): Likewise.
5718 (ssaddv8qi3): Likewise.
5719 (ssaddv4hi3): Likewise.
5720 (ssaddv2si3): Likewise.
5721 (usaddv8qi3): Likewise.
5722 (usaddv4hi3): Likewise.
5723 (usaddv2si3): Likewise.
5724 (sub<mode>3_iwmmxt): Likewise.
5725 (sssubv8qi3): Likewise.
5726 (sssubv4hi3): Likewise.
5727 (sssubv2si3): Likewise.
5728 (ussubv8qi3): Likewise.
5729 (ussubv4hi3): Likewise.
5730 (ussubv2si3): Likewise.
5731 (mulv4hi3_iwmmxt): Likewise.
5732 (smulv4hi3_highpart): Likewise.
5733 (umulv4hi3_highpart): Likewise.
5734 (iwmmxt_wmacs): Likewise.
5735 (iwmmxt_wmacsz): Likewise.
5736 (iwmmxt_wmacu): Likewise.
5737 (iwmmxt_wmacuz): Likewise.
5738 (iwmmxt_clrdi): Likewise.
5739 (iwmmxt_clrv8qi): Likewise.
5740 (iwmmxt_clr4hi): Likewise.
5741 (iwmmxt_clr2si): Likewise.
5742 (iwmmxt_uavgrndv8qi3): Likewise.
5743 (iwmmxt_uavgrndv4hi3): Likewise.
5744 (iwmmxt_uavgv8qi3): Likewise.
5745 (iwmmxt_uavgv4hi3): Likewise.
5746 (iwmmxt_tinsrb): Likewise.
5747 (iwmmxt_tinsrh): Likewise.
5748 (iwmmxt_tinsrw): Likewise.
5749 (iwmmxt_textrmub): Likewise.
5750 (iwmmxt_textrmsb): Likewise.
5751 (iwmmxt_textrmuh): Likewise.
5752 (iwmmxt_textrmsh): Likewise.
5753 (iwmmxt_textrmw): Likewise.
5754 (iwmxxt_wshufh): Likewise.
5755 (eqv8qi3): Likewise.
5756 (eqv4hi3): Likewise.
5757 (eqv2si3): Likewise.
5758 (gtuv8qi3): Likewise.
5759 (gtuv4hi3): Likewise.
5760 (gtuv2si3): Likewise.
5761 (gtv8qi3): Likewise.
5762 (gtv4hi3): Likewise.
5763 (gtv2si3): Likewise.
5764 (smax<mode>3_iwmmxt): Likewise.
5765 (umax<mode>3_iwmmxt): Likewise.
5766 (smin<mode>3_iwmmxt): Likewise.
5767 (umin<mode>3_iwmmxt): Likewise.
5768 (iwmmxt_wpackhss): Likewise.
5769 (iwmmxt_wpackwss): Likewise.
5770 (iwmmxt_wpackdss): Likewise.
5771 (iwmmxt_wpackhus): Likewise.
5772 (iwmmxt_wpackwus): Likewise.
5773 (iwmmxt_wpackdus): Likewise.
5774 (iwmmxt_wunpckihb): Likewise.
5775 (iwmmxt_wunpckihh): Likewise.
5776 (iwmmxt_wunpckihw): Likewise.
5777 (iwmmxt_wunpckilb): Likewise.
5778 (iwmmxt_wunpckilh): Likewise.
5779 (iwmmxt_wunpckilw): Likewise.
5780 (iwmmxt_wunpckehub): Likewise.
5781 (iwmmxt_wunpckehuh): Likewise.
5782 (iwmmxt_wunpckehuw): Likewise.
5783 (iwmmxt_wunpckehsb): Likewise.
5784 (iwmmxt_wunpckehsh): Likewise.
5785 (iwmmxt_wunpckehsw): Likewise.
5786 (iwmmxt_wunpckelub): Likewise.
5787 (iwmmxt_wunpckeluh): Likewise.
5788 (iwmmxt_wunpckeluw): Likewise.
5789 (iwmmxt_wunpckelsb): Likewise.
5790 (iwmmxt_wunpckelsh): Likewise.
5791 (iwmmxt_wunpckelsw): Likewise.
5792 (ror<mode>3): Likewise.
5793 (ashr<mode>3_iwmmxt): Likewise.
5794 (lshr<mode>3_iwmmxt): Likewise.
5795 (ashl<mode>3_iwmmxt): Likewise.
5796 (ror<mode>3_di): Likewise.
5797 (ashr<mode>3_di): Likewise.
5798 (lshr<mode>3_di): Likewise.
5799 (ashl<mode>3_di): Likewise.
5800 (iwmmxt_wmadds): Likewise.
5801 (iwmmxt_wmaddu): Likewise.
5802 (iwmmxt_tmia): Likewise.
5803 (iwmmxt_tmiaph): Likewise.
5804 (iwmmxt_tmiabb): Likewise.
5805 (iwmmxt_tmiatb): Likewise.
5806 (iwmmxt_tmiabt): Likewise.
5807 (iwmmxt_tmiatt): Likewise.
5808 (iwmmxt_tmovmskb): Likewise.
5809 (iwmmxt_tmovmskh): Likewise.
5810 (iwmmxt_tmovmskw): Likewise.
5811 (iwmmxt_waccb): Likewise.
5812 (iwmmxt_wacch): Likewise.
5813 (iwmmxt_waccw): Likewise.
5814 (iwmmxt_waligni): Likewise.
5815 (iwmmxt_walignr): Likewise.
5816 (iwmmxt_walignr0): Likewise.
5817 (iwmmxt_walignr1): Likewise.
5818 (iwmmxt_walignr2): Likewise.
5819 (iwmmxt_walignr3): Likewise.
5820 (iwmmxt_wsadb): Likewise.
5821 (iwmmxt_wsadh): Likewise.
5822 (iwmmxt_wsadbz): Likewise.
5823 (iwmmxt_wsadhz): Likewise.
5824 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5825 (iwmmxt_wabsdiffb): Likewise.
5826 (iwmmxt_wabsdiffh): Likewise.
5827 (iwmmxt_wabsdiffw): Likewise.
5828 (iwmmxt_waddsubhx): Likewise
5829 (iwmmxt_wsubaddhx): Likewise.
5830 (addc<mode>3): Likewise.
5831 (iwmmxt_avg4): Likewise.
5832 (iwmmxt_avg4r): Likewise.
5833 (iwmmxt_wmaddsx): Likewise.
5834 (iwmmxt_wmaddux): Likewise.
5835 (iwmmxt_wmaddsn): Likewise.
5836 (iwmmxt_wmaddun): Likewise.
5837 (iwmmxt_wmulwsm): Likewise.
5838 (iwmmxt_wmulwum): Likewise.
5839 (iwmmxt_wmulsmr): Likewise.
5840 (iwmmxt_wmulumr): Likewise.
5841 (iwmmxt_wmulwsmr): Likewise.
5842 (iwmmxt_wmulwumr): Likewise.
5843 (iwmmxt_wmulwl): Likewise.
5844 (iwmmxt_wqmulm): Likewise.
5845 (iwmmxt_wqmulwm): Likewise.
5846 (iwmmxt_wqmulmr): Likewise.
5847 (iwmmxt_wqmulwmr): Likewise.
5848 (iwmmxt_waddbhusm): Likewise.
5849 (iwmmxt_waddbhusl): Likewise.
5850 (iwmmxt_wqmiabb): Likewise.
5851 (iwmmxt_wqmiabt): Likewise.
5852 (iwmmxt_wqmiatb): Likewise.
5853 (iwmmxt_wqmiatt): Likewise.
5854 (iwmmxt_wqmiabbn): Likewise.
5855 (iwmmxt_wqmiabtn): Likewise.
5856 (iwmmxt_wqmiatbn): Likewise.
5857 (iwmmxt_wqmiattn): Likewise.
5858 (iwmmxt_wmiabb): Likewise.
5859 (iwmmxt_wmiabt): Likewise.
5860 (iwmmxt_wmiatb): Likewise.
5861 (iwmmxt_wmiatt): Likewise.
5862 (iwmmxt_wmiabbn): Likewise.
5863 (iwmmxt_wmiabtn): Likewise.
5864 (iwmmxt_wmiatbn): Likewise.
5865 (iwmmxt_wmiattn): Likewise.
5866 (iwmmxt_wmiawbb): Likewise.
5867 (iwmmxt_wmiawbt): Likewise.
5868 (iwmmxt_wmiawtb): Likewise.
5869 (iwmmxt_wmiawtt): Likewise.
5870 (iwmmxt_wmiawbbn): Likewise.
5871 (iwmmxt_wmiawbtn): Likewise.
5872 (iwmmxt_wmiawtbn): Likewise.
5873 (iwmmxt_wmiawttn): Likewise.
5874 (iwmmxt_wmerge): Likewise.
5875 (iwmmxt_tandc<mode>3): Likewise.
5876 (iwmmxt_torc<mode>3): Likewise.
5877 (iwmmxt_torvsc<mode>3): Likewise.
5878 (iwmmxt_textrc<mode>3): Likewise.
5879 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5880 (wmmxt_pack): Likewise.
5881 (wmmxt_mult_c1): Likewise.
5882 (wmmxt_mult_c2): Likewise.
5883 (wmmxt_alu_c1): Likewise.
5884 (wmmxt_alu_c2): Likewise.
5885 (wmmxt_alu_c3): Likewise.
5886 (wmmxt_transfer_c1): Likewise.
5887 (wmmxt_transfer_c2): Likewise.
5888 (wmmxt_transfer_c3): Likewise.
5889 (marvell_f_iwmmxt_wstr): Likewise.
5890 (marvell_f_iwmmxt_wldr): Likewise.
5892 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
5894 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5896 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5899 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5900 * lra-constraints.c (need_for_split_p): Check call used hard regs
5901 living through calls.
5903 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5904 call used regs for call insn.
5906 2013-06-28 Jakub Jelinek <jakub@redhat.com>
5909 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5910 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5912 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
5914 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5916 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5918 * lra-constraints.c (need_for_split_p): Check call used hard regs
5919 living through calls.
5921 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
5924 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5925 to tie with any other modes. Eliminate Altivec vector mode tests,
5926 since these are a subset of ALTIVEC or VSX vector modes. Simplify
5927 code, to return 0 if testing MODE2 for a condition, if we've
5928 already tested MODE1 for the same condition.
5930 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5932 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5935 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5937 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5938 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5940 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
5942 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5944 (aarch64_symbolic_constant_p): Remove.
5945 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5946 static. Fix line length and white space.
5947 (aarch64_symbolic_constant_p): Remove.
5948 * config/aarch64/predicates.md (aarch64_valid_symref):
5949 Use aarch64_classify_symbol_expression.
5951 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5953 * config/arm/constraints.md (Ts): New constraint.
5954 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5956 (compare_scc): Use "Ts" constraint for operand 0.
5957 (ior_scc_scc): Likewise.
5958 (and_scc_scc): Likewise.
5959 (and_scc_scc_nodom): Likewise.
5960 (ior_scc_scc_cmp): Likewise for operand 7.
5961 (and_scc_scc_cmp): Likewise.
5962 * config/arm/thumb2.md (thumb2_movsi_insn):
5963 Add alternatives for 16-bit encodings.
5964 (thumb2_movhi_insn): Likewise.
5965 (thumb2_movsicc_insn): Likewise.
5966 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
5967 (thumb2_negscc): Use "Ts" constraint.
5968 Move mvn instruction outside cond_exec block.
5969 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5970 for 16-bit encodings.
5972 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5974 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5976 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5977 (mulsi3subsi): Likewise.
5978 (mulsidi3adddi): Likewise.
5979 (mulsidi3_v6): Likewise.
5980 (umulsidi3_v6): Likewise.
5981 (umulsidi3adddi_v6): Likewise.
5982 (smulsi3_highpart_v6): Likewise.
5983 (umulsi3_highpart_v6): Likewise.
5984 (mulhisi3tb): Likewise.
5985 (mulhisi3bt): Likewise.
5986 (mulhisi3tt): Likewise.
5987 (maddhisi4): Likewise.
5988 (maddhisi4tb): Likewise.
5989 (maddhisi4tt): Likewise.
5990 (maddhidi4): Likewise.
5991 (maddhidi4tb): Likewise.
5992 (maddhidi4tt): Likewise.
5993 (zeroextractsi_compare0_scratch): Likewise.
5994 (insv_zero): Likewise.
5995 (insv_t2): Likewise.
5996 (anddi_notzesidi_di): Likewise.
5997 (anddi_notsesidi_di): Likewise.
5998 (andsi_notsi_si): Likewise.
5999 (iordi_zesidi_di): Likewise.
6000 (xordi_zesidi_di): Likewise.
6001 (andsi_iorsi3_notsi): Likewise.
6003 (smax_m1): Likewise.
6005 (not_shiftsi): Likewise.
6006 (unaligned_loadsi): Likewise.
6007 (unaligned_loadhis): Likewise.
6008 (unaligned_loadhiu): Likewise.
6009 (unaligned_storesi): Likewise.
6010 (unaligned_storehi): Likewise.
6011 (extv_reg): Likewise.
6012 (extzv_t2): Likewise.
6014 (udivsi3): Likewise.
6015 (arm_zero_extendhisi2addsi): Likewise.
6016 (arm_zero_extendqisi2addsi): Likewise.
6017 (compareqi_eq0): Likewise.
6018 (arm_extendhisi2_v6): Likewise.
6019 (arm_extendqisi2addsi): Likewise.
6020 (arm_movt): Likewise.
6021 (thumb2_ldrd): Likewise.
6022 (thumb2_ldrd_base): Likewise.
6023 (thumb2_ldrd_base_neg): Likewise.
6024 (thumb2_strd): Likewise.
6025 (thumb2_strd_base): Likewise.
6026 (thumb2_strd_base_neg): Likewise.
6027 (arm_negsi2): Add alternative for 16-bit encoding.
6028 (arm_one_cmplsi2): Likewise.
6030 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6032 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
6033 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
6034 (movdfcc): Likewise.
6035 * config/arm/vfp.md (*thumb2_movsf_vfp):
6036 Disable predication for arm_restrict_it.
6037 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
6038 (*thumb2_movdfcc_vfp): Likewise.
6039 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
6040 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
6041 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
6042 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
6043 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
6044 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
6045 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
6046 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
6047 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
6048 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
6049 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
6050 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
6051 Disable predication for arm_restrict_it.
6053 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
6055 * config/i386/bmiintrin.h (_bextr_u32): New.
6056 (_bextr_u64): Ditto.
6058 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
6060 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
6061 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
6062 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
6063 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
6064 * config/mips/n32-elf.h: ...this new file.
6066 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
6069 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
6070 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
6072 2013-06-27 Catherine Moore <clm@codesourcery.com>
6074 * config/mips/mips-tables.opt: Regenerate.
6075 * config/mips/mips-cpus.def: Add m14ke and m14kec.
6076 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
6077 * doc/invoke.texi: Add -m14kc.
6079 2013-06-27 Jakub Jelinek <jakub@redhat.com>
6082 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
6083 constraints of operand 1 and 2.
6086 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
6087 to match RTL canonicalization. Swap predicates and
6088 constraints of operand 1 and 2.
6090 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
6092 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
6093 Process OP_INOUT regs for splitting too.
6095 2013-06-27 Jakub Jelinek <jakub@redhat.com>
6097 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
6098 decl before the loop, initialize to NULL.
6099 (vectorizable_load): Initialize ptr_incr to NULL.
6101 2013-06-27 Martin Jambor <mjambor@suse.cz>
6104 * ipa-ref.h (ipa_maybe_record_reference): Declare.
6105 * ipa-ref.c (ipa_maybe_record_reference): New function.
6106 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
6107 * ipa-cp.c (create_specialized_node): Record potential references from
6109 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
6111 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
6113 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
6114 parameter 'mode' of type 'enum machine_mode mode'; change to pass
6115 'mode' to force_reg.
6116 (aarch64_add_offset): Update calls to aarch64_force_temporary.
6117 (aarch64_expand_mov_immediate): Likewise.
6119 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
6121 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
6122 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
6124 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6126 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
6127 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6128 (struct machine_function): Add tbegin_p.
6129 (s390_canonicalize_comparison): Fold CC mode compares to
6130 conditional jump if possible.
6131 (s390_emit_jump): Return the emitted jump.
6132 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
6133 Handle CCRAWmode compares.
6134 (s390_option_override): Default to -mhtm if available.
6135 (s390_reg_clobbered_rtx): Handle floating point regs as well.
6136 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
6137 FPRs instead of df_regs_ever_live_p.
6138 (s390_optimize_nonescaping_tx): New function.
6139 (s390_init_frame_layout): Extend clobbered_regs array to cover
6141 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
6142 (s390_expand_tbegin): New function.
6143 (enum s390_builtin): New enum definition.
6144 (code_for_builtin): New array definition.
6145 (s390_init_builtins): New function.
6146 (s390_expand_builtin): New function.
6147 (TARGET_INIT_BUILTINS): Define.
6148 (TARGET_EXPAND_BUILTIN): Define.
6149 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
6150 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
6151 (s390_alc_comparison): Likewise.
6152 * config/s390/s390-modes.def: Add CCRAWmode.
6153 * config/s390/s390.h (processor_flags): Add PF_TX.
6154 (TARGET_CPU_HTM): Define macro.
6155 (TARGET_HTM): Define macro.
6156 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
6157 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
6158 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6159 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
6160 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
6161 (TBEGIN_MASK, TBEGINC_MASK): New constants.
6162 ("*cc_to_int"): Move up.
6163 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
6164 constants other than 0.
6165 ("*ccraw_to_int"): New insn and splitter definition.
6166 ("tbegin", "tbegin_nofloat", "tbegin_retry")
6167 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
6168 ("tx_assist"): New expander.
6169 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
6170 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
6171 * config/s390/s390.opt: Add -mhtm option.
6172 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
6173 * config/s390/htmxlintrin.h: New file.
6174 * config/s390/htmintrin.h: New file.
6175 * config/s390/s390intrin.h: New file.
6176 * doc/extend.texi: Document htm builtins.
6177 * config.gcc: Add the new header files to extra_headers.
6179 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
6181 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
6182 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
6184 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6185 Pat Haugen <pthaugen@us.ibm.com>
6186 Peter Bergner <bergner@vnet.ibm.com>
6188 * config/rs6000/power8.md: New.
6189 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
6190 setting for power8 entry.
6191 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
6192 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
6193 test for Power4/Power5 only.
6194 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
6196 (force_new_group): Adjust comment.
6197 * config/rs6000/rs6000.md: Include power8.md.
6199 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
6201 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
6202 * config/arm/arm-protos.h (arm_max_conditional_execute): New
6204 (tune_params): Update comment.
6205 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
6206 (arm_max_conditional_execute): New function.
6207 (thumb2_final_prescan_insn): Use max_insn_skipped and
6208 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
6210 2013-06-25 Jakub Jelinek <jakub@redhat.com>
6212 PR tree-optimization/57705
6213 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
6214 SSA_NAME step, provided that it is not defined inside the loop.
6215 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
6216 (get_initial_def_for_induction): Handle SSA_NAME IV step.
6218 2013-06-25 Martin Jambor <mjambor@suse.cz>
6221 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
6222 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
6224 (ipa_note_param_call): Initialize member_ptr flag.
6225 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
6226 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
6227 (ipa_write_indirect_edge_info): Stream member_ptr flag.
6228 (ipa_read_indirect_edge_info): Likewise.
6230 2013-06-25 Richard Biener <rguenther@suse.de>
6233 * passes.c (init_optimization_passes): Move pass_fold_builtins
6234 and pass_dce earlier with -Og.
6236 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
6238 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
6239 <BIT_FIELD_REF>: Remove trailing TAB.
6240 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
6243 2013-06-24 Martin Jambor <mjambor@suse.cz>
6245 PR tree-optimization/57358
6246 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
6247 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
6248 (ipa_analyze_params_uses): Generate pessimistic info when true.
6250 2013-06-24 Martin Jambor <mjambor@suse.cz>
6252 PR tree-optimization/57539
6253 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
6254 global.inlined_to of the new node to it. All callers changed.
6255 * ipa-inline-transform.c (clone_inlined_nodes): New variable
6256 inlining_into, pass it to cgraph_clone_node.
6257 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
6258 ipa_free_edge_args_substructures.
6259 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
6260 rdesc linked list. Do not assert rdesc edges have inlined caller.
6261 Assert we have found an rdesc in the rdesc list.
6263 2013-06-24 Richard Biener <rguenther@suse.de>
6265 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
6266 (pointer_set_lookup): Declare.
6267 (class pointer_map): New template class implementing a
6268 generic pointer to T map.
6269 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
6270 pointer_map<T>::contains, pointer_map<T>::insert,
6271 pointer_map<T>::traverse): New functions.
6272 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
6273 (pointer_set_lookup): New function.
6274 (pointer_set_contains): Use pointer_set_lookup.
6275 (pointer_set_insert): Likewise.
6276 (insert_aux): Remove.
6277 (struct pointer_map_t): Embed a pointer_set_t.
6278 (pointer_map_create): Adjust.
6279 (pointer_map_destroy): Likewise.
6280 (pointer_map_contains): Likewise.
6281 (pointer_map_insert): Likewise.
6282 (pointer_map_traverse): Likewise.
6283 * tree-streamer.h (struct streamer_tree_cache_d): Use a
6284 pointer_map<unsigned> instead of a pointer_map_t.
6285 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
6286 (streamer_tree_cache_lookup): Likewise.
6287 (streamer_tree_cache_create): Likewise.
6288 (streamer_tree_cache_delete): Likewise.
6289 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
6290 pointer_map<unsigned> instead of a pointer_map_t.
6291 (lto_init_tree_ref_encoder): Adjust.
6292 (lto_destroy_tree_ref_encoder): Likewise.
6293 * lto-section-out.c (lto_output_decl_index): Likewise.
6294 (lto_record_function_out_decl_state): Likewise.
6295 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
6297 2013-06-24 Richard Biener <rguenther@suse.de>
6299 PR tree-optimization/57488
6300 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
6302 2013-06-24 Alan Modra <amodra@gmail.com>
6304 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
6305 (gen_easy_altivec_constant): Likewise.
6306 * config/rs6000/predicates.md (easy_vector_constant_add_self,
6307 easy_vector_constant_msb): Likewise.
6309 2013-06-23 Jakub Jelinek <jakub@redhat.com>
6312 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
6313 add missing return true.
6315 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
6318 * config/sh/predicates.md (general_extend_operand): Invoke
6319 general_movsrc_operand for memory operands.
6320 (general_movsrc_operand): Allow reg+reg addressing, do not use
6321 general_operand for memory operands.
6323 2013-06-23 Sriraman Tallam <tmsriram@google.com>
6325 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
6326 when current target options does not apply.
6327 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
6328 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
6329 * config/i386/bmiintrin.h: Pass appropriate target
6330 attributes to header.
6331 * config/i386/mmintrin.h: Ditto.
6332 * config/i386/nmmintrin.h: Ditto.
6333 * config/i386/avx2intrin.h: Ditto.
6334 * config/i386/fxsrintrin.h: Ditto.
6335 * config/i386/tbmintrin.h: Ditto.
6336 * config/i386/xsaveintrin.h: Ditto.
6337 * config/i386/f16cintrin.h: Ditto.
6338 * config/i386/xtestintrin.h: Ditto.
6339 * config/i386/xsaveoptintrin.h: Ditto.
6340 * config/i386/bmi2intrin.h: Ditto.
6341 * config/i386/lzcntintrin.h: Ditto.
6342 * config/i386/smmintrin.h: Ditto.
6343 * config/i386/wmmintrin.h: Ditto.
6344 * config/i386/x86intrin.h: Remove all header include guards.
6345 * config/i386/prfchwintrin.h: Ditto.
6346 * config/i386/pmmintrin.h: Ditto.
6347 * config/i386/tmmintrin.h: Ditto.
6348 * config/i386/xmmintrin.h: Ditto.
6349 * config/i386/popcntintrin.h: Ditto.
6350 * config/i386/rdseedintrin.h: Ditto.
6351 * config/i386/ammintrin.h: Ditto.
6352 * config/i386/emmintrin.h: Ditto.
6353 * config/i386/immintrin.h: Remove all header include guards.
6354 * config/i386/fma4intrin.h: Ditto.
6355 * config/i386/lwpintrin.h: Ditto.
6356 * config/i386/xopintrin.h: Ditto.
6357 * config/i386/ia32intrin.h: Ditto.
6358 * config/i386/avxintrin.h: Ditto.
6359 * config/i386/rtmintrin.h: Ditto.
6360 * config/i386/fmaintrin.h: Ditto.
6361 * config/i386/mm3dnow.h: Ditto.
6363 2013-06-22 Sriraman Tallam <tmsriram@google.com>
6365 * common/config/i386/i386-common.c: Handle LZCNT.
6367 2013-06-22 Andi Kleen <ak@linux.intel.com>
6369 * doc/extend.texi: Use __atomic_store_n instead of
6370 __atomic_store in HLE example.
6372 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
6374 * config/sh/sh.c: Remove <cstdlib> workaround.
6376 2013-06-21 Andi Kleen <ak@linux.intel.com>
6378 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
6380 2013-06-21 Andi Kleen <ak@linux.intel.com>
6382 * doc/extend.texi: Document that __atomic_clear and
6383 __atomic_test_and_set should only be used with bool.
6385 2013-06-20 Jan Hubicka <jh@suse.cz>
6387 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
6389 * tree.c (decls_same_for_odr): New function.
6390 (same_for_edr): New function.
6391 (types_same_for_odr): New function.
6392 (get_binfo_at_offset): Use it.
6393 * tree.h (types_same_for_odr): Declare.
6395 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
6396 Jason Merrill <jason@redhat.com>
6398 * system.h: Include <cstdlib> as well as <stdlib.h>.
6400 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
6403 * config/i386/i386.c (construct_container): Report error if
6404 long double is used with disabled x87 float returns.
6406 2013-06-20 Jan Hubicka <jh@suse.cz>
6408 * lto-cgraph.c (input_symtab): Do not set cgraph state.
6410 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
6412 PR rtl-optimization/57425
6413 PR rtl-optimization/57569
6414 * alias.c (write_dependence_p): Remove parameters mem_mode and
6415 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
6416 Changed all callers.
6417 (canon_anti_dependence): Get comments and semantics in sync.
6418 Add parameter mem_canonicalized. Changed all callers.
6419 * rtl.h (canon_anti_dependence): Update prototype.
6421 2013-06-20 Richard Biener <rguenther@suse.de>
6423 * data-streamer-in.c (streamer_read_uhwi): Optimize single
6424 byte case, inline streamer_read_uchar and defer section
6427 2013-06-20 Richard Biener <rguenther@suse.de>
6429 PR tree-optimization/57584
6430 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
6431 SSA names into the expanded expression that take part in
6432 abnormal coalescing.
6434 2013-06-19 Sharad Singhai <singhai@google.com>
6436 * gcov.c (print_usage): Handle new option.
6437 (process_args): Ditto.
6438 (get_gcov_intermediate_filename): New function.
6439 (output_intermediate_file): New function.
6440 (output_gcov_file): New function
6441 (generate_results): Handle new option.
6442 (release_function): Relase demangled name.
6443 (read_graph_file): Handle demangled name.
6444 (output_lines): Ditto.
6445 * doc/gcov.texi: Document gcov intermediate format.
6447 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
6450 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6451 (lra_emit_add): Use the functions. Add comment about Y as an
6454 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
6457 * collect2.c (collect_atexit): New.
6458 (collect_exit): Delete.
6459 (main): Register collect_atexit with atexit.
6460 (collect_wait): Change collect_exit to exit.
6462 * collect2.h (collect_exit): Delete.
6463 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
6465 2013-06-19 Wei Mi <wmi@google.com>
6467 PR rtl-optimization/57518
6468 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
6469 if regno is used in paradoxical subreg.
6470 (update_equiv_regs): Check pdx_subregs[regno] before
6471 set a reg to be equivalent with a mem.
6473 2013-06-19 Matthias Klose <doko@ubuntu.com>
6476 * file-find.h (find_a_file): Add a mode parameter.
6477 * file-find.c (find_a_file): Likewise.
6478 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
6479 with X_OK for the executables.
6480 * collect2.c (main): Call find_a_file with X_OK.
6482 2013-06-19 Steve Ellcey <sellcey@mips.com>
6485 * config/mips/mips.md (casesi_internal_mips16_<mode>):
6486 Use NEXT_INSN instead of next_real_insn.
6488 2013-06-19 Jan Hubicka <jh@suse.cz>
6490 * cgraph.h (const_value_known_p): Replace by ...
6491 (ctor_for_folding): .. this one.
6492 * cgraphunit.c (process_function_and_variable_attributes): Use it.
6493 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
6494 * expr.c (expand_expr_real_1): Likewise.
6495 (string_constant): Likewise.
6496 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
6497 * ipa.c (process_references): Likewise.
6498 (symtab_remove_unreachable_nodes): Likewise.
6499 * ipa-inline-analysis.c (param_change_prob): Likewise.
6500 * gimple-fold.c (canonicalize_constructor_val): Likewise.
6501 (get_base_constructor): Likwise.
6502 * varpool.c (varpool_remove_node): Likewise.
6503 (varpool_remove_initializer): LIkewise.
6504 (dump_varpool_node): LIkwise.
6505 (const_value_known_p): Rewrite to ...
6506 (ctor_for_folding): ... this one.
6508 2013-06-19 Jakub Jelinek <jakub@redhat.com>
6511 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
6512 PERSONALITY in $PATH derived prefixes.
6514 2013-06-19 Jeff Law <law@redhat.com>
6516 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
6519 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6520 (simplify_bitwise_binary): Use it to simpify certain binary ops on
6523 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
6525 * config/arm/vfp.md: Move VFP instruction classification documentation
6527 * config/arm/arm.md: ... here. Update instruction classification
6530 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
6532 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6533 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6534 pattern. Use more efficient sequences on ARMv5 and Thumb2.
6536 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
6539 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6540 with NEXT_INSN. Use tablejump_p to check for jump table data
6543 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
6546 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6547 that now in C++ the value is correct per the C++ standards.
6549 2013-06-19 Richard Biener <rguenther@suse.de>
6551 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6554 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6557 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6560 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6561 with next_active_insn.
6563 2013-06-18 Sriraman Tallam <tmsriram@google.com>
6565 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6566 functions are inlined during failures to flag an error.
6567 * tree-inline.c (expand_call_inline): Allow the error to be flagged
6568 in early inline pass.
6570 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
6572 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6575 2013-06-18 Julian Brown <julian@codesourcery.com>
6577 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6578 Permit virtual register pre-reload if !strict.
6579 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6581 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6583 * config/arm/neon.md (movmisalign<mode>): Use
6584 neon_perm_struct_or_reg_operand instead of
6585 neon_struct_or_register_operand.
6586 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6587 neon_permissive_struct_operand instead of neon_struct_operand.
6588 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6589 neon_vector_mem_operand.
6590 * config/arm/predicates.md (neon_struct_operand): Adjust call to
6591 neon_vector_mem_operand.
6592 (neon_permissive_struct_operand): New.
6593 (neon_struct_or_register_operand): Rename to...
6594 (neon_perm_struct_or_reg_operand): This. Adjust call to
6595 neon_vector_mem_operand.
6597 2013-06-18 Richard Biener <rguenther@suse.de>
6599 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6600 * lto-streamer.h: Include pointer-set.h.
6601 (struct lto_decl_slot): Remove.
6602 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6603 Remove next_index entry.
6604 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6605 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6606 (lto_init_tree_ref_encoder): Adjust.
6607 (lto_destroy_tree_ref_encoder): Likewise.
6608 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6609 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6610 (lto_output_decl_index): Adjust.
6611 (lto_new_out_decl_state): Likewise.
6612 (lto_record_function_out_decl_state): Likewise.
6613 * lto-streamer-out.c (copy_function): Likewise.
6615 2013-06-18 Richard Biener <rguenther@suse.de>
6617 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6618 * cgraphunit.c: Include cfgloop.h.
6619 (init_lowered_empty_function): Initialize the loop tree.
6620 (assemble_thunk): Insert new BBs into loops.
6622 2013-06-18 Richard Biener <rguenther@suse.de>
6624 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6625 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6626 the map from cache entry to cache index optional.
6627 (streamer_tree_cache_replace_tree): Adjust accordingly.
6628 (streamer_tree_cache_append): Likewise.
6629 (streamer_tree_cache_delete): Likewise.
6630 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6631 streamer cache map from cache entry to cache index.
6632 * lto-streamer-out.c (create_output_block): Adjust.
6634 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
6636 * config/arm/arm.md (attribute "insn"): Move multiplication and
6637 division attributes to...
6638 (attribute "type"): ... here. Remove mult.
6639 (attribute "mul32"): New attribute.
6640 (attribute "mul64"): Add umaal.
6641 (*arm_mulsi3): Update attributes.
6642 (*arm_mulsi3_v6): Likewise.
6643 (*thumb_mulsi3): Likewise.
6644 (*thumb_mulsi3_v6): Likewise.
6645 (*mulsi3_compare0): Likewise.
6646 (*mulsi3_compare0_v6): Likewise.
6647 (*mulsi_compare0_scratch): Likewise.
6648 (*mulsi_compare0_scratch_v6): Likewise.
6649 (*mulsi3addsi): Likewise.
6650 (*mulsi3addsi_v6): Likewise.
6651 (*mulsi3addsi_compare0): Likewise.
6652 (*mulsi3addsi_compare0_v6): Likewise.
6653 (*mulsi3addsi_compare0_scratch): Likewise.
6654 (*mulsi3addsi_compare0_scratch_v6): Likewise.
6655 (*mulsi3subsi): Likewise.
6656 (*mulsidi3adddi): Likewise.
6657 (*mulsi3addsi_v6): Likewise.
6658 (*mulsidi3adddi_v6): Likewise.
6659 (*mulsidi3_nov6): Likewise.
6660 (*mulsidi3_v6): Likewise.
6661 (*umulsidi3_nov6): Likewise.
6662 (*umulsidi3_v6): Likewise.
6663 (*umulsidi3adddi): Likewise.
6664 (*umulsidi3adddi_v6): Likewise.
6665 (*smulsi3_highpart_nov6): Likewise.
6666 (*smulsi3_highpart_v6): Likewise.
6667 (*umulsi3_highpart_nov6): Likewise.
6668 (*umulsi3_highpart_v6): Likewise.
6669 (mulhisi3): Likewise.
6670 (*mulhisi3tb): Likewise.
6671 (*mulhisi3bt): Likewise.
6672 (*mulhisi3tt): Likewise.
6673 (maddhisi4): Likewise.
6674 (*maddhisi4tb): Likewise.
6675 (*maddhisi4tt): Likewise.
6676 (maddhidi4): Likewise.
6677 (*maddhidi4tb): Likewise.
6678 (*maddhidi4tt): Likewise.
6680 (udivsi3): Likewise.
6681 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6682 (thumb2_mulsi_short_compare0): Likewise.
6683 (thumb2_mulsi_short_compare0_scratch): Likewise.
6684 * config/arm/arm1020e.md (1020mult1): Update attribute change.
6685 (1020mult2): Likewise.
6686 (1020mult3): Likewise.
6687 (1020mult4): Likewise.
6688 (1020mult5): Likewise.
6689 (1020mult6): Likewise.
6690 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6692 (cortex_a15_mult64): Likewise.
6693 (cortex_a15_sdiv): Likewise.
6694 (cortex_a15_udiv): Likewise.
6695 * config/arm/arm1026ejs.md (mult1): Update attribute change.
6701 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6702 (pj4_ir_div): Likewise.
6703 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6704 (11_mult2): Likewise.
6705 (11_mult3): Likewise.
6706 (11_mult4): Likewise.
6707 (11_mult5): Likewise.
6708 (11_mult6): Likewise.
6709 (11_mult7): Likewise.
6710 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6711 (cortex_a8_mla): Likewise.
6712 (cortex_a8_mull): Likewise.
6713 (cortex_a8_smulwy): Likewise.
6714 (cortex_a8_smlald): Likewise.
6715 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6716 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6717 (cortex_r4_mul_3): Likewise.
6718 (cortex_r4_mla_4): Likewise.
6719 (cortex_r4_mla_3): Likewise.
6720 (cortex_r4_smlald): Likewise.
6721 (cortex_r4_mull): Likewise.
6722 (cortex_r4_sdiv): Likewise.
6723 (cortex_r4_udiv): Likewise.
6724 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6725 (cortex_a7_idiv): Likewise.
6726 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6727 (9_mult2): Likewise.
6728 (9_mult3): Likewise.
6729 (9_mult4): Likewise.
6730 (9_mult5): Likewise.
6731 (9_mult6): Likewise.
6732 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6733 (cortex_a53_sdiv): Likewise.
6734 (cortex_a53_udiv): Likewise.
6735 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6736 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6737 (mp626_mult2): Likewise.
6738 (mp626_mult3): Likewise.
6739 (mp626_mult4): Likewise.
6740 * config/arm/fa526.md (526_mult1): Update attribute change.
6741 (526_mult2): Likewise.
6742 * config/arm/arm-generic.md (mult): Update attribute change.
6743 (mult_ldsched_strongarm): Likewise.
6744 (mult_ldsched): Likewise.
6745 (multi_cycle): Likewise.
6746 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6747 * config/arm/fa606te.md (606te_mult1): Update attribute change.
6748 (606te_mult2): Likewise.
6749 (606te_mult3): Likewise.
6750 (606te_mult4): Likewise.
6751 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6752 (cortex_a9_mac16): Likewise.
6753 (cortex_a9_multiply): Likewise.
6754 (cortex_a9_mac): Likewise.
6755 (cortex_a9_multiply_long): Likewise.
6756 * config/arm/fa626te.md (626te_mult1): Update attribute change.
6757 (626te_mult2): Likewise.
6758 (626te_mult3): Likewise.
6759 (626te_mult4): Likewise.
6761 2013-06-18 Richard Biener <rguenther@suse.de>
6764 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6766 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6769 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6770 with next_active_insn.
6772 2013-06-18 Alan Modra <amodra@gmail.com>
6774 * config/rs6000/rs6000.h (enum data_align): New.
6775 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6776 (DATA_ABI_ALIGNMENT): Define.
6777 (CONSTANT_ALIGNMENT): Correct comment.
6778 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6779 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6781 2013-06-17 David Malcolm <dmalcolm@redhat.com>
6783 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6784 ATTRIBUTE_UNUSED marking.
6786 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
6788 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6789 alternative and update.
6790 (aarch64_dup_lanedi): Delete.
6791 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6792 * config/aarch64/aarch64-simd-builtins.def: Update.
6794 2013-06-17 Richard Biener <rguenther@suse.de>
6796 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6797 (lto_input_scc): Declare.
6798 (lto_input_tree_1): Likewise.
6799 (struct lto_stats_d): Add num_tree_bodies_output and
6800 num_pickle_refs_output.
6801 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6802 (lto_read_tree_1): Split out from ...
6803 (lto_read_tree): ... this.
6804 (lto_input_scc): New function.
6805 (lto_input_tree_1): Split out from ...
6806 (lto_input_tree): ... this. Handle LTO_tree_scc.
6807 (lto_data_in_create): Create the streamer cache without hashes.
6808 * lto-streamer-out.c (create_output_block): Create the streamer
6809 cache with hashes when not doing WPA.
6810 (lto_write_tree_1): Split out from ...
6811 (lto_write_tree): ... this.
6812 (get_symbol_initial_value): New function.
6813 (lto_output_tree_1): Split out from ...
6814 (lto_output_tree): ... this. Write trees as series of SCCs
6815 using a DFS walk via DFS_write_tree.
6816 (struct sccs, struct scc_entry): New types.
6817 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6818 (DFS_write_tree_body): New function.
6819 (DFS_write_tree): Likewise.
6820 (hash_tree): Likewise.
6821 (scc_entry_compare): Likewise.
6822 (hash_scc): Likewise.
6823 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6824 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6825 TREE_CHAIN as regular reference.
6826 (streamer_read_integer_cst): Remove.
6827 (streamer_get_pickled_tree): Adjust.
6828 * tree-streamer-out.c (streamer_write_chain): Disable streaming
6829 of DECL_EXTERNALs in BLOCK_VARS for now.
6830 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6832 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6833 Add hash value argument and record that if hashes are recorded
6835 (streamer_tree_cache_insert_1): Adjust.
6836 (streamer_tree_cache_insert): Likewise.
6837 (streamer_tree_cache_insert_at): Rename to ...
6838 (streamer_tree_cache_replace_tree): ... this and adjust.
6839 (streamer_tree_cache_append): Adjust.
6840 (record_common_node): Likewise.
6841 (streamer_tree_cache_create): Add argument whether to
6842 record hash values together with trees.
6843 (streamer_tree_cache_delete): Adjust.
6844 * tree-streamer.h (struct streamer_tree_cache_d): Add
6846 (streamer_read_integer_cst): Remove.
6847 (streamer_tree_cache_insert): Adjust.
6848 (streamer_tree_cache_append): Likewise.
6849 (streamer_tree_cache_insert_at): Rename to ...
6850 (streamer_tree_cache_replace_tree): ... this and adjust.
6851 (streamer_tree_cache_create): Add argument whether to record hashes.
6852 (streamer_tree_cache_get): Rename to ...
6853 (streamer_tree_cache_get_tree): ... this.
6854 (streamer_tree_cache_get_hash): New function.
6855 * tree.c (cache_integer_cst): New function.
6856 * tree.h (cache_integer_cst): Declare.
6857 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6858 * lto-symtab.c (lto_varpool_replace_node): Only release
6859 DECL_INITIAL of non-prevailing decls.
6860 * varpool.c (varpool_remove_initializer): Do not release
6861 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6863 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
6865 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6866 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6867 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6868 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6869 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6870 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6871 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6872 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6873 instead of TARGET_64BIT.
6874 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6875 Require ISA_HAS_<D>DIV.
6877 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
6879 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6880 (mips*-*-linux*): Move default with_llsc setting to where other
6882 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6884 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6885 Likewise. Remove default with_tune setting. Move default float
6886 setting to its own block. Handle with_llsc in the same block as above.
6888 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
6890 PR rtl-optimization/57425
6891 PR rtl-optimization/57569
6892 * alias.c (write_dependence_p): Add new parameters mem_mode,
6893 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
6894 Changed all callers.
6895 (canon_anti_dependence): New function.
6896 * cse.c (check_dependence): Use canon_anti_dependence.
6897 * cselib.c (cselib_invalidate_mem): Likewise.
6898 * rtl.h (canon_anti_dependence): Declare.
6900 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
6902 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6903 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6904 ".set mips3" for 64-bit targets.
6906 2013-06-15 Dehao Chen <dehao@google.com>
6908 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6909 * gimple-low.c (gimple_check_call_matching_types): Likewise.
6910 (gimple_check_call_args): Likewise.
6911 * value-prof.c (check_ic_target): Likewise.
6912 * ipa-inline.c (early_inliner): Likewise.
6913 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6914 * cgraph.c (cgraph_create_edge_1): Likewise.
6915 (cgraph_make_edge_direct): Likewise.
6917 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6920 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6921 rs6000_output_move_128bit to handle emitting quad memory
6922 operations. Set attribute length to 8 bytes.
6924 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
6926 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6928 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6929 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6930 (aarch64_<su>mlsl<mode>): Likewise.
6932 2013-06-14 Mike Stump <mikestump@comcast.net>
6934 * Makefile.in (TARGET_H): Add insn-codes.h.
6936 2013-06-14 Alan Modra <amodra@gmail.com>
6940 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6941 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
6942 bitfield expansion when EXPAND_MEMORY.
6943 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6945 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
6947 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6948 test for clearing quad memory on 32-bit later.
6950 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
6952 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6953 (fold_negate_expr): Likewise.
6954 (fold_real_zero_addition_p): Handle vectors.
6955 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6957 2013-06-14 Alan Modra <amodra@gmail.com>
6959 * varasm.c (force_const_mem): Revert 2013-06-07 change.
6961 2013-06-13 Jan Hubicka <jh@suse.cz>
6963 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6964 Local comdats are not externally visible.
6965 * symtab.c (dump_symtab_base): Dump externally visible.
6966 (verify_symtab_base): Verify back links in the symtab hash.
6968 2013-06-13 Bin Cheng <bin.cheng@arm.com>
6970 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6971 CONVERT_EXPR as equal nodes.
6973 2013-06-13 Bin Cheng <bin.cheng@arm.com>
6975 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6977 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
6979 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6980 Generalize to complex and vector.
6981 * tree.c (build_all_ones_cst): New function.
6982 * tree.h (build_all_ones_cst): Declare it.
6984 2013-06-13 Alan Modra <amodra@gmail.com>
6986 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6987 * config/rs6000/rs6000.md (signbittf2): New insn.
6988 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6989 (abstf2_internal, cmptf_internal2): Likewise.
6990 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6992 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6993 Pat Haugen <pthaugen@us.ibm.com>
6994 Peter Bergner <bergner@vnet.ibm.com>
6996 * config/rs6000/rs6000.c (emit_load_locked): Add support for
6997 power8 byte, half-word, and quad-word atomic instructions.
6998 (emit_store_conditional): Likewise.
6999 (rs6000_expand_atomic_compare_and_swap): Likewise.
7000 (rs6000_expand_atomic_op): Likewise.
7002 * config/rs6000/sync.md (larx): Add new modes for power8.
7004 (AINT): New mode iterator to include TImode as well as normal
7005 integer modes on power8.
7006 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
7007 that VSX registers are not considered. Use AINT mode iterator
7008 instead of INT1 to allow inclusion of quad word atomic operations
7010 (load_locked<mode>): Likewise.
7011 (store_conditional<mode>): Likewise.
7012 (atomic_compare_and_swap<mode>): Likewise.
7013 (atomic_exchange<mode>): Likewise.
7014 (atomic_nand<mode>): Likewise.
7015 (atomic_fetch_<fetchop_name><mode>): Likewise.
7016 (atomic_nand_fetch<mode>): Likewise.
7017 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
7019 (ATOMIC): On power8, add QImode, HImode modes.
7020 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
7021 modes that promote to SImode.
7022 (load_lockedti): Convert TImode arguments to PTImode, so that we
7023 get a guaranteed even/odd register pair.
7024 (load_lockedpti): Likewise.
7025 (store_conditionalti): Likewise.
7026 (store_conditionalpti): Likewise.
7028 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
7029 atomic load/store instructions.
7032 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
7034 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
7036 (insn_count): New attribute, with most cases extracted from...
7037 (length): ...here. Redefine most cases in terms of insn_count.
7038 (single_insn): Delete.
7039 (can_delay): Use insn_count to check for single instructions.
7040 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
7041 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
7042 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
7043 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
7044 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
7045 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
7046 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
7047 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
7048 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
7049 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
7050 rather than "length".
7051 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
7052 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
7053 Use "insn_count" rather than "length".
7054 * config/mips/mips-dsp.md
7055 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
7056 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
7059 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
7061 PR tree-optimization/57361
7062 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
7064 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
7066 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
7068 (aarch64_simd_combine<mode>): New instruction expansion.
7069 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
7071 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
7072 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
7074 2013-06-12 Jan Hubicka <jh@suse.cz>
7076 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
7077 decl has when in streaming stage.
7078 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
7079 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
7081 2013-06-12 Roland Stigge <stigge@antcom.de>
7084 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
7086 2013-06-12 Jakub Jelinek <jakub@redhat.com>
7088 PR tree-optimization/57537
7089 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
7090 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
7092 2013-06-12 Richard Biener <rguenther@suse.de>
7094 * data-streamer.h (streamer_write_char_stream): CSE
7095 obs->current_pointer.
7096 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
7097 streamer_write_char_stream manually and optimize the resulting loop.
7098 (streamer_write_hwi_stream): Likewise.
7100 2013-06-12 Jan Hubicka <jh@suse.cz>
7102 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
7103 * cgraph.h (varpool_create_empty_node): Declare.
7104 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
7106 * symtab.c (symtab_unregister_node): Be lax about missin entries
7108 (symtab_get_node): Update comment.
7109 * varpool.c (varpool_create_empty_node): Break out from ...
7110 (varpool_node_for_decl): ... here.
7111 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
7113 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
7115 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
7116 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
7117 part. Use straight-line flow at the end.
7118 <COMPONENT_REF>: Remove superfluous else.
7119 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
7121 2013-06-12 Jakub Jelinek <jakub@redhat.com>
7124 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
7125 target hook even for !TREE_PUBLIC decls. If no resolution info
7126 is available, return false for common and external decls.
7128 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
7130 * config/rl78/constraints.md (U): New constraint.
7131 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
7134 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
7137 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
7138 to allow returning address to AT_PLATFORM name.
7140 2013-06-11 Jan Hubicka <jh@suse.cz>
7142 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
7143 * cgraph.h (symtab_node_base): Add weakref flag.
7144 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
7145 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
7146 (output_weakrefs): Use weakref flag.
7147 * fold-const.c (simple_operand_p): Handle WEAK.
7148 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
7149 * ipa.c (varpool_externally_visible_p): Drop weakref.
7150 (function_and_variable_visibility): Update comment; fix weakref
7151 sanity checks; do not clear DECL_WEAK on them.
7152 * lto-cgraph.c (lto_output_node): update.
7153 (lto_output_varpool_node): Update.
7154 (input_overwrite_node): Update.
7155 (input_node): Update.
7156 (input_varpool_node): Update.
7157 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
7158 (lto_symtab_merge_symbols): Add sanity check.
7159 (lto_symtab_prevailing_decl): Do not special case weakrefs.
7160 * passes.c (rest_of_decl_compilation): Set static flag, too.
7161 * symtab.c (dump_symtab_base): Dump weakref.
7162 (verify_symtab_base): Sanity check weakrefs.
7163 (symtab_make_decl_local): Remove duplicated code.
7164 (symtab_alias_ultimate_target): Simplify.
7165 * varpool.c (varpool_create_variable_alias): Set weakref flag.
7167 2013-06-11 DJ Delorie <dj@redhat.com>
7169 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
7170 (rl78_unwind_word_mode): New.
7172 2013-06-11 David Malcolm <dmalcolm@redhat.com>
7174 * final.c (debug_prefix_maps): Make static.
7176 2013-06-11 David Malcolm <dmalcolm@redhat.com>
7178 * function.c (initial_trampoline): Remove stray copy.
7180 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
7182 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
7184 2013-06-11 Martin Jambor <mjambor@suse.cz>
7186 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
7187 within bounds at the beginning of the function.
7189 2013-06-11 Alan Modra <amodra@gmail.com>
7191 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
7193 (get_named_section): Don't NULL !DECL_P decl.
7195 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
7197 * doc/invoke.texi (core-avx2): Document.
7199 (atom): Updated with MOVBE.
7201 2013-06-11 Richard Biener <rguenther@suse.de>
7203 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
7205 2013-06-11 Anton Blanchard <anton@samba.org>
7207 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
7208 correct shift value in little-endian mode.
7210 2013-06-11 Jakub Jelinek <jakub@redhat.com>
7213 * varasm.c (get_variable_align): Move #endif to the right place.
7215 2013-06-10 Cary Coutant <ccoutant@google.com>
7217 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
7218 for hash so that hash table traversal order is deterministic.
7220 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7221 Pat Haugen <pthaugen@us.ibm.com>
7222 Peter Bergner <bergner@vnet.ibm.com>
7224 * config/rs6000/vector.md (GPR move splitter): Do not split moves
7225 of vectors in GPRS if they are direct moves or quad word load or
7228 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
7230 (direct_move_p): Likewise.
7231 (quad_load_store_p): Likewise.
7233 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
7234 classes into bins based on the physical register type.
7235 (reg_class_to_reg_type): Likewise.
7236 (IS_STD_REG_TYPE): Likewise.
7237 (IS_FP_VECT_REG_TYPE): Likewise.
7238 (reload_fpr_gpr): Arrays to determine what insn to use if we can
7239 use direct move instructions.
7240 (reload_gpr_vsx): Likewise.
7241 (reload_vsx_gpr): Likewise.
7242 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
7243 information that is a simplification of register classes. Also
7244 precalculate direct move reload helpers.
7245 (direct_move_p): New function to return true if the operation can
7246 be done as a direct move instruciton.
7247 (quad_load_store_p): New function to return true if the operation
7248 is a quad memory operation.
7249 (rs6000_legitimize_address): If quad memory, only allow register
7250 indirect for TImode addresses.
7251 (rs6000_legitimate_address_p): Likewise.
7252 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
7253 (rs6000_reload_register_type): Likewise.
7254 (register_to_reg_type): Return register type.
7255 (rs6000_secondary_reload_simple_move): New helper function for
7256 secondary reload and secondary memory needed to identify anything
7257 that is a simple move, and does not need reloading.
7258 (rs6000_secondary_reload_direct_move): New helper function for
7259 secondary reload to identify cases that can be done with several
7260 instructions via the direct move instructions.
7261 (rs6000_secondary_reload_move): New helper function for secondary
7262 reload to identify moves between register types that can be done.
7263 (rs6000_secondary_reload): Add support for quad memory operations
7264 and for direct move.
7265 (rs6000_secondary_memory_needed): Likewise.
7266 (rs6000_debug_secondary_memory_needed): Change argument names.
7267 (rs6000_output_move_128bit): New function to return the move to
7268 use for 128-bit moves, including knowing about the various
7269 limitations of quad memory operations.
7271 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
7272 memory operations. call rs6000_output_move_128bit for the actual
7273 instruciton(s) to generate.
7274 (vsx_movti_64bit): Likewise.
7276 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
7277 (UNSPEC_P8V_MTVSRWZ): Likewise.
7278 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
7279 (UNSPEC_P8V_MTVSRD): Likewise.
7280 (UNSPEC_P8V_XXPERMDI): Likewise.
7281 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
7282 (UNSPEC_FUSION_GPR): Likewise.
7283 (FMOVE128_GPR): New iterator for direct move.
7284 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
7287 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
7288 loads and direct move instructions.
7289 (zero_extendsidi2_lfiwzx): Likewise.
7290 (extendsidi2_lfiwax): Likewise.
7291 (extendsidi2_nocell): Likewise.
7292 (floatsi<mode>2_lfiwax): Likewise.
7294 (floatunssi<mode>2_lfiwzx): Likewise.
7296 (fix_trunc<mode>_stfiwx): Likewise.
7297 (fixuns_trunc<mode>_stfiwx): Likewise.
7298 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
7299 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
7300 (parity<mode>2_cmpb): Set length/type attr.
7301 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
7302 for 'mr.' to fast_compare.
7303 (bpermd_<mode>): Change type attr to popcnt.
7304 (p8_fmrgow_<mode>): New insns for power8 direct move support.
7305 (p8_mtvsrwz_1): Likewise.
7306 (p8_mtvsrwz_2): Likewise.
7307 (reload_fpr_from_gpr<mode>): Likewise.
7308 (p8_mtvsrd_1): Likewise.
7309 (p8_mtvsrd_2): Likewise.
7310 (p8_xxpermdi_<mode>): Likewise.
7311 (reload_vsx_from_gpr<mode>): Likewise.
7312 (reload_vsx_from_gprsf): Likewise.
7313 (p8_mfvsrd_3_<mode>): LIkewise.
7314 (reload_gpr_from_vsx<mode>): Likewise.
7315 (reload_gpr_from_vsxsf): Likewise.
7316 (p8_mfvsrd_4_disf): Likewise.
7317 (multi-word GPR splits): Do not split direct moves or quad memory
7320 2013-06-10 David Malcolm <dmalcolm@redhat.com>
7322 * tree-into-ssa.c (interesting_blocks): Make static.
7324 2013-06-10 Jakub Jelinek <jakub@redhat.com>
7327 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
7328 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
7329 Use DATA_ABI_ALIGNMENT for that case instead if defined.
7330 (get_variable_align): New function.
7331 (get_variable_section, emit_bss, emit_common,
7332 assemble_variable_contents, place_block_symbol): Use
7333 get_variable_align instead of DECL_ALIGN.
7334 (assemble_noswitch_variable): Add align argument, use it
7335 instead of DECL_ALIGN.
7336 (assemble_variable): Adjust caller. Use get_variable_align
7337 instead of DECL_ALIGN.
7338 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
7340 (DATA_ABI_ALIGNMENT): Define.
7341 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
7342 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
7343 opt is false, only return the psABI mandated alignment increase.
7344 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
7345 (DATA_ABI_ALIGNMENT): ... this.
7346 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
7347 (DATA_ABI_ALIGNMENT): ... this.
7348 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
7349 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
7350 (DATA_ABI_ALIGNMENT): ... this.
7351 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
7352 * doc/tm.texi: Regenerated.
7354 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
7356 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
7357 cmp_code to construct REG_EQUAL note.
7359 2013-06-09 Jakub Jelinek <jakub@redhat.com>
7362 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
7363 that operands[2] doesn't overlap with operands[0].
7365 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
7366 Jan Hubicka <jh@suse.cz>
7368 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
7369 hack to mark symbols as used.
7371 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
7373 PR rtl-optimization/57559
7374 * lra-constraints.c (process_alt_operands): Don't discourage
7375 memory with known offset for offsetable memory constraint.
7376 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
7378 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
7380 * varasm.c (struct oc_local_state): Reorder fields.
7381 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
7382 and adjust accordingly.
7383 (output_constructor): Reorder initialization code and adjust call to
7384 output_constructor_bitfield.
7386 2013-06-07 Jan Hubicka <jh@suse.cz>
7388 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
7390 2013-06-07 David Malcolm <dmalcolm@redhat.com>
7392 * tree-object-size.c (unknown): Make const.
7394 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
7396 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
7397 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
7398 for last alternative in the cpu_facility attribute.
7400 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7403 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
7404 (xordi3): Change operand 2 constraint to arm_xordi_operand.
7405 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
7406 * config/arm/constraints.md (Dg): New constraint.
7407 * config/arm/neon.md (xordi3_neon): Remove.
7408 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
7409 * config/arm/predicates.md (arm_xordi_operand): New predicate.
7411 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7413 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
7414 Clean up alternatives.
7416 2013-06-07 Alan Modra <amodra@gmail.com>
7418 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
7421 2013-06-07 Alan Modra <amodra@gmail.com>
7423 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
7425 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7427 * config/arm/constraints.md (Df): New constraint.
7428 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
7429 Correct length attribute for last two alternatives.
7431 2013-06-07 Alan Modra <amodra@gmail.com>
7433 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7434 override user -mfp-in-toc.
7435 (offsettable_ok_by_alignment): Consider just the current access
7436 rather than the whole object, unless BLKmode. Handle
7437 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
7438 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
7439 for -mcmodel=medium.
7440 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
7441 override user -mfp-in-toc or -msum-in-toc. Default to
7442 -mno-fp-in-toc for -mcmodel=medium.
7444 2013-06-06 DJ Delorie <dj@redhat.com>
7446 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
7447 TARGET_VALID_POINTER_MODE.
7449 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7450 Pat Haugen <pthaugen@us.ibm.com>
7451 Peter Bergner <bergner@vnet.ibm.com>
7453 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7454 Document new power8 builtins.
7456 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
7457 condition code register, to allow 128-bit logical operations to be
7458 done in the VSX or GPR registers.
7459 (nor<mode>3): Use the canonical form for nor.
7460 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
7461 vclz*, and vpopcnt* vector instructions.
7462 (nand<mode>3): Likewise.
7463 (orc<mode>3): Likewise.
7464 (clz<mode>2): LIkewise.
7465 (popcount<mode>2): Likewise.
7467 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
7468 that only the GPRs are recognized.
7470 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7471 support for new power8 builtins.
7473 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
7475 (xscvdpspn): Likewise.
7481 (vpopcnt): Likewise.
7482 (vpopcntb): Likewise.
7483 (vpopcnth): Likewise.
7484 (vpopcntw): Likewise.
7485 (vpopcntd): Likewise.
7490 (eqv_v16qi3): Likewise.
7491 (eqv_v8hi3): Likewise.
7492 (eqv_v4si3): Likewise.
7493 (eqv_v2di3): Likewise.
7494 (eqv_v4sf3): Likewise.
7495 (eqv_v2df3): Likewise.
7497 (nand_v16qi3): Likewise.
7498 (nand_v8hi3): Likewise.
7499 (nand_v4si3): Likewise.
7500 (nand_v2di3): Likewise.
7501 (nand_v4sf3): Likewise.
7502 (nand_v2df3): Likewise.
7504 (orc_v16qi3): Likewise.
7505 (orc_v8hi3): Likewise.
7506 (orc_v4si3): Likewise.
7507 (orc_v2di3): Likewise.
7508 (orc_v4sf3): Likewise.
7509 (orc_v2df3): Likewise.
7511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
7512 allow power8 quad mode in 64-bit.
7513 (rs6000_builtin_vectorized_function): Add support to vectorize
7514 ISA 2.07 count leading zeros, population count builtins.
7515 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
7516 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
7517 (builtin_function_type): Add vgbbd builtin function which takes an
7519 (altivec_expand_vec_perm_const): Add support for new power8 merge
7522 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7523 that does not include TImdoe for use with 32-bit.
7524 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7526 (UNSPEC_VSX_CVDPSPN): Likewise.
7527 (vsx_xscvdpspn): Likewise.
7528 (vsx_xscvspdpn): Likewise.
7529 (vsx_xscvdpspn_scalar): Likewise.
7530 (vsx_xscvspdpn_directmove): Likewise.
7531 (vsx_and<mode>3): Split logical operations into 32-bit and
7532 64-bit. Add support to do logical operations on TImode as well as
7533 VSX vector types. Allow logical operations to be done in either
7534 VSX registers or in general purpose registers in 64-bit mode. Add
7535 splitters if GPRs were used. For AND, add clobber of CCmode to
7536 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
7538 (vsx_and<mode>3_32bit): Likewise.
7539 (vsx_and<mode>3_64bit): Likewise.
7540 (vsx_ior<mode>3): Likewise.
7541 (vsx_ior<mode>3_32bit): Likewise.
7542 (vsx_ior<mode>3_64bit): Likewise.
7543 (vsx_xor<mode>3): Likewise.
7544 (vsx_xor<mode>3_32bit): Likewise.
7545 (vsx_xor<mode>3_64bit): Likewise.
7546 (vsx_one_cmpl<mode>2): Likewise.
7547 (vsx_one_cmpl<mode>2_32bit): Likewise.
7548 (vsx_one_cmpl<mode>2_64bit): Likewise.
7549 (vsx_nor<mode>3): Likewise.
7550 (vsx_nor<mode>3_32bit): Likewise.
7551 (vsx_nor<mode>3_64bit): Likewise.
7552 (vsx_andc<mode>3): Likewise.
7553 (vsx_andc<mode>3_32bit): Likewise.
7554 (vsx_andc<mode>3_64bit): Likewise.
7555 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7556 and xxlorc instructions.
7557 (vsx_eqv<mode>3_64bit): Likewise.
7558 (vsx_nand<mode>3_32bit): Likewise.
7559 (vsx_nand<mode>3_64bit): Likewise.
7560 (vsx_orc<mode>3_32bit): Likewise.
7561 (vsx_orc<mode>3_64bit): Likewise.
7563 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7565 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7567 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7568 (p8_vmrgow): Likewise.
7569 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7570 GPRs to be split under VSX.
7571 (p8v_clz<mode>2): Add power8 count leading zero support.
7572 (p8v_popcount<mode>2): Add power8 population count support.
7573 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7576 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7579 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7581 (vec_nand): Likewise.
7582 (vec_vclz): Likewise.
7583 (vec_vclzb): Likewise.
7584 (vec_vclzd): Likewise.
7585 (vec_vclzh): Likewise.
7586 (vec_vclzw): Likewise.
7587 (vec_vgbbd): Likewise.
7588 (vec_vmrgew): Likewise.
7589 (vec_vmrgow): Likewise.
7590 (vec_vpopcnt): Likewise.
7591 (vec_vpopcntb): Likewise.
7592 (vec_vpopcntd): Likewise.
7593 (vec_vpopcnth): Likewise.
7594 (vec_vpopcntw): Likewise.
7596 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7598 PR rtl-optimization/57468
7599 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7602 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7604 PR rtl-optimization/57459
7605 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7606 type when setting live regs.
7608 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7610 * config/s390/s390.opt (mlra): New option.
7611 * config/s390/s390.c (s390_decompose_address): Check displacement
7612 for all registers for LRA.
7613 (s390_secondary_reload): Don't used secondary reloads for LRA.
7614 (s390_lra_p): New function.
7615 (TARGET_LRA_P): Define.
7616 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7617 of attribute cpu_facility to zarch for the last alternative.
7618 (*cmpmem_short): Ditto.
7620 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
7622 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7623 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
7624 (arm_expand_prologue): Likewise.
7626 2013-06-06 Teresa Johnson <tejohnson@google.com>
7629 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7630 as this is now done by redirect_edge_and_branch_force.
7631 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7632 barriers, and fix interaction with splitting.
7633 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7634 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7635 reflect changes made in the routine.
7636 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7637 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7638 since this is called in cfglayout mode, and replace partition fixup
7639 with assert as that is now done by force_nonfallthru_and_redirect.
7640 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7641 already be marked with region crossing note.
7642 (insert_section_boundary_note): Make non-static, gate on flag
7643 has_bb_partition, rewrite to also check for multiple partitions.
7644 (rest_of_handle_reorder_blocks): Remove call to
7645 insert_section_boundary_note, now done later during free_cfg.
7646 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7647 * bb-reorder.h (insert_section_boundary_note): Declare.
7648 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7649 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7650 invoke insert_section_boundary_note.
7651 (try_redirect_by_replacing_jump): Remove unnecessary
7652 check for region crossing note.
7653 (fixup_partition_crossing): New function.
7654 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7655 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7656 in non-cfglayout mode.
7657 (force_nonfallthru_and_redirect): Fixup partition boundaries,
7658 remove old code that tried to do this. Emit barrier correctly
7659 when we are in cfglayout mode.
7660 (last_bb_in_partition): New function.
7661 (rtl_split_edge): Correctly fixup partition boundaries.
7662 (commit_one_edge_insertion): Remove old code that tried to
7663 fixup region crossing edge since this is now handled in
7664 split_block, and set up insertion point correctly since
7665 block may now end in a jump.
7666 (verify_hot_cold_block_grouping): Guard against checking when not in
7667 linearized RTL mode.
7668 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7670 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7671 rtl_verify_flow_info, so not called in cfglayout mode.
7672 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7673 (fixup_reorder_chain): Remove old code that attempted to fixup region
7674 crossing note as this is now handled in force_nonfallthru_and_redirect.
7675 (duplicate_insn_chain): Don't duplicate switch section notes.
7676 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7678 * basic-block.h (emit_barrier_after_bb): Declare.
7680 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7682 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7683 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7684 arm_usatsihi): Adjust alternatives for arm_restrict_it.
7686 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7688 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7690 * config/arm/ldmstm.md: Regenerate.
7692 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7694 * config/arm/sync.md (atomic_loaddi_1):
7695 Disable predication for arm_restrict_it.
7696 (arm_load_exclusive<mode>): Likewise.
7697 (arm_load_exclusivesi): Likewise.
7698 (arm_load_exclusivedi): Likewise.
7699 (arm_load_acquire_exclusive<mode>): Likewise.
7700 (arm_load_acquire_exclusivesi): Likewise.
7701 (arm_load_acquire_exclusivedi): Likewise.
7702 (arm_store_exclusive<mode>): Likewise.
7703 (arm_store_exclusive<mode>): Likewise.
7704 (arm_store_release_exclusivedi): Likewise.
7705 (arm_store_release_exclusive<mode>): Likewise.
7707 2013-06-06 Richard Biener <rguenther@suse.de>
7709 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7711 (lto_tag_is_tree_code_p): Adjust.
7712 (lto_tag_is_gimple_code_p): Likewise.
7713 (lto_gimple_code_to_tag): Likewise.
7714 (lto_tag_to_gimple_code): Likewise.
7715 (lto_tree_code_to_tag): Likewise.
7716 (lto_tag_to_tree_code): Likewise.
7717 * data-streamer.h (streamer_write_hwi_in_range): Use
7718 uhwi streaming to stream the normalized range.
7719 (streamer_read_hwi_in_range): Likewise.
7721 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7723 * config/arm/arm.md (enabled_for_depr_it): New attribute.
7724 (predicable_short_it): Likewise.
7725 (predicated): Likewise.
7726 (enabled): Handle above.
7727 (define_cond_exec): Set predicated attribute to yes.
7729 2013-06-05 Mike Stump <mikestump@comcast.net>
7731 * gdbinit.in (__FUNCTION__): Add.
7733 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
7735 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7736 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7738 2013-06-05 Jan Hubicka <jh@suse.cz>
7740 * varasm.c (mark_decl_referenced): Revert the removal until targets
7743 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
7745 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7746 instead of mark_decl_referenced.
7748 2013-06-05 Jan Hubicka <jh@suse.cz>
7750 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7751 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7752 and symtab_used_from_object_file_p.
7753 (cgraph_make_node_local_1): Clear forced_by_abi.
7754 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7755 * cgraph.h (symtab_node_base): Add forced_by_abi.
7756 (decide_is_variable_needed): Remove.
7757 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7758 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7759 (decide_is_symbol_needed): ... this one; handle symbols in general;
7760 always analyze virtuals; honnor forced_by_abi.
7761 (cgraph_finalize_function): Update.
7762 (varpool_finalize_decl): Update.
7763 (symbol_defined_and_needed): Remove.
7764 (analyze_functions): Update.
7765 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7766 output_refs, input_overwrite_node): Handle forced_by_abi.
7767 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7768 (address_taken_from_non_vtable_p): ... this one.
7769 (comdat_can_be_unshared_p_1): New function.
7770 (cgraph_comdat_can_be_unshared_p): Rename to ...
7771 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7772 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7773 (function_and_variable_visibility): Clear forced_by_abi as needed.
7774 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7775 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7776 * symtab.c (dump_symtab_base): Dump forced_by_abi.
7777 * varpool.c (decide_is_variable_needed): Remove.
7779 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7781 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7782 (arm_option_override): Override arm_restrict_it where appropriate.
7783 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7784 * config/arm/arm.opt (mrestrict-it): New command-line option.
7785 * doc/invoke.texi: Document -mrestrict-it.
7787 2013-06-05 David Malcolm <dmalcolm@redhat.com>
7789 * tsan.c (tsan_atomic_table): Make const.
7791 2013-06-05 Richard Biener <rguenther@suse.de>
7793 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7794 index associated with the tree we are supposed to replace.
7795 * tree-streamer-out.c (pack_ts_base_value_fields): Output
7796 TREE_ASM_WRITTEN as zero for everything but SSA names.
7798 2013-06-05 David Malcolm <dmalcolm@redhat.com>
7800 * tree-ssa-structalias.c (call_stmt_vars): Make static.
7802 2013-06-04 Jan Hubicka <jh@suse.cz>
7804 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7805 (input_node, input_varpool_node): Handle correctly external same
7807 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7808 nodes at ltrans stage.
7810 2013-06-04 Jan Hubicka <jh@suse.cz>
7812 * ipa-inline.c (update_caller_keys): Fix availability test.
7813 (update_callee_keys): Likewise.
7814 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7815 to follow ELF standard.
7817 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
7819 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7820 (mips64r5900el-*-elf*): New configurations.
7821 * config/mips/mips-cpus.def (r5900): New processor.
7822 * config/mips/mips-tables.opt: Regenerate.
7823 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7824 (mips_issue_rate): Handle PROCESSOR_R5900.
7825 (mips_reorg_process_insns): Force reorder mode for the R5900.
7826 * config/mips/mips.h (TARGET_MIPS5900): Define.
7827 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7829 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7831 * config/mips/mips.md (processor): Add r5900.
7832 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7834 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7836 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7837 into function to generate MOVI instruction.
7838 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7839 (aarch64_preferred_simd_mode): Turn into wrapper.
7840 (aarch64_output_scalar_simd_mov_immediate): New function.
7841 * config/aarch64/aarch64-protos.h: Add prototype for above.
7843 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7845 * config/aarch64/aarch64.c (simd_immediate_info): Remove
7846 element_char member.
7847 (sizetochar): Return signed char.
7848 (aarch64_simd_valid_immediate): Remove elchar and other
7849 unnecessary variables.
7850 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7851 Calculate element_char as required.
7852 * config/aarch64/aarch64-protos.h: Update and move prototype
7853 for aarch64_output_simd_mov_immediate.
7854 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7857 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7859 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7860 information completed by aarch64_simd_valid_immediate.
7861 (aarch64_legitimate_constant_p): Update arguments.
7862 (aarch64_simd_valid_immediate): Work with struct rather than many
7864 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7865 (aarch64_simd_make_constant): Update arguments.
7866 (aarch64_output_simd_mov_immediate): Work with struct rather than
7867 many pointers. Output immediate directly rather than as operand.
7868 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7870 * config/aarch64/constraints.md (Dn): Update arguments.
7872 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7874 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7876 (aarch64_simd_immediate_valid_for_move): Remove.
7877 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7878 (aarch64_simd_make_constant): Update call.
7879 (aarch64_output_simd_mov_immediate): Update call.
7880 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7882 * config/aarch64/constraints.md (Dn): Update call.
7884 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7886 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7887 return type to bool for prototype.
7888 (aarch64_legitimate_constant_p): Check for true instead of not -1.
7889 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7890 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7892 2013-06-04 Catherine Moore <clm@codesourcery.com>
7894 * config/mips/mips.opt (meva): New.
7895 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7896 (ASM_SPEC): Handle -meva.
7897 * doc/invoke.texi (meva): Document.
7899 2013-06-04 Alan Modra <amodra@gmail.com>
7901 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7904 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7906 * rtl.def: Add extra fourth optional field to define_cond_exec.
7907 * gensupport.c (process_one_cond_exec): Process attributes from
7909 * doc/md.texi: Document fourth field in define_cond_exec.
7911 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
7913 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7914 out the processing order as in store_bit_field_1.
7916 2013-06-04 Jan Hubicka <jh@suse.cz>
7919 * cgraphunit.c (cgraph_process_same_body_aliases): Create
7920 non-VAR_DECL node if it does not exist yet.
7922 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
7924 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7925 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7926 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7927 target_cpu_default setting.
7929 2013-06-03 Teresa Johnson <tejohnson@google.com>
7931 * dumpfile.c (opt_info_switch_p): Change -fopt-info
7932 default to -fopt-info=optimized instead of all.
7933 * doc/invoke.texi: Ditto.
7934 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7935 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7936 (execute_vect_slp): Emit BB vectorization success under
7937 MSG_OPTIMIZED_LOCATIONS.
7938 * tree-vect-slp.c (vect_slp_transform_bb): Change
7939 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7940 * tree-vect-loop.c (vect_transform_loop): Ditto.
7942 2013-06-03 Jason Merrill <jason@redhat.com>
7945 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7946 Use TARGET_EXPR for C++.
7948 2013-06-03 Jakub Jelinek <jakub@redhat.com>
7950 PR rtl-optimization/57268
7951 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7952 if DEBUG_INSN_P (insn).
7955 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
7957 PR rtl-optimization/57268
7958 * sched-deps.c (sched_analyze_2): Flush dependence lists if
7959 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7961 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
7963 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7964 (ix86_avoid_lea_for_addr): Likewise.
7965 (exact_dependency_1): Likewise.
7966 (ix86_adjust_cost): Likewise.
7967 (swap_top_of_ready_list): Fix formatting and !reload_completed check
7969 (do_reorder_for_imul): Fix typo, formatting and
7970 !reload_completed check removed.
7971 (ix86_sched_reorder): Fix typo and formatting.
7972 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7975 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
7977 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7979 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
7981 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7982 <STRING_CST>: Likewise.
7983 <VECTOR_CST>: Likewise.
7985 2013-06-01 Janus Weil <janus@gcc.gnu.org>
7986 Mikael Morin <mikael@gcc.gnu.org>
7988 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7989 * config.in: Regenerated.
7990 * configure: Regenerated.
7992 2013-06-01 Jan Hubicka <jh@suse.cz>
7995 * cgraphunit.c (compile): When weakref is not supported,
7996 set up transparent aliases before final output pass.
7997 * varasm.c (assemble_alias): Do not try to do it here.
7999 2013-06-01 Jan Hubicka <jh@suse.cz>
8002 * passes.c (for_per_function): Skip unanalyzed functions.
8004 2013-06-01 Jan Hubicka <jh@suse.cz>
8006 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
8007 (lto_symtab_merge_symbols_1): ... this one.
8008 (lto_symtab_merge_cgraph_nodes): Rename to ...
8009 (lto_symtab_merge_symbols): ... this one; simplify.
8010 * cgraph.c (same_body_aliases_done): Rename to ...
8011 (cpp_implicit_aliases_done): ... this one.
8012 (cgraph_create_function_alias): Update.
8013 (cgraph_same_body_alias): Update.
8014 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
8015 (verify_edge_corresponds_to_fndecl): Simplify.
8016 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
8017 (cgraph_node): Remove same_body_alias.
8018 (varpool_node): Remove alias_of and extra_name_alias.
8019 (same_body_aliases_done): Rename to ..
8020 (cpp_implicit_aliases_done): ... this one.
8021 (symtab_alias_ultimate_target): Add default parameter.
8022 (symtab_resolve_alias): New function.
8023 (fixup_same_cpp_alias_visibility): Declare.
8024 (cgraph_function_node): Add default parameter.
8025 (cgraph_node_asm_name): Likewise.
8026 (cgraph_function_or_thunk_node): Add default parameter; do
8027 not ICE when it is NULL.
8028 (varpool_variable_node): Likewise.
8029 * tree-emutls.c (create_emultls_var): Update.
8030 (ipa_lower_emutls): Update.
8031 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
8032 (cgraph_reset_node): Reset alias info.
8033 (cgraph_finalize_function): Update.
8034 (fixup_same_cpp_alias_visibility): Move to symtab.c.
8035 (analyze_function): Simplify.
8036 (cgraph_process_same_body_aliases): Simplify.
8037 (analyze_functions): Fixup same body aliases.
8038 (handle_alias_pairs): Simplify.
8039 (assemble_thunk): Update.
8040 (assemble_thunks_and_aliases): Update.
8041 (output_weakrefs): Rewrite.
8042 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
8043 (lto_output_varpool_node): Likewise.
8044 (compute_ltrans_boundary): Remve assert.
8045 (get_alias_symbol): New functoin.
8046 (input_node): Rewrite alias handling.
8047 (input_varpool_node): Likewise.
8048 * ipa-pure-const.c (propagate_pure_const): Fix formating.
8049 * ipa.c (process_references): Handle weakrefs correctly.
8050 (symtab_remove_unreachable_nodes): Likewise.
8051 * trans-mem.c (get_cg_data): Update.
8052 (ipa_tm_create_version_alias): Update.
8053 (ipa_tm_execute): Update.
8054 * symtab.c (dump_symtab_base): Dump aliases.
8055 (verify_symtab_base): Verify aliases.
8056 (symtab_node_availability): New function.
8057 (symtab_alias_ultimate_target): Simplify.
8058 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
8059 handle all the fixup cases.
8060 (symtab_resolve_alias): New function.
8061 * passes.c (ipa_write_summaries): Handle weakrefs.
8062 * varpool.c (varpool_analyze_node): Simplify.
8063 (assemble_aliases): Update.
8064 (varpool_create_variable_alias): Simplify.
8065 (varpool_extra_name_alias): Simplify.
8066 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
8067 (lto_symtab_merge_symbols): ... this one.
8069 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
8072 PR rtl-optimization/57268
8073 * sched-deps.c (sched_analyze_2): Flush dependence lists if
8074 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8076 2013-06-01 Tobias Burnus <burnus@net-b.de>
8079 2013-05-31 Tobias Burnus <burnus@net-b.de>
8082 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
8085 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
8087 PR rtl-optimization/57268
8088 * sched-deps.c (sched_analyze_2): Flush dependence lists if
8089 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8091 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
8093 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
8094 unordered comparison operators when -fno-trapping-math is in effect
8096 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
8097 and implement unordered comparison operators properly on the e500.
8099 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
8101 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
8102 for constant scalar integers.
8103 (simplify_relational_operation_1): Likewise.
8105 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
8107 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
8108 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
8111 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
8112 Igor Zamyatin <igor.zamyatin@intel.com>
8114 Silvermont (SLM) architecture performance tuning.
8115 * config/i386/i386.h (enum ix86_tune_indices): Add
8116 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
8117 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
8119 * config/i386/i386.c (initial_ix86_tune_features)
8120 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
8121 (ix86_lea_outperforms): Handle Silvermont tuning.
8122 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
8124 (ix86_use_lea_for_mov): Likewise.
8125 (ix86_avoid_lea_for_addr): Likewise.
8126 (ix86_lea_for_add_ok): Likewise.
8127 (exact_dependency_1): New function.
8128 (exact_store_load_dependency): Likewise.
8129 (ix86_adjust_cost): Handle Silvermont tuning.
8130 (do_reoder_for_imul): Likewise.
8131 (swap_top_of_ready_list): New function.
8132 (ix86_sched_reorder): Changed to handle Silvermont tuning.
8134 * config/i386/i386.md (peepholes that split memory operand in fp
8137 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
8139 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8140 Remove un-necessary braces.
8142 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
8144 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8145 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
8147 2013-05-31 Tobias Burnus <burnus@net-b.de>
8150 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
8152 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8155 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
8156 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
8157 * config/arm/neon.md (iordi3_neon): Remove.
8158 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
8159 * config/arm/predicates.md (imm_for_neon_logic_operand):
8160 Move to earlier in the file.
8161 (neon_logic_op2): Likewise.
8162 (arm_iordi_operand_neon): New predicate.
8164 2013-05-31 Richard Biener <rguenther@suse.de>
8166 PR tree-optimization/57478
8167 PR tree-optimization/57453
8168 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
8171 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
8173 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
8174 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
8176 2013-05-30 Tobias Burnus <burnus@net-b.de>
8177 Thomas Koenig <tkoenig@gcc.gnu.org>
8180 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
8181 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
8183 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
8185 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
8187 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
8189 * target.def (register_usage_leveling_p): New hook.
8190 * targhooks.c (default_register_usage_leveling_p): New.
8191 * targhooks.h (default_register_usage_leveling_p): New prototype.
8192 * lra-assigns.c (register_usage_leveling_p): Use the hook.
8193 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
8194 * doc/tm.texi: Update.
8195 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
8197 2013-05-30 Ian Bolton <ian.bolton@arm.com>
8199 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
8200 (*insv_reg<mode>): New define_insn.
8202 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
8204 PR rtl-optimization/57439
8205 * postreload.c (move2add_valid_value_p): Check that we have
8206 a zero subreg_regno_offset when accessing the register in
8209 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
8210 Igor Zamyatin <igor.zamyatin@intel.com>
8212 Silvermont (SLM) architecture pipeline model, tuning and
8214 * config.gcc: Add slm config options and target.
8216 * config/i386/slm.md: New.
8218 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
8220 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): New case
8222 (ix86_target_macros_internal): Likewise.
8224 * gcc/config/i386/i386.c (slm_cost): New cost.
8225 (m_SLM): New macro flag.
8226 (initial_ix86_tune_features): Set m_SLM.
8227 (x86_accumulate_outgoing_args): Likewise.
8228 (x86_arch_always_fancy_math_387): Likewise.
8229 (processor_target_table): Add slm cost.
8230 (cpu_names): Add slm cpu name.
8231 (x86_option_override_internal): Set SLM ISA.
8232 (ix86_issue_rate): New case PROCESSOR_SLM.
8233 (ia32_multipass_dfa_lookahead): Likewise.
8234 (fold_builtin_cpu): Add slm.
8236 * config/i386/i386.h (TARGET_SLM): New target macro.
8237 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
8238 (processor_type): Add PROCESSOR_SLM.
8240 * config/i386/i386.md (cpu): Add new value "slm".
8241 (slm.md): Include slm.md.
8243 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
8244 Zhenqiang Chen <zhenqiang.chen@linaro.org>
8246 * config/arm/arm-protos.h: Add and update function protos.
8247 * config/arm/arm.c (use_simple_return_p): New added.
8248 (thumb2_expand_return): Check simple_return flag.
8249 * config/arm/arm.md: Add simple_return and conditional simple_return.
8250 * config/arm/iterators.md: Add iterator for return and simple_return.
8252 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
8254 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
8255 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
8256 (arm_emit_vfp_multi_reg_pop): Likewise.
8257 (thumb2_emit_ldrd_pop): Likewise.
8258 (arm_expand_epilogue): Add misc REG_CFA notes.
8259 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
8261 2013-05-29 Lawrence Crowl <crowl@google.com>
8263 * config/arm/t-arm: Update for below.
8265 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
8266 Change type to hash_table. Update dependent calls and types.
8268 * config/i386/t-cygming: Update for below.
8270 * config/i386/t-interix: Update for below.
8272 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
8273 Change type to hash_table. Update dependent calls and types.
8274 (i386_find_on_wrapper_list::wrappers): Likewise.
8276 * config/ia64/t-ia64: Update for below.
8278 * config/ia64/ia64.c (bundle_state_table):
8279 Change type to hash_table. Update dependent calls and types.
8281 * config/mips/mips.c (mips_reorg_process_insns::htab):
8282 Change type to hash_table. Update dependent calls and types.
8284 * config/sol2.c (solaris_comdat_htab):
8285 Change type to hash_table. Update dependent calls and types.
8287 * config/t-sol2: Update for above.
8289 2013-05-29 Teresa Johnson <tejohnson@google.com>
8291 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
8292 functions are not yet marked as defined.
8294 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8295 Pat Haugen <pthaugen@us.ibm.com>
8296 Peter Bergner <bergner@vnet.ibm.com>
8298 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
8302 (vrotl<mode>3): Likewise.
8303 (vashl<mode>3): Likewise.
8304 (vlshr<mode>3): Likewise.
8305 (vashr<mode>3): Likewise.
8307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8308 support for power8 V2DI builtins.
8310 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
8311 power8 V2DI builtins.
8312 (vupkhsw): Likewise.
8313 (vupklsw): Likewise.
8314 (vaddudm): Likewise.
8319 (vpkudum): Likewise.
8320 (vpksdss): Likewise.
8321 (vpkudus): Likewise.
8322 (vpksdus): Likewise.
8327 (vsubudm): Likewise.
8328 (vcmpequd): Likewise.
8329 (vcmpgtsd): Likewise.
8330 (vcmpgtud): Likewise.
8331 (vcmpequd_p): Likewise.
8332 (vcmpgtsd_p): Likewise.
8333 (vcmpgtud_p): Likewise.
8334 (vupkhsw): Likewise.
8335 (vupklsw): Likewise.
8336 (vaddudm): Likewise.
8341 (vpksdss): Likewise.
8342 (vpksdus): Likewise.
8343 (vpkudum): Likewise.
8344 (vpkudus): Likewise.
8349 (vsubudm): Likewise.
8351 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
8352 support for power8 V2DI instructions.
8354 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
8355 power8 V2DI instructions. Combine pack and unpack insns to use an
8356 iterator for each mode. Check whether a particular mode supports
8357 Altivec instructions instead of just checking TARGET_ALTIVEC.
8358 (UNSPEC_VPKUWUM): Likewise.
8359 (UNSPEC_VPKSHSS): Likewise.
8360 (UNSPEC_VPKSWSS): Likewise.
8361 (UNSPEC_VPKUHUS): Likewise.
8362 (UNSPEC_VPKSHUS): Likewise.
8363 (UNSPEC_VPKUWUS): Likewise.
8364 (UNSPEC_VPKSWUS): Likewise.
8365 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
8366 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
8367 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
8368 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
8369 (UNSPEC_VUPKHSB): Likewise.
8370 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
8371 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
8372 (UNSPEC_VUPKHSH): Likewise.
8373 (UNSPEC_VUPKLSB): Likewise.
8374 (UNSPEC_VUPKLSH): Likewise.
8376 (VI_char): Likewise.
8377 (VI_scalar): Likewise.
8378 (VI_unit): Likewise.
8380 (VP_small): Likewise.
8381 (VP_small_lc): Likewise.
8382 (VU_char): Likewise.
8383 (add<mode>3): Likewise.
8384 (altivec_vaddcuw): Likewise.
8385 (altivec_vaddu<VI_char>s): Likewise.
8386 (altivec_vadds<VI_char>s): Likewise.
8387 (sub<mode>3): Likewise.
8388 (altivec_vsubcuw): Likewise.
8389 (altivec_vsubu<VI_char>s): Likewise.
8390 (altivec_vsubs<VI_char>s): Likewise.
8391 (altivec_vavgs<VI_char>): Likewise.
8392 (altivec_vcmpbfp): Likewise.
8393 (altivec_eq<mode>): Likewise.
8394 (altivec_gt<mode>): Likewise.
8395 (altivec_gtu<mode>): Likewise.
8396 (umax<mode>3): Likewise.
8397 (smax<mode>3): Likewise.
8398 (umin<mode>3): Likewise.
8399 (smin<mode>3): Likewise.
8400 (altivec_vpkuhum): Likewise.
8401 (altivec_vpkuwum): Likewise.
8402 (altivec_vpkshss): Likewise.
8403 (altivec_vpkswss): Likewise.
8404 (altivec_vpkuhus): Likewise.
8405 (altivec_vpkshus): Likewise.
8406 (altivec_vpkuwus): Likewise.
8407 (altivec_vpkswus): Likewise.
8408 (altivec_vpks<VI_char>ss): Likewise.
8409 (altivec_vpks<VI_char>us): Likewise.
8410 (altivec_vpku<VI_char>us): Likewise.
8411 (altivec_vpku<VI_char>um): Likewise.
8412 (altivec_vrl<VI_char>): Likewise.
8413 (altivec_vsl<VI_char>): Likewise.
8414 (altivec_vsr<VI_char>): Likewise.
8415 (altivec_vsra<VI_char>): Likewise.
8416 (altivec_vsldoi_<mode>): Likewise.
8417 (altivec_vupkhsb): Likewise.
8418 (altivec_vupkhs<VU_char>): Likewise.
8419 (altivec_vupkls<VU_char>): Likewise.
8420 (altivec_vupkhsh): Likewise.
8421 (altivec_vupklsb): Likewise.
8422 (altivec_vupklsh): Likewise.
8423 (altivec_vcmpequ<VI_char>_p): Likewise.
8424 (altivec_vcmpgts<VI_char>_p): Likewise.
8425 (altivec_vcmpgtu<VI_char>_p): Likewise.
8426 (abs<mode>2): Likewise.
8427 (vec_unpacks_hi_v16qi): Likewise.
8428 (vec_unpacks_hi_v8hi): Likewise.
8429 (vec_unpacks_lo_v16qi): Likewise.
8430 (vec_unpacks_hi_<VP_small_lc>): Likewise.
8431 (vec_unpacks_lo_v8hi): Likewise.
8432 (vec_unpacks_lo_<VP_small_lc>): Likewise.
8433 (vec_pack_trunc_v8h): Likewise.
8434 (vec_pack_trunc_v4si): Likewise.
8435 (vec_pack_trunc_<mode>): Likewise.
8437 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
8439 (vec_vmaxsd): Likewise.
8440 (vec_vmaxud): Likewise.
8441 (vec_vminsd): Likewise.
8442 (vec_vminud): Likewise.
8443 (vec_vpksdss): Likewise.
8444 (vec_vpksdus): Likewise.
8445 (vec_vpkudum): Likewise.
8446 (vec_vpkudus): Likewise.
8447 (vec_vrld): Likewise.
8448 (vec_vsld): Likewise.
8449 (vec_vsrad): Likewise.
8450 (vec_vsrd): Likewise.
8451 (vec_vsubudm): Likewise.
8452 (vec_vupkhsw): Likewise.
8453 (vec_vupklsw): Likewise.
8455 2013-05-29 Jan Hubicka <jh@suse.cz>
8457 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
8458 flags; reorder rest of fields in more consistent way.
8459 (varpool_node): Remove analyzed, finalized and alias.
8460 (cgraph_ndoe): Likewise.
8461 (symtab_alias_ultimate_target): New function.
8462 (cgraph_function_node): Move offline.
8463 (cgraph_reset_node): Declare.
8464 (cgraph_comdat_can_be_unshared_p): Remove.
8465 (varpool_remove_initializer): Declare.
8466 (varpool_first_defined_variable, varpool_next_defined_variable
8467 cgraph_first_defined_function, cgraph_next_defined_function): Update.
8468 (cgraph_function_with_gimple_body_p): Update.
8469 (varpool_all_refs_explicit_p): Update.
8470 (symtab_alias_target): New function.
8471 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
8472 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
8473 (cgraph_function_or_thunk_node): Simplify using
8474 symtab_alias_ultimate_target.
8475 (varpool_variable_node): Likewise.
8476 * cgraph.c (cgraph_create_function_alias): Update.
8477 (cgraph_add_thunk): Update.
8478 (cgraph_remove_node): Update.
8479 (dump_cgraph_node): Do not dump removed flags.
8480 (cgraph_function_body_availability): Update.
8481 (cgraph_propagate_frequency): Update.
8482 (verify_cgraph_node): Check sanity of local flag.
8483 (cgraph_function_node): Move here from cgraph.h; revamp for
8484 cgraph_function_or_thunk_node.
8485 * lto-symtab.c (lto_varpool_replace_node): Update.
8486 (lto_symtab_resolve_can_prevail_p): Update.
8487 (lto_symtab_merge_cgraph_nodes): Update.
8488 * ipa-cp.c (determine_versionability, initialize_node_lattices,
8489 propagate_constants_accross_call, devirtualization_time_bonus,
8490 ipcp_propagate_stage): Update.
8491 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
8492 * ipa-inline-transform.c (clone_inlined_nodes,
8493 preserve_function_body_p): Update.
8494 * ipa-reference.c (propagate): Update.
8495 (write_node_summary_p): Update.
8496 * toplev.c (wrapup_global_declaration_2): Update.
8497 * cgraphunit.c (cgraph_analyze_function): Rename to ...
8498 (analyze_function) ... this one.
8499 (cgraph_process_new_functions): Update.
8500 (cgraph_reset_node): Export.
8501 (cgraph_finalize_function): Update.
8502 (cgraph_add_new_function): Update.
8503 (process_function_and_variable_attributes): Update.
8504 (varpool_finalize_decl): Update.
8505 (symbol_finalized): Remove.
8506 (symbol_finalized_and_needed): Rename to ...
8507 (symbol_defined_and_needed): ... update.
8508 (cgraph_analyze_functions): Update.
8509 (handle_alias_pairs): Update.
8510 (mark_functions_to_output): Update.
8511 (assemble_thunk): Update.
8512 (output_in_order): Update.
8513 (output_weakrefs): Update.
8514 (finalize_compilation_unit): Update.
8515 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
8516 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
8517 input_node, input_varpool_node): Update.
8518 * dbxout.c (dbxout_expand_expr): Update.
8519 * cgraphclones.c (cgraph_clone_node): Update.
8520 (cgraph_copy_node_for_versioning): Update.
8521 (cgraph_materialize_clone): Update.
8522 (cgraph_materialize_all_clones): Update.
8523 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8524 propagate_pure_const, propagate_nothrow): Update.
8525 * lto-streamer-out.c (lto_output, write_symbol): Update.
8526 * ipa-utils.c (ipa_reverse_postorder): Update.
8527 * ipa-inline.c (can_inline_edge_p): Update.
8528 (update_caller_keys, ipa_inline): Update.
8529 * dwarf2out.c (reference_to_unused,
8530 premark_types_used_by_global_vars_helper): Update.
8531 * tree-eh.c (tree_could_trap_p): Update.
8532 * ipa-split.c (consider_split, execute_split_functions): Update.
8533 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8534 has_addr_references_p): Update; move ahead in file for better
8536 (process_references): Simplify.
8537 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8539 (cgraph_comdat_can_be_unshared_p): Make static.
8540 (cgraph_externally_visible_p): Update.
8541 (varpool_externally_visible_p): Update.
8542 (function_and_variable_visibility): Update.
8543 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8544 ipa_tm_mark_force_output_node): Update.
8545 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8546 estimate_edge_devirt_benefit, inline_generate_summary,
8547 inline_write_summary): Update.
8548 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8549 * ipa-prop.c (ipa_compute_jump_functions): Update.
8550 (ipa_print_node_params, ipa_prop_read_section,
8551 ipa_update_after_lto_read, read_replacements_section): Update.
8552 * varasm.c (mark_decl_referenced): Update.
8553 (assemble_alias, dump_tm_clone_pairs): Update.
8554 * tree-inline.c (copy_bb): Update.
8555 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8557 * symtab.c (dump_symtab_base): Print new flags.
8558 (verify_symtab_base): Verify new flags.
8559 (symtab_alias_ultimate_target): New function.
8560 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8561 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8563 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8565 * i386.c (ix86_get_function_versions_dispatcher,
8566 ix86_generate_version_dispatcher_body): Update.
8567 (fold_builtin_cpu): Use varpool_add_new_variable.
8568 * varpool.c (varpool_remove_initializer): Break out from ...
8569 (varpool_remove_node): ... this one.
8570 (dump_varpool_node, varpool_node_for_asm,
8571 cgraph_variable_initializer_availability, varpool_analyze_node,
8572 varpool_assemble_decl, varpool_remove_unreferenced_decls,
8573 varpool_finalize_named_section_flags, varpool_create_variable_alias):
8576 2013-05-29 Jan Hubicka <jh@suse.cz>
8578 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8580 2013-05-29 Easwaran Raman <eraman@google.com>
8582 PR tree-optimization/57442
8583 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8584 when control exits the main loop.
8586 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
8588 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8590 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8591 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8592 * rx/t-rx: Add rx100 under multi library matches option for nofpu
8595 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8597 PR tree-optimization/57441
8598 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8599 Don't limit size of incr_vec to number of candidates.
8601 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
8603 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8604 and mips16 directories.
8605 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8606 (MULTILIB_DIRNAMES): Ditto.
8607 (MULTILIB_EXCEPTIONS): Add new exceptions.
8608 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8609 (MULTILIB_DIRNAMES): Ditto.
8610 (MULTILIB_EXCEPTIONS): Add new exceptions.
8612 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8613 Marcus Shawcroft <marcus.shawcroft@arm.com>
8615 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8616 SYMBOL_TINY_ABSOLUTE.
8617 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8618 SYMBOL_TINY_ABSOLUTE.
8619 (aarch64_expand_mov_immediate): Likewise.
8620 (aarch64_classify_symbol): Likewise.
8621 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8622 Permit SYMBOL_TINY_ABSOLUTE.
8623 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8625 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8626 Marcus Shawcroft <marcus.shawcroft@arm.com>
8628 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8629 Refactor if/switch. Replace gcc_assert with if.
8631 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8633 * config/i386/i386.c (initial_ix86_tune_features): Enable
8634 FP Reassociation for AMD bdver1 and bdver2.
8636 2013-05-29 Martin Jambor <mjambor@suse.cz>
8638 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8639 and IMAGPART_EXPR do not occur within other handled_components.
8641 2013-05-29 Richard Biener <rguenther@suse.de>
8643 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8644 access on whether the use is in the BB we currently try to
8646 (vect_bb_vectorization_profitable_p): Pass the BB we currently
8647 vectorize to vect_bb_slp_scalar_cost.
8649 2013-05-29 Richard Biener <rguenther@suse.de>
8651 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8652 computing scalar cost offsetted by stmts that are kept live
8654 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8655 for computation of scalar cost.
8657 2013-05-28 Steve Ellcey <sellcey@mips.com>
8659 * config/mips/mips-cpus.def (mips32r2): Change processor type.
8661 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
8663 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8664 array notation built-in reduction functions.
8665 * doc/passes.texi (Passes): Added documentation about changes done
8667 * doc/invoke.texi (C Dialect Options): Added documentation about
8668 the -fcilkplus flag.
8669 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8670 (BUILTINS_DEF): Depend on cilkplus.def.
8671 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
8672 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8673 * cilkplus.def: New file.
8675 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
8677 PR rtl-optimization/57439
8678 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8680 2013-05-28 Easwaran Raman <eraman@google.com>
8682 PR tree-optimization/57337
8683 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8684 (find_insert_point): Correctly identify the insertion point
8685 when two statements with the same UID is compared.
8687 2013-05-28 Richard Biener <rguenther@suse.de>
8689 PR tree-optimization/56787
8690 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8691 from the list of data references.
8692 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8694 (vect_analyze_loop_operations): Likewise.
8695 (vect_transform_loop): Remove clobbers.
8697 2013-05-28 Martin Jambor <mjambor@suse.cz>
8699 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8700 and REALPART_EXPRs have scalar type.
8702 2013-05-28 Richard Biener <rguenther@suse.de>
8704 PR tree-optimization/57411
8705 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8707 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8708 virtual operand propagation.
8710 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
8712 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8713 destination register for bmasksi_vis.
8714 (vector_init_bshuffle): Likewise.
8715 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8717 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
8719 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8720 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8721 mode if the instruction isn't available in the original mode.
8722 * config/sparc/sparc.opt (mfix-ut699): New option.
8723 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8724 (divdf3): Turn into expander.
8725 (divdf3_nofix): New insn.
8726 (divdf3_fix): Likewise.
8727 (divsf3): Disable if -mfix-ut699.
8728 (sqrtdf2): Turn into expander.
8729 (sqrtdf2_nofix): New insn.
8730 (sqrtdf2_fix): Likewise.
8731 (sqrtsf2): Disable if -mfix-ut699.
8733 2013-05-27 Richard Biener <rguenther@suse.de>
8736 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8737 block for the new loop.
8739 2013-05-27 Richard Biener <rguenther@suse.de>
8741 PR tree-optimization/57343
8742 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8743 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8744 (number_of_iterations_cond): Do not build the folded tree.
8746 2013-05-27 Richard Biener <rguenther@suse.de>
8750 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8751 OEP_CONSTANT_ADDRESS_OF retained.
8753 PR tree-optimization/57417
8754 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8756 (set_ssa_val_to): Compare addresses using
8757 get_addr_base_and_unit_offset.
8759 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
8761 PR rtl-optimization/56833
8762 * postreload.c (move2add_record_mode): New function.
8763 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8764 (move2add_use_add2_insn): Use move2add_record_sym_value.
8765 (move2add_use_add3_insn): Likewise.
8766 (reload_cse_move2add): Use move2add_valid_value_p and
8767 move2add_record_mode. Invalidate call-clobbered and REG_INC
8768 affected regs by setting reg_mode to VOIDmode.
8769 (move2add_note_store): Don't pretend the inside of a SUBREG is
8770 the actual destination. Invalidate single/leading registers by
8771 setting reg_mode to VOIDmode.
8772 Use move2add_record_sym_value, move2add_valid_value_p and
8773 move2add_record_mode.
8775 2013-05-27 Richard Biener <rguenther@suse.de>
8777 PR tree-optimization/57396
8778 * tree-affine.c (double_int_constant_multiple_p): Properly
8779 return false for val == 0 and div != 0.
8781 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8783 * config/mips/mips.h: Use #elif in preprocessor conditions.
8785 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8788 * config/mips/constraints.md (kl): New constraint.
8789 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8790 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
8791 constraint for operand 0. Split after CSE for MIPS16. Emit a move
8793 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8795 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8798 * config/mips/mips.c (mips_can_inline_p): New function.
8799 (TARGET_CAN_INLINE_P): Define.
8801 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
8803 * sched-int.h (ds_t, dw_t): Make unsigned int.
8804 Fix documentation that describes how all the ds_t bits are used.
8805 Reserve the last bit for delayed-branch scheduling.
8806 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8807 (BITS_PER_DEP_WEAK): Fix definition and documentation.
8808 (gen_dep_weak_1): Remove prototype.
8809 * sched-deps.c (get_dep_weak_1): Make static.
8810 * target.def (speculate_insn, needs_block_p, gen_spec_check,
8811 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8812 * doc/tm.texi: Regenerate.
8813 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8815 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
8818 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8820 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
8821 Sandra Loosemore <sandra@codesourcery.com>
8823 * config.gcc (powerpc-*): Allow native for with-cpu.
8825 2013-05-24 Jeff Law <law@redhat.com>
8827 PR tree-optimization/57124
8828 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8829 conversion feeding a condition if the range has an overflow
8830 if -fstrict-overflow. Add warnings for when we do make the
8833 2013-05-24 Dehao Chen <dehao@google.com>
8835 * tree-cfg.c (locus_discrim_map): Fix the typo.
8836 (locus_discrim_hasher): Likewise.
8837 (locus_discrim_hasher::hash): Likewise.
8838 (locus_discrim_hasher::equal): Likewise.
8840 2013-05-24 Martin Jambor <mjambor@suse.cz>
8842 PR tree-optimization/57294
8843 * cgraph.h (ipa_record_stmt_references): Declare.
8844 * cgraphbuild.c (ipa_record_stmt_references): New function.
8845 (build_cgraph_edges): Use ipa_record_stmt_references.
8846 (rebuild_cgraph_edges): Likewise.
8847 (cgraph_rebuild_references): Likewise.
8848 * ipa-prop.c (ipa_modify_call_arguments): Discard references
8849 associated with the old statement and build references from the
8850 newly built statements.
8851 * ipa-ref.c (ipa_remove_stmt_references): New function.
8852 * ipa-ref.h (ipa_remove_stmt_references): Declare.
8854 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
8856 * lra-constraints.c (emit_spill_move): Use smaller mode for
8858 (check_and_process_move): Consider mem-reg moves for secondary
8860 (curr_insn_transform): Don't lose insns emitted before for
8861 secondary memory moves.
8862 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
8863 reg set up in the current insn.
8865 2013-05-24 Dehao Chen <dehao@google.com>
8867 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8869 (locus_descrim_hasher::equal): Likewise.
8870 (build_gimple_cfg): New discriminator assignment algorithm.
8871 (make_edges): Likewise.
8872 (next_discriminator_for_locus): Likewise.
8873 (same_line_p): Likewise.
8874 (assign_discriminators): Likewise.
8875 (make_cond_expr_edges): Likewise.
8876 (make_gimple_switch_edges): Likewise.
8877 (make_goto_expr_edges): Likewise.
8878 (make_gimple_asm_edges): Likewise.
8880 2013-05-24 Ian Bolton <ian.bolton@arm.com>
8882 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8883 X format specifier to only display bottom 16 bits.
8884 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8885 immediate to match for operand 2, since it will be masked.
8887 2013-05-24 Richard Biener <rguenther@suse.de>
8889 PR tree-optimization/57287
8890 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8891 all SSA names that occur in abnormal PHIs.
8893 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
8896 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8897 that index is not negative.
8899 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
8901 PR rtl-optimization/55177
8902 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8903 (simplify_byte_swapping_operation): New.
8904 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8905 (simplify_relational_operation_1): Deal with BSWAP.
8907 2013-05-23 Richard Henderson <rth@redhat.com>
8910 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8911 (ix86_reorg): Call it.
8913 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
8916 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8917 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8918 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8920 2013-05-23 Christian Bruel <christian.bruel@st.com>
8923 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8925 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8926 Marcus Shawcroft <marcus.shawcroft@arm.com>
8928 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8929 * config/aarch64/constraints.md (Usa): Remove.
8930 * doc/md.texi (AArch64 Usa): Remove.
8932 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8933 Marcus Shawcroft <marcus.shawcroft@arm.com>
8935 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8936 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8937 * config/aarch64/predicates.md (aarch64_const_address): Remove.
8938 (aarch64_mov_operand): Use aarch64_mov_operand_p.
8940 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
8942 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8943 instruction (AdvSIMD).
8944 * config/aarch64/aarch64-builtins.c
8945 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8946 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8948 2013-05-23 Martin Jambor <mjambor@suse.cz>
8951 * tree.h (contains_bitfld_component_ref_p): Declare.
8952 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8953 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
8955 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8956 not access a bit-field. Assert all final offsets are byte-aligned.
8958 2013-05-23 Richard Biener <rguenther@suse.de>
8960 PR tree-optimization/57380
8961 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8962 least one invariant or re-used load.
8963 * passes.c (init_optimization_passes): Move pass_phiprop before
8966 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
8968 * config/aarch64/aarch64-simd.md
8969 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8971 2013-05-23 Richard Biener <rguenther@suse.de>
8974 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8975 OEP_CONSTANT_ADDRESS_OF retained.
8977 2013-05-23 Jakub Jelinek <jakub@redhat.com>
8980 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8981 don't lower unit. Handle unit not being always BITS_PER_WORD.
8983 2013-05-23 Richard Biener <rguenther@suse.de>
8985 PR rtl-optimization/57341
8986 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8987 instead of true_dependence.
8989 2013-05-22 David Malcolm <dmalcolm@redhat.com>
8991 * bb-reorder.c (branch_threshold): Make const.
8992 (exec_threshold): Ditto.
8994 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8995 Pat Haugen <pthaugen@us.ibm.com>
8996 Peter Bergner <bergner@vnet.ibm.com>
8998 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8999 documentation for the power8 crypto builtins.
9001 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
9003 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
9004 macros for defining power8 builtin functions.
9005 (BU_P8V_AV_2): Likewise.
9006 (BU_P8V_AV_P): Likewise.
9007 (BU_P8V_VSX_1): Likewise.
9008 (BU_P8V_OVERLOAD_1): Likewise.
9009 (BU_P8V_OVERLOAD_2): Likewise.
9010 (BU_CRYPTO_1): Likewise.
9011 (BU_CRYPTO_2): Likewise.
9012 (BU_CRYPTO_3): Likewise.
9013 (BU_CRYPTO_OVERLOAD_1): Likewise.
9014 (BU_CRYPTO_OVERLOAD_2): Likewise.
9015 (XSCVSPDP): Fix typo, point to the correct instruction.
9016 (VCIPHER): Add power8 crypto builtins.
9017 (VCIPHERLAST): Likewise.
9018 (VNCIPHER): Likewise.
9019 (VNCIPHERLAST): Likewise.
9020 (VPMSUMB): Likewise.
9021 (VPMSUMH): Likewise.
9022 (VPMSUMW): Likewise.
9023 (VPERMXOR_V2DI): Likewise.
9024 (VPERMXOR_V4SI: Likewise.
9025 (VPERMXOR_V8HI: Likewise.
9026 (VPERMXOR_V16QI: Likewise.
9027 (VSHASIGMAW): Likewise.
9028 (VSHASIGMAD): Likewise.
9030 (VPERMXOR): Likewise.
9031 (VSHASIGMA): Likewise.
9033 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
9034 __CRYPTO__ if the crypto instructions are available.
9035 (altivec_overloaded_builtins): Add support for overloaded power8
9038 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
9039 support for power8 crypto builtins.
9040 (builtin_function_type): Likewise.
9041 (altivec_init_builtins): Add support for builtins that take vector
9042 long long (V2DI) arguments.
9044 * config/rs6000/crypto.md: New file, define power8 crypto
9047 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
9048 Pat Haugen <pthaugen@us.ibm.com>
9049 Peter Bergner <bergner@vnet.ibm.com>
9051 * doc/invoke.texi (Option Summary): Add power8 options.
9052 (RS/6000 and PowerPC Options): Likewise.
9054 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
9055 constraints.md instead of rs6000.h. Reorder w* constraints. Add
9056 wm, wn, wr documentation.
9058 * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
9059 registers if direct move instructions are enabled.
9060 (wn): New constraint for no registers.
9061 (wq): New constraint for quad word even GPR registers.
9062 (wr): New constraint if 64-bit instructions are enabled.
9063 (wv): New constraint if power8 vector instructions are enabled.
9064 (wQ): New constraint for quad word memory locations.
9066 * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
9067 constraint for 0..15 for crypto instructions.
9068 (gpc_reg_operand): If VSX allow registers in VSX registers as well
9069 as GPR and floating point registers.
9070 (int_reg_operand): New predicate to match only GPR registers.
9071 (base_reg_operand): New predicate to match base registers.
9072 (quad_int_reg_operand): New predicate to match even GPR registers
9073 for quad memory operations.
9074 (vsx_reg_or_cint_operand): New predicate to allow vector logical
9075 operations in both GPR and VSX registers.
9076 (quad_memory_operand): New predicate for quad memory operations.
9077 (reg_or_indexed_operand): New predicate for direct move support.
9079 * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
9080 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
9081 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
9082 (POWERPC_MASKS): Add power8 options.
9083 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
9086 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
9087 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
9089 * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
9090 (-mpower8-fusion): New power8 options.
9091 (-mpower8-fusion-sign): Likewise.
9092 (-mpower8-vector): Likewise.
9093 (-mcrypto): Likewise.
9094 (-mdirect-move): Likewise.
9095 (-mquad-memory): Likewise.
9097 * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
9099 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
9101 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
9102 (rs6000_debug_vector_unit): Add p8_vector.
9103 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
9104 definitions. Also print fusion state.
9105 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
9106 (rs6000_builtin_mask_calculate): Add power8 builtin support.
9107 (rs6000_option_override_internal): Add support for power8.
9108 (rs6000_common_init_builtins): Add debugging for skipped builtins
9110 (rs6000_adjust_cost): Add power8 support.
9111 (rs6000_issue_rate): Likewise.
9112 (insn_must_be_first_in_group): Likewise.
9113 (insn_must_be_last_in_group): Likewise.
9114 (force_new_group): Likewise.
9115 (rs6000_register_move_cost): Likewise.
9116 (rs6000_opt_masks): Likewise.
9118 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
9119 power8 capable assembler, default to power7 options.
9120 (TARGET_DIRECT_MOVE): Likewise.
9121 (TARGET_CRYPTO): Likewise.
9122 (TARGET_P8_VECTOR): Likewise.
9123 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
9124 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
9125 (VECTOR_MEM_P8_VECTOR_P): Likewise.
9126 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
9127 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
9128 (TARGET_XSCVDPSPN): Likewise.
9129 (TARGET_XSCVSPDPN): Likewsie.
9130 (TARGET_SYNC_HI_QI): Likewise.
9131 (TARGET_SYNC_TI): Likewise.
9132 (MASK_CRYPTO): Likewise.
9133 (MASK_DIRECT_MOVE): Likewise.
9134 (MASK_P8_FUSION): Likewise.
9135 (MASK_P8_VECTOR): Likewise.
9136 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
9137 temporary used by some of the direct move instructions to get two FP
9138 temporary registers does not force creation of a stack frame.
9139 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
9140 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
9141 that any VSX registers are tieable, even if they are also an
9142 Altivec vector mode.
9143 (r6000_reg_class_enum): Add wm, wr, wv constraints.
9144 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
9145 (RS6000_BTM_CRYPTO): Likewise.
9146 (RS6000_BTM_COMMON): Likewise.
9148 * config/rs6000/rs6000.md (cpu attribute): Add power8.
9149 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
9150 (enum rs6000_vector): Add power8 vector support.
9152 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9156 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
9157 (any_sibcall_could_use_r3): this and handle indirect calls.
9158 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
9160 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9162 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
9164 2013-05-22 Richard Biener <rguenther@suse.de>
9167 * profile.c (branch_prob): Do not split blocks that are
9168 abnormally receiving from ECF_RETURNS_TWICE functions.
9170 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
9172 * recog.c (offsettable_address_addr_space_p): Fix calculation of
9173 address mode. Move pointer mode initialization to the same place.
9175 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
9177 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
9178 while it has any effect.
9180 2013-05-21 Easwaran Raman <eraman@google.com>
9182 PR tree-optimization/57322
9183 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
9184 UID of the statement added to the BB to be 1.
9186 2013-05-21 Jakub Jelinek <jakub@redhat.com>
9188 PR tree-optimization/57331
9189 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
9190 of conversion from pointer type to integral type with integer.
9192 2013-05-21 Martin Jambor <mjambor@suse.cz>
9195 * ipa-prop.c (ipa_read_node_info): Process param_used and
9196 controlled_uses in the same order as when writing.
9198 2013-05-21 Magnus Granberg <baldrick@free.fr>
9201 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
9203 2013-05-21 Richard Biener <rguenther@suse.de>
9205 PR tree-optimization/57318
9206 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
9207 estimate stmts with side-effects as likely eliminated.
9209 2013-05-21 Richard Biener <rguenther@suse.de>
9211 PR tree-optimization/57330
9212 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
9213 preserve the call stmts fntype.
9215 2013-05-21 Richard Biener <rguenther@suse.de>
9217 PR tree-optimization/57303
9218 * tree-ssa-sink.c (statement_sink_location): Improve killing
9219 stmt detection and properly handle self-assignments.
9221 2013-05-21 Christian Bruel <christian.bruel@st.com>
9223 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
9224 spanning registers. LEAF_REG_REMAP is supported only for contiguous
9225 registers. Set register size out of the PARALLEL loop.
9227 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
9230 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
9231 (*fmasf4, *fmasf4_media): New insns.
9233 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9235 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
9236 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
9237 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
9238 (mips_idiv_insns): Update the comments to say that the returned
9239 instruction counts are in units of BASE_INSN_LENGTH.
9240 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
9241 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
9242 using 2 rather than 4 as the length of indirect MIPS16 and
9243 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
9244 length of a NOP. Don't divide MIPS16 lengths by 2.
9245 (mips16_split_long_branches): Assume a branch is long if the
9246 length is greater than 4 rather than 8.
9247 * config/mips/mips.md (length): Give MIPS16 lengths directly,
9248 rather than multiplying them by 2. Multiply instruction counts
9249 by BASE_INSN_LENGTH rather than 4.
9250 (*jump_mips16, tls_get_tp_mips16_<mode>)
9251 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
9253 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9255 * config/mips/mips.md (extended_mips16): Remove branch case.
9256 (length): Remove duplicated extended_mips16 test.
9258 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9260 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
9262 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
9264 * recog.h (Recog_data): Rename to...
9265 (recog_data_d): ...this.
9266 (recog_data): Update accordingly.
9267 * recog.c (recog_data): Likewise.
9268 * reload.c (save_recog_data): Likewise.
9269 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
9270 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
9272 2013-05-17 Julian Brown <julian@codesourcery.com>
9274 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
9275 found in a REG_EQUAL note, invalidate it.
9277 2013-05-17 Easwaran Raman <eraman@google.com>
9279 * tree-ssa-reassoc.c (find_insert_point): New function.
9280 (insert_stmt_after): Likewise.
9281 (get_def_stmt): Likewise.
9282 (ensure_ops_are_available): Likewise.
9283 (not_dominated_by): Likewise.
9284 (rewrite_expr_tree): Do not move statements beyond what is
9285 necessary. Remove call to swap_ops_for_binary_stmt...
9286 (reassociate_bb): ... and move it here.
9287 (build_and_add_sum): Assign UIDs for new statements.
9288 (linearize_expr): Likewise.
9289 (do_reassoc): Renumber gimple statement UIDs.
9291 2013-05-17 Jan Hubicka <jh@suse.cz>
9293 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
9295 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
9296 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
9298 (output_weakrefs): Update.
9300 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
9301 Martin Jambor <mjambor@suse.cz>
9304 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
9305 value that corresponds to the given aggval is found in values vector.
9307 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
9309 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
9310 sse, sse2, sse3, ssse3 and sse4a flags to options.
9312 2013-05-17 David Malcolm <dmalcolm@redhat.com>
9314 * gengtype-state.c: (s_expr_writer): New class, to handle
9315 prettifying of output layout of s-expressions.
9316 (state_writer): New class, to write out gtype.state.
9317 (state_written_type_count): Move this variable into member data of
9319 (s_expr_writer::s_expr_writer): New code: constructor for new class
9320 (state_writer::state_writer(): ditto
9321 (s_expr_writer::write_new_line): New function
9322 (s_expr_writer::write_any_indent): ditto
9323 (s_expr_writer::begin_s_expr): ditto
9324 (s_expr_writer::end_s_expr): ditto
9325 (write_state_fileloc): convert to method of state_writer...
9326 (state_writer:: write_state_fileloc): ...and use methods of
9327 s_expr_writer to write indentation into the gtype.state output file
9328 to visually represent the hierarchical structure of the list
9330 (write_state_fields): ditto, renaming to...
9331 (state_writer::write_state_fields)
9332 (write_state_a_string): ditto, renaming to...
9333 (state_writer::write_state_a_string)
9334 (write_state_string_option): ditto, renaming to...
9335 (state_writer::write_state_string_option)
9336 (write_state_type_option): ditto, renaming to...
9337 (state_writer::write_state_type_option)
9338 (write_state_nested_option): ditto, renaming to...
9339 (state_writer::write_state_nested_option)
9340 (write_state_option): ditto, renaming to...
9341 (state_writer::write_state_option)
9342 (write_state_options): ditto, renaming to...
9343 (state_writer::write_state_options)
9344 (write_state_lang_bitmap): ditto, renaming to...
9345 (state_writer::write_state_lang_bitmap)
9346 (write_state_version): ditto, renaming to...
9347 (state_writer::write_state_version)
9348 (write_state_scalar_type): ditto, renaming to...
9349 (state_writer::write_state_scalar_type)
9350 (write_state_string_type): ditto, renaming to...
9351 (state_writer::write_state_string_type)
9352 (write_state_undefined_type): ditto, renaming to...
9353 (state_writer::write_state_undefined_type)
9354 (write_state_struct_union_type): ditto, renaming to...
9355 (state_writer::write_state_struct_union_type)
9356 (write_state_struct_type): ditto, renaming to...
9357 (state_writer::write_state_struct_type)
9358 (write_state_user_struct_type): ditto, renaming to...
9359 (state_writer::write_state_user_struct_type)
9360 (write_state_lang_struct_type): ditto, renaming to...
9361 (state_writer::write_state_lang_struct_type)
9362 (write_state_param_struct_type): ditto, renaming to...
9363 (state_writer::write_state_param_struct_type)
9364 (write_state_pointer_type): ditto, renaming to...
9365 (state_writer::write_state_pointer_type)
9366 (write_state_array_type): ditto, renaming to...
9367 (state_writer::write_state_array_type)
9368 (write_state_gc_used): ditto, renaming to...
9369 (state_writer::write_state_gc_used)
9370 (write_state_common_type_content): ditto, renaming to...
9371 (state_writer::write_state_common_type_content)
9372 (write_state_type): ditto, renaming to...
9373 (state_writer::write_state_type)
9374 (write_state_pair_list): ditto, renaming to...
9375 (state_writer::write_state_pair_list)
9376 (write_state_pair): ditto, renaming to...
9377 (state_writer::write_state_pair)
9378 (write_state_typedefs): ditto, renaming to...
9379 (state_writer::write_state_typedefs)
9380 (write_state_structures): ditto, renaming to...
9381 (state_writer::write_state_structures)
9382 (write_state_param_structs): ditto, renaming to...
9383 (state_writer::write_state_param_structs)
9384 (write_state_variables): ditto, renaming to...
9385 (state_writer::write_state_variables)
9386 (write_state_srcdir): ditto, renaming to...
9387 (state_writer::write_state_srcdir)
9388 (write_state_files_list): ditto, renaming to...
9389 (state_writer::write_state_files_list)
9390 (write_state_languages): ditto, renaming to...
9391 (state_writer::write_state_languages)
9392 (write_state): create a state_writer instance and use it when
9393 writing out the state file
9395 2013-05-17 Mike Stump <mikestump@comcast.net>
9397 PR rtl-optimization/57304
9398 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
9399 accessing DF_REF_REAL_LOC.
9401 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9403 PR rtl-optimization/57281
9404 PR rtl-optimization/57300
9405 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
9406 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
9407 what the other splitter did if the registers are dead.
9409 2013-05-17 Richard Biener <rguenther@suse.de>
9411 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
9414 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9416 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
9419 2013-05-17 Marek Polacek <polacek@redhat.com>
9421 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
9422 length when doing non-zero store of storing '\0' to '\0'.
9424 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9426 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
9427 vect_external_def oprnd1 with loop_vinfo, try to emit
9428 optional cast, negation and and stmts on the loop preheader
9429 edge instead of into the pattern def seq.
9431 PR tree-optimization/57051
9432 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
9433 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
9435 2013-05-16 Nick Clifton <nickc@redhat.com>
9437 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
9438 (rl78_is_naked_func): New function.
9439 (rl78_expand_prologue): Skip prologue generation for naked functions.
9440 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
9441 * doc/extend.texi (naked): Add RL78 to the list of processors
9442 that supports this attribute.
9444 2013-05-16 Jeff Law <law@redhat.com>
9446 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
9448 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
9450 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
9451 cache parameters using detect_caches_amd also for CYRIX,
9452 NSC and TM2 signatures.
9454 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
9455 Dzianis Kahanovich <mahatma@eu.by>
9459 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9460 VIA/Centaur processors and determine their cache parameters
9461 using detect_caches_amd.
9463 2013-05-16 Teresa Johnson <tejohnson@google.com>
9465 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
9466 (rtl_verify_edges): New function.
9467 (rtl_verify_bb_insns): Ditto.
9468 (rtl_verify_bb_pointers): Ditto.
9469 (rtl_verify_bb_insn_chain): Ditto.
9470 (rtl_verify_fallthru): Ditto.
9471 (rtl_verify_bb_layout): Ditto.
9472 (rtl_verify_flow_info_1): Outline checks into new functions.
9473 (rtl_verify_flow_info): Ditto.
9475 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
9477 * cfghooks.c (copy_bbs): Add update_dominance argument.
9478 * cfghooks.h (copy_bbs): Update prototype.
9479 * tree-cfg.c (gimple_duplicate_sese_region):
9480 Add update_dominance argument.
9481 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
9482 * tree-ssa-loop-ch.c (copy_loop_headers): Update
9483 gimple_duplicate_sese_region call.
9484 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
9485 Update copy_bbs call.
9486 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
9487 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
9489 2013-05-16 Jakub Jelinek <jakub@redhat.com>
9491 * tree-vectorizer.h (NUM_PATTERNS): Increment.
9492 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
9493 vect_recog_rotate_pattern.
9494 (vect_recog_rotate_pattern): New function.
9496 2013-05-16 Jason Merrill <jason@redhat.com>
9498 * Makefile.in (LLINKER): New variable.
9499 (mostlyclean): Remove link mutex.
9500 * configure.ac: Handle --enable-link-mutex.
9501 * lock-and-run.sh: New script.
9503 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9506 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
9509 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9511 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
9513 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
9515 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
9516 * config/arm/arm.c (next_consecutive_mem): New function.
9517 (gen_movmem_ldrd_strd): Likewise.
9518 * config/arm/arm.md (movmemqi): Update condition and code.
9519 (unaligned_loaddi, unaligned_storedi): New patterns.
9521 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9523 * config.gcc: Obsolete *-*-solaris2.9*.
9524 * doc/install.texi (Specific, *-*-solaris2*): Document it.
9526 2013-05-16 Richard Biener <rguenther@suse.de>
9528 * passes.c (init_optimization_passes): Move pass_parallelize_loops
9529 earlier, after GRAPHITE transforms and IV canonicalization.
9531 2013-05-16 Jakub Jelinek <jakub@redhat.com>
9533 * omp-low.c (extract_omp_for_data): For collapsed loops,
9534 if at least one of the loops is known at compile time to
9535 iterate zero times, set count to 0.
9536 (expand_omp_regimplify_p): New function.
9537 (expand_omp_for_generic): For collapsed loops, if at least
9538 one of the loops isn't known to iterate at least once,
9539 add runtime check with setting count to 0.
9540 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9541 For unsigned types if it isn't known at compile time that
9542 the loop will iterate at least once, add runtime check to bypass
9543 the whole loop if initial condition isn't true.
9545 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
9547 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9549 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
9552 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9553 transformations to avoid an infinite loop.
9555 2013-05-16 Marek Polacek <polacek@redhat.com>
9557 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9559 2013-05-15 Leif Ekblad <leif@rdos.net>
9561 * config/i386/i386.c (ix86_decompose_address): Use
9562 DEFAULT_TLS_SEG_REG to access TLS segment register.
9563 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9564 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9565 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9567 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
9570 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9571 sibling calls to functions that would normally be lazily bound,
9572 unless $gp is call-clobbered.
9574 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
9576 * config/i386/i386.c (ix86_option_override_internal): Update
9577 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
9578 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9579 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9580 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9582 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9584 2013-05-15 Andreas Schwab <schwab@suse.de>
9586 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9587 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
9590 2013-05-15 Teresa Johnson <tejohnson@google.com>
9592 * loop-unroll.c (report_unroll_peel): Check decision before
9593 emitting unroll/peel message.
9595 2013-05-15 Teresa Johnson <tejohnson@google.com>
9597 * function.h (has_bb_partition): New rtl_data flag.
9598 (bb_reorder_complete): Ditto.
9599 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9600 instead of flag_reorder_blocks_and_partition.
9601 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9602 with some enhancements.
9603 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9604 * bb-reorder.c (connect_traces): Check for has_bb_partition
9605 instead of flag_reorder_blocks_and_partition.
9606 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9607 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9608 verify_hot_cold_block_grouping.
9609 (partition_hot_cold_basic_blocks): Set has_bb_partition.
9611 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9614 * config/arm/predicates.md (call_insn_operand): New predicate.
9615 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
9616 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9617 if insn is not a tail call.
9618 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9620 * config/arm/arm.h (enum reg_class): New caller save register class.
9621 (REG_CLASS_NAMES): Likewise.
9622 (REG_CLASS_CONTENTS): Likewise.
9623 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9626 2013-05-15 Richard Biener <rguenther@suse.de>
9628 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9629 of MSG_OPTIMIZED_LOCATIONS.
9630 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9631 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9633 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9634 of MSG_OPTIMIZED_LOCATIONS.
9635 (execute_vect_slp): Likewise.
9636 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9637 (vect_create_cond_for_alias_checks): Likewise.
9638 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9639 (vect_recog_widen_mult_pattern): Likewise.
9640 (vect_recog_widen_sum_pattern): Likewise.
9641 (vect_recog_over_widening_pattern): Likewise.
9642 (vect_recog_widen_shift_pattern): Likewise.
9643 (vect_recog_vector_vector_shift_pattern): Likewise.
9644 (vect_recog_divmod_pattern): Likewise.
9645 (vect_recog_mixed_size_cond_pattern): Likewise.
9646 (vect_recog_bool_pattern): Likewise.
9647 (vect_pattern_recog_1): Likewise.
9649 2013-05-15 Martin Jambor <mjambor@suse.cz>
9651 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9652 non-functions to builtin_unreachable.
9653 * ipa-inline-transform.c (inline_call): Do not assert estimates were
9654 correct when new direct edges were discovered.
9656 2013-05-15 Martin Jambor <mjambor@suse.cz>
9658 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9659 header, print symbol order instead of node uid, print more information
9660 about indirect edge targets.
9661 (ipa_make_edge_direct_to_target): Print symbol order instead of node
9663 (ipa_make_edge_direct_to_target): Likewise.
9664 (remove_described_reference): Likewise.
9665 (propagate_controlled_uses): Likewise.
9666 (ipa_print_node_params): Also print symbol order.
9667 (ipcp_transform_function): Print symbol order instead of node uids.
9668 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9669 (cgraph_get_create_real_symbol_node): Likewise.
9670 * ipa-cp.c (print_lattice): Likewise.
9671 (print_all_lattices): Likewise.
9672 (determine_versionability): Likewise.
9673 (initialize_node_lattices): Likewise.
9674 (estimate_local_effects): Likewise.
9675 (update_profiling_info): Likewise.
9676 (create_specialized_node): Likewise.
9677 (perhaps_add_new_callers): Likewise.
9678 (decide_about_value): Likewise.
9679 (decide_whether_version_node): Likewise.
9680 (identify_dead_nodes): Likewise.
9681 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9682 (dump_inline_summary): Likewise.
9683 (estimate_node_size_and_time): Likewise.
9684 (inline_analyze_function): Likewise.
9685 * ipa-inline.c (report_inline_failed_reason): Likewise.
9686 (want_early_inline_function_p): Likewise.
9687 (edge_badness): Likewise.
9688 (update_edge_key): Likewise.
9689 (inline_small_functions): Likewise. Add dumping of order to two other
9691 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9692 instead of node uids.
9693 (propagate_pure_const): Likewise.
9694 (propagate_pure_const): Likewise.
9695 * ipa-utils.c (dump_cgraph_node_set): Likewise.
9696 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9697 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9699 * tree-pretty-print.c (dump_function_header): Likewise.
9700 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9701 Print symbol order instead of node uids.
9703 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9705 * config/s390/s390.c (s390_register_move_cost): Don't impose the
9706 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9708 2013-05-15 Richard Biener <rguenther@suse.de>
9710 PR tree-optimization/57275
9711 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9712 return value for fail to do runtime alias checks for gather loads.
9714 2013-05-15 Jan Hubicka <jh@suse.cz>
9718 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9719 weakrefs are not external.
9720 (lto_symtab_merge_decls): Fix thinko when dealing with
9721 non-lto_symtab decls.
9722 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9723 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9724 * varpool.c (dump_varpool_node): Dump more flags.
9726 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
9728 * config/i386/i386.c (processor_alias_table): Add instruction
9729 FSGSBASE for AMD bdver3 architecture.
9731 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9733 * tree.c (warn_deprecated_use): Print file:line using locus color.
9734 * diagnostic.c (diagnostic_report_current_module): Print file:line
9735 and file:line:column using locus color.
9737 2013-05-14 Mike Stump <mikestump@comcast.net>
9739 * gdbinit.in: Add __null.
9741 2013-05-14 Mike Stump <mikestump@comcast.net>
9743 * recog.h: Rename struct recog_data to Recog_data.
9744 * recog.c: Likewise.
9745 * reload.c (can_reload_into): Likewise.
9746 * config/picochip/picochip.c: Likewise.
9748 2013-05-14 Mike Stump <mikestump@comcast.net>
9750 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9752 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
9754 * resource.h (struct resources): Remove unch_memory member.
9755 (CLEAR_RESOURCE): Don't clear unch_memory.
9756 * resource.c (mark_referenced_resources): Don't set it.
9757 (mark_set_resources): Likewise.
9758 (mark_target_live_regs): Don't clear it.
9759 (init_resource_info): Likewise.
9760 * reorg.c (resource_conflicts_p): Don't compare it.
9761 (redundant_insn): Don't set it.
9763 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9765 * emit-rtl.c (next_label): Remove unused function.
9766 (skip_consecutive_labels, link_cc0_insns): Move to ...
9767 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9768 only place where these functions are used, and make them static.
9770 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
9772 * fold-const.c (fold_negate_expr): Handle vectors.
9773 (fold_truth_not_expr): Make it static.
9774 (fold_invert_truthvalue): New static function.
9775 (invert_truthvalue_loc): Handle vectors. Do not call
9776 fold_truth_not_expr directly.
9777 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9778 <TRUTH_NOT_EXPR>: Do not cast to boolean.
9779 (fold_comparison): Handle vector constants.
9780 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9781 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9782 * tree.h (fold_truth_not_expr): Remove declaration.
9784 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
9786 * config/aarch64/aarch64-simd.md
9787 (aarch64_vcond_internal<mode>): Rename to...
9788 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9789 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9790 float modes. Clarify all iterator modes.
9791 (vcond<mode><mode>): Use new name for vcond expanders.
9792 (vcond<v_cmp_result><mode>): Likewise.
9793 (vcondu<mode><mode>: Likewise.
9794 * config/aarch64/iterators.md (VDQF_COND): New.
9796 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
9799 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9800 variable for the shift amount. Check that we shift by non-negative
9803 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
9806 * config/arm/arm.h (EPILOGUE_USES): Only return true
9807 for LR_REGNUM after epilogue_completed.
9809 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
9811 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9814 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9817 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9818 and Solaris 11+/x86 with gld.
9819 * configure: Regenerate.
9821 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9823 * expmed.c (expand_shift_1): Canonicalize rotates by
9824 constant bitsize / 2 to bitsize - 1.
9825 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9826 case ROTATERT>: Likewise.
9829 2013-05-10 Jakub Jelinek <jakub@redhat.com>
9831 * config/i386/i386.md (rotateinv): New code attr.
9832 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9833 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9834 roll $31, %eax, etc.
9836 2013-05-14 Richard Biener <rguenther@suse.de>
9839 * tree-eh.c (sink_clobbers): Give up for successors with
9840 multiple predecessors and no virtual uses.
9842 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
9844 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9845 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9847 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9850 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9851 the case when both op0 and op1 have VOIDmode.
9853 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
9855 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9856 in multiply-accumulate mode.
9858 2013-05-13 Guozhi Wei <carrot@google.com>
9860 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9862 2013-05-13 Kai Tietz <ktietz@redhat.com>
9865 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9866 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9867 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9868 * config/i386/i386.c (ix86_option_override_internal): Likewise.
9869 (ix86_expand_prologue): Likewise.
9870 (ix86_expand_split_stack_prologue): Likewise.
9871 (legitimate_pic_address_disp_p): Likewise.
9872 (legitimize_pic_address): Likewise.
9873 (legitimize_tls_address): Likewise.
9874 (legitimize_pe_coff_symbol): Likewise.
9875 (output_pic_addr_const): Likewise.
9876 (construct_plt_address): Likewise.
9877 (ix86_expand_call): Likewise.
9878 (x86_output_mi_thunk): Likewise.
9879 (x86_function_profiler): Likewise.
9881 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
9883 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9884 similar switch cases.
9885 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9886 (aarch64_simd_mov_to_<mode>low): Delete.
9887 (aarch64_simd_mov_to_<mode>high): Delete.
9888 (move_lo_quad_<mode>): Add w<-r alternative.
9889 (aarch64_simd_move_hi_quad_<mode>): Likewise.
9890 (aarch64_simd_mov_from_*): Update type attribute.
9891 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9894 2013-05-13 Jan Hubicka <jh@suse.cz>
9896 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9897 * config/i386/i386.c (ix86_compute_frame_layout,
9898 ix86_expand_epilogue, emit_i387_cw_initialization,
9899 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9900 ix86_local_alignment): Fix use of size/speed predicates.
9902 2013-05-13 Jakub Jelinek <jakub@redhat.com>
9904 PR tree-optimization/45216
9905 PR tree-optimization/57157
9906 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9907 the (-Y) & (B - 1) variant if OP is |.
9908 * expmed.c (expand_shift_1): For rotations by const0_rtx just
9909 return shifted. Use (-op1) & (prec - 1) as other_amount
9910 instead of prec - op1.
9912 2013-05-13 Martin Jambor <mjambor@suse.cz>
9915 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9916 (ipa_constant_data): New type.
9917 (ipa_jump_func): Use ipa_constant_data to hold information about
9918 constant jump functions.
9919 (ipa_get_jf_constant): Adjust to jump function type changes.
9920 (ipa_get_jf_constant_rdesc): New function.
9921 (ipa_param_descriptor): New field controlled_uses.
9922 (ipa_get_controlled_uses): New function.
9923 (ipa_set_controlled_uses): Likewise.
9924 * ipa-ref.h (ipa_find_reference): Declare.
9925 * ipa-prop.c (ipa_cst_ref_desc): New type.
9926 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9928 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
9929 New parameter cs. Adjust all callers.
9930 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9931 (remove_described_reference): New function.
9932 (jfunc_rdesc_usable): Likewise.
9933 (try_make_edge_direct_simple_call): Decrement controlled use count,
9934 attempt to remove reference if it hits zero.
9935 (combine_controlled_uses_counters): New function.
9936 (propagate_controlled_uses): Likewise.
9937 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9938 (ipa_edge_duplication_hook): Duplicate reference descriptions.
9939 (ipa_print_node_params): Print described use counter.
9940 (ipa_write_jump_function): Adjust to jump function type changes.
9941 (ipa_read_jump_function): New parameter CS, pass it to
9942 ipa_set_jf_constant. Adjust caller.
9943 (ipa_write_node_info): Stream controlled use count
9944 (ipa_read_node_info): Likewise.
9945 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9947 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9948 count. Remove cloning-added reference if it reaches zero.
9949 * ipa-ref.c (ipa_find_reference): New function.
9951 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
9953 * config/i386/i386.c (processor_target_table): Modified default
9954 alignment values for AMD BD and BT architectures.
9956 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
9958 * tree-vect-generic.c (uniform_vector_p): Move ...
9959 * tree.c (uniform_vector_p): ... here.
9960 * tree.h (uniform_vector_p): Declare it.
9961 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9964 2013-05-13 Jakub Jelinek <jakub@redhat.com>
9966 PR tree-optimization/57230
9967 * tree-ssa-strlen.c (handle_char_store): Record length for
9968 array store from STRING_CST.
9970 PR tree-optimization/57230
9971 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9974 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
9976 * config/epiphany/epiphany.c (epiphany_init): Check size of
9977 NUM_MODES_FOR_MODE_SWITCHING.
9978 (epiphany_expand_prologue):
9979 Remove CONFIG_REGNUM initial value handling code.
9980 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9981 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9982 (emit_set_fp_mode, epiphany_mode_after): Likewise.
9983 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9984 Don't return 1 for FP_MODE_NONE.
9985 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9986 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9987 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9988 * config/epiphany/epiphany.md (save_config): New pattern.
9990 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
9992 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9994 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
9996 * config/i386/i386.md (memory): Handle sseishft1.
9997 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9998 (*vec_extractv2di_1): Ditto.
10000 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
10002 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
10005 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
10007 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
10008 Add mthumb/march=armv7-a multilib.
10009 Add mthumb/march=armv7-r multilib.
10010 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
10012 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
10014 * config/v850/t-rtems: Add more multilibs.
10016 2013-05-10 Richard Biener <rguenther@suse.de>
10018 PR tree-optimization/57214
10019 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
10020 not propagate from SSA names that occur in abnormal PHI nodes.
10022 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
10024 * stor-layout.c (element_precision): New function.
10025 * machmode.h (element_precision): Declare it.
10026 * tree.c (build_minus_one_cst): New function.
10027 (element_precision): Likewise.
10028 * tree.h (build_minus_one_cst): Declare new function.
10029 (element_precision): Likewise.
10030 * fold-const.c (operand_equal_p): Use element_precision.
10031 (fold_binary_loc): Handle vector types.
10032 * convert.c (convert_to_integer): Use element_precision.
10033 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
10036 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
10038 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
10039 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
10040 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
10041 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10042 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
10043 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
10044 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
10045 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10046 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
10047 (Uuw8): New constraints.
10048 (Usb4): Move into alphabetical order.
10049 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
10050 (sd8_operand, ub8_operand, uw8_operand): New predicates.
10051 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
10052 previously unnamed patterns.
10053 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
10054 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
10055 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
10056 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
10057 of set_attr_alternative/if_then_else. Use extended_mips16 instead
10058 of specific lengths.
10060 2013-05-10 Jakub Jelinek <jakub@redhat.com>
10062 * config/i386/i386.md (rotateinv): New code attr.
10063 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
10064 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
10065 roll $31, %eax, etc.
10067 PR tree-optimization/45216
10068 PR tree-optimization/57157
10069 * tree-ssa-forwprop.c (simplify_rotate): New function.
10070 (ssa_forward_propagate_and_combine): Call it.
10072 2013-05-10 Richard Biener <rguenther@suse.de>
10074 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
10075 disable peeling when we version for aliasing.
10076 (vector_alignment_reachable_p): Honor explicit user alignment.
10077 (vect_supportable_dr_alignment): Likewise.
10078 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
10079 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
10080 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
10081 then peeling to arrange for the cost-model check to come first.
10083 2013-05-10 Alan Modra <amodra@gmail.com>
10085 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
10086 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
10087 * configure: Regenerate.
10089 2013-05-10 Alan Modra <amodra@gmail.com>
10092 * varasm.c (default_elf_select_section): Move !DECL_P check..
10093 (get_named_section): ..to here before calling get_section_name.
10095 (default_section_type_flags): Add DECL_P check.
10096 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
10097 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
10099 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
10101 * config/epiphany/epiphany.c (epiphany_expand_prologue):
10102 When using gen_stack_adjust_str with a register offset, add a
10103 REG_FRAME_RELATED_EXPR note.
10105 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
10107 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
10108 (*vec_extractv4si_zext_mem): Ditto.
10109 (*vec_extractv2di): Add 0->x and x->x alternatives.
10110 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
10111 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
10113 2013-05-09 Jason Merrill <jason@redhat.com>
10115 N3639 C++1y VLA support
10116 * gimplify.c (gimplify_vla_decl): Don't touch an existing
10119 * tree.c (build_constructor_va): New.
10120 * tree.h: Declare it.
10122 2013-05-09 Martin Jambor <mjambor@suse.cz>
10125 * gimple-fold.c (canonicalize_constructor_val): Call
10126 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
10128 2013-05-09 Jan Hubicka <jh@suse.cz>
10129 Richard Biener <rguenther@suse.de>
10132 * symtab.c (symtab_make_decl_local): Do not add private names.
10134 2013-05-09 Jan Hubicka <jh@suse.cz>
10137 * symtab.c (insert_to_assembler_name_hash): Handle clones.
10138 (unlink_from_assembler_name_hash): Likewise.
10139 (symtab_prevail_in_asm_name_hash, symtab_register_node,
10140 symtab_unregister_node, symtab_initialize_asm_name_hash,
10141 change_decl_assembler_name): Update.
10143 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
10145 * config/aarch64/aarch64.md: New movtf split.
10146 (*movtf_aarch64): Update.
10147 (aarch64_movdi_tilow): Handle TF modes and rename to
10148 aarch64_movdi_<mode>low.
10149 (aarch64_movdi_tihigh): Handle TF modes and rename to
10150 aarch64_movdi_<mode>high
10151 (aarch64_movtihigh_di): Handle TF modes and rename to
10152 aarch64_mov<mode>high_di
10153 (aarch64_movtilow_di): Handle TF modes and rename to
10154 aarch64_mov<mode>low_di
10155 (aarch64_movtilow_tilow): Remove spurious whitespace.
10156 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
10158 (aarch64_print_operand): Update.
10160 2013-05-09 Alan Modra <amodra@gmail.com>
10162 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
10164 * configure: Regenerate.
10166 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
10168 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
10169 splitter preparation statements.
10170 * config/i386/sse.md (*vec_extract* splitters): Ditto.
10171 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
10174 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10176 * gimple-ssa-strength-reduction.c (count_candidates): Change
10177 return value to int.
10178 (analyze_candidates_and_replace): Change type of length to int.
10180 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
10182 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
10183 (*vec_extract<mode>): Use VI12_128 mode iterator.
10184 (*vec_extract<mode>_mem): Ditto.
10185 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
10188 2013-05-08 Diego Novillo <dnovillo@google.com>
10193 2012-08-17 Diego Novillo <dnovillo@google.com>
10196 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
10197 * config.in: Regenerate.
10198 * configure: Regenerate.
10199 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
10201 2013-05-08 Jan Hubicka <jh@suse.cz>
10204 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
10205 * cgraph.h (symtab_node_base): Add unique_name.
10206 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10207 input_overwrite_node, input_varpool_node): Stream unique_name.
10208 * cgraphclones.c (cgraph_create_virtual_clone,
10209 cgraph_function_versioning): Set unique_name.
10210 * ipa.c (function_and_variable_visibility): Set unique_name.
10212 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10214 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
10215 (alloc_cand_and_find_basis): Restrict conditional candidate
10216 processing to CAND_MULTs.
10218 2013-05-08 Jan Hubicka <jh@suse.cz>
10221 lto-symtab.c (lto_symtab_symbol_p): New function.
10222 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
10223 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
10224 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
10225 Skip static symbols.
10227 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
10229 PR tree-optimization/57200
10230 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
10231 Only call inform if the preceding warning_at returns true.
10233 2013-05-07 Han Shen <shenhan@google.com>
10235 * cfgexpand.c (record_or_union_type_has_array_p): New function.
10236 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
10237 * common.opt (fstack-protector-strong): New option.
10238 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
10239 * doc/invoke.texi (Optimization Options): Document
10240 "-fstack-protector-strong".
10241 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
10243 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
10245 * config/mips/mips.c (mips_machine_reorg2): Return 0.
10247 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
10249 * ira.c (update_equiv_regs): Add insn having equiv memory even if
10250 it is not lhs of the insn.
10251 (setup_reg_equiv): Remove insn having equiv memory which it is not
10253 * lra-constraints.c (process_address): Try to improve generation
10254 code for address base + disp.
10255 (lra_constraints): Make correct the code for checking insn setting
10256 up backward equivalence. Remove insn only if it is in the init
10258 * lra-eliminations.c (update_reg_eliminate): Change return value.
10259 (lra_eliminate): Use the result.
10261 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
10263 * config/i386/sse.md (ssescalarnummask): New mode attribute.
10264 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
10265 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
10266 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
10267 register target operands.
10268 (*vec_extractv8hi_sse2): New pattern.
10269 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
10270 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
10271 (*vec_extract<mode>_mem): New insn and split pattern.
10273 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
10275 * config/arm/arm.c (arm_asan_shadow_offset): New function.
10276 (TARGET_ASAN_SHADOW_OFFSET): Define.
10277 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
10278 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
10280 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10282 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
10283 (incr_vec_index): Return -1 if increment not found.
10284 (create_add_on_incoming_edge): Assert if increment not found.
10285 (record_increment): Limit number of increments recorded.
10286 (all_phi_incrs_profitable): Return false if an increment not found.
10287 (replace_profitable_candidates): Don't process increments that were
10289 (analyze_candidates_and_replace): Limit size of incr_vec.
10291 2013-05-07 Richard Biener <rguenther@suse.de>
10293 * calls.c (special_function_p): setjmp-like functions are leaf.
10294 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
10295 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
10297 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
10299 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
10300 (aarch64_simd_mov<mode>): New expander.
10301 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
10302 (aarch64_simd_mov_to_<mode>high): Likewise.
10303 (aarch64_simd_mov_from_<mode>low): Likewise.
10304 (aarch64_simd_mov_from_<mode>high): Likewise.
10305 (aarch64_dup_lane<mode>): Update.
10306 (aarch64_dup_lanedi): New instruction pattern.
10307 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
10308 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
10310 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10312 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
10313 (replace_mult_candidate): Remove unnecessary argument; remove
10314 unnecessary parameter from call to introduce_cast_before_cand.
10315 (replace_unconditional_candidate): Remove unnecessary parameter
10316 from call to replace_mult_candidate.
10317 (replace_conditional_candidate): Likewise.
10318 (insert_initializers): Use make_temp_ssa_name.
10319 (introduce_cast_before_cand): Remove unnecessary argument; use
10320 make_temp_ssa_name.
10321 (replace_one_candidate): Remove unnecessary argument; remove
10322 unnecessary parameter from calls to introduce_cast_before_cand.
10323 (replace_profitable_candidates): Remove unnecessary parameters
10324 from calls to replace_one_candidate.
10326 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10328 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10329 phi def as possibly hiding a basis for a CAND_ADD whose operands
10330 have been commuted in the analysis.
10331 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
10333 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10335 * config/aarch64/aarch64.md
10336 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
10337 shift value between 0-4.
10339 2013-05-07 Richard Biener <rguenther@suse.de>
10341 * double-int.h (rshift): New overload.
10342 * double-int.c (rshift): New function.
10343 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
10344 (create_reference_ops_from_ref): Remove.
10345 (vn_reference_insert): Use shared ops for constructing the
10346 reference and copy it.
10348 2013-05-07 Richard Biener <rguenther@suse.de>
10350 PR middle-end/57190
10351 * tree-eh.c (sink_clobbers): Properly propagate
10352 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
10354 2013-05-07 Jakub Jelinek <jakub@redhat.com>
10356 PR tree-optimization/57149
10357 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
10358 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
10359 collect_phi_def_edges, execute_late_warn_uninitialized): Use
10360 uninit_undefined_value_p instead of ssa_undefined_value_p.
10363 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
10364 for modifier == EXPAND_INITIALIZER.
10366 2013-05-07 Anton Blanchard <anton@samba.org>
10368 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
10369 for powerpc64 little endian.
10370 * configure: Regenerate.
10372 2013-05-06 Graham Stott <grahams@btinternet.com>
10374 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
10375 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
10376 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
10377 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
10379 2013-05-06 Graham Stott <grahams@btinternet.com>
10381 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
10382 codes which allow non-lvalues.
10384 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
10386 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
10387 components are all 1s.
10388 (integer_minus_onep): New function.
10389 * tree.h (integer_minus_onep): Declare it.
10390 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
10391 integer_minus_onep instead of integer_all_onesp.
10393 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10396 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
10397 variations of these patterns.
10399 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
10401 * config/i386/i386.md (isa): Add x64_sse4 member.
10402 (enabled): Handle x64_sse4.
10403 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
10404 instruction for 64bit SSE4_1 targets. Update insn attributes.
10405 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
10406 instruction for SSE4_1 targets. Update insn attributes.
10407 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10408 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
10409 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
10411 (*vec_extractv4si): Rename from *sse4_1_pextrd.
10412 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
10413 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
10415 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10418 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
10420 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
10422 * final.c (do_assembler_dialects): Don't handle curly braces and
10423 vertical bar escaped by % as dialect delimiters.
10424 (output_asm_insn): Print curly braces and vertical bar if escaped
10425 by % and ASSEMBLER_DIALECT defined.
10426 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
10427 * doc/tm.texi: Regenerated.
10429 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
10431 * config/mips/mips.c: Include tree-pass.h.
10432 (mips_reorg): Split in pre- and post-dbr_schedule parts.
10433 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
10434 (pass_mips_machine_reorg2): New machine specific pass.
10435 (insert_pass_mips_machine_reorg2): New pass plugin definition.
10436 (mips_option_override): Register the new pass.
10437 * rtl.h (cleanup_barriers): Remove prototype.
10438 (dbr_schedule): Likewise.
10439 * jump.c (cleanup_barriers): Make static.
10440 * reorg.c (dbr_schedule): Likewise.
10442 2013-05-06 Richard Biener <rguenther@suse.de>
10444 PR tree-optimization/57185
10445 * tree-parloops.c (add_field_for_reduction): Handle anonymous
10446 SSA names properly.
10448 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
10451 * config/i386/i386.c (add_parameter_dependencies): Add dependence
10452 between "first_arg" and "insn", not "last" and "insn".
10454 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10456 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
10457 (find_candidates_in_block): Re-enable slsr_process_phi.
10458 (create_phi_basis): Fix double counting of candidate adjustment.
10460 2013-05-06 Richard Biener <rguenther@suse.de>
10462 PR middle-end/57147
10463 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
10464 the edge is also fallthru, preserve it and just clear the
10466 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
10467 also complex, preserve that and just clear the fallthru flag.
10468 * tree-inline.c (update_ssa_across_abnormal_edges): Also
10469 update virtual operands.
10471 2013-05-06 Alan Modra <amodra@gmail.com>
10473 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
10474 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
10475 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
10476 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
10477 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10478 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
10480 2013-05-06 Alan Modra <amodra@gmail.com>
10482 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
10483 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
10484 (DEFAULT_ASM_ENDIAN): Define.
10485 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10486 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
10487 Update -K PIC clause from sysv4.h.
10488 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
10489 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
10491 2013-05-06 Alan Modra <amodra@gmail.com>
10493 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
10494 twice for little-endian.
10495 (ashrdi3_no_power, ashrdi3): Support little-endian.
10497 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10500 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
10501 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
10503 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
10504 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
10505 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
10508 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
10509 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
10511 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
10512 * config/pa/pa.opt: Make mbig-switch a no-op.
10513 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
10514 (CASE_VECTOR_MODE): Always return SImode.
10515 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
10516 for the !TARGET_BIG_SWITCH case.
10517 * config/pa/pa-linux.h: Likewise.
10518 * config/pa/pa-openbsd.h: Likewise.
10519 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
10520 * config/pa/pa.md (short_jump): Remove define_insn.
10521 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
10522 (casesi0): Remove define_insn.
10523 (type): Remove btable_branch.
10524 (pa_combine_type): Likewise.
10525 (in_nullified_branch_delay): Likewise.
10526 (in_call_delay): Likewise.
10527 (define_delay): Likewise.
10528 (define_insn_reservation "Z3"): Likewise.
10529 (define_insn_reservation "Z4"): Likewise.
10530 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10531 (pa_adjust_insn_length): Remove adjustment for btable branches.
10532 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10535 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
10537 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10538 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10539 Add m->r,x alternatives.
10540 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10541 splitters using SWI48x mode iterator.
10542 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
10543 TARGET_64BIT. Add m->x alternative.
10544 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10545 Add o->x alternative. Enable for TARGET_SSE.
10546 (sse_storeq): Remove expander.
10547 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10548 with memory input operand.
10549 (*vec_extractv2di_1 splitter): New.
10550 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10551 * config/i386/i386.md (ssevecmodelower): New mode attribute.
10553 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
10555 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
10556 (INT_LOWPART): Delete.
10557 (extract_MB): Adjust.
10558 (extract_ME): Adjust.
10559 (print_operand): Adjust.
10561 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
10563 * config/rs6000/predicates.md (reg_or_add_cint_operand,
10564 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10565 (reg_or_logical_cint_operand, easy_fp_constant,
10566 logical_const_operand): Delete "CONST_DOUBLE" case.
10567 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10568 "HOST_BITS_PER_WIDE_INT == 64" test.
10569 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
10570 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10571 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10572 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10574 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10575 CONST_DOUBLE DImode/VOIDmode case.
10576 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10577 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
10578 CONST_DOUBLE VOIDmode case.
10579 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10580 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10581 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10582 Delete CONST_DOUBLE case.
10583 (splitters for mov FMOVE64 const_double): Delete
10584 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
10585 "HOST_BITS_PER_WIDE_INT >= 64" test.
10586 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10588 (mov DI const_double): Delete.
10590 2013-05-04 Jakub Jelinek <jakub@redhat.com>
10592 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10593 on op shows all bits zero in mode of a lowpart subreg, return zero.
10595 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10598 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10599 to save TFmode registers and DImode to save TImode registers for
10600 caller save operations.
10601 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10602 mark being partially clobbered since they only use the first
10605 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10606 and TDmode only use the upper 64-bits of each VSX register.
10608 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10610 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10611 (find_candidates_in_block): Disable slsr_process_phi.
10613 2013-05-03 Guozhi Wei <carrot@google.com>
10615 * coverage.c (coverage_obj_init): Move the construction of gcov
10617 (build_init_ctor): ... here.
10619 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10621 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10622 (slsr_cand_d): Redefine def_phi.
10623 (stride_status, phi_adjust_status, count_phis_status): New enums.
10624 (find_phi_def): New.
10625 (find_basis_for_base_expr): New.
10626 (find_basis_for_candidate): Handle hidden bases.
10627 (alloc_cand_and_find_basis): Handle phi candidates.
10628 (slsr_process_phi): New.
10629 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10630 (create_mul_imm_cand): Likewise.
10631 (create_add_ssa_cand): Exclude phi base candidates.
10632 (create_add_imm_cand): Likewise.
10633 (slsr_process_cast): Likewise.
10634 (slsr_process_copy): Likewise.
10635 (find_candidates_in_block): Handle phi candidates.
10636 (dump_candidate): Likewise.
10637 (unconditional_cands): Delete.
10638 (unconditional_cands_with_known_stride_p): Delete.
10639 (phi_dependent_cand_p): New.
10640 (cand_increment): Handle phi-dependent candidates.
10641 (replace_dependent): Delete.
10642 (replace_mult_candidate): New.
10643 (replace_unconditional_candidate): New.
10644 (incr_vec_index): Move to avoid forward reference.
10645 (create_add_on_incoming_edge): New.
10646 (create_phi_basis): New.
10647 (replace_dependents): Delete.
10648 (replace_conditional_candidate): New.
10649 (phi_add_costs): New.
10650 (replace_uncond_cands_and_profitable_phis): New.
10651 (record_increment): Handle phi adjustments.
10652 (record_phi_increments): New.
10653 (record_increments): Handle phi adjustments.
10654 (phi_incr_cost): New.
10655 (lowest_cost_path): Handle phis.
10656 (total_savings): Likewise.
10657 (analyze_increments): Likewise.
10658 (ncd_with_phi): New.
10659 (ncd_of_cand_and_phis): New.
10660 (nearest_common_dominator_for_cands): Handle phi increments.
10661 (all_phi_incrs_profitable): New.
10662 (replace_profitable_candidates): Handle phi-dependent candidates.
10663 (analyze_candidates_and_replace): Likewise.
10665 2013-05-03 Teresa Johnson <tejohnson@google.com>
10668 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10669 do not exceed REG_BR_PROB_BASE.
10671 2013-05-03 Jeff Law <law@redhat.com>
10673 PR tree-optimization/57144
10674 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10675 operand of the condition will bit into the new type when eliminating
10676 a cast feeding a condition.
10678 2013-05-03 Jakub Jelinek <jakub@redhat.com>
10680 PR rtl-optimization/57130
10681 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10682 of COMPARE as in_code to the recursive call if needed.
10684 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
10686 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10687 (enabled): Handle new members.
10688 * config/i386/sse.md (*vec_concatv2si): Merge from
10689 *vec_concatv2si_sse2 and vec_concatv2si_sse.
10690 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10692 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
10694 PR tree-optimization/57027
10695 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10696 for fnms opportunity, check we got the prerequisite kind
10697 of tree / gimple before using accessor functions.
10699 2013-05-03 Richard Biener <rguenther@suse.de>
10701 * double-int.h (lshift): New overload without precision
10702 and arith argument.
10703 (operator *=, operator +=, operator -=): Move ...
10704 * double-int.c (operator *=, operator +=, operator -=): ... here
10705 and implement more efficiently.
10706 (mul_double_with_sign): Remove.
10707 (lshift_double): Adjust to take unsinged shift argument, push
10708 dispatching code to callers.
10709 (mul_double_wide_with_sign): Add early out for callers that
10710 are not interested in high parts or overflow.
10711 (lshift): New function.
10712 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10713 dispatch code here.
10714 (lrotate, rrotate): Use logical shifts.
10715 * expr.c (get_inner_reference): Use lshift.
10716 * fixed-value.c (do_fixed_divide): Likewise.
10717 * tree-dfa.c (get_ref_base_and_extent): Likewise.
10718 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10719 (indirect_refs_may_alias_p): Likewise.
10720 (stmt_kills_ref_p_1): Likewise.
10722 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
10724 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10726 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
10728 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10729 scalar form of FABD instruction.
10731 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
10733 * lra-constraints.c (process_alt_operands): Add checking alt
10734 number to choose the best alternative.
10736 2013-05-02 Richard Biener <rguenther@suse.de>
10738 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10739 bitmap and its handling.
10740 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10742 2013-05-02 Richard Biener <rguenther@suse.de>
10744 PR middle-end/57140
10745 * tree-inline.c (copy_loops): Properly handle removed loops.
10746 (copy_cfg_body): Mark destination loops for fixup if source
10747 loops needed fixup.
10749 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
10752 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10753 generating simple_return for naked functions.
10755 2013-05-02 Martin Jambor <mjambor@suse.cz>
10757 PR middle-end/56988
10758 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10759 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10761 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10762 ipa_agg_replacement_value structures.
10763 (known_aggs_to_agg_replacement_list): Likewise.
10764 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10765 (read_agg_replacement_chain): Likewise.
10766 (ipcp_transform_function): Also check that by_ref flags match.
10768 2013-05-02 Richard Biener <rguenther@suse.de>
10770 * graphds.h (struct graph): Add obstack member.
10771 * graphds.c (new_graph): Initialize obstack and allocate
10773 (add_edge): Allocate edge from the obstack.
10774 (free_graph): Free the obstack instead of all edges and vertices.
10776 2013-05-02 Teresa Johnson <tejohnson@google.com>
10778 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10780 * cfg.c (update_bb_profile_for_threading): Ditto.
10781 * tree-inline.c (copy_bb): Ditto.
10782 (copy_edges_for_bb): Ditto.
10783 (initialize_cfun): Ditto.
10784 (copy_cfg_body): Ditto.
10785 (expand_call_inline): Ditto.
10786 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10787 (estimate_node_size_and_time): Ditto.
10788 (inline_merge_summary): Ditto.
10789 * cgraphclones.c (cgraph_clone_edge): Ditto.
10790 (cgraph_clone_node): Ditto.
10791 * sched-rgn.c (compute_dom_prob_ps): Ditto.
10792 (compute_trg_info): Ditto.
10794 2013-05-02 Ian Bolton <ian.bolton@arm.com>
10796 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10797 S reg when fp attribute set.
10798 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10800 2013-05-02 Ian Bolton <ian.bolton@arm.com>
10802 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10804 (*and_one_cmplsi3_compare0_uxtw): Likewise.
10805 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10806 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10808 2013-05-02 Richard Biener <rguenther@suse.de>
10810 * tree-scalar-evolution.c (scev_info_hasher): Remove.
10811 (struct instantiate_cache_entry): New type.
10812 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10813 (struct instantiate_cache_type): New type.
10814 (set_instantiated_value, get_instantiated_value): Remove.
10815 (get_instantiated_value_entry): New function.
10816 (instantiate_scev_name): Use the new cache and adjust.
10817 (instantiate_scev_poly): Adjust.
10818 (instantiate_scev_binary): Likewise.
10819 (instantiate_array_ref): Likewise.
10820 (instantiate_scev_convert): Likewise.
10821 (instantiate_scev_not): Likewise.
10822 (instantiate_scev_3): Likewise.
10823 (instantiate_scev_2): Likewise.
10824 (instantiate_scev_r): Likewise.
10825 (instantiate_scev): Likewise.
10826 (resolve_mixers): Likewise.
10828 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
10831 * lra-constraints.c (best_small_class_operands_num): Remove.
10832 (process_alt_operands): Remove small_class_operands_num. Take
10833 small classes operands into losers and only if the operand is not
10834 matched. Modify debugging output.
10835 (curr_insn_transform): Remove best_small_class_operands_num.
10838 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10840 * config/aarch64/aarch64-builtins.c
10841 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10842 * config/aarch64/aarch64-simd-builtins.def
10843 (reduc_splus_): Add new modes.
10844 (reduc_uplus_): New.
10845 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10846 (reduc_uplus_v4sf): Likewise.
10847 (reduc_splus_v4sf): Likewise.
10848 (aarch64_addv<mode>): Likewise.
10849 (reduc_uplus_<mode>): Likewise.
10850 (reduc_splus_<mode>): Likewise.
10851 (aarch64_addvv2di): Likewise.
10852 (reduc_uplus_v2di): Likewise.
10853 (reduc_splus_v2di): Likewise.
10854 (aarch64_addvv2si): Likewise.
10855 (reduc_uplus_v2si): Likewise.
10856 (reduc_splus_v2si): Likewise.
10857 (reduc_<sur>plus_<mode>): New.
10858 (reduc_<sur>plus_v2di): Likewise.
10859 (reduc_<sur>plus_v2si): Likewise.
10860 (reduc_<sur>plus_v4sf): Likewise.
10861 (aarch64_addpv4sf): Likewise.
10862 * config/aarch64/arm_neon.h
10863 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10864 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10865 add UNSPEC_SADDV, UNSPEC_UADDV.
10867 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10869 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10871 * config/aarch64/arm_neon.h
10872 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10874 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10876 * config/aarch64/aarch64-builtins
10877 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10879 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10881 * config/aarch64/aarch64-simd-builtins.def
10882 (reduc_smax_): New.
10883 (reduc_smin_): Likewise.
10884 (reduc_umax_): Likewise.
10885 (reduc_umin_): Likewise.
10886 (reduc_smax_nan_): Likewise.
10887 (reduc_smin_nan_): Likewise.
10890 (smax): Update for V2SF, V4SF and V2DF modes.
10893 (smin_nan): Likewise.
10894 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10895 (<su><maxmin><mode>3): ...This, refactor.
10896 (s<maxmin><mode>3): New.
10897 (<maxmin_uns><mode>3): Likewise.
10898 (reduc_<maxmin_uns>_<mode>): Refactor.
10899 (reduc_<maxmin_uns>_v4sf): Likewise.
10900 (reduc_<maxmin_uns>_v2si): Likewise.
10901 (aarch64_<fmaxmin><mode>: Remove.
10902 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10904 (vmin<q>_f<32,64>): Likewise.
10905 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10907 (su): Add mappings for smax, smin, umax, umin.
10909 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10910 (FMAXMIN): Rename as...
10911 (FMAXMIN_UNS): ...This.
10913 (fmaxminv): Likewise.
10914 (fmaxmin): Likewise.
10916 (maxmin_uns_op): Likewise.
10918 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10920 * config/aarch64/arm_neon.h
10921 (vac<ge, gt><sd>_f<32, 64>): Rename to...
10922 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10923 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10925 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10927 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10928 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10930 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10932 * config/aarch64/aarch64-simd.md
10933 (vcond<mode>_internal): Handle special cases for constant masks.
10934 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10935 (vcondu<mode><mode>): Likewise.
10936 (vcond<v_cmp_result><mode>): New.
10938 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10940 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10941 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10942 * config/aarch64/aarch64-simd-builtins.def
10943 (cmeq): Update to BUILTIN_VALLDI.
10948 * config/aarch64/arm_neon.h
10949 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10950 to builtins or C as appropriate.
10952 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10954 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10956 (cmhi): Rename to...
10958 * config/aarch64/aarch64-simd.md
10959 (simd_mode): Add SF.
10960 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10961 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10962 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10963 (cstore<mode>_neg): ...This.
10964 * config/aarch64/iterators.md
10966 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10967 (COMPARISONS): New.
10968 (UCOMPARISONS): Likewise.
10969 (optab): Add missing comparisons.
10975 (VCMP_S): Likewise.
10976 (VCMP_U): Likewise.
10977 (V_cmp_result): Add DF, SF modes.
10978 (v_cmp_result): Likewise.
10980 (vmtype): Likewise.
10981 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10983 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
10985 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10986 define_insn to define_insn_and_split.
10987 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10988 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10989 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10990 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10991 (thumb2_negscc): Likewise.
10993 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
10995 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10997 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
10999 * config/arm/thumb2.md: Remove trailing whitespaces.
11001 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
11003 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
11004 Use gen_int_mode rather than GEN_INT.
11006 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
11008 * value-prof.c (stream_in_histogram_value): Remove the strayed
11011 2013-04-30 Richard Biener <rguenther@suse.de>
11013 PR middle-end/57122
11014 * cfghooks.c (split_edge): Properly check for the loop latch edge.
11016 2013-04-30 Richard Biener <rguenther@suse.de>
11018 PR middle-end/57107
11019 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
11021 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
11023 PR rtl-optimization/56957
11024 PR rtl-optimization/57105
11025 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
11026 variable. Use just INSN_UID for determining whether an insn
11027 should be only disconnected from the insn stream.
11028 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
11030 2013-04-30 Jakub Jelinek <jakub@redhat.com>
11032 PR tree-optimization/57104
11033 * tsan.c (instrument_expr): Don't instrument accesses to
11034 DECL_HARD_REGISTER VAR_DECLs.
11036 2013-04-30 Richard Biener <rguenther@suse.de>
11038 * function.h (loops_for_fn): New inline function.
11039 (set_loops_for_fn): Likewise.
11040 * cfgloop.h (place_new_loop): Add struct function parameter.
11041 (get_loop): Likewise.
11042 (get_loops): Likewise.
11043 (number_of_loops): Likewise.
11044 (fel_next): Adjust.
11045 (fel_init): Likewise.
11046 * cfg.c (get_loop_copy): Adjust.
11047 * cfgloop.c (flow_loops_dump): Likewise.
11048 (record_loop_exits): Likewise.
11049 (verify_loop_structure): Likewise.
11050 * cfgloopanal.c (mark_irreducible_loops): Likewise.
11051 (estimate_reg_pressure_cost): Likewise.
11052 (mark_loop_exit_edges): Likewise.
11053 * cfgloopmanip.c (place_new_loop): Likewise.
11054 (add_loop): Likewise.
11055 (duplicate_loop): Likewise.
11056 * graph.c (draw_cfg_nodes): Likewise.
11057 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
11058 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
11059 (extract_affine_chrec): Likewise.
11060 (build_scop_iteration_domain): Likewise.
11061 * graphite.c (graphite_initialize): Likewise.
11062 * ira-build.c (create_loop_tree_nodes): Likewise.
11063 (more_one_region_p): Likewise.
11064 (rebuild_regno_allocno_maps): Likewise.
11065 (mark_loops_for_removal): Likewise.
11066 (mark_all_loops_for_removal): Likewise.
11067 (remove_unnecessary_regions): Likewise.
11068 (ira_build): Likewise.
11069 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
11070 * loop-init.c (fix_loop_structure): Likewise.
11071 (gate_rtl_move_loop_invariants): Likewise.
11072 (gate_rtl_unswitch): Likewise.
11073 (gate_rtl_unroll_and_peel_loops): Likewise.
11074 (rtl_doloop): Likewise.
11075 * lto-streamer-in.c (input_cfg): Likewise.
11076 * lto-streamer-out.c (output_cfg): Likewise.
11077 * modulo-sched.c (sms_schedule): Likewise.
11078 * predict.c (tree_estimate_probability): Likewise.
11079 (tree_estimate_probability_driver): Likewise.
11080 (estimate_loops): Likewise.
11081 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
11082 (move_sese_region_to_fn): Likewise.
11083 (debug_loop_num): Likewise.
11084 * tree-chrec.c (chrec_evaluate): Likewise.
11085 (hide_evolution_in_other_loops_than_loop): Likewise.
11086 (chrec_component_in_loop_num): Likewise.
11087 (reset_evolution_in_loop): Likewise.
11088 (evolution_function_is_invariant_rec_p): Likewise.
11089 * tree-if-conv.c (main_tree_if_conversion): Likewise.
11090 * tree-inline.c (copy_loops): Likewise.
11091 (copy_cfg_body): Likewise.
11092 (tree_function_versioning): Likewise.
11093 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
11094 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
11096 (add_to_evolution_1): Likewise.
11097 (scev_const_prop): Likewise.
11098 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
11099 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
11100 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
11101 (tree_ssa_lim_initialize): Likewise.
11102 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
11103 (verify_loop_closed_ssa): Likewise.
11104 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
11105 (tree_ssa_loop_im): Likewise.
11106 (tree_ssa_loop_unswitch): Likewise.
11107 (tree_vectorize): Likewise.
11108 (check_data_deps): Likewise.
11109 (tree_ssa_loop_ivcanon): Likewise.
11110 (tree_ssa_loop_bounds): Likewise.
11111 (tree_complete_unroll): Likewise.
11112 (tree_complete_unroll_inner): Likewise.
11113 (tree_parallelize_loops): Likewise.
11114 (tree_ssa_loop_prefetch): Likewise.
11115 (tree_ssa_loop_ivopts): Likewise.
11116 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
11117 * tree-vectorizer.c (vectorize_loops): Likewise.
11119 2013-04-29 Mike Frysinger <vapier@gentoo.org>
11121 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
11122 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
11123 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
11124 with EABI_LINK_SPEC.
11126 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11129 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
11132 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
11135 * lra-constraints.c (process_alt_operands): Discourage a bit more
11136 using memory for pseudos. Print cost dump for alternatives.
11137 Modify cost values for conflicts with early clobbers.
11138 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
11140 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11143 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
11145 2013-04-29 Ian Bolton <ian.bolton@arm.com>
11147 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
11148 from/to S register.
11149 (movdi_aarch64): Support LDR/STR from/to D register.
11151 2013-04-29 Ian Bolton <ian.bolton@arm.com>
11153 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
11154 or higher by default.
11156 2013-04-29 Richard Biener <rguenther@suse.de>
11158 PR middle-end/57075
11159 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
11160 even if not adding abnormal edges for calls that can make
11163 2013-04-29 Richard Biener <rguenther@suse.de>
11165 PR middle-end/57103
11166 * tree-cfg.c (move_stmt_op): Fix condition under which to update
11168 (move_stmt_r): Remove redundant checking.
11170 2013-04-29 Teresa Johnson <tejohnson@google.com>
11173 * basic-block.h (apply_scale): New function.
11174 (apply_probability): Use apply_scale.
11175 * gimple-streamer-in.c (input_bb): Ditto.
11176 * lto-streamer-in.c (input_cfg): Ditto.
11177 * lto-cgraph.c (merge_profile_summaries): Ditto.
11178 * tree-optimize.c (execute_fixup_cfg): Ditto.
11179 * tree-inline.c (copy_bb): Update comment to use apply_scale.
11180 (copy_edges_for_bb): Ditto.
11181 (copy_cfg_body): Ditto.
11183 2013-04-29 Tom de Vries <tom@codesourcery.com>
11185 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
11186 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
11187 (tail_merge_optimize): Handle current_loops == NULL.
11189 2013-04-26 Jeff Law <law@redhat.com>
11191 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
11192 (simplify_cond_using_ranges): Generalize code to simplify
11193 COND_EXPRs where one argument is a constant and the other
11194 is an SSA_NAME created by an integral type conversion.
11196 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11198 * config/arm/arm.md (store_minmaxsi): Use only when
11199 optimize_insn_for_size_p.
11201 2013-04-29 Christian Bruel <christian.bruel@st.com>
11204 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
11206 2013-04-29 Richard Biener <rguenther@suse.de>
11208 PR middle-end/57089
11209 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
11210 loop tree make sure to schedule a fixup for the child as well.
11211 (expand_omp_for_generic): Properly add loops.
11212 (expand_omp_for_static_nochunk): Likewise.
11213 (expand_omp_for_static_chunk): Likewise.
11214 (expand_omp_for): For the degenerate case fixup loops.
11215 (expand_omp_sections): Fix default bb placement in loops.
11216 (expand_omp_atomic_pipeline): Properly add loops.
11218 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11220 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
11222 2013-04-29 Tom de Vries <tom@codesourcery.com>
11224 * tree-ssa-tail-merge.c: Update header comment.
11226 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11228 * config/aarch64/arm_neon.h
11229 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
11230 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
11231 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
11232 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
11233 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
11234 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
11235 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
11236 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
11238 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11240 * config/aarch64/aarch64-simd.md
11241 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
11242 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
11243 fix_trunc, fixuns_trunc.
11244 (ftrunc<VDQF:mode>2): New.
11245 * config/aarch64/iterators.md (optab): Add fix, fixuns.
11246 (fix_trunc_optab): New.
11248 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11250 * config/aarch64/aarch64-builtins.c
11251 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
11252 iceilf, lround, iroundf.
11254 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11257 * config/i386/i386.h (enum ix86_tune_indices)
11258 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
11259 New, split from X86_TUNE_INTER_UNIT_MOVES.
11260 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
11261 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
11262 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
11263 (TARGET_INTER_UNIT_MOVES): Remove.
11264 * config/i386/i386.c (initial_ix86_tune_features): Update.
11265 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
11266 (ix86_expand_convert_uns_didf_sse): Use
11267 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11268 (ix86_expand_vector_init_one_nonzero): Ditto.
11269 (ix86_expand_vector_init_interleave): Ditto.
11270 (inline_secondary_memory_needed): Return true for moves from SSE class
11271 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
11272 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11273 * config/i386/constraints.md (Yi, Ym): Depend on
11274 TARGET_INTER_UNIT_MOVES_TO_VEC.
11275 (Yj, Yn): New constraints.
11276 * config/i386/i386.md (*movdi_internal): Change constraints of
11277 operand 1 from Yi to Yj and from Ym to Yn.
11278 (*movsi_internal): Ditto.
11279 (*movdf_internal): Ditto.
11280 (*movsf_internal): Ditto.
11281 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
11282 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11283 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
11284 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
11285 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
11286 * config/i386/sse.md (movdi_to_sse): Ditto.
11287 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
11288 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
11289 TARGET_INTER_UNIT_MOVES.
11290 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
11291 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
11292 instead of TARGET_INTER_UNIT_MOVES.
11293 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
11294 operand 1 from Yi to Yj and from Ym to Yn.
11296 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11298 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
11299 (float_truncate_hi_): Likewise.
11300 (float_extend_lo_): Likewise.
11301 (float_truncate_lo_): Likewise.
11302 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
11303 (aarch64_float_extend_lo_v2df): Likewise.
11304 (vec_unpacks_hi_v4sf): Likewise.
11305 (aarch64_float_truncate_lo_v2sf): Likewise.
11306 (aarch64_float_truncate_hi_v4sf): Likewise.
11307 (vec_pack_trunc_v2df): Likewise.
11308 (vec_pack_trunc_df): Likewise.
11310 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11312 * config/aarch64/aarch64-builtins.c
11313 (aarch64_fold_builtin): Fold float conversions.
11314 * config/aarch64/aarch64-simd-builtins.def
11315 (floatv2si, floatv4si, floatv2di): New.
11316 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
11317 * config/aarch64/aarch64-simd.md
11318 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
11319 * config/aarch64/iterators.md (FLOATUORS): New.
11320 (optab): Add float, floatuns.
11321 (su_optab): Likewise.
11323 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11325 * config/aarch64/aarch64-builtins.c
11326 (aarch64_builtin_vectorized_function): Use new names for
11328 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
11329 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
11330 (fcvtzu): Split as...
11331 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
11332 (fcvtas): Split as...
11333 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
11334 (fcvtau): Split as...
11335 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
11336 (fcvtps): Split as...
11337 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
11338 (fcvtpu): Split as...
11339 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
11340 (fcvtms): Split as...
11341 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
11342 (fcvtmu): Split as...
11343 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
11344 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
11345 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
11346 (lfrintnusf, lfrintnudf): Likewise.
11347 * config/aarch64/aarch64-simd.md
11348 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
11350 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
11351 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
11352 (fcvt_pattern): Likewise.
11354 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11356 * config/aarch64/aarch64-simd.md
11357 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
11358 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
11360 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11362 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
11363 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
11364 (vrnd<a,m,n,p>_f32): Implement using builtins.
11365 (vrnd<i,x><q>_f<32, 64>): New.
11367 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11369 * config/aarch64/aarch64-builtins.c
11370 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
11371 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
11372 (frintz): Rename to...
11374 (frintp): Rename to...
11376 (frintm): Rename to...
11378 (frinti): Rename to...
11379 (nearbyint): ...this.
11380 (frintx): Rename to...
11382 (frinta): Rename to...
11384 * config/aarch64/aarch64-simd.md
11385 (aarch64_frint<frint_suffix><mode>): Delete.
11386 (<frint_pattern><mode>2): Convert to insn.
11387 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
11388 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
11389 (frint_pattern): Likewise.
11390 (frint_suffix): Likewise.
11392 2013-04-29 Richard Biener <rguenther@suse.de>
11394 PR tree-optimization/57081
11395 * loop-init.c: Include tree-flow.h.
11396 (loop_optimizer_finalize): Free number of iteration estimates.
11397 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
11399 2013-04-29 Jakub Jelinek <jakub@redhat.com>
11401 PR tree-optimization/57083
11402 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
11403 non-singleton shift count range, zero extend low_bound for uns case.
11405 * config/i386/predicates.md (general_vector_operand): New predicate.
11406 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
11407 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
11408 if they aren't nonimmediate operands. If their original values
11409 satisfy const_vector_equal_evenodd_p, don't shift them.
11410 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
11411 predicates. For the SSE4.1 case force operands[{1,2}] into registers
11412 if not nonimmediate_operand.
11413 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
11414 instead of register_operand.
11415 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
11417 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
11419 * stor-layout.c (finalize_size_functions): Allocate a structure and
11420 reset cfun before dumping the functions.
11422 2013-04-27 Jakub Jelinek <jakub@redhat.com>
11424 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
11427 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
11428 use xop_pmacsdqh if uns_p.
11429 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
11430 the immediate rotate count.
11432 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
11434 * rtl.h (struct rtx_def): Add comment for field jump.
11435 (LRA_SUBREG_P): New macro.
11436 * recog.c (register_operand): Check LRA_SUBREG_P.
11437 * lra.c (lra): Add note at the end of RTL code. Align non-empty
11439 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11440 (lra_final_code_change): Skip subreg change for operators.
11441 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11442 if there are no operand changes.
11443 * lra-constraints.c (curr_insn_set): New.
11444 (match_reload): Set LRA_SUBREG_P.
11445 (emit_spill_move): Ditto.
11446 (check_and_process_move): Use curr_insn_set. Process only single
11447 set insns. Don't initialize sec_mem_p and change_p.
11448 (simplify_operand_subreg): Use LRA_SUBREG_P.
11449 (reg_in_class_p): New function.
11450 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
11451 of #ifdef. Add code to remove cycling.
11452 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11453 non-null disp. Reload inner instead of disp when base and index
11454 are null. Try to put lo_sum into register.
11455 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11456 (check_and_process_move): Move code for move cost check to
11457 simple_move_p. Remove equiv_substitution.
11458 (simple_move_p): New function.
11459 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
11460 curr_insn_set. Call check_and_process_move only for single set
11461 insns. Use the new function. Move call of check_and_process_move
11462 after operand equiv substitution and address process.
11464 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11467 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
11468 with nonlocal goto receivers or returns twice calls, ignore
11469 unininitialized values from abnormal edges to nl goto receiver
11470 or returns twice call.
11472 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11474 PR tree-optimization/57051
11475 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
11476 and VEC_RSHIFT_EXPR if shift count is a multiple of element
11479 2013-04-26 Richard Biener <rguenther@suse.de>
11481 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
11482 (expand_omp_taskreg): Likewise. Mark loops for fixup.
11483 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
11484 (fixup_loop_arrays_after_move): New function.
11485 (move_sese_region_to_fn): Properly outline the loop tree parts
11486 of the SESE region.
11488 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
11490 * config/i386/i386.md (type, unit): Fix long lines.
11492 2013-04-26 Richard Biener <rguenther@suse.de>
11494 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
11495 (lto-streamer-out.o): Likewise.
11496 * cfgloop.c (init_loops_structure): Export, add struct function
11497 argument and adjust.
11498 (flow_loops_find): Adjust.
11499 * cfgloop.h (enum loop_estimation): Add EST_LAST.
11500 (init_loops_structure): Declare.
11501 * lto-streamer-in.c: Include cfgloop.h.
11502 (input_cfg): Input the loop tree.
11503 * lto-streamer-out.c: Include cfgloop.h.
11504 (output_cfg): Output the loop tree.
11505 (output_struct_function_base): Do not drop PROP_loops.
11507 2013-03-26 Richard Biener <rguenther@suse.de>
11509 * tree-cfg.c (execute_build_cfg): Build the loop tree.
11510 (pass_build_cfg): Provide PROP_loops.
11511 (move_sese_region_to_fn): Remove loops that are outlined into fn
11513 * tree-inline.c: Include cfgloop.h.
11514 (initialize_cfun): Do not drop PROP_loops.
11515 (copy_loops): New function.
11516 (copy_cfg_body): Copy loop structure.
11517 (tree_function_versioning): Initialize destination loop tree.
11518 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
11519 (pass_parallelize_loops): Do IL verification.
11520 * loop-init.c (loop_optimizer_init): Fixup loops if required.
11521 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
11522 the CFG make sure we fixup loops as well.
11523 * tree-ssa-tail-merge.c: Include cfgloop.h.
11524 (replace_block_by): When merging loop latches mark loops for fixup.
11525 * lto-streamer-out.c (output_struct_function_base): Drop
11526 PROP_loops for now.
11527 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11528 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11529 * ipa-split.c: Include cfgloop.h.
11530 (split_function): Add the new return block to the loop tree root.
11531 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11532 whether we have removed the forwarder block.
11533 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11534 * cfgloop.h (place_new_loop): Declare.
11535 * cfgloopmanip.c (place_new_loop): Export.
11536 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11537 (tree-switch-conversion.o): Likewise.
11538 (tree-complex.o): Likewise.
11539 (tree-inline.o): Likewise.
11540 (tree-ssa-tailmerge.o): Likewise.
11541 (ipa-split.o): Likewise.
11542 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11543 (tree-ssa-copy.o): Likewise.
11544 * tree-switch-conversion.c: Include cfgloop.h
11545 (process_switch): If we emit a bit-test cascade, schedule loops
11547 * tree-complex.c: Include cfgloop.h.
11548 (expand_complex_div_wide): Properly add new basic-blocks to loops.
11549 * asan.c: Include cfgloop.h.
11550 (create_cond_insert_point): Properly add new basic-blocks to
11551 loops, schedule loop fixup.
11552 * cfgloop.c (verify_loop_structure): Check that looks are not
11554 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11556 (expand_omp_for_generic): Likewise.
11557 (expand_omp_sections): Likewise.
11558 (expand_omp_atomic_pipeline): Schedule loops for fixup.
11559 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11560 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11561 is initialized, not when loops are present.
11562 * tree-parloops.c (parallelize_loops): Remove checking here.
11563 * passes.c (init_optimization_passes): Schedule a copy-propagation
11564 pass before complete unrolling of inner loops.
11566 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11568 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11569 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11570 (colorize_init): Add argument to _WIN32 version.
11571 * toplev.c: Include diagnostic-color.h.
11572 (process_options): Default to -fdiagnostics-color=auto if
11573 GCC_COLORS env var is in the environment.
11574 * common.opt (fdiagnostics-color=): Add Var and Init.
11575 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11576 env var is in the environment, the default is auto rather than never.
11578 * diagnostic.h (file_name_as_prefix): Add context argument.
11579 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
11580 the string as locus.
11581 * langhooks.c (lhd_print_error_function): Adjust caller.
11583 2013-04-25 Lawrence Crowl <crowl@google.com>
11585 * var-tracking.c (shared_hash_def::htab):
11586 Change type to hash_table. Update dependent calls and types.
11588 2013-04-25 Lawrence Crowl <crowl@google.com>
11590 * Makefile.in: Update as needed below.
11592 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11593 Move declaration to after the type's method definitons.
11595 * attribs.c (htab_t scoped_attributes::attribute_hash):
11596 Change type to hash_table. Update dependent calls and types.
11598 * bitmap.c (htab_t bitmap_desc_hash):
11599 Change type to hash_table. Update dependent calls and types.
11601 * cselib.c (htab_t cselib_hash_table):
11602 Change type to hash_table. Update dependent calls and types.
11604 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11605 (hash_string_slot_node): Move implementation into lto-streamer.h
11606 struct string_slot_hasher.
11607 (eq_string_slot_node): Likewise.
11609 * data-streamer-out.c: Update output_block::string_hash_table
11610 dependent calls and types.
11612 * dwarf2cfi.c (htab_t trace_index):
11613 Change type to hash_table. Update dependent calls and types.
11615 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11616 Change type to hash_table. Update dependent calls and types.
11617 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11618 (htab_t optimize_external_refs::map): Likewise.
11619 (htab_t output_comp_unit::extern_map): Likewise.
11620 (htab_t output_comdat_type_unit::extern_map): Likewise.
11621 (htab_t output_macinfo::macinfo_htab): Likewise.
11622 (htab_t optimize_location_lists::htab): Likewise.
11623 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11625 * except.c (htab_t ehspec_hash_type):
11626 Change type to hash_table. Update dependent calls and types.
11627 (assign_filter_values::ttypes): Likewise.
11628 (assign_filter_values::ehspec): Likewise.
11629 (sjlj_assign_call_site_values::ar_hash): Likewise.
11630 (convert_to_eh_region_ranges::ar_hash): Likewise.
11632 * gcse.c (htab_t pre_ldst_table):
11633 Change type to hash_table. Update dependent calls and types.
11635 * ggc-common.c (htab_t saving_htab):
11636 Change type to hash_table. Update dependent calls and types.
11637 (htab_t loc_hash): Likewise.
11638 (htab_t ptr_hash): Likewise.
11639 (call_count): Rename ggc_call_count.
11640 (call_alloc): Rename ggc_call_alloc.
11641 (loc_descriptor): Rename make_loc_descriptor.
11642 (add_statistics): Rename ggc_add_statistics.
11644 * ggc-common.c (saving_htab):
11645 Change type to hash_table. Update dependent calls and types.
11647 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11648 (push_gimplify_context): Likewise.
11649 (pop_gimplify_context): Likewise.
11650 (struct gimple_temp_hash_elt): Added.
11651 (struct gimplify_hasher): Likewise.
11652 (struct gimplify_ctx.temp_htab):
11653 Change type to hash_table. Update dependent calls and types.
11655 * gimple-fold.c: Include gimplify-ctx.h.
11657 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11658 Change type to hash_table. Update dependent calls and types.
11659 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11660 avoid potential global name collision.
11662 * gimplify.c: Include gimplify-ctx.h.
11663 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11664 (htab_t gimplify_ctx::temp_htab):
11665 Update dependent calls and types for new type hash_table.
11666 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11667 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11669 * gimplify-ctx.h: New.
11670 (struct gimple_temp_hash_elt): Move from gimplify.c.
11671 (class gimplify_hasher): New.
11672 (struct gimplify_ctx): Move from gimple.h.
11673 (htab_t gimplify_ctx::temp_htab):
11674 Change type to hash_table. Update dependent calls and types.
11676 * graphite-clast-to-gimple.c: Include graphite-htab.h.
11677 (htab_t ivs_params::newivs_index):
11678 Change type to hash_table. Update dependent calls and types.
11679 (htab_t ivs_params::params_index): Likewise.
11680 (htab_t print_generated_program::params_index): Likewise.
11681 (htab_t gloog::newivs_index): Likewise.
11682 (htab_t gloog::params_index): Likewise.
11684 * graphite.c: Include graphite-htab.h.
11685 4htab_t graphite_transform_loops::bb_pbb_mapping):
11686 Change type to hash_table. Update dependent calls and types.
11688 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11689 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11690 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11692 * graphite-dependences.c: Include graphite-htab.h.
11693 (loop_is_parallel_p): Change hash table type of parameter.
11695 * graphite-htab.h: New.
11696 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11697 (extern find_pbb_via_hash): Move from graphite-poly.h.
11698 (extern loop_is_parallel_p): Move from graphite-poly.h.
11699 (extern get_loop_body_pbbs): Move from graphite-poly.h.
11701 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11702 (extern loop_is_parallel_p): Move to graphite-htab.h.
11703 (extern get_loop_body_pbbs): Move to graphite-htab.h.
11705 * haifa-sched.c (htab_t delay_htab):
11706 Change type to hash_table. Update dependent calls and types.
11707 (htab_t delay_htab_i2): Likewise.
11709 * ira-color.c (htab_t allocno_hard_regs_htab):
11710 Change type to hash_table. Update dependent calls and types.
11712 * ira-costs.c (htab_t cost_classes_htab):
11713 Change type to hash_table. Update dependent calls and types.
11715 * loop-invariant.c (htab_t merge_identical_invariants::eq):
11716 Change type to hash_table. Update dependent calls and types.
11718 * loop-iv.c (htab_t bivs):
11719 Change type to hash_table. Update dependent calls and types.
11721 * loop-unroll.c (htab_t opt_info::insns_to_split):
11722 Change type to hash_table. Update dependent calls and types.
11723 (htab_t opt_info::insns_with_var_to_expand): Likewise.
11725 * lto-streamer.h (struct string_slot): Move from data-streamer.h
11726 (struct string_slot_hasher): New.
11727 (htab_t output_block::string_hash_table):
11728 Change type to hash_table. Update dependent calls and types.
11730 * lto-streamer-in.c (freeing_string_slot_hasher): New.
11731 (htab_t file_name_hash_table):
11732 Change type to hash_table. Update dependent calls and types.
11734 * lto-streamer-out.c: Update output_block::string_hash_table dependent
11737 * lto-streamer.c (htab_t tree_htab):
11738 Change type to hash_table. Update dependent calls and types.
11740 * omp-low.c: Include gimplify-ctx.h.
11742 * passes.c (htab_t name_to_pass_map):
11743 Change type to hash_table. Update dependent calls and types.
11744 (pass_traverse): Rename to passes_pass_traverse.
11746 * plugin.c (htab_t event_tab):
11747 Change type to hash_table. Update dependent calls and types.
11749 * postreload-gcse.c (htab_t expr_table):
11750 Change type to hash_table. Update dependent calls and types.
11751 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11753 * sese.c (debug_rename_map_1): Make extern.
11754 (htab_t copy_bb_and_scalar_dependences::rename_map):
11755 Change type to hash_table. Update dependent calls and types.
11757 * sese.h (extern debug_rename_map): Move to .c file.
11759 * store-motion.c (htab_t store_motion_mems_table):
11760 Change type to hash_table. Update dependent calls and types.
11762 * trans-mem.c (htab_t tm_new_mem_hash):
11763 Change type to hash_table. Update dependent calls and types.
11765 * tree-browser.c (htab_t TB_up_ht):
11766 Change type to hash_table. Update dependent calls and types.
11768 * tree-cfg.c (htab_t discriminator_per_locus):
11769 Change type to hash_table. Update dependent calls and types.
11771 * tree-complex.c: Include tree-hasher.h
11772 (htab_t complex_variable_components):
11773 Change type to hash_table. Update dependent calls and types.
11775 * tree-eh.c (htab_t finally_tree):
11776 Change type to hash_table. Update dependent calls and types.
11778 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11779 struct int_tree_hasher.
11780 (extern int_tree_map_eq): Likewise.
11781 (uid_decl_map_hash): Removed.
11782 (extern decl_tree_map_eq): Likewise.
11784 * tree-hasher.h: New.
11785 (struct int_tree_hasher): New.
11786 (typedef int_tree_htab_type): New.
11788 * tree-inline.c: Include gimplify-ctx.h.
11790 * tree-mudflap.c: Include gimplify-ctx.h.
11792 * tree-parloops.c: Include tree-hasher.h.
11793 (htab_t eliminate_local_variables_stmt::decl_address):
11794 Change type to hash_table. Update dependent calls and types.
11795 (htab_t separate_decls_in_region::decl_copies): Likewise.
11797 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11798 Change type to hash_table. Update dependent calls and types.
11800 * tree-sra.c (candidates):
11801 Change type to hash_table. Update dependent calls and types.
11803 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11805 (int_tree_map_hash): Likewise.
11807 * tree-ssa-dom.c (htab_t avail_exprs):
11808 Change type to hash_table. Update dependent calls and types.
11810 * tree-ssa-live.c (var_map_base_init::tree_to_index):
11811 Change type to hash_table. Update dependent calls and types.
11813 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11814 Change type to hash_table. Update dependent calls and types.
11816 * tree-ssa-phiopt.c (seen_ssa_names):
11817 Change type to hash_table. Update dependent calls and types.
11819 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11820 Change type to hash_table. Update dependent calls and types.
11822 * tree-ssa-uncprop.c (equiv):
11823 Change type to hash_table. Update dependent calls and types.
11825 2013-04-25 Jakub Jelinek <jakub@redhat.com>
11827 PR rtl-optimization/57003
11828 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11829 call note_stores with kill_clobbered_value callback again after
11830 killing regs_invalidated_by_call.
11832 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11834 * config/aarch64/aarch64-simd.md
11835 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11836 (aarch64_simd_bsl<mode>): Likewise.
11837 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11839 2013-04-25 Marek Polacek <polacek@redhat.com>
11841 PR tree-optimization/57066
11842 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11844 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
11846 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11848 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11850 * config/aarch64/aarch64-builtins.c
11851 (aarch64_fold_builtin): New.
11852 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11853 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11854 * config/aarch64/aarch64-simd-builtins.def (abs): New.
11855 * config/aarch64/arm_neon.h
11856 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11858 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11859 Tejas Belagod <tejas.belagod@arm.com>
11861 * config/aarch64/aarch64-builtins.c
11862 (aarch64_gimple_fold_builtin): New.
11863 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11864 * config/aarch64/aarch64-simd-builtins.def (addv): New.
11865 * config/aarch64/aarch64-simd.md (addpv4sf): New.
11866 (addvv4sf): Update.
11867 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11869 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11871 * config/aarch64/aarch64.md
11872 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11874 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11876 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11877 (*ngcsi_uxtw): New pattern.
11879 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11880 Julian Brown <julian@codesourcery.com>
11882 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11883 (TB_DREG): Add T_V4HF.
11884 (v4hf_UP): New macro.
11885 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11886 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11887 Handle initialisation of V4HF. Adjust initialisation of reinterpret
11889 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11890 (arm_vector_mode_supported_p): Handle V4HF.
11891 (arm_mangle_map): Handle V4HFmode.
11892 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11893 * config/arm/arm_neon_builtins.def: Add entries for
11894 vcvtv4hfv4sf, vcvtv4sfv4hf.
11895 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11896 (neon_vcvtv4hfv4sf): Likewise.
11897 * config/arm/neon-gen.ml: Handle half-precision floating point
11899 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11900 * config/arm/arm_neon.h: Regenerate.
11901 * config/arm/neon.ml (type elts): Add F16.
11902 (type vectype): Add T_float16x4, T_floatHF.
11903 (type vecmode): Add V4HF.
11904 (type features): Add Requires_FP_bit feature.
11905 (elt_width): Handle F16.
11906 (elt_class): Likewise.
11907 (elt_of_class_width): Likewise.
11908 (mode_of_elt): Refactor.
11909 (type_for_elt): Handle F16, fix error messages.
11910 (vectype_size): Handle T_float16x4.
11911 (vcvt_sh): New function.
11912 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11913 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11914 (string_of_mode): Handle V4HF.
11915 * doc/arm-neon-intrinsics.texi: Regenerate.
11917 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11919 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11920 format specifier in 'X' case.
11922 2013-04-25 Alan Modra <amodra@gmail.com>
11925 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11926 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11927 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11928 Repeat for many other rotate/shift and mask patterns using subregs.
11929 Name lshiftrt insns.
11930 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11931 on WORDS_BIG_ENDIAN.
11933 2013-04-25 Alan Modra <amodra@gmail.com>
11935 * config.gcc: Support little-endian powerpc-linux targets.
11936 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11937 (LINK_OS_LINUX_SPEC): Define.
11938 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11939 Preserve MASK_LITTLE_ENDIAN.
11940 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11941 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11942 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11943 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11944 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11945 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11946 Correct fp word order for little-endian. Don't shift toc entries
11947 smaller than a word for little-endian.
11948 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11949 (bswapdi2 splits): Correct low-part subreg for little-endian.
11950 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11951 low/high where such is correct only for be.
11952 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11953 little-endian for -mcall-aixdesc.
11955 2013-04-25 Alan Modra <amodra@gmail.com>
11957 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11958 replace_equiv_address_nv.
11960 2013-04-25 Alan Modra <amodra@gmail.com>
11962 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11964 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11967 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11968 * rtl.h (struct rtx_def): ...
11970 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
11972 PR rtl-optimizations/57046
11973 * lra-constraints (split_reg): Set up lra_risky_transformations_p
11974 for multi-reg splits.
11976 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
11978 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11980 2013-04-24 Sterling Augustine <saugustine@google.com>
11982 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11983 (comp_dir_string, debug_str_dwo_section): New.
11984 (DEBUG_STR_DWO_SECTION): Rename to ...
11985 (DEBUG_DWO_STR_SECTION): ... this.
11986 (DEBUG_NORM_STR_SECTION): Delete.
11987 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11988 (DEBUG_STR_DWO_SECTION_FLAGS): New.
11989 (find_AT_string): Move most logic to ...
11990 (find_AT_string_in_table): ... here. New.
11991 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11992 add_skeleton_AT_string. Delete logic.
11993 (output_skeleton_debug_sections): Remove call to
11994 add_top_level_skeleton_die_attrs.
11995 (add_comp_dir_attribute): Move logic to comp_dir_string.
11996 (dwarf2out_init): Initialize debug_str_dwo_section.
11997 (output_indirect_string): Call find_string_form.
11998 (output_indirect_strings): Rewrite.
11999 (prune_unused_types): Empty skeleton_debug_str_hash.
12000 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
12001 (dwarf2out_finish): Call output_indirect_strings.
12003 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
12005 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
12007 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
12009 * rtl.h (struct rtx_def): Add comment for field jump.
12010 (LRA_SUBREG_P): New macro.
12011 * recog.c (register_operand): Check LRA_SUBREG_P.
12012 * lra.c (lra): Add note at the end of RTL code. Align non-empty
12014 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
12015 (lra_final_code_change): Skip subreg change for operators.
12016 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
12017 if there are no operand changes.
12018 * lra-constraints.c (curr_insn_set): New.
12019 (match_reload): Set LRA_SUBREG_P.
12020 (emit_spill_move): Ditto.
12021 (check_and_process_move): Use curr_insn_set. Process only single
12022 set insns. Don't initialize sec_mem_p and change_p.
12023 (simplify_operand_subreg): Use LRA_SUBREG_P.
12024 (reg_in_class_p): New function.
12025 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
12026 of #ifdef. Add code to remove cycling.
12027 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
12028 non-null disp. Reload inner instead of disp when base and index
12029 are null. Try to put lo_sum into register.
12030 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
12031 (check_and_process_move): Move code for move cost check to
12032 simple_move_p. Remove equiv_substitution.
12033 (simple_move_p): New function.
12034 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
12035 curr_insn_set. Call check_and_process_move only for single set
12036 insns. Use the new function. Move call of check_and_process_move
12037 after operand equiv substitution and address process.
12039 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
12041 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
12042 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
12043 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
12045 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
12047 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
12049 2013-04-24 Marek Polacek <polacek@redhat.com>
12051 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
12052 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
12053 (select_loops_exit_conditions): Likewise.
12054 (number_of_iterations_for_all_loops): Likewise.
12055 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
12056 (scev_analysis): Likewise.
12058 2013-04-02 Catherine Moore <clm@codesourcery.com>
12059 Chao-ying Fu <fu@mips.com>
12061 * config/mips/micromips.md (jraddiusp): New pattern.
12062 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
12063 instruction if possible.
12065 2013-04-24 Alan Modra <amodra@gmail.com>
12067 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
12069 2013-04-24 Julian Brown <julian@codesourcery.com>
12070 Chung-Lin Tang <cltang@codesourcery.com>
12072 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
12073 dependency behavior in enumeration type DIE generation. Add TODO note
12074 to comments about future DW_FORM_sdata/udata re-work of related code.
12076 2013-04-23 Lawrence Crowl <crowl@google.com>
12078 * Makefile.in: Update as needed below.
12080 * hash-table.h (class hash_table):
12081 Correct many methods with parameter types compare_type to the correct
12082 value_type. (Correct code was unlikely to notice the change.)
12083 (hash_table::elements_with_deleted) New.
12084 (class hashtable::iterator): New.
12085 (hashtable::begin()): New.
12086 (hashtable::end()): New.
12087 (FOR_EACH_HASH_TABLE_ELEMENT): New.
12089 * statistics.c (statistics_hashes):
12090 Change type to hash_table. Update dependent calls and types.
12092 * tree-into-ssa.c (var_infos):
12093 Change type to hash_table. Update dependent calls and types.
12095 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
12096 Change type to hash_table. Update dependent calls and types.
12098 * tree-ssa-loop-im.c (struct mem_ref.refs):
12099 Change type to hash_table. Update dependent calls and types.
12101 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
12102 Change type to hash_table. Update dependent calls and types.
12104 * tree-ssa-sccvn.c (vn_tables_s::nary):
12105 Change type to hash_table. Update dependent calls and types.
12106 (vn_tables_s::phis): Likewise.
12107 (vn_tables_s::references): Likewise.
12109 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
12110 (vn_reference_eq): Update parameter and return types.
12112 * tree-ssa-structalias.c (pointer_equiv_class_table):
12113 Change type to hash_table. Update dependent calls and types.
12114 (location_equiv_class_table): Likewise.
12116 * tree-vect-data-refs.c: Consequential changes for making
12117 peeling a hash_table.
12119 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
12120 (destroy_loop_vec_info): Dependent hash_table update.
12122 * tree-vectorizer.h (peeling_htab):
12123 Change type to hash_table. Update dependent calls and types.
12125 2013-04-23 Shiva Chen <shiva0217@gmail.com>
12127 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
12128 to check the register content is equal or not.
12129 * lra-constraints.c (match_reload): Use lra_assign_reg_val
12130 to assign register content record.
12131 * lra-eliminations.c (update_reg_eliminate): Use
12132 lra_update_reg_val_offset to update register content offset.
12133 * lra-int.h (struct lra_reg): Add offset member.
12134 (lra_reg_val_equal_p): New static inline function.
12135 (lra_update_reg_val_offset): New static inline function.
12136 (lra_assign_reg_val): New static inline function.
12137 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
12138 to assign register content record.
12139 (initialize_lra_reg_info_element): Initial offset to zero.
12141 2013-04-23 Catherine Moore <clm@codesourcery.com>
12143 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
12144 operands. Record compression.
12146 2013-04-23 Xinliang David Li <davidxl@google.com>
12148 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
12150 2013-04-23 Richard Biener <rguenther@suse.de>
12152 PR middle-end/57036
12153 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
12154 parameter, only add abnormal goto edges from the copied body
12155 if the call could perform abnormal gotos.
12156 (copy_cfg_body): Adjust.
12158 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
12160 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
12162 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
12164 * coretypes.h (gimple_stmt_iterator): Add struct to make
12167 2013-04-23 Richard Biener <rguenther@suse.de>
12169 PR tree-optimization/57026
12170 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
12171 from SSA names occuring in abnormal PHI nodes.
12173 2013-04-22 Andi Kleen <ak@linux.intel.com>
12175 * lto/lto.c (print_lto_report_1): Fix LTO report names.
12177 2013-04-22 Andi Kleen <ak@linux.intel.com>
12179 * lto/lto.c (print_lto_report_1): Declare early.
12180 (read_cgraph_and_symbols): Call print_lto_report_1 early.
12182 2013-04-22 Andi Kleen <ak@linux.intel.com>
12184 * common.opt (-flto-report-wpa): Add.
12185 * doc/invoke.texi (-flto-report-wpa): Add.
12186 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
12189 2013-04-22 Xinliang David Li <davidxl@google.com>
12191 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
12192 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
12193 * Makefile.in: New dependency
12195 David Daney <ddaney.cavm@gmail.com>
12197 * configure.ac (gcc_cv_as_micromips_support): Use the
12198 --fatal-warnings option.
12199 * configure: Regenerate.
12201 2013-04-22 Marek Polacek <polacek@redhat.com>
12204 * tsan.c (instrument_expr): Don't instrument expression
12205 in case its size is zero.
12207 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
12211 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
12213 * config/alpha/alpha.c (TARGET_LRA_P): New define.
12215 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12217 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
12218 (gimple_stmt_iterator): New typedef.
12219 * gimple.h (gimple_stmt_iterator): Rename to...
12220 (gimple_stmt_iterator_d): ... This.
12221 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
12222 trees be valid for GIMPLE and GENERIC.
12223 (TARGET_GIMPLE_FOLD_BUILTIN): New.
12224 * gimple-fold.c (gimple_fold_call): Call target hook
12225 gimple_fold_builtin.
12226 * hooks.c (hook_bool_gsiptr_false): New.
12227 * hooks.h (hook_bool_gsiptr_false): New.
12228 * target.def (fold_stmt): New.
12229 * doc/tm.texi: Regenerate.
12231 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
12234 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
12235 a set sp if no stack realignment.
12237 2013-04-22 Nick Clifton <nickc@redhat.com>
12239 * config.gcc (tilegx-linux): Extend extra_objs rather than
12241 (tilepro-linux): Likewise.
12243 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12245 * config/aarch64/aarch64-builtins.c
12247 (CF0, CF1, CF2, CF3, CF4, CF10): New.
12248 (VAR<1-12>): Add MAP parameter.
12249 (BUILTIN_*): Likewise.
12250 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
12251 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
12252 (aarch64_ushl_n<mode>): Likewise.
12253 (aarch64_sshr_n<mode>): Likewise.
12254 (aarch64_ushr_n<mode>): Likewise.
12255 (aarch64_<maxmin><mode>): Likewise.
12256 (aarch64_sqrt<mode>): Likewise.
12257 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
12258 (vshr<q>_n_*): Likewise.
12260 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12262 * config/aarch64/aarch64-builtins.c
12263 (aarch64_simd_builtin_type_mode): Handle SF types.
12265 (BUILTIN_GPF): Define.
12266 (aarch64_init_simd_builtins): Handle SF types.
12267 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
12268 (frecps): Likewise.
12269 (frecpx): Likewise.
12270 * config/aarch64/aarch64-simd.md
12271 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
12272 (aarch64_frecpe<mode>): New.
12273 (aarch64_frecps<mode>): Likewise.
12274 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
12275 (v8type): Add frecp<esx>.
12276 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
12277 (aarch64_frecps<mode>): Likewise.
12278 * config/aarch64/iterators.md (FRECP): New.
12279 (frecp_suffix): Likewise.
12280 * config/aarch64/arm_neon.h
12281 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
12283 2013-04-22 Christian Bruel <christian.bruel@st.com>
12286 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
12287 (REG_CLASS_NAMES): Idem.
12288 (REG_CLASS_CONTENTS): Idem.
12289 (REGCLASS_HAS_FP_REG): Idem.
12290 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
12291 (sh_conditional_register_usage): Idem.
12293 2013-04-21 Jeff Law <law@redhat.com>
12295 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
12296 (ssa_forward_propagate_and_combine): Use it.
12298 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
12300 * lra.c: Update the flow chart diagram.
12302 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
12304 PR rtl-optimization/56847
12305 * lra-constraints.c (process_alt_operands): Discourage alternative
12306 with non-matche doffsettable memory constraint fro memory with
12309 2013-04-19 Richard Biener <rguenther@suse.de>
12311 PR tree-optimization/56982
12312 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
12314 * gimplify.c (gimplify_call_expr): Notice special calls.
12315 (gimplify_modify_expr): Likewise.
12316 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
12317 abnormal control flow receivers.
12318 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
12319 in the same way as cfun->has_nonlocal_labels.
12320 (gimple_purge_dead_abnormal_call_edges): Likewise.
12321 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
12322 receivers start a basic-block.
12324 2013-04-19 Richard Biener <rguenther@suse.de>
12326 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
12328 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
12329 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
12330 (SLP_TREE_LOAD_PERMUTATION): Add.
12331 (vect_transform_slp_perm_load): Adjust prototype.
12332 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
12333 (vect_free_slp_instance): Likewise.
12334 (vect_create_new_slp_node): Likewise.
12335 (vect_supported_slp_permutation_p): Remove.
12336 (vect_slp_rearrange_stmts): Adjust.
12337 (vect_supported_load_permutation_p): Likewise. Inline
12338 vect_supported_slp_permutation_p here.
12339 (vect_analyze_slp_instance): Compute load permutations per
12340 slp node instead of per instance.
12341 (vect_get_slp_defs): Adjust.
12342 (vect_transform_slp_perm_load): Likewise.
12343 (vect_schedule_slp_instance): Remove redundant code.
12344 (vect_schedule_slp): Remove hack for PR56270, add it ...
12345 * tree-vect-stmts.c (vectorizable_load): ... here, do not
12346 CSE loads for SLP. Adjust.
12348 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
12350 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
12351 spelling in two comments.
12353 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
12356 * config/arm/arm.c (load_multiple_sequence): Require SP
12357 as base register for loads if SP is in the register list.
12359 2013-04-19 Martin Jambor <mjambor@suse.cz>
12361 PR tree-optimization/56718
12362 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
12363 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
12364 and made public. Adjusted all callers.
12365 (ipa_intraprocedural_devirtualization): New function.
12366 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
12367 (ipa_intraprocedural_devirtualization): Likewise.
12368 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
12370 2013-04-19 Richard Biener <rguenther@suse.de>
12372 PR tree-optimization/57000
12373 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
12375 2013-04-19 Terry Guo <terry.guo@arm.com>
12377 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
12379 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
12380 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
12381 (cortex_m4_fmacs): Use new reservations.
12382 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
12384 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
12386 PR rtl-optimization/56999
12387 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
12389 (lra_coalesce): Remove split_origin_bitmap and related code.
12390 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
12391 ranges if necessary.
12393 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
12395 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
12397 (ix86_expand_call): Remove clobbered_registers array and use
12398 x86_64_ms_sysv_extra_clobbered_registers instead.
12399 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
12401 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
12403 * config/i386/i386.md (*call_rex64_ms_sysv): Use
12404 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
12405 (*call_value_rex64_ms_sysv): Ditto.
12407 2013-04-18 Cary Coutant <ccoutant@google.com>
12409 * dwarf2out.c (output_pubnames): Check die_perennial_p of
12410 parent instead of die_mark.
12412 2013-04-18 Diego Novillo <dnovillo@google.com>
12414 * gimple.c (create_gimple_tmp): New.
12415 (get_expr_type): New.
12416 (build_assign): New.
12417 (build_type_cast): New.
12418 * gimple.h (enum ssa_mode): Define.
12419 (gimple_seq_set_location): New.
12420 * asan.c (build_check_stmt): Change some gimple_build_* calls
12421 to use build_assign and build_type_cast.
12423 2013-04-18 Richard Biener <rguenther@suse.de>
12425 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
12426 handle negative step. Remove redundant checks.
12427 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
12428 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
12429 for negative step and grouped loads fail to vectorize.
12431 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
12433 * emit-rtl.c (reset_insn_used_flags): New function.
12434 (reset_all_used_flags): Use it.
12435 (verify_insn_sharing): New function.
12436 (verify_rtl_sharing): Fix verification for SEQUENCEs.
12438 2013-04-18 Jakub Jelinek <jakub@redhat.com>
12440 PR tree-optimization/56984
12441 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
12442 and (x >> M) >= N don't register any assertion if N << M is the
12445 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
12447 * lower-subreg.c (resolve_simple_move): If called self-recursive,
12448 do not delete_insn insns that have not yet been emitted, only
12449 unlink them with remove_insn.
12450 * df-scan.c (df_insn_delete): Revert r197492.
12452 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
12454 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
12455 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
12457 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
12459 * config/arm/arm.md (movsicc_insn): Convert define_insn into
12460 define_insn_and_split.
12461 (and_scc,ior_scc,negscc): Likewise.
12462 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
12464 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
12466 * config/arm/arm.c (use_return_insn): Return 0 for targets that
12467 can benefit from using a sequence of LDRD instructions in epilogue
12468 instead of a single LDM instruction.
12470 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
12473 * doc/extend.texi: Fix typo.
12475 2013-04-17 Richard Biener <rguenther@suse.de>
12477 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
12478 (vect_build_slp_tree): ... here.
12479 (vect_build_slp_tree_1): Compute which stmts of the SLP group
12480 match. Remove special-casing of mismatched complex loads.
12481 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
12482 re-try the match with swapped commutative operands.
12483 (vect_supported_load_permutation_p): Remove special-casing of
12484 mismatched complex loads.
12485 (vect_analyze_slp_instance): Adjust.
12487 2013-04-17 Richard Biener <rguenther@suse.de>
12489 PR rtl-optimization/56921
12490 * cfgloop.h (struct loop): Add simple_loop_desc member.
12491 (struct niter_desc): Mark with GTY(()).
12492 (simple_loop_desc): Do not use aux field but simple_loop_desc.
12493 * loop-iv.c (get_simple_loop_desc): Likewise.
12494 (free_simple_loop_desc): Likewise.
12497 2013-04-16 Richard Biener <rguenther@suse.de>
12499 PR rtl-optimization/56921
12500 * loop-init.c (pass_rtl_move_loop_invariants): Add
12501 TODO_do_not_ggc_collect to todo_flags_finish.
12502 (pass_rtl_unswitch): Same.
12503 (pass_rtl_unroll_and_peel_loops): Same.
12504 (pass_rtl_doloop): Same.
12506 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
12508 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
12509 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
12510 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
12512 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
12513 * tree-streamer.c (record_common_node): Adjust reference in comment.
12515 2013-04-17 Terry Guo <terry.guo@arm.com>
12517 * config/arm/cortex-m4.md: Add a new bypass.
12519 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12521 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12523 (*subs_<optab><mode>_multp2): New pattern.
12524 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12525 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12527 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12529 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12530 (*subs_mul_imm_<mode>): New pattern.
12532 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
12535 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12536 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
12537 (vsx_movti_32bit): Change j->wa to O->wa.
12539 2013-04-16 Richard Biener <rguenther@suse.de>
12541 PR rtl-optimization/56921
12542 * loop-init.c (pass_rtl_move_loop_invariants): Add
12543 TODO_do_not_ggc_collect to todo_flags_finish.
12544 (pass_rtl_unswitch): Same.
12545 (pass_rtl_unroll_and_peel_loops): Same.
12546 (pass_rtl_doloop): Same.
12548 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
12550 * config/arm/arm.c (emit_multi_reg_push): New declaration
12551 for an existing function.
12552 (arm_emit_strd_push): New function.
12553 (arm_expand_prologue): Used here.
12554 (arm_emit_ldrd_pop): New function.
12555 (arm_expand_epilogue): Used here.
12556 (arm_get_frame_offsets): Update condition.
12557 (arm_emit_multi_reg_pop): Add a special case for load of a single
12558 register with writeback.
12560 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
12562 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12565 2013-04-16 Richard Biener <rguenther@suse.de>
12567 PR tree-optimization/56756
12568 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12569 (first_mem_ref_loc): New.
12570 (execute_sm): Place the load temporarily before a previous
12571 access instead of in the latch edge to ensure its SSA dependencies
12572 are defined at points dominating the load.
12574 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
12576 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12577 correct fix by moving header and footer insn to the footer of
12578 the merged basic block. Clear BB_END of the merged-away block.
12580 PR middle-end/43631
12581 * emit-rtl.c (make_note_raw): New function.
12582 (link_insn_into_chain): New static inline function.
12583 (add_insn): Use it.
12584 (add_insn_before, add_insn_after): Factor insn chain linking code...
12585 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12586 using link_insn_into_chain.
12587 (note_outside_basic_block_p): New helper function for emit_note_after
12588 and emit_note_before.
12589 (emit_note_after): Use nobb variant of add_insn_after if the note
12590 should not be contained in a basic block.
12591 (emit_note_before): Use nobb variant of add_insn_before if the note
12592 should not be contained in a basic block.
12593 (emit_note_copy): Use make_note_raw.
12594 (emit_note): Likewise.
12595 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12596 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12597 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12598 the moved barrier the tail of the basic block it follows.
12599 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12601 2013-04-15 Jakub Jelinek <jakub@redhat.com>
12603 PR tree-optimization/56962
12604 * gimple-ssa-strength-reduction.c (record_increment): Only set
12605 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12606 either rhs1 or rhs2 is equal to c->base_expr.
12608 2013-04-15 Richard Biener <rguenther@suse.de>
12610 PR tree-optimization/56933
12611 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12613 (GROUP_READ_WRITE_DEPENDENCE): Remove.
12614 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12615 * tree-vect-data-refs.c (vect_analyze_group_access): Move
12616 dependence check ...
12617 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12619 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12620 GROUP_READ_WRITE_DEPENDENCE.
12622 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12624 * emit-rtl.c (reset_all_used_flags): New function.
12625 (verify_rtl_sharing): Call reset_all_used_flags before and after
12626 performing the checks.
12628 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12630 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12631 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12632 * config/arm/constraints.md (De): New constraint.
12633 * config/arm/neon.md (anddi3_neon): Delete.
12634 (neon_vand<mode>): Expand to standard anddi3 pattern.
12635 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12636 Move earlier in the file.
12637 (neon_inv_logic_op2): Likewise.
12638 (arm_anddi_operand_neon): New predicate.
12640 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12642 * configure.ac (gcc_cv_ld_as_needed): Set
12643 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12644 Use -z ignore, -z record on *-*-solaris2*.
12645 (HAVE_LD_AS_NEEDED): Update comment.
12646 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12647 * configure: Regenerate.
12648 * config.in: Regenerate.
12649 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12650 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12651 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12652 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12653 equivalents. Fix markup.
12654 * doc/tm.texi: Regenerate.
12656 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
12658 * config/i386/i386.opt: New option mstack-protector-guard=.
12659 * config/i386/i386-opts.h: Add enum stack_protector_guard.
12660 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12661 TARGET_SSP_TLS_GUARD.
12662 * config/i386/i386.c (ix86_option_override_internal): Set
12663 ix86_stack_protector_guard.
12664 * config/i386/i386.md (stack_protect_set): Enable for
12665 TARGET_SSP_TLS_GUARD only.
12666 (stack_protect_set_<mode>): Ditto.
12667 (stack_protect_test): Ditto.
12668 (stack_protect_test_<mode>): Ditto.
12669 * doc/invoke.texi (i386 Option): Document.
12671 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
12674 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12675 (S_MODES): Set H_MODE bit.
12676 (SF_MODES): Set only S_MODE and SF_MODE bits.
12677 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12678 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12679 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12680 <MODE_FLOAT>: Likewise.
12682 2013-04-15 Joey Ye <joey.ye@arm.com>
12684 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12686 2013-04-15 Joey Ye <joey.ye@arm.com>
12688 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12690 (thumb_far_jump_used_p): Count instruction size and set
12693 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
12695 * reorg.c (fill_simple_delay_slots): Reindent block of code.
12696 * resource.c (mark_target_live_regs): Reformat conditional block.
12698 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
12700 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12701 notes, they are emitted only just before final.
12702 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12704 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
12706 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12707 * cfgrtl.c (delete_insn): Call it here instead.
12708 * lra-spills.c (lra_final_code_change): Use delete_insn.
12709 * haifa-sched.c (sched_remove_insn): Likewise.
12710 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12711 returning to the nop pool.
12712 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12713 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
12715 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
12717 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12718 * doc/tm.texi: Regenerated.
12720 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
12722 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12725 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
12727 * df-core.c (df_find_def): Compare register numbers.
12728 (df_find_use): Likewise.
12730 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
12733 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12734 lra_in_progress for return.
12736 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
12738 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12739 define_insn into define_insn_and_split and emit movsicc patterns.
12741 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
12743 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12745 2013-04-12 Richard Biener <rguenther@suse.de>
12747 * tree-pass.h (TODO_do_not_ggc_collect): New.
12748 * passes.c (execute_one_ipa_transform_pass): Honor
12749 TODO_do_not_ggc_collect.
12750 (execute_one_pass): Likewise.
12753 2013-04-10 Richard Biener <rguenther@suse.de>
12755 * passes.c (init_optimization_passes): Remove reload pass.
12756 * ira.c (do_reload): Merge into ...
12758 (rest_of_handle_reload): Remove.
12759 (pass_reload): Likewise.
12760 * config/i386/i386.c (ix86_option_override): Refer to ira instead
12761 of reload for vzeroupper pass placement.
12763 2013-04-12 Jakub Jelinek <jakub@redhat.com>
12765 PR tree-optimization/56918
12766 PR tree-optimization/56920
12767 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12768 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
12769 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
12770 use wide_mul_with_sign method.
12772 2013-04-12 Richard Biener <rguenther@suse.de>
12774 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12775 not be considered a gimple constant.
12777 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
12779 * fold-const.c (const_binop): Handle vector shifts by a scalar.
12780 (fold_binary_loc): Call const_binop also for mixed vector-scalar
12783 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
12784 Jakub Jelinek <jakub@redhat.com>
12786 * opts.c: Include diagnostic-color.h.
12787 (common_handle_option): Handle OPT_fdiagnostics_color_.
12788 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12789 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12790 (diagnostic-color.o): New.
12791 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12792 (diagnostic_color_rule): New enum.
12793 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12794 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12795 the location string.
12796 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12797 either NULL, or color kind.
12798 * diagnostic-color.c: New file.
12799 * diagnostic-color.h: New file.
12800 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12802 * doc/invoke.texi (-fdiagnostics-color): Document.
12803 * pretty-print.h (pp_show_color): Define.
12804 (struct pretty_print_info): Add show_color field.
12805 * diagnostic.c: Include diagnostic-color.h.
12806 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12807 macros. Colorize error:, warning: etc. strings and also the location
12809 (diagnostic_show_locus): Colorize the caret line.
12810 * pretty-print.c: Include diagnostic-color.h.
12811 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
12812 inside of %< %> quotes or quoted through q format modifier.
12814 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12816 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12818 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12820 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12821 code in CC_NZ mode.
12822 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12825 2013-04-11 Marek Polacek <polacek@redhat.com>
12827 PR tree-optimization/48184
12828 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12830 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
12832 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12833 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12834 (skip_simple_arithmetic): Tidy up.
12835 * tree.h (skip_simple_constant_arithmetic): Declare.
12837 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12839 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12841 2013-04-11 Richard Biener <rguenther@suse.de>
12843 * tree-vect-loop.c (get_initial_def_for_induction): Properly
12844 generate vector constants.
12846 2013-04-11 Richard Biener <rguenther@suse.de>
12848 PR tree-optimization/56878
12849 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12850 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12852 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12853 Prefer to align the DR with the most invariant base address.
12855 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
12857 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12860 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
12862 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12863 floating-point vector comparisons against 0.
12865 2013-04-11 Jakub Jelinek <jakub@redhat.com>
12867 PR tree-optimization/56899
12868 * fold-const.c (extract_muldiv_1): Apply distributive law
12869 only if TYPE_OVERFLOW_WRAPS (ctype).
12871 2013-04-11 Bin Cheng <bin.cheng@arm.com>
12874 * ira-costs.c (scan_one_insn): Check whether the source rtx of
12875 loading has side effect.
12877 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
12879 * config/sparc/sparc.c: Include tree-pass.h.
12880 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12881 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
12882 head of file. Change return type. Split off gate function.
12883 (sparc_gate_work_around_errata): New function.
12884 (pass_work_around_errata): New pass definition.
12885 (insert_pass_work_around_errata) New pass insert definition to
12886 insert pass_work_around_errata just after delayed-branch scheduling.
12887 (sparc_option_override): Insert the pass.
12888 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12890 2013-04-10 David S. Miller <davem@davemloft.net>
12892 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12893 or -mcpu=hypersparc.
12895 * target.def (cstore_mode): New hook.
12896 * target.h: Include insn-codes.h
12897 * targhooks.c: Likewise.
12898 (default_cstore_mode): New function.
12899 * targhooks.h: Declare it.
12900 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12901 * doc/tm.texi: Rebuild.
12902 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12903 target hook, rather than inspecting the insn_data.
12904 * config/sparc/sparc.c (sparc_cstore_mode): New function.
12905 (TARGET_CSTORE_MODE): Redefine.
12906 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12908 * config/sparc/predicates.md (cstore_result_operand): New special
12910 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12911 Use it for operand 0.
12912 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12913 (*snesi_special): Likewise.
12914 (*snesi_zero): Likewise.
12915 (*seqsi_zero): Likewise.
12916 (*sltu_insn): Likewise.
12917 (*sgeu_insn): Likewise.
12918 (*seqdi_special): Make operand 0 and comparison operation be of
12920 (*snedi_special): Likewise.
12921 (*snedi_special_vis3): Likewise.
12922 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12923 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12924 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12925 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12926 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12927 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12928 (*sltu_extend_sp64): Likewise.
12929 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12930 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12931 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12932 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12933 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12935 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
12937 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12938 (aarch64_start_file): Use the new function.
12940 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
12941 Jason Merrill <jason@redhat.com>
12943 * common.opt: Add -gdwarf.
12944 * opts.c (common_handle_option): Handle it.
12945 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12947 2013-04-10 Richard Biener <rguenther@suse.de>
12949 * passes.c (execute_todo): Do not call ggc_collect conditional here.
12950 (execute_one_ipa_transform_pass): But unconditionally here.
12951 (execute_one_pass): And here.
12952 (init_optimization_passes): Remove reload pass.
12953 * tree-pass.h (TODO_ggc_collect): Remove.
12954 (pass_reload): Likewise.
12955 * ira.c (do_reload): Merge into ...
12957 (rest_of_handle_reload): Remove.
12958 (pass_reload): Likewise.
12959 * config/i386/i386.c (ix86_option_override): Refer to ira instead
12960 of reload for vzeroupper pass placement.
12961 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12962 and todo_flags_finish of all passes.
12964 2013-04-10 Richard Biener <rguenther@suse.de>
12966 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12967 first_const_oprnd field, rename first_def_type to first_op_type.
12968 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12969 (vect_get_and_check_slp_defs): Always use the type of the
12970 operand. Allow mixed vect_external_def, vect_constant_def types.
12971 (vect_get_constant_vectors): Handle mixed vect_external_def,
12972 vect_constant_def types.
12974 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
12976 PR tree-optimization/55524
12977 * tree-ssa-math-opts.c
12978 (convert_mult_to_fma): Don't use an fms construct
12979 when we don't have an fms operation, but fnma, and it looks
12980 likely that we'll be able to use the latter.
12982 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
12984 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12986 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12987 inline fail caused by overwritable functions.
12989 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
12991 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12992 unnecessary bits in the constant power of two case.
12994 2013-04-10 Richard Biener <rguenther@suse.de>
12996 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12997 broken code swapping operands.
12998 (vect_build_slp_tree): Do not compute load permutations here.
12999 (vect_analyze_slp_instance): Compute load permutations here,
13000 after building the SLP tree.
13002 2013-04-09 Christian Bruel <christian.bruel@st.com>
13004 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
13005 of next/prev_real_insn.
13007 2013-04-09 Jan Hubicka <jh@suse.cz>
13009 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
13010 Drop aliased parameter.
13011 (function_and_variable_visibility): Do not handle alias pairs.
13012 * cgraph.c (varpool_externally_visible_p): Update prototype.
13013 * varpool.c (varpool_add_new_variable): Update.
13015 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13017 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
13019 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
13021 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
13023 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
13025 2013-04-09 Marek Polacek <polacek@redhat.com>
13027 PR tree-optimization/48762
13028 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
13030 2013-04-09 Richard Biener <rguenther@suse.de>
13032 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
13034 (vect_build_slp_tree): Likewise.
13035 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
13036 calculating the cost of a SLP instance.
13037 (vect_analyze_slp_instance): Use it from here, after building
13040 2013-04-09 Jakub Jelinek <jakub@redhat.com>
13042 PR middle-end/56883
13043 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
13044 expand_omp_for_static_chunk): Use simple_p = true in
13045 force_gimple_operand_gsi calls when assigning to addressable decls.
13047 2013-04-09 Jeff Law <law@redhat.com>
13049 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
13050 when the boolean was created by converting a wider object which
13051 had a boolean range.
13053 2013-04-09 Richard Biener <rguenther@suse.de>
13055 * tree-vectorizer.h (slp_void_p): Remove.
13056 (slp_tree): Typedef before _slp_tree declaration.
13057 (struct _slp_tree): Use a vector of slp_tree as children.
13058 (vect_get_place_in_interleaving_chain): Remove.
13059 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
13061 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
13063 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
13064 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
13065 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
13066 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
13067 Use slp_node instead of slp_void_p and adjust.
13069 2013-04-09 Richard Biener <rguenther@suse.de>
13071 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
13072 work that is not necessary.
13074 2013-04-09 Jakub Jelinek <jakub@redhat.com>
13076 PR tree-optimization/56854
13077 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
13078 forward into clobber stmts if it would change MEM_REF lhs into
13081 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
13083 * tree.c (type_hash_lookup, type_hash_add): Make static.
13084 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
13086 2013-04-09 Richard Biener <rguenther@suse.de>
13088 * tree.h (unsave_expr_now): Remove.
13089 * tree-inline.c (mark_local_for_remap_r): Remove.
13090 (unsave_expr_1): Likewise.
13091 (unsave_r): Likewise.
13092 (unsave_expr_now): Likewise.
13093 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
13094 (propagate_tree_value): Likewise.
13096 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
13098 * doc/rtl.texi (sequence): Rewrite documentation to match the
13099 current use of SEQUENCE rtl objects.
13100 * rtl.def (SEQUENCE): Likewise.
13102 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
13103 Update documentation.
13104 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
13105 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
13107 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
13109 2013-04-08 Teresa Johnson <tejohnson@google.com>
13111 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
13112 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
13114 (estimate_edge_size_and_time): Add comment to suggest using rounding
13116 (estimate_node_size_and_time): Ditto.
13117 (remap_edge_change_prob): Use helper rounding divide methods.
13118 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
13119 (gimple_mod_pow2_value_transform): Ditto.
13120 (gimple_mod_subtract_transform): Ditto.
13121 (gimple_ic_transform): Ditto.
13122 (gimple_stringops_transform): Ditto.
13123 * stmt.c (conditional_probability): Ditto.
13124 (emit_case_dispatch_table): Ditto.
13125 * lto-cgraph.c (merge_profile_summaries): Ditto.
13126 * tree-optimize.c (execute_fixup_cfg): Ditto.
13127 * cfgcleanup.c (try_forward_edges): Ditto.
13128 * cfgloopmanip.c (scale_loop_profile): Ditto.
13130 (duplicate_loop_to_header_edge): Ditto.
13131 (lv_adjust_loop_entry_edge): Ditto.
13132 * tree-vect-loop.c (vect_transform_loop): Ditto.
13133 * profile.c (compute_branch_probabilities): Ditto.
13134 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
13135 * lto-streamer-in.c (input_cfg): Ditto.
13136 * gimple-streamer-in.c (input_bb): Ditto.
13137 * ipa-cp.c (update_profiling_info): Ditto.
13138 (update_specialized_profile): Ditto.
13139 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
13140 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
13142 * sched-rgn.c (compute_dom_prob_ps): Ditto.
13143 (compute_trg_info): Ditto.
13144 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
13145 (purge_dead_edges): Ditto.
13146 * loop-unswitch.c (unswitch_loop): Ditto.
13147 * cgraphclones.c (cgraph_clone_edge): Ditto.
13148 (cgraph_clone_node): Ditto.
13149 * tree-inline.c (copy_bb): Ditto.
13150 (copy_edges_for_bb): Ditto.
13151 (initialize_cfun): Ditto.
13152 (copy_cfg_body): Ditto.
13153 (expand_call_inline): Ditto.
13155 2013-04-08 Kai Tietz <ktietz@redhat.com>
13157 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
13158 TARGET_CYGWIN64 by TARGET_64BIT.
13160 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
13162 * config/epiphany/epiphany.md (GPR_1): New constant.
13163 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
13164 * config/epiphany/epiphany.c (gen_compare_reg):
13165 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
13166 is already in place.
13167 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
13168 Don't require being called during rtl expansion; If y operlaps r0,
13170 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
13171 (epiphany_expand_epilogue): Likewise.
13173 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
13174 Don't use CC_FPmode for ORDERED / UNORDERED.
13175 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
13177 * config/epiphany/constraints.md (CnL): New constraint.
13178 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
13179 * config/epiphany/predicates.md (add_operand): Allow 1024.
13181 * config/epiphany/epiphany.md (logical_op): New code iterator.
13182 (op_mnc): New code attribute.
13183 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
13184 (mov_f+1, mov_f+2): New peephole2 patterns.
13186 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
13187 (cstoresi4): Also allow re-use of zero result when doing a NE
13188 comparison to a non-zero operand.
13189 Use (clobber (scratch)) for first insn if the gpr output is not needed.
13191 * config/epiphany/epiphany.md (<insn_opname>v2si3):
13192 Use gen_addsi3_i / gen_subsi3_i.
13194 2013-04-08 Jakub Jelinek <jakub@redhat.com>
13198 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
13199 contain anything but clobbers, at most one __builtin_stack_restore,
13200 optionally debug stmts and final resx, and if it has at least one
13201 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
13202 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
13203 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
13204 which isn't defaut definition, remove them.
13205 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
13206 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
13207 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
13208 with MEM_REF LHS with SSA_NAME address.
13210 2013-04-08 Jeff Law <law@redhat.com>
13212 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
13214 2013-04-08 Richard Biener <rguenther@suse.de>
13216 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
13218 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
13219 determined vector type.
13220 (vect_analyze_data_refs): Likewise.
13221 (vect_get_new_vect_var): Adjust.
13222 (vect_create_destination_var): Preserve SSA name versions.
13223 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
13224 not dump anything here.
13226 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
13228 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
13229 Add member lr_slot_known.
13230 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
13232 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
13233 Remove code that sets lr_slot_offset according to what a previous
13234 version of epiphany_emit_save_restore used to do.
13235 (epiphany_emit_save_restore): When doing an lr save or restore,
13236 set/verify lr_slot_known and lr_slot_offset.
13238 2013-04-08 Xinyu Qi <xyqi@marvell.com>
13241 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
13244 2013-04-08 Richard Biener <rguenther@suse.de>
13246 * alias.c (find_base_term): Fix thinko in previous change.
13248 2013-04-08 Jakub Jelinek <jakub@redhat.com>
13250 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
13251 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
13252 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
13253 if possible to compute val.
13254 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
13255 For QImode integers don't require anything about precision. Use
13256 const_with_all_bytes_same to find out if the constant doesn't have
13257 repeated bytes in it.
13259 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
13261 * config/s390/s390.c (s390_expand_insv): Only accept insertions
13264 2013-04-08 Marek Polacek <polacek@redhat.com>
13266 PR rtl-optimization/48182
13267 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
13270 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
13273 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
13274 nuses, make sure we have a label.
13276 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13279 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
13280 (rs6000_emit_swdiv_low_precision): Remove.
13281 (rs6000_emit_swdiv): Rewrite to handle between one and four
13282 iterations of Newton-Raphson generally; modify required number of
13283 iterations for some cases.
13284 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
13286 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
13288 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
13289 set-but-unused variable.
13291 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
13292 basic blocks of released function bodies garbage-collectable.
13294 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
13295 (struct rtl_opt_pass): Add TODO_df_finish.
13297 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
13299 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13301 * config/arm/constraints.md (q): New constraint.
13302 * config/arm/ldrdstrd.md: New file.
13303 * config/arm/arm.md (ldrdstrd.md) New include.
13304 (arm_movdi): Use "q" instead of "r" constraint
13305 for double-word memory access.
13306 (movdf_soft_insn): Likewise.
13307 * config/arm/vfp.md (movdi_vfp): Likewise.
13308 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
13309 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
13310 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
13311 (mem_ok_for_ldrd_strd): Likewise.
13312 (output_move_double): Update assertion.
13314 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13316 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
13318 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13320 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
13321 define_insn_and_split.
13322 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
13324 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13326 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
13327 define_insn_and_split.
13328 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
13329 (shiftsi3_compare): New pattern.
13330 (rrx): New pattern.
13331 * config/arm/unspecs.md (UNSPEC_RRX): New.
13333 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13335 * config/arm/arm.md (negdi_extendsidi): New pattern.
13336 (negdi_zero_extendsidi): Likewise.
13338 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13340 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
13341 define_insn_and_split.
13342 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
13343 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
13345 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13347 * config/arm/arm.md (arm_subdi3): Convert define_insn into
13348 define_insn_and_split.
13349 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
13350 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
13352 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13354 * config/arm/arm.md (subsi3_carryin): New pattern.
13355 (subsi3_carryin_const): Likewise.
13356 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
13357 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
13359 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13361 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
13363 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13365 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
13366 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
13368 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13370 * config/arm/arm.c (arm_expand_builtin): Change fcode
13371 type to unsigned int.
13373 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13375 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
13377 2013-04-04 Ian Lance Taylor <iant@google.com>
13379 * doc/standards.texi (Standards): The Go frontend supports the Go 1
13382 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
13384 PR middle-end/56729
13385 * df-scan.c (df_insn_delete): Disable failing assert.
13387 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13389 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
13390 New function prototype.
13391 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
13392 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
13393 (arm_builtin_vectorized_function): New function.
13395 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13397 * config/arm/arm_neon_builtins.def: New file.
13398 * config/arm/arm.c (neon_builtin_data): Move contents to
13399 arm_neon_builtins.def.
13400 (enum arm_builtins): Include neon builtin definitions.
13401 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
13402 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
13404 2013-04-04 Marek Polacek <polacek@redhat.com>
13406 PR tree-optimization/48186
13407 * predict.c (maybe_hot_frequency_p): Return false if
13408 HOT_BB_FREQUENCY_FRACTION is 0.
13409 (cgraph_maybe_hot_edge_p): Likewise.
13411 2013-04-04 Richard Biener <rguenther@suse.de>
13413 PR tree-optimization/56826
13414 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
13417 2013-04-04 Richard Biener <rguenther@suse.de>
13419 PR tree-optimization/56213
13420 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
13421 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
13423 2013-04-04 Richard Biener <rguenther@suse.de>
13425 PR tree-optimization/56837
13426 * tree-loop-distribution.c (classify_partition): For non-zero
13427 values require that the value has the same precision as its
13428 mode to be useful as memset value.
13430 2013-04-03 Nick Clifton <nickc@redhat.com>
13432 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
13433 (fmssf4): Use fmsf.s on E3V5 architectures.
13434 (fnmasf4): Use fnmaf.s on E3V5 architectures.
13435 (fnmssf4): Use fnmsf.s on E3V5 architectures.
13437 2013-04-03 Jeff Law <law@redhat.com>
13439 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
13440 (lra-eliminations.o): Likewise.
13442 2013-04-03 Teresa Johnson <tejohnson@google.com>
13444 * gcov-io.c (compute_working_sets): Moved most of body of old
13445 compute_working_sets here from profile.c.
13446 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
13447 (gcov_working_set_t): Moved typedef here from basic-block.h
13448 (compute_working_set): Declare.
13449 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
13450 (get_working_sets): Renamed from compute_working_set,
13451 replace most of body with call to new compute_working_sets.
13452 (get_exec_counts): Replace call to compute_working_sets
13453 to get_working_sets.
13454 * profile.h (get_working_sets): Renamed from compute_working_set.
13455 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
13456 to get_working_sets.
13457 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
13458 * gcov-dump.c (dump_working_sets): New function.
13460 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
13462 * hwint.c (sext_hwi, zext_hwi): New functions.
13463 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
13464 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
13465 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
13466 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
13467 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
13468 (sext_hwi, zext_hwi): New functions.
13470 2013-04-03 Jeff Law <law@redhat.com>
13472 PR tree-optimization/56799
13473 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
13474 back test for widening conversion erroneously dropped in prior change.
13476 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13479 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
13480 instead of next_real_insn.
13482 2013-04-03 Marek Polacek <polacek@redhat.com>
13485 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13487 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13490 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
13492 (thumb1_output_casesi): Likewise.
13493 (thumb2_output_casesi): Likewise.
13495 2013-04-03 Richard Biener <rguenther@suse.de>
13497 PR tree-optimization/56817
13498 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
13500 (tree_unroll_loops_completely_1): ... new function to manually
13501 walk the loop tree, properly defering outer loops of unrolled
13502 loops to later iterations.
13504 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13506 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
13507 (vectorizable_load): Likewise.
13508 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
13509 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
13511 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13513 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
13516 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
13518 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
13520 2013-04-03 Bin Cheng <bin.cheng@arm.com>
13522 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13524 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13526 PR tree-optimization/56790
13527 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13530 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13532 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13534 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13535 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13538 2013-04-03 Jakub Jelinek <jakub@redhat.com>
13541 * tree.h (force_folding_builtin_constant_p): New decl.
13542 * builtins.c (force_folding_builtin_constant_p): New variable.
13543 (fold_builtin_constant_p): Fold immediately also if
13544 force_folding_builtin_constant_p.
13546 2013-04-03 Richard Biener <rguenther@suse.de>
13548 PR tree-optimization/56812
13549 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13550 DRs of the same interleaving chain are independent.
13552 2013-04-02 Jason Merrill <jason@redhat.com>
13554 * gdbinit.in (pbb): Use debug fn.
13556 2013-04-02 Lawrence Crowl <crowl@google.com>
13558 * sese.h (struct ivtype_map_elt_s): Remove unused.
13559 (extern debug_ivtype_map): Remove unused.
13560 (extern eq_ivtype_map_elts): Remove unused.
13561 * sese.c (debug_ivtype_map): Removed unused.
13562 (debug_ivtype_map_1): Removed unused.
13563 (debug_ivtype_elt): Remove unused.
13564 (eq_ivtype_map_elts): Remove unused.
13567 2013-04-02 Kai Tietz <ktietz@redhat.com>
13570 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13571 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13572 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13574 (legitimize_pe_coff_symbol): Likewise.
13575 (is_imported_p): New helper-function.
13576 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13577 for Windows x64 targets.
13578 (ix86_expand_prologue): Optimize for pe-coff targets.
13579 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13580 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13581 medium/large code-model.
13582 (legitimize_pic_address): Likewise.
13583 (legitimize_tls_address): Likewise.
13584 (ix86_expand_call): Likewise.
13585 (x86_output_mi_thunk): Likewise.
13586 (get_dllimport_decl): Add new beimport argument.
13587 (construct_plt_address): Don't assert for x64 pe-coff targets.
13588 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13590 (SYMBOL_FLAG_STUBVAR): New macro.
13591 (SYMBOL_REF_STUBVAR_P): Likewise.
13592 * config/i386/winnt.c (stub_list): New structure.
13593 (stub_head): New local variable.
13594 (i386_pe_record_stub): New function.
13595 (i386_pe_file_end): Emit refptr-stubs.
13597 2013-04-02 Jakub Jelinek <jakub@redhat.com>
13599 PR rtl-optimization/56745
13600 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13601 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13604 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13605 and both of them are MEM_REFs, just compare first argument for
13606 equality and attempt to deal even with differing offsets.
13609 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13610 of gimple_clobber_p to be MEM_REF.
13611 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13612 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
13613 after gimplification.
13614 * asan.c (get_mem_ref_of_assignment): Don't instrument
13615 gimple_clobber_p stmts.
13616 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13617 gimple_clobber_p stmt if they have MEM_REF lhs and
13618 are dead because of another gimple_clobber_p stmt.
13619 * tree-ssa-live.c (clear_unused_block_pointer): Treat
13620 gimple_clobber_p stmts like debug stmts.
13621 (remove_unused_locals): Remove clobbers with MEM_REF lhs
13622 that refer to unused VAR_DECLs or uninitialized values.
13623 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13624 gimple_clobber_p stmts if they refer to removed parameters.
13625 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13628 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
13630 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13631 using SWI48 mode attribute.
13633 2013-04-02 Wei Mi <wmi@google.com>
13635 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13636 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13637 *<rotate_insn><mode>3_mask in i386.md.
13639 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
13641 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13643 2013-04-02 Richard Biener <rguenther@suse.de>
13645 PR tree-optimization/56778
13646 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13647 Runtime alias tests are not supported for gather loads.
13648 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13649 stmts referenced from SSA operands before updating SSA form.
13651 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
13652 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13654 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13655 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13656 * config/arm/cortex-a53.md: New file.
13657 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13658 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13659 * config/arm/arm.c (arm_issue_rate): Likewise.
13660 * config/arm/arm-tune.md: Regenerate
13661 * config/arm/arm-tables.opt: Regenerate.
13662 * config/arm/arm-cores.def: Add cortex-a53.
13664 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
13666 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13669 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
13671 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13672 scalar load/store operations using B/H registers.
13673 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13675 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
13677 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13679 * config/aarch64/aarch64.c
13680 (aarch64_simd_scalar_immediate_valid_for_move): New.
13681 * config/aarch64/aarch64-protos.h
13682 (aarch64_simd_scalar_immediate_valid_for_move): New.
13683 * config/aarch64/constraints.md (Dh, Dq): New.
13684 * config/aarch64/iterators.md (hq): New.
13686 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
13688 * reorg.c (get_branch_condition): Deal with conditional returns.
13689 (fill_simple_delay_slots): Remove dead code dealing with jumps.
13691 2013-04-01 Wei Mi <wmi@google.com>
13693 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13694 Truncate operand 2 using %b asm operand modifier.
13695 (*<shift_insn><mode>3_mask): Ditto.
13696 (*<rotate_insn><mode>3_mask): Ditto.
13698 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
13700 PR middle-end/56798
13701 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13703 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
13705 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13707 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13709 2013-03-30 Lawrence Crowl <crowl@google.com>
13711 * dse.c (clear_alias_sets): Remove never set.
13712 (disqualified_clear_alias_sets): Remove never set.
13713 (clear_alias_mode_pool): Remove never set.
13714 (dse_step0): Remove condition that is never true.
13715 (canon_address): Remove condition that is never true.
13716 (dse_step7): Remove condition that is never true.
13717 (rest_of_handle_dse): Remove condition that is never true.
13718 (rest_of_handle_dse::did_global): Remove never read from above.
13719 (dse_step2_spill): Remove never called from above.
13720 (dse_step5_spill): Remove never called from above.
13722 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
13724 * doc/md.texi (Standard Names) <casesi>: Update documentation for
13725 JUMP_TABLE_DATA changes.
13726 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13727 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13728 (Insns) <jump_table_data>: New entry.
13729 * doc/tm.texi: Regenerate.
13731 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13733 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13734 for table jump at the end of a basic block using tablejump_p.
13735 * targhooks.c (default_invalid_within_doloop): Likewise.
13736 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13737 target hook implementation that is identical to the default hook.
13738 (rs6000_invalid_within_doloop): Remove.
13740 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13741 unused variable from tablejump_p call.
13743 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13744 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13745 (INSN_DELETED_P): Likewise.
13746 (emit_jump_table_data): New prototype.
13747 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13748 after 4th as unused.
13749 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13750 * sched-vis.c (print_insn): Likewise.
13751 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13752 insn for compatibility with back ends that use next_active_insn to
13753 identify jump table data.
13754 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13755 (remove_insn): Likewise.
13756 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13758 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13759 (emit_jump_table_data): New function.
13761 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13762 basic block, a JUMP_TABLE_DATA never is.
13763 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13764 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13765 off from code handling real insns.
13766 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13767 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13768 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13769 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13770 is not a NONDEBUG_INSN_P.
13771 * ira-costs.c (scan_one_insn): Likewise.
13772 * jump.c (mark_all_labels): Likewise.
13773 (mark_jump_label_1): Likewise.
13774 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13775 * lra.c (get_insn_freq): Expect all insns reaching here to be in
13777 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13778 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13779 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13780 JUMP_TABLE_DATA_P insns.
13781 (calculate_elim_costs_all_insns): Likewise.
13782 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13783 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13784 (delete_output_reload): Code style fixups.
13785 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13786 insn flags on this non-insn.
13787 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13788 as scheduling barriers, for pre-change compatibility.
13789 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13790 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13792 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13793 redundant JUMP_TABLE_DATA_P test.
13794 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13795 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13796 (frv_for_each_packet): Likewise.
13797 * config/i386/i386.c (min_insn_size): Likewise.
13798 (ix86_avoid_jump_mispredicts): Likewise.
13799 * config/m32r/m32r.c (m32r_is_insn): Likewise.
13800 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13801 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13802 (mips16_insn_length): Robustify.
13803 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13804 (mips16_split_long_branches): Likewise.
13805 * config/pa/pa.c (pa_combine_instructions): Likewise.
13806 * config/rs6000/rs6000.c (get_next_active_insn): Treat
13807 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13808 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13809 as contributing to pool range lengths.
13810 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13811 Remove redundant JUMP_TABLE_DATA_P test.
13812 (sh_loop_align): Likewise.
13813 (split_branches): Likewise.
13814 (sh_insn_length_adjustment): Likewise.
13815 * config/spu/spu.c (get_branch_target): Likewise.
13817 2013-03-29 Jan Hubicka <jh@suse.cz>
13819 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13820 gcov streaming; stream hot bb threshold to ltrans.
13821 * predict.c (get_hot_bb_threshold): Break out from ....
13822 (maybe_hot_count_p): ... here.
13823 (set_hot_bb_threshold): New function.
13824 * lto-section-in.c (lto_section_name): Add profile.
13825 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13826 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13827 and data-streamer.h
13828 (histogram_entry): New structure.
13829 (histogram, histogram_pool): New global vars.
13830 (histogram_hash): New structure.
13831 (histogram_hash::hash): New method.
13832 (histogram_hash::equal): Likewise.
13833 (account_time_size): New function.
13834 (cmp_counts): New function.
13835 (dump_histogram): New function.
13836 (ipa_profile_generate_summary): New function.
13837 (ipa_profile_write_summary): New function.
13838 (ipa_profile_read_summary): New function.
13839 (ipa_profile): Decide on threshold.
13840 (pass_ipa_profile): Add ipa_profile_write_summary and
13841 ipa_profile_read_summary.
13842 * Makefile.in (ipa.o): Update dependencies.
13843 * lto-streamer.h (LTO_section_ipa_profile): New section.
13845 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
13847 * tree.h (VAR_P): New.
13849 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
13852 * doc/invoke.texi ([-fwhole-program]): Fix typo.
13854 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
13856 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13857 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13858 (control_flow_insn_p): Likewise.
13859 * cfgrtl.c (duplicate_insn_chain): Likewise.
13860 * final.c (get_attr_length_1): Likewise.
13861 (shorten_branches): Likewise.
13862 (final_scan_insn): Likewise.
13863 * function.c (instantiate_virtual_regs): Likewise.
13864 * gcse.c (insert_insn_end_basic_block): Likewise.
13865 * ira-costs.c (scan_one_insn): Likewise.
13866 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13867 * lra.c (check_rtl): Likewise.
13868 * reload1.c (elimination_costs_in_insn): Likewise.
13869 * reorg.c (follow_jumps): Likewise.
13871 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13872 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13873 (thumb_far_jump_used_p): Likewise.
13874 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13875 (workaround_speculation): Likewise.
13876 (add_sched_insns_for_speculation): Likewise.
13877 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13878 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13879 (frv_for_each_packet): Likewise.
13880 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13881 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13882 (final_emit_insn_group_barriers): Likewise.
13883 * config/m32r/m32r.c (m32r_is_insn): Likewise.
13884 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13885 (mips16_insn_length): Likewise.
13886 * config/pa/pa.c (pa_reorg): Likewise.
13887 (pa_combine_instructions): Likewise.
13888 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13889 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13890 (sh_reorg): Likewise.
13891 (split_branches): Likewise.
13892 * config/spu/spu.c (get_branch_target): Likewise.
13894 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13897 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
13899 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13900 Fix declaration name.
13902 2013-03-28 Lawrence Crowl <crowl@google.com>
13904 * graphds.h (struct graph.indicies): Remove unused.
13905 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13906 (SCOP_ORIGINAL_PDDRS): Remove unused.
13907 * sese.h (extern insert_loop_close_phis): Removed unused.
13908 (extern insert_guard_phis): Removed unused.
13909 (extern ivtype_map_elt_info): Removed unused.
13910 (new_ivtype_map_elt): Removed unused.
13911 * sese.c (ivtype_map_elt_info): Removed unused.
13913 2013-03-28 Lawrence Crowl <crowl@google.com>
13915 * Makefile.in: Add several missing include dependences.
13917 (test-dump.o): New. This object is not added to any executable,
13918 but is present for ad-hoc testing.
13920 (debug (const bitmap_head_def &)): New.
13921 (debug (const bitmap_head_def *)): New.
13923 (extern debug (const bitmap_head_def &)): New.
13924 (extern debug (const bitmap_head_def *)): New.
13926 (debug (edge_def &)): New.
13927 (debug (edge_def *)): New.
13929 (debug (basic_block_def &)): New.
13930 (debug (basic_block_def *)): New.
13932 (dump_node (const_tree, int, FILE *)): Correct source file.
13934 (debug (die_struct &)): New.
13935 (debug (die_struct *)): New.
13937 (extern debug (die_struct &)): New.
13938 (extern debug (die_struct *)): New.
13939 * gimple-pretty-print.c
13940 (debug (gimple_statement_d &)): New.
13941 (debug (gimple_statement_d *)): New.
13942 * gimple-pretty-print.h
13943 (extern debug (gimple_statement_d &)): New.
13944 (extern debug (gimple_statement_d *)): New.
13946 (debug (ira_allocno_copy &)): New.
13947 (debug (ira_allocno_copy *)): New.
13948 (debug (ira_allocno &)): New.
13949 (debug (ira_allocno *)): New.
13951 (extern debug (ira_allocno_copy &)): New.
13952 (extern debug (ira_allocno_copy *)): New.
13953 (extern debug (ira_allocno &)): New.
13954 (extern debug (ira_allocno *)): New.
13956 (debug (live_range &)): New.
13957 (debug (live_range *)): New.
13959 (debug (lra_live_range &)): New.
13960 (debug (lra_live_range *)): New.
13962 (debug (lra_live_range &)): New.
13963 (debug (lra_live_range *)): New.
13965 (debug (omega_pb_d &)): New.
13966 (debug (omega_pb_d *)): New.
13968 (extern debug (omega_pb_d &)): New.
13969 (extern debug (omega_pb_d *)): New.
13971 (debug (const rtx_def &)): New.
13972 (debug (const rtx_def *)): New.
13974 (debug_tree (tree): Move within file.
13975 (debug_raw (const tree_node &)): New.
13976 (debug_raw (const tree_node *)): New.
13977 (dump_tree_via_hooks (const tree_node *, int)): New.
13978 (debug (const tree_node &)): New.
13979 (debug (const tree_node *)): New.
13980 (debug_verbose (const tree_node &)): New.
13981 (debug_verbose (const tree_node *)): New.
13982 (debug_head (const tree_node &)): New.
13983 (debug_head (const tree_node *)): New.
13984 (debug_body (const tree_node &)): New.
13985 (debug_body (const tree_node *)): New.
13986 (debug_vec_tree (tree): Move and reimplement in terms of dump.
13987 (debug (vec<tree, va_gc> &)): New.
13988 (debug (vec<tree, va_gc> *)): New.
13990 (extern debug (const rtx_def &)): New.
13991 (extern debug (const rtx_def *)): New.
13993 (debug_raw (simple_bitmap_def &)): New.
13994 (debug_raw (simple_bitmap_def *)): New.
13995 (debug (simple_bitmap_def &)): New.
13996 (debug (simple_bitmap_def *)): New.
13998 (extern debug (simple_bitmap_def &)): New.
13999 (extern debug (simple_bitmap_def *)): New.
14000 (extern debug_raw (simple_bitmap_def &)): New.
14001 (extern debug_raw (simple_bitmap_def *)): New.
14003 (debug (vinsn_def &)): New.
14004 (debug (vinsn_def *)): New.
14005 (debug_verbose (vinsn_def &)): New.
14006 (debug_verbose (vinsn_def *)): New.
14007 (debug (expr_def &)): New.
14008 (debug (expr_def *)): New.
14009 (debug_verbose (expr_def &)): New.
14010 (debug_verbose (expr_def *)): New.
14011 (debug (vec<rtx> &)): New.
14012 (debug (vec<rtx> *)): New.
14014 (extern debug (vinsn_def &)): New.
14015 (extern debug (vinsn_def *)): New.
14016 (extern debug_verbose (vinsn_def &)): New.
14017 (extern debug_verbose (vinsn_def *)): New.
14018 (extern debug (expr_def &)): New.
14019 (extern debug (expr_def *)): New.
14020 (extern debug_verbose (expr_def &)): New.
14021 (extern debug_verbose (expr_def *)): New.
14022 (extern debug (vec<rtx> &)): New.
14023 (extern debug (vec<rtx> *)): New.
14025 (_list_iter_cond_expr): Make inline instead of static.
14027 (debug (sreal &)): New.
14028 (debug (sreal *)): New.
14030 (extern debug (sreal &)): New.
14031 (extern debug (sreal *)): New.
14033 (extern debug_raw (const tree_node &)): New.
14034 (extern debug_raw (const tree_node *)): New.
14035 (extern debug (const tree_node &)): New.
14036 (extern debug (const tree_node *)): New.
14037 (extern debug_verbose (const tree_node &)): New.
14038 (extern debug_verbose (const tree_node *)): New.
14039 (extern debug_head (const tree_node &)): New.
14040 (extern debug_head (const tree_node *)): New.
14041 (extern debug_body (const tree_node &)): New.
14042 (extern debug_body (const tree_node *)): New.
14043 (extern debug (vec<tree, va_gc> &)): New.
14044 (extern debug (vec<tree, va_gc> *)): New.
14046 (debug (struct loop &)): New.
14047 (debug (struct loop *)): New.
14048 (debug_verbose (struct loop &)): New.
14049 (debug_verbose (struct loop *)): New.
14050 * tree-dump.c: Add header dependence.
14052 (extern debug (struct loop &)): New.
14053 (extern debug (struct loop *)): New.
14054 (extern debug_verbose (struct loop &)): New.
14055 (extern debug_verbose (struct loop *)): New.
14057 (debug (data_reference &)): New.
14058 (debug (data_reference *)): New.
14059 (debug (vec<data_reference_p> &)): New.
14060 (debug (vec<data_reference_p> *)): New.
14061 (debug (vec<ddr_p> &)): New.
14062 (debug (vec<ddr_p> *)): New.
14064 (extern debug (data_reference &)): New.
14065 (extern debug (data_reference *)): New.
14066 (extern debug (vec<data_reference_p> &)): New.
14067 (extern debug (vec<data_reference_p> *)): New.
14068 (extern debug (vec<ddr_p> &)): New.
14069 (extern debug (vec<ddr_p> *)): New.
14071 (debug (pt_solution &)): New.
14072 (debug (pt_solution *)): New.
14074 (extern debug (pt_solution &)): New.
14075 (extern debug (pt_solution *)): New.
14077 (debug (_var_map &)): New.
14078 (debug (_var_map *)): New.
14079 (debug (tree_live_info_d &)): New.
14080 (debug (tree_live_info_d *)): New.
14082 (extern debug (_var_map &)): New.
14083 (extern debug (_var_map *)): New.
14084 (extern debug (tree_live_info_d &)): New.
14085 (extern debug (tree_live_info_d *)): New.
14087 2013-03-28 Jan Hubicka <jh@suse.cz>
14089 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
14091 2013-03-28 Ian Bolton <ian.bolton@arm.com>
14093 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
14094 record only when desired or required.
14096 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
14098 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
14099 *vec_extractv2di_1_rex64. Use x64 isa attribute.
14101 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14103 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
14104 (*andsi3_compare0_uxtw): New pattern.
14105 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
14106 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
14108 2013-03-28 Jan Hubicka <jh@suse.cz>
14110 * data-streamer-in.c (streamer_read_gcov_count): New function.
14111 * gimple-streamer-out.c: Include value-prof.h.
14112 (output_gimple_stmt): Output histogram.
14113 (output_bb): Use streamer_write_gcov_count.
14114 * value-prof.c: Include data-streamer.h
14115 (dump_histogram_value): Add HIST_TYPE_MAX.
14116 (stream_out_histogram_value): New function.
14117 (stream_in_histogram_value): New function.
14118 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
14119 (stream_out_histogram_value, stream_in_histogram_value): Declare.
14120 * data-streamer-out.c (streamer_write_gcov_count): New function.
14121 (streamer_write_gcov_count_stream): New function.
14122 * lto-cgraph.c (lto_output_edge): Update counter streaming.
14123 (lto_output_node): Likewise.
14124 (input_node, input_edge): Likewise.
14125 * lto-streamer-out.c (output_cfg): Update streaming.
14126 * lto-streamer-in.c (input_cfg): Likewise.
14127 * data-streamer.h (streamer_write_gcov_count,
14128 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
14129 * gimple-streamer-in.c: Include value-prof.h
14130 (input_gimple_stmt): Input histograms.
14131 (input_bb): Update profile streaming.
14133 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
14135 * genmodes.c (emit_max_int): New function.
14136 (emit_insn_modes_h): Added call to emit_max_function.
14137 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
14139 * machmode.def: Fixed comment.
14141 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
14143 * combine.c (try_combine): Removed useless assert.
14144 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
14146 2013-03-28 Marek Polacek <polacek@redhat.com>
14147 Richard Biener <rguenther@suse.de>
14149 PR tree-optimization/56695
14150 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
14151 build signed result of a vector comparison.
14152 * tree-cfg.c (verify_gimple_comparison): Check that a result
14153 of a vector comparison has signed type.
14155 2013-03-28 Richard Biener <rguenther@suse.de>
14157 PR tree-optimization/37021
14158 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
14159 do not restrict gaps between groups.
14160 * tree-vect-stmts.c (vectorizable_load): Properly account for
14161 a gap between groups.
14163 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
14165 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
14166 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
14169 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
14171 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
14172 * doc/extend.texi (Named Address Spaces): Ditto.
14173 (Variable Attributes): Ditto.
14175 2013-03-27 Kai Tietz <ktietz@redhat.com>
14177 * config.build: Add support for cygwin x64 target.
14178 * config.gcc: Likewise.
14179 * config.host: Likewise.
14180 * configure.ac: Likewise
14181 * configure: Regenerated.
14183 2013-03-27 Kai Tietz <ktietz@redhat.com>
14185 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
14186 * config/i386/t-cygwin-w64: New file.
14187 * config/i386/cygwin-w64.h: New file.
14188 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
14189 and add support for x64-cygwin target.
14190 (CPP_SPEC): Likewise.
14191 (CXX_WRAP_SPEC_LIST): Undefine before define.
14192 (LIBGCJ_SONAME): Use 15 as version.
14194 2013-03-27 Richard Biener <rguenther@suse.de>
14196 PR tree-optimization/56716
14197 * tree-ssa-structalias.c (perform_var_substitution): Adjust
14198 dumping for ref nodes.
14200 2013-03-27 Martin Jambor <mjambor@suse.cz>
14202 PR tree-optimization/55334
14203 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
14204 restricted pointers to arrays.
14206 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
14208 * Makefile.in (.SUFFIXES): Add .cc.
14209 (.c.o): Apply same recipe for implicit rule .cc.o.
14211 2013-03-27 Richard Biener <rguenther@suse.de>
14213 PR tree-optimization/37021
14214 * tree-vect-data-refs.c (vect_check_strided_load): Allow
14215 REALPART/IMAGPART_EXPRs around the supported refs.
14216 * tree-ssa-structalias.c (find_func_aliases): Assume that
14217 floating-point values are not used to transfer pointers.
14219 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
14221 * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
14222 * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
14223 * gcc/doc/tm.texi: Regenerate.
14224 * gcc/targhooks.h (default_has_ifunc_p): New.
14225 * gcc/targhooks.c (default_has_ifunc_p): Ditto.
14226 * gcc/config/linux-protos.h: New file.
14227 * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
14228 this hook for linux which disables support of indirect functions in
14230 * gcc/config/linux-android.c: New file.
14231 * gcc/config/t-linux-android.c: Ditto.
14232 * gcc/config.gcc: Added new object file linux-android.o.
14233 * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
14234 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
14235 * gcc/varasm.c (do_assemble_alias): Likewise.
14236 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
14237 doesn't support indirect functions.
14238 * configure: Regenerate.
14240 2013-03-27 Bin Cheng <bin.cheng@arm.com>
14243 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
14244 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
14247 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
14249 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
14251 2013-03-27 Terry Guo <terry.guo@arm.com>
14253 * config/arm/arm-cores.def: Added core cortex-r7.
14254 * config/arm/arm-tune.md: Regenerated.
14255 * config/arm/arm-tables.opt: Regenerated.
14256 * doc/invoke.texi: Added entry for core cortex-r7.
14258 2013-03-27 Walter Lee <walt@tilera.com>
14260 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
14261 double-decrement of next_scratch_regno.
14263 2013-03-27 Walter Lee <walt@tilera.com>
14265 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
14267 (insn_v1mulus): Ditto.
14268 (insn_v2muls): Ditto.
14270 2013-03-27 Walter Lee <walt@tilera.com>
14272 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
14273 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
14275 2013-03-27 Walter Lee <walt@tilera.com>
14277 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
14278 (*sibcall_value): Ditto.
14280 2013-03-27 Walter Lee <walt@tilera.com>
14282 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
14283 (insn_mnz_v8qi): ... this ...
14284 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
14286 (insn_v<n>mnz): Replaced by ...
14287 (insn_v1mnz): ... this ...
14288 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
14290 (insn_mz_<mode>): Replaced by ...
14291 (insn_mz_v8qi): ... this ...
14292 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
14294 (insn_v<n>mz): Replaced by ...
14295 (insn_v1mz): ... this ...
14296 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
14299 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
14301 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
14303 2013-03-26 Roland McGrath <mcgrathr@google.com>
14305 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
14306 than fprintf with a non-constant, non-format string.
14308 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
14310 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
14311 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
14312 operand 0 predicate.
14313 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
14314 attribute. Use general_x64nomem_operand as operand 1 predicate.
14315 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
14316 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
14317 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
14318 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
14319 (mov<mode>_insv_1): Remove expander. Merge insn with
14320 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
14321 Use general_x64nomem_operand as operand 1 predicate.
14322 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
14323 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
14324 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
14325 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
14326 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
14327 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
14328 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
14329 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
14330 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
14331 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
14332 (general_x64nomem_operand): Ditto.
14334 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14336 * config/rtems.opt: Add -pthread option.
14338 2013-03-26 Richard Biener <rguenther@suse.de>
14340 * alias.c (find_base_term): Avoid redundant and not used recursion.
14341 (base_alias_check): Get the initial base term from the caller.
14342 (true_dependence_1): Compute and pass base terms to base_alias_check.
14343 (write_dependence_p): Likewise.
14344 (may_alias_p): Likewise.
14346 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
14348 * config/aarch64/aarch64.c (aarch64_classify_address): Support
14349 PC-relative load in SI modes and above only.
14351 2013-03-26 Xinyu Qi <xyqi@marvell.com>
14353 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
14354 * config/arm/iwmmxt.md (WCGR0): Update.
14355 (WCGR1, WCGR2, WCGR3): Likewise.
14357 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
14359 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
14360 Use x64 and nox64 isa attributes.
14362 2013-03-26 Richard Biener <rguenther@suse.de>
14364 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
14365 alignment computations and rely on get_object_alignment_1
14366 for the !TYPE_P case.
14367 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
14369 2013-03-26 Walter Lee <walt@tilera.com>
14371 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
14372 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
14374 2013-03-25 Jeff Law <law@redhat.com>
14376 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
14377 check for INTEGRAL_TYPE_P that was missing due to checking in
14378 wrong version of prior patch.
14380 2013-03-25 Walter Lee <walt@tilera.com>
14382 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
14383 TILEGX_INSN_SHUFFLEBYTES1.
14384 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
14386 (tilegx_builtins): Ditto.
14387 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
14389 2013-03-25 Walter Lee <walt@tilera.com>
14391 * config/tilegx/tilegx.md (floatsisf2): New pattern.
14392 (floatunssisf2): New pattern.
14393 (floatsidf2): New pattern.
14394 (floatunssidf2): New pattern.
14396 2013-03-25 Walter Lee <walt@tilera.com>
14398 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
14399 tests for constraint J, K, N, P.
14401 2013-03-25 Walter Lee <walt@tilera.com>
14403 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
14404 Use indirect/pcrel encoding.
14405 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
14408 2013-03-25 Steve Ellcey <sellcey@mips.com>
14410 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
14411 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
14412 * config/mips/mips.c (mips_option_override): Set IMADD default.
14413 * config/mips/mips.h (PTF_AVOID_IMADD): New.
14414 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
14415 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
14416 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
14417 * doc/invoke.texi (-mimadd/-mno-imadd): New.
14419 2013-03-25 Jeff Law <law@redhat.com>
14421 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
14422 slightly to avoid creating and folding useless trees. Simplify
14423 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
14425 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
14427 * config/i386/i386.md (*zero_extendsidi2): Merge with
14428 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
14429 * config/i386/predicates.md (x86_64_zext_operand): Rename from
14430 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
14431 targets. Clarify comment.
14433 2013-03-25 Martin Jambor <mjambor@suse.cz>
14435 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
14436 pass-through jump functions differently.
14437 (ipa_read_jump_function): Likewise. Also use setter functions to set
14440 2013-03-25 Martin Jambor <mjambor@suse.cz>
14442 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
14443 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
14445 (ipa_get_indirect_edge_target): New function.
14446 (devirtualization_time_bonus): New parameter known_aggs, pass it to
14447 ipa_get_indirect_edge_target. Update all callers.
14448 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
14449 ipa_get_indirect_edge_target_1 instead of calling
14450 ipa_get_indirect_edge_target.
14451 (create_specialized_node): Pass aggvlas to
14452 ipcp_discover_new_direct_edges.
14454 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14456 * config/arm/arm.md (f_sels, f_seld): New types.
14457 (*cmov<mode>): New pattern.
14458 * config/arm/predicates.md (arm_vsel_comparison_operator): New
14461 2013-03-25 Kai Tietz <ktietz@redhat.com>
14463 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
14464 POSIX-printf for mingw-hosted builds.
14466 2013-03-25 Richard Biener <rguenther@suse.de>
14468 PR middle-end/56694
14469 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
14470 must-not-throw stmt location.
14472 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14474 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
14475 Emit load-acquire versions when acq is true.
14476 (arm_emit_store_exclusive): Add rel parameter.
14477 Emit store-release versions when rel is true.
14478 (arm_split_compare_and_swap): Use acquire-release instructions
14480 of barriers when appropriate.
14481 (arm_split_atomic_op): Likewise.
14482 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
14483 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
14484 (VUNSPEC_SLX): Likewise.
14485 (VUNSPEC_LDA): Likewise.
14486 (VUNSPEC_STL): Likewise.
14487 * config/arm/sync.md (atomic_load<mode>): New pattern.
14488 (atomic_store<mode>): Likewise.
14489 (arm_load_acquire_exclusive<mode>): Likewise.
14490 (arm_load_acquire_exclusivesi): Likewise.
14491 (arm_load_acquire_exclusivedi): Likewise.
14492 (arm_store_release_exclusive<mode>): Likewise.
14494 2013-03-25 Catherine Moore <clm@codesourcery.com>
14496 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
14497 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
14498 * config/mip/predicates.md (lwsp_swsp_operand,
14499 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
14500 sb16_operand, db4_operand, db7_operand, ib3_operand,
14501 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
14502 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
14503 andi16_operand): New predicates.
14504 * config/mips/mips.md (compression): New attribute.
14505 (enabled): New attribute.
14506 (length): Consider compression in computing length.
14507 (shift_compression): New code attribute.
14508 (*add<mode>3): New operands. Record compression.
14509 (sub<mode>3): Likewise.
14510 (one_cmpl<mode>2): Likewise.
14511 (*and<mode>3): Likewise.
14512 (*ior<mode>3): Likewise.
14513 (unnamed pattern for xor): Likewise.
14514 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
14515 (*<optab><mode>3): Likewise.
14516 (*mov<mode>_internal: Likewise.
14517 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
14518 (mips_unsigned_immediate_p): New.
14519 (umips_lwsp_swsp_address_p): New.
14520 (m16_based_address_p): New.
14521 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
14522 (mips_unsigned_immediate_p): New prototype.
14523 (lwsp_swsp_address_p): New prototype.
14524 (m16_based_address_p): New prototype.
14525 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14526 (mips_signed_immediate_p): New function.
14527 (m16_based_address_p): New function.
14528 (lwsp_swsp_address_p): New function.
14529 (mips_print_operand_punctuation): Recognize short delay slot insns
14530 for microMIPS.add<mode>3"
14532 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14535 * config/arm/iterators.md (v_cmp_result): New mode attribute.
14536 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14538 2013-03-25 Richard Biener <rguenther@suse.de>
14540 PR tree-optimization/56689
14541 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14544 2013-03-25 Richard Biener <rguenther@suse.de>
14546 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14548 (memory_references): Likewise.
14549 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14550 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14551 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14552 (gather_mem_refs_in_loops): Fold into ...
14553 (analyze_memory_references): ... this. Move initialization
14554 to tree_ssa_lim_initialize.
14555 (fill_always_executed_in): Rename to ...
14556 (fill_always_executed_in_1): ... this.
14557 (fill_always_executed_in): Move contains_call computation to
14558 this new function from ...
14559 (tree_ssa_lim_initialize): ... here.
14560 (tree_ssa_lim): Call fill_always_executed_in.
14562 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
14564 * postreload.c (reload_combine): Fix code detecting returns.
14566 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
14568 * function.c (emit_use_return_register_into_block): On cc0 targets,
14569 do not emit the sequence between cc0 setter and user.
14571 2013-03-25 Kai Tietz <ktietz@redhat.com>
14573 * config/i386/predicates.md (local_symbolic_operand): Interpret
14574 dll-imported symbols as none-local.
14576 2013-03-25 Richard Biener <rguenther@suse.de>
14578 * tree-ssa-loop-im.c (struct depend): Remove.
14579 (struct lim_aux_data): Make depends a vec of gimples.
14580 (free_lim_aux_data): Adjust.
14581 (add_dependency): Likewise.
14582 (set_level): Likewise.
14584 2013-03-25 Richard Biener <rguenther@suse.de>
14586 PR middle-end/56434
14587 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14588 the pointer returned by calls with ECF_MALLOC set.
14590 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
14592 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14594 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
14596 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14597 using MMXMODE mode iterator.
14598 (*move<mode>_internal): Merge with *movv2sf_internal and
14599 *movv2sf_internal_rex64 using MMXMODE mode iterator.
14601 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
14603 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14604 (record_last_mem_set_info): Likewise.
14606 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14607 of XNEWVEC followed by memset.
14608 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14610 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
14612 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14613 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14614 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14615 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14616 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14617 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14618 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14619 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14620 BARRIER_P instead of GET_CODE.
14622 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14624 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14625 inaccuracy in the probing code.
14627 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14628 (ctrapdi4): Likewise.
14630 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14632 * calls.c (expand_call): Add missing guard to code handling return
14633 of non-BLKmode structures in MSB.
14634 * function.c (expand_function_end): Likewise.
14636 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14638 * combine.c (try_combine): Adjust comment. Do not add the set of
14639 insn #0 if the destination indirectly is set or dies in insn #2.
14640 Tidy up code to distribute a new note.
14642 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
14644 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14645 also for alternatives 16 and 17.
14647 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
14649 * config/i386/sse.md (*mov<mode>_internal): Merge with
14650 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
14651 Emit insn template depending on type attribute. Use
14652 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14653 movd instead of movq mnemonic for interunit moves. Rewrite mode
14654 attribute calculation. Remove unit attribute calculation.
14655 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14656 Set prefix_data16 attribute for DImode ssemov types.
14657 Use Ym instead of y for SSE-MMX conversion alternatives.
14658 Reorder operand constraints.
14660 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
14662 * df.h (df_insn_delete): Adjust prototype.
14663 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14664 and let it decide whether mark the basic block dirty.
14665 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14666 * df-scan.c (df_insn_info_delete): New helper function, split
14667 off from df_insn_delete.
14668 (df_scan_free_bb_info): Use it.
14669 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14671 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
14672 that the insn is actually an insn and it has a non-NULL basic block.
14673 Do not mark basic block dirty if only deleting a DEBUG_INSN.
14675 2013-03-22 Richard Biener <rguenther@suse.de>
14677 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14679 (mem_ref_alloc): Do not allocate them.
14680 (refs_independent_p): Do not query or maintain a cache.
14682 2013-03-22 Richard Biener <rguenther@suse.de>
14684 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14685 (gather_mem_refs_in_loops): Do not compute it.
14686 (analyze_memory_references): Do not allocate it.
14687 (tree_ssa_lim_finalize): Do not free it.
14688 (for_all_locs_in_loop): Do not query all_refs_in_loop.
14690 2013-03-22 Richard Biener <rguenther@suse.de>
14692 * is-a.h (as_a): Use gcc_checking_assert.
14694 2013-03-22 Ian Bolton <ian.bolton@arm.com>
14696 * config/aarch64/aarch64.c (aarch64_print_operand): New
14697 format specifier for printing a constant in hex.
14698 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14699 format specifier for printing second operand.
14701 2013-03-22 Richard Biener <rguenther@suse.de>
14703 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14705 (gather_mem_refs_in_loops): Perform store accumulation here.
14706 (create_vop_ref_mapping_loop): Remove.
14707 (create_vop_ref_mapping): Likewise.
14708 (analyze_memory_references): Initialize refs_stored_in_loop.
14709 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14710 (record_indep_loop): Remove.
14711 (record_dep_loop): New function.
14712 (ref_indep_loop_p_1): Adjust to only walk over references
14713 in the loop, not its subloops.
14714 (ref_indep_loop_p): Rename to ...
14715 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14716 maintaining a more fine-grained cache.
14717 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14718 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14720 2013-03-22 Richard Biener <rguenther@suse.de>
14722 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14723 (struct mem_ref): Make accesses_in_loop a vec of a vec of
14724 aggregate mem_ref_loc.
14725 (free_mem_ref_locs): Inline into ...
14726 (memref_free): ... this and adjust.
14727 (mem_ref_alloc): Adjust.
14728 (mem_ref_locs_alloc): Remove.
14729 (record_mem_ref_loc): Adjust.
14730 (get_all_locs_in_loop): Rewrite into ...
14731 (for_all_locs_in_loop): ... this iterator.
14732 (rewrite_mem_ref_loc): New functor.
14733 (rewrite_mem_refs): Use for_all_locs_in_loop.
14734 (sm_set_flag_if_changed): New functor.
14735 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14736 (ref_always_accessed): New functor.
14737 (ref_always_accessed_p): Use for_all_locs_in_loop.
14739 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14741 * tree-pass.h (PROP_gimple_lvec): New.
14742 * passes.c (dump_properties): Handle PROP_gimple_lvec.
14743 (init_optimization_passes): Move pass_lower_vector.
14744 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14746 (pass_lower_vector): Provide PROP_gimple_lvec.
14747 (pass_lower_vector_ssa): Likewise.
14748 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14750 2013-03-21 Mark Wielaard <mjw@redhat.com>
14752 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14754 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
14756 * config/i386/i386.md (*movdi_internal): Disparage slightly
14757 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
14758 conversion alternatives.
14760 2013-03-21 Jakub Jelinek <jakub@redhat.com>
14762 PR middle-end/48087
14763 * diagnostic.def (DK_WERROR): New kind.
14764 * diagnostic.h (werrorcount): Define.
14765 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14766 promoted to DK_ERROR, increment DK_WERROR counter instead of
14768 * toplev.c (toplev_main): Call print_ignored_options even if
14769 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
14770 even if just werrorcount is non-zero.
14773 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14775 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14776 (string_cst_pool_decl): New function.
14777 (optimize_one_addr_into_implicit_ptr): New function.
14778 (resolve_addr_in_expr): Optimize DWARF location expression
14779 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14780 which doesn't live in memory, but has DW_AT_location or
14781 DW_AT_const_value, or refers to a string literal, into
14782 DW_OP_GNU_implicit_pointer.
14783 (optimize_location_into_implicit_ptr): New function.
14784 (resolve_addr): If removing DW_AT_location of a variable because
14785 it was DW_OP_addr of address of the variable, but the variable doesn't
14786 live in memory, try to emit const value attribute for the initializer.
14788 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14790 * tree.h (VECTOR_TYPE_P): New macro.
14791 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14792 TYPE_MODE): Use it.
14793 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14794 VEC_COND_EXPR cannot be lvalues.
14795 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14797 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14799 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14800 Restrict the transformation to equal modes.
14802 2013-03-21 Richard Biener <rguenther@suse.de>
14804 PR tree-optimization/39326
14805 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14806 (MEM_ANALYZABLE): Adjust.
14807 (record_mem_ref_loc): Move bitmap ops ...
14808 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
14809 unanalyzable refs, do not record locations for it.
14810 (analyze_memory_references): Allocate ref zero as shared
14812 (refs_independent_p): Do not test for unanalyzed mems here.
14813 (ref_indep_loop_p_1): Special-case disambiguation against
14814 the unanalyzed ref.
14815 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14817 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
14819 * config/arm/arm-protos.h (tune_params): Add
14820 prefer_neon_for_64bits field.
14821 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14822 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14823 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14824 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14825 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14826 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14827 (arm_option_override): Handle -mneon-for-64bits new option.
14828 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14829 (prefer_neon_for_64bits): Declare new variable.
14830 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14831 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14832 (arch_enabled): Handle new arch types. Remove support for onlya8
14834 (one_cmpldi2): Use new arch names.
14835 (zero_extend<mode>di2, extend<mode>di2): Ditto.
14836 * config/arm/arm.opt (mneon-for-64bits): Add option.
14837 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14838 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14839 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14841 * doc/invoke.texi (-mneon-for-64bits): Document.
14843 2013-03-21 Richard Biener <rguenther@suse.de>
14845 PR tree-optimization/39326
14846 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14847 (sort_bbs_in_loop_postorder_cmp): New function.
14848 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14850 2013-03-21 Richard Biener <rguenther@suse.de>
14852 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14853 (vect_insert_into_interleaving_chain): Likewise.
14854 (vect_drs_dependent_in_basic_block): Inline ...
14855 (vect_slp_analyze_data_ref_dependence): ... here. New function,
14857 (vect_analyze_data_ref_dependence): ... here. Simplify.
14858 (vect_check_interleaving): Simplify.
14859 (vect_analyze_data_ref_dependences): Likewise. Split out ...
14860 (vect_slp_analyze_data_ref_dependences): ... this new function.
14861 (dr_group_sort_cmp): New function.
14862 (vect_analyze_data_ref_accesses): Compute data-reference groups
14863 here instead of in vect_analyze_data_ref_dependence. Use
14864 a more efficient algorithm.
14865 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14866 vect_slp_analyze_data_ref_dependences. Call
14867 vect_analyze_data_ref_accesses earlier.
14868 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14869 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14870 (vect_slp_analyze_data_ref_dependences): New prototype.
14872 2013-03-21 Richard Biener <rguenther@suse.de>
14874 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14875 ref is stored in the loop.
14876 (find_refs_for_sm): Walk only over all stores.
14877 (store_motion_loop): Allocate from lim_bitmap_obstack.
14878 (store_motion): Likewise.
14880 2013-03-21 Richard Biener <rguenther@suse.de>
14882 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14883 Update virtual SSA form.
14885 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14887 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14888 * configure: Regenerate.
14889 * config.in: Regenerate.
14890 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14891 if !HAVE_LD_EH_FRAME_CIEV3.
14893 2013-03-21 Richard Biener <rguenther@suse.de>
14895 * tree-cfg.c (verify_expr_no_block): New function.
14896 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14897 nor DECL_VALUE_EXPR have locations with associated blocks.
14898 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14899 (clear_unused_block_pointer): Remove code dealing with
14900 blocks in DECL_DEBUG_EXPR locations.
14902 2013-03-21 Richard Biener <rguenther@suse.de>
14904 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14905 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
14906 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14907 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14908 instead of DECL_DEBUG_EXPR_IS_FROM.
14909 * gimplify.c (gimplify_modify_expr): Likewise.
14910 * tree-cfg.c (verify_expr_location_1): Likewise.
14911 * tree-complex.c (create_one_component_var): Likewise.
14912 * tree-sra.c (create_access_replacement): Likewise.
14913 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14914 (clear_unused_block_pointer): Likewise.
14915 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14916 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14917 * var-tracking.c (var_debug_decl): Likewise.
14918 (track_expr_p): Likewise.
14919 * tree-inline.c (add_local_variables): Likewise. Set
14920 DECL_HAS_DEBUG_EXPR_P after copying it.
14921 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14922 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
14924 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
14927 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14928 * configure: Regenerate.
14929 * config.in: Regenerate.
14930 * config/i386/i386.md (*movdf_internal): Use
14931 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14932 movd instead of movq mnemonic for interunit moves.
14933 (*movdi_internal): Ditto.
14935 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14937 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14938 (abd<mode>_3): New pattern.
14939 (aba<mode>_3): New pattern.
14940 (fabd<mode>_3): New pattern.
14942 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14944 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14945 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14946 occurrence of REGISTER_PREFIX as its empty string.
14948 2013-03-20 Jeff Law <law@redhat.com>
14950 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14951 addititional equivalences for equality comparisons between an SSA_NAME
14952 and a constant where the SSA_NAME was set from a widening conversion.
14954 2013-03-20 Walter Lee <walt@tilera.com>
14956 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14958 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14960 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14961 depending on type attribute.
14962 (*movti_internal): Ditto.
14963 (*movtf_internal): Ditto.
14964 (*movxf_internal): Ditto.
14965 (*movdf_internal): Ditto.
14966 (*movsf_internal): Ditto.
14968 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
14970 * config/i386/i386.md (*movti_internal): Set prefix attribute to
14971 maybe_vex for sselog1 and ssemov types.
14972 (*movdi_internal): Reorder operand constraints.
14973 (*movsi_internal): Ditto. Set prefix attribute to
14974 maybe_vex for sselog1 and ssemov types.
14975 (*movtf_internal): Set prefix attribute to maybe_vex
14976 for sselog1 and ssemov types.
14977 (*movdf_internal): Ditto. Set prefix_data16 attribute for
14978 DImode ssemov types. Reorder operand constraints.
14979 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
14980 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
14981 attribute for SImode ssemov types. Reorder operand constraints.
14983 2013-03-20 Martin Jambor <mjambor@suse.cz>
14985 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14986 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14988 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
14990 * config/rs6000/predicates.md (indexed_address, update_address_mem
14991 update_indexed_address_mem): New predicates.
14992 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14993 attribute for load/store instructions.
14994 * config/rs6000/dfp.md (movsd_store): Likewise.
14995 (movsd_load): Likewise.
14996 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14997 (unnamed HI->DI extend define_insn): Likewise.
14998 (unnamed SI->DI extend define_insn): Likewise.
14999 (unnamed QI->SI extend define_insn): Likewise.
15000 (unnamed QI->HI extend define_insn): Likewise.
15001 (unnamed HI->SI extend define_insn): Likewise.
15002 (unnamed HI->SI extend define_insn): Likewise.
15003 (extendsfdf2_fpr): Likewise.
15004 (movsi_internal1): Likewise.
15005 (movsi_internal1_single): Likewise.
15006 (movhi_internal): Likewise.
15007 (movqi_internal): Likewise.
15008 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
15009 attribute for load/store instructions.
15010 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
15012 (mov<mode>_softfloat): Likewise.
15013 (mov<mode>_hardfloat32): Likewise.
15014 (mov<mode>_hardfloat64): Likewise.
15015 (mov<mode>_softfloat64): Likewise.
15016 (movdi_internal32): Likewise.
15017 (movdi_internal64): Likewise.
15018 (probe_stack_<mode>): Likewise.
15020 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15022 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
15023 floating point, and decimal floating point to reload iterator.
15025 * config/rs6000/constraints.md (wl constraint): New constraints to
15026 return FLOAT_REGS if certain options are used to reduce the number
15027 of separate patterns that exist in the file.
15028 (wx constraint): Likewise.
15029 (wz constraint): Likewise.
15031 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15032 -mdebug=reg, print wg, wl, wx, and wz constraints.
15033 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15034 Initialize the reload functions for 64-bit binary/decimal floating
15036 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15037 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15038 create the buffer on the stack to overcome not having a 32-bit
15040 (rs6000_emit_move): Likewise.
15041 (rs6000_secondary_memory_needed_rtx): Likewise.
15042 (rs6000_alloc_sdmode_stack_slot): Likewise.
15043 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15044 via xxlxor, just like DFmode 0.0.
15046 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
15047 define as 1 if we are running on a power7 or newer.
15048 (enum r6000_reg_class_enum): Add new constraints.
15050 * config/rs6000/dfp.md (movsd): Delete, combine with binary
15051 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
15052 with other moves by using conditional constraits (wg). Use LFIWZX
15053 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
15054 (movsd splitter): Likewise.
15055 (movsd_hardfloat): Likewise.
15056 (movsd_softfloat): Likewise.
15058 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
15059 binary and decimal floating point moves.
15060 (fmove_ok): New attributes to combine binary and decimal floating
15061 point moves, and to combine power6x (mfpgpr) moves along normal
15063 (real_value_to_target): Likewise.
15064 (f32_lr): Likewise.
15065 (f32_lm): Likewise.
15066 (f32_li): Likewise.
15067 (f32_sr): Likewise.
15068 (f32_sm): Likewise.
15069 (f32_si): Likewise.
15070 (movsf): Combine binary and decimal floating point moves. Combine
15071 power6x (mfpgpr) moves with other moves by using conditional
15072 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
15073 (mov<mode> for SFmode/SDmode); Likewise.
15074 (SFmode/SDmode splitters): Likewise.
15075 (movsf_hardfloat): Likewise.
15076 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
15077 (movsf_softfloat): Likewise.
15078 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
15080 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
15081 wx and wz constraints.
15083 * config/rs6000/constraints.md (wg constraint): New constraint to
15084 return FLOAT_REGS if -mmfpgpr (power6x) was used.
15086 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
15089 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15090 -mdebug=reg, print wg, wl, wx, and wz constraints.
15091 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15092 Initialize the reload functions for 64-bit binary/decimal floating
15094 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15095 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15096 create the buffer on the stack to overcome not having a 32-bit
15098 (rs6000_emit_move): Likewise.
15099 (rs6000_secondary_memory_needed_rtx): Likewise.
15100 (rs6000_alloc_sdmode_stack_slot): Likewise.
15101 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15102 via xxlxor, just like DFmode 0.0.
15104 * config/rs6000/dfp.md (movdd): Delete, combine with binary
15105 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
15106 with other moves by using conditional constraits (wg). Use LFIWZX
15107 and STFIWX for loading SDmode on power7.
15108 (movdd splitters): Likewise.
15109 (movdd_hardfloat32): Likewise.
15110 (movdd_softfloat32): Likewise.
15111 (movdd_hardfloat64_mfpgpr): Likewise.
15112 (movdd_hardfloat64): Likewise.
15113 (movdd_softfloat64): Likewise.
15115 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
15116 64-bit binary and decimal floating point moves.
15117 (FMOVE64X): Likewise.
15118 (movdf): Combine 64-bit binary and decimal floating point moves.
15119 Combine power6x (mfpgpr) moves with other moves by using
15120 conditional constraits (wg).
15121 (mov<mode> for DFmode/DDmode): Likewise.
15122 (DFmode/DDmode splitters): Likewise.
15123 (movdf_hardfloat32): Likewise.
15124 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
15125 (movdf_softfloat32): Likewise.
15126 (movdf_hardfloat64_mfpgpr): Likewise.
15127 (movdf_hardfloat64): Likewise.
15128 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
15129 (movdf_softfloat64): Likewise.
15130 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
15131 (reload_<mode>_load): Move to later in the file so they aren't in
15132 the middle of the floating point move insns.
15133 (reload_<mode>_store): Likewise.
15135 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
15138 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
15139 constraint if -mdebug=reg.
15140 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
15141 Enable using dd reload support if needed.
15143 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
15144 binary and decimal floating point moves in rs6000.md.
15145 (movtd_internal): Likewise.
15147 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
15148 decimal floating point moves.
15150 (movtf_internal): Likewise.
15151 (mov<mode>_internal, TDmode/TFmode): Likewise.
15152 (movtf_softfloat): Likewise.
15153 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
15155 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
15156 movdi_internal64, using wg constraint for move direct operations.
15157 (movdi_internal64): Likewise.
15159 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
15160 MODES_TIEABLE_P for selected modes. Print the numerical value of
15161 the various virtual registers. Use GPR/FPR first/last values,
15162 instead of hard coding the register numbers. Print which modes
15163 have reload functions registered.
15164 (rs6000_option_override_internal): If -mdebug=reg, trace the options
15165 settings before/after setting cpu, target and subtarget settings.
15166 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
15167 and for secondary reload failures in rs6000_secondary_reload_inner.
15168 (rs6000_secondary_reload_fail): Likewise.
15169 (rs6000_secondary_reload_inner): Likewise.
15171 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
15172 macros for first/last GPR and FPR registers.
15173 (LAST_GPR_REGNO): Likewise.
15174 (FIRST_FPR_REGNO): Likewise.
15175 (LAST_FPR_REGNO): Likewise.
15177 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
15178 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
15179 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
15180 (vcond<mode><mode>): Likewise.
15181 (vcondu<mode><mode>): Likewise.
15182 (vector_gtu<mode>): Likewise.
15183 (vector_gte<mode>): Likewise.
15184 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
15185 to prevent the compiler from converting DImode operations to TImode.
15186 (ior<mode>3): Likewise.
15187 (and<mode>3): Likewise.
15188 (one_cmpl<mode>2): Likewise.
15189 (nor<mode>3): Likewise.
15190 (andc<mode>3): Likewise.
15192 * config/rs6000/constraints.md (wt constraint): New constraint
15193 that returns VSX_REGS if TImode is allowed in VSX registers.
15195 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
15196 constant under VSX.
15198 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
15199 similar to TImode, but it is restricted to being in the GPRs.
15201 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
15202 TImode to occupy a single VSX register.
15204 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
15205 -mvsx-timode for power7/power8.
15206 (power7 cpu): Likewise.
15207 (power8 cpu): Likewise.
15209 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
15210 sure that TFmode/TDmode take up two registers if they are ever
15211 allowed in the upper VSX registers.
15212 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
15214 (rs6000_init_hard_regno_mode_ok): Likewise.
15215 (rs6000_debug_reg_global): Add debugging for PTImode and wt
15216 constraint. Print if LRA is turned on.
15217 (rs6000_option_override_internal): Give an error if -mvsx-timode
15218 and VSX is not enabled.
15219 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
15220 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
15221 to reg+offset addressing. Use PTImode when checking offset
15222 addresses for validity.
15223 (reg_offset_addressing_ok_p): Likewise.
15224 (rs6000_legitimate_offset_address_p): Likewise.
15225 (rs6000_legitimize_address): Likewise.
15226 (rs6000_legitimize_reload_address): Likewise.
15227 (rs6000_legitimate_address_p): Likewise.
15228 (rs6000_eliminate_indexed_memrefs): Likewise.
15229 (rs6000_emit_move): Likewise.
15230 (rs6000_secondary_reload): Likewise.
15231 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
15232 reloads to fpr registers to continue to use reg+offset addressing,
15233 but 64-bit reloads to altivec registers need reg+reg addressing.
15234 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
15235 it. Treat LO_SUM like a PLUS operation.
15236 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
15237 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
15238 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
15239 registers to share a register with a smaller sized type, since VSX
15240 puts scalars in the upper 64-bits.
15241 (print_operand): Add support for PTImode.
15242 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
15243 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
15244 registers, but don't have arithmetic support.
15245 (rs6000_memory_move_cost): Add test for VSX.
15246 (rs6000_opt_masks): Add -mvsx-timode.
15248 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
15251 (VSr): Use wt constraint for TImode.
15252 (VSv): Drop TImode support.
15253 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
15254 (vsx_movti_64bit): Likewise.
15255 (vsx_movti_32bit): Likewise.
15256 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
15257 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
15258 one '?' on the appropriate output constraint. Do not allow TImode
15259 logical operations on 32-bit systems.
15260 (vsx_ior<mode>3): Likewise.
15261 (vsx_xor<mode>3): Likewise.
15262 (vsx_one_cmpl<mode>2): Likewise.
15263 (vsx_nor<mode>3): Likewise.
15264 (vsx_andc<mode>3): Likewise.
15265 (vsx_concat_<mode>): Likewise.
15266 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
15268 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
15269 OPTION_MASK_VSX_TIMODE.
15270 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
15271 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
15273 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
15274 (TI2 iterator): New iterator for TImode, PTImode.
15275 (wd mode attribute): Add values for vector types.
15276 (movti_string): Replace TI move operations with operations for TImode
15277 and PTImode. Add support for TImode being allowed in VSX registers.
15278 (mov<mode>_string, TImode/PTImode): Likewise.
15279 (movti_ppc64): Likewise.
15280 (mov<mode>_ppc64, TImode/PTImode): Likewise.
15281 (TI mode splitters): Likewise.
15283 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
15286 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
15288 PR tree-optimization/56355
15289 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
15290 Also handle integers with undefined overflow.
15292 2013-03-20 Catherine Moore <clm@codesourcery.com>
15293 Maciej W. Rozycki <macro@codesourcery.com>
15294 Tom de Vries <tom@codesourcery.com>
15295 Nathan Sidwell <nathan@codesourcery.com>
15296 Iain Sandoe <iain@codesourcery.com>
15297 Nathan Froyd <froydnj@codesourcery.com>
15298 Chao-ying Fu <fu@mips.com>
15300 * doc/extend.texi: (micromips, nomicromips, nocompression):
15301 Document new function attributes.
15302 * doc/invoke.texi (minterlink-compressed, mmicromips,
15303 m14k, m14ke, m14kec): Document new options.
15304 (minterlink-mips16): Update documentation.
15305 * doc/md.texi (ZC, ZD): Document new constraints.
15306 * configure.ac (gcc_cv_as_micromips): Check if linker
15307 supports the .set micromips directive.
15308 * configure: Regenerate.
15309 * config.in: Regenerate.
15310 * config/mips/mips-tables.opt: Regenerate.
15311 * config/mips/micromips.md: New file.
15312 * constraints.md (ZC, ZD): New constraints.
15313 * config/mips/predicates.md (movep_src_register): New predicate.
15314 (movep_src_operand): New predicate.
15315 (non_volatile_mem_operand): New predicate.
15316 * config/mips/mips.md (multimem): New type.
15317 (length): Differentiate between 17-bit and 18-bit branch offsets.
15318 (MOVEP1, MOVEP2): New mode iterator.
15319 (mov_<load>l): Use ZC constraint.
15320 (mov_<load>r): Likewise.
15321 (mov_<store>l): Likewise.
15322 (mov_<store>r): Likewise.
15323 (*branch_equality<mode>_inverted): Add microMIPS support.
15324 (*branch_equality<mode>): Likewise.
15325 (*jump_absolute): Likewise.
15326 (indirect_jump_<mode>): Likewise.
15327 (tablejump_<mode>): Likewise.
15328 (<optab>_internal): Likewise.
15329 (sibcall_internal): Likewise.
15330 (sibcall_value_internal): Likewise.
15331 (prefetch): Use constraint ZD.
15332 * config/mips/mips.opt (minterlink-compressed): New option.
15333 (minterlink-mips16): Now an alias for minterlink-compressed.
15334 (mmicromips): New option.
15335 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15336 (compare_and_swap_12): Likewise.
15337 (sync_add<mode>): Likewise.
15338 (sync_<optab>_12): Likewise.
15339 (sync_old_<optab>_12): Likewise.
15340 (sync_new_<optab>_12): Likewise.
15341 (sync_nand_12): Likewise.
15342 (sync_old_nand_12): Likewise.
15343 (sync_new_nand_12): Likewise.
15344 (sync_sub<mode>): Likewise.
15345 (sync_old_add<mode>): Likewise.
15346 (sync_old_sub<mode>): Likewise.
15347 (sync_new_add<mode>): Likewise.
15348 (sync_new_sub<mode>): Likewise.
15349 (sync_<optab><mode>): Likewise.
15350 (sync_old_<optab><mode>): Likewise.
15351 (sync_new_<optab><mode>): Likewise.
15352 (sync_nand<mode>): Likewise.
15353 (sync_old_nand<mode>): Likewise.
15354 (sync_new_nand<mode>): Likewise.
15355 (sync_lock_test_and_set<mode>): Likewise.
15356 (test_and_set_12): Likewise.
15357 (atomic_compare_and_swap<mode>): Likewise.
15358 (atomic_exchange<mode>_llsc): Likewise.
15359 (atomic_fetch_add<mode>_llsc): Likewise.
15360 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15361 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15362 (umips_save_restore_pattern_p): Likewise.
15363 (umips_load_store_pair_p): Likewise.
15364 (umips_output_load_store_pair): Likewise.
15365 (umips_movep_target_p): Likewise.
15366 (umips_12bit_offset_address_p): Likewise.
15367 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15368 (mips_base_mips16): Rename this...
15369 (mips_base_compression_flags): ...to this. Update all uses.
15370 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15371 (mips_mips16_decl_p): Delete.
15372 (mips_nomips16_decl_p): Delete.
15373 (mips_get_compress_on_flags): New function.
15374 (mips_get_compress_off_flags): New function.
15375 (mips_get_compress_mode): New function.
15376 (mips_get_compress_on_name): New function.
15377 (mips_get_compress_off_name): New function.
15378 (mips_insert_attributes): Support multiple compression types.
15379 (mips_merge_decl_attributes): Likewise.
15380 (umips_12bit_offset_address_p): New function.
15381 (mips_start_function_definition): Emit .set micromips directive.
15382 (mips_call_may_need_jalx_p): New function.
15383 (mips_function_ok_for_sibcall): Add microMIPS support.
15384 (mips_print_operand_punctuation): Support short delay slots and
15386 (umips_swm_mask, umips_swm_encoding): New.
15387 (umips_build_save_restore): New function.
15388 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15389 (was_mips16_p): Remove.
15390 (old_compression_mode): New.
15391 (mips_set_compression_mode): New function.
15392 (mips_set_current_function): Add microMIPS support.
15393 (mips_option_override): Likewise.
15394 (umips_save_restore_pattern_p): New function.
15395 (umips_output_save_restore): New function.
15396 (umips_load_store_pair_p_1): New function.
15397 (umips_load_store_pair_p): New function.
15398 (umips_output_load_store_pair_1): New function.
15399 (umips_output_load_store_pair): New function.
15400 (umips_movep_target_p) New function.
15401 (mips_prepare_pch_save): Add microMIPS support.
15402 * config/mips/mips.h (TARGET_COMPRESSION): New.
15403 (TARGET_CPU_CPP_BUILTINS): Update macro
15404 to use new compression flags and to support microMIPS.
15405 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15406 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15407 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15408 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15409 (ASM_SPEC): Support mmicromips and mno-micromips.
15410 (M16STORE_REG_P): New macro.
15411 (MIPS_CALL): Support TARGET_MICROMIPS.
15412 (MICROMIPS_J): New macro.
15413 (mips_base_mips16): Rename this...
15414 (mips_base_compression_flags): ...to this.
15415 (UMIPS_12BIT_OFFSET_P): New macro.
15416 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15417 (MULTILIB_DIRNAMES): Likewise.
15418 2013-03-20 Richard Biener <rguenther@suse.de>
15420 PR tree-optimization/56661
15421 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
15422 the result does not have to be distinct.
15424 2013-03-20 Richard Biener <rguenther@suse.de>
15426 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
15429 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15430 Steven Bosscher <steven@gcc.gnu.org>
15432 PR rtl-optimization/56605
15433 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
15435 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
15438 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
15439 that require movd instead of movq.
15441 2013-03-20 Richard Biener <rguenther@suse.de>
15443 * tree-ssa-structalias.c (struct variable_info): Add pointer
15444 to the first field of an aggregate with sub-vars. Make
15445 this and the pointer to the next subfield its ID.
15446 (vi_next): New function.
15447 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
15448 storedanything_id, integer_id): Increment by one.
15449 (new_var_info, get_call_vi, lookup_call_clobber_vi,
15450 get_call_clobber_vi): Adjust.
15451 (solution_set_expand): Simplify and speedup.
15452 (solution_set_add): Inline into ...
15453 (set_union_with_increment): ... this. Adjust accordingly.
15454 (do_sd_constraint): Likewise.
15455 (do_ds_constraint): Likewise.
15456 (do_complex_constraint): Simplify.
15457 (build_pred_graph): Adjust.
15458 (solve_graph): Likewise. Simplify and speedup.
15459 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
15460 get_constraint_for_component_ref, get_constraint_for_1,
15461 first_vi_for_offset, first_or_preceding_vi_for_offset,
15462 create_function_info_for, create_variable_info_for_1,
15463 create_variable_info_for, intra_create_variable_infos): Adjust.
15464 (init_base_vars): Push NULL for ID zero.
15465 (compute_points_to_sets): Adjust.
15467 2013-03-20 Richard Biener <rguenther@suse.de>
15469 * cfgloop.c (verify_loop_structure): Streamline and avoid
15470 ICEing on corrupt loop tree.
15471 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
15474 2013-03-20 Richard Biener <rguenther@suse.de>
15476 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
15477 check whether an SSA update is needed.
15479 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
15481 * config/mips/constraints.md (T): Rename to...
15485 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
15486 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
15488 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15490 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
15491 (*subsi3_carryin_uxtw): Likewise.
15493 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15495 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
15496 (*rorsi3_insn_uxtw): Likewise.
15498 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15500 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
15501 (*extrsi5_insn_uxtw): Likewise.
15503 2013-03-19 Richard Biener <rguenther@suse.de>
15505 PR tree-optimization/56273
15506 * passes.c (init_optimization_passes): Move second VRP after DOM.
15508 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
15510 * config/i386/i386.md (*movti_internal): Merge from
15511 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
15512 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
15513 nox64 isa attributes.
15515 2013-03-18 Richard Biener <rguenther@suse.de>
15517 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
15519 (merge_node_constraints): Likewise.
15520 (build_succ_graph): Likewise.
15521 (valid_graph_edge): Inline into single caller.
15522 (unify_nodes): Likewise. Use bitmap_set_bit return value
15524 (scc_visit): Fix formatting and variable use.
15525 (do_sd_constraint): Use gcc_checking_assert.
15526 (do_ds_constraint): Likewise.
15527 (do_complex_constraint): Likewise.
15528 (condense_visit): Likewise. Cleanup.
15529 (dump_pred_graph): New function.
15530 (perform_var_substitution): Dump the pred-graph before
15531 variable substitution.
15532 (find_equivalent_node): Use gcc_checking_assert.
15533 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15535 2013-03-18 Richard Biener <rguenther@suse.de>
15537 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15538 Remove cond_expr_stmt_list argument and do not gimplify the
15540 (vect_loop_versioning): Adjust.
15541 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15542 Cleanup to use less temporaries.
15543 (vect_create_data_ref_ptr): Cleanup.
15545 2013-03-18 Jakub Jelinek <jakub@redhat.com>
15547 PR tree-optimization/56635
15548 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15549 require types_compatible_p types.
15551 2013-03-18 Nick Clifton <nickc@redhat.com>
15553 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15554 spurious backslash.
15556 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15557 Add missing line to comment describing function.
15559 2013-03-18 Richard Biener <rguenther@suse.de>
15561 PR tree-optimization/56210
15562 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15563 Handle string / character search functions.
15564 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15566 2013-03-18 Richard Biener <rguenther@suse.de>
15568 PR middle-end/56483
15569 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15570 and implement properly.
15571 * gimple.h (gimple_cond_single_var_p): Remove.
15573 2013-03-18 Richard Biener <rguenther@suse.de>
15575 * tree-data-ref.h (find_data_references_in_loop): Declare.
15576 * tree-data-ref.c (get_references_in_stmt): Use a stack
15577 vector pre-allocated in the callers.
15578 (find_data_references_in_stmt): Adjust.
15579 (graphite_find_data_references_in_stmt): Likewise.
15580 (create_rdg_vertices): Likewise.
15581 (find_data_references_in_loop): Export.
15582 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15583 Compute dependences here...
15584 (vect_analyze_data_refs): ...not here. When we encounter
15585 a non-vectorizable data reference in basic-block vectorization
15586 truncate the data reference vector. Do not bother to
15587 fixup data-dependence information for gather loads.
15588 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15589 of data references, as reported.
15591 2013-03-18 Richard Biener <rguenther@suse.de>
15593 PR tree-optimization/3713
15594 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
15595 has_constants and expr.
15596 (stmt_has_constants): Properly valueize SSA names when deciding
15597 whether the stmt has constants.
15599 2013-03-18 Richard Biener <rguenther@suse.de>
15601 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15602 whole function when there is nothing to do.
15603 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15604 * tree-vectorizer.c (vectorize_loops): Update virtual and
15605 loop-closed SSA once.
15606 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15608 2013-03-18 Richard Biener <rguenther@suse.de>
15610 PR middle-end/56113
15611 * domwalk.c (bb_postorder): New global static.
15612 (cmp_bb_postorder): New function.
15613 (walk_dominator_tree): Replace scheme imposing an order for
15614 visiting dominator sons by one sorting them at the time they
15615 are pushed on the stack.
15617 2013-03-18 Richard Biener <rguenther@suse.de>
15619 PR tree-optimization/39326
15620 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15621 (struct mem_ref): Replace mem member with ao_ref typed member.
15622 (MEM_ANALYZABLE): Adjust.
15623 (memref_eq): Likewise.
15624 (mem_ref_alloc): Likewise.
15625 (gather_mem_refs_stmt): Likewise.
15626 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15627 (execute_sm_if_changed_flag_set): Adjust.
15628 (execute_sm): Likewise.
15629 (ref_always_accessed_p): Likewise.
15630 (refs_independent_p): Likewise.
15631 (can_sm_ref_p): Likewise.
15633 2013-03-18 Jakub Jelinek <jakub@redhat.com>
15636 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15637 return 1 even for !unsignedp.
15639 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15641 * config/i386/i386.md (isa): Add x64 and nox64.
15642 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15643 (*pushtf): Enable *roF alternative for x64 isa only.
15644 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
15645 mode attribute of integer alternatives to DImode for TARGET_64BIT.
15646 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
15647 (*movtf_internal): Merge from *movtf_internal_rex64 and
15648 *movtf_internal_sse. Use x64 and nox64 isa attributes.
15649 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
15650 nox64 isa attributes.
15651 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
15652 nox64 isa attributes.
15653 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15655 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15657 * config/alpha/alpha.c (TARGET_LRA_P): New define.
15659 2013-03-17 Jakub Jelinek <jakub@redhat.com>
15662 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15663 class names. Remove trailing comma after "ALL_REGS".
15665 2013-03-16 Jan Hubicka <jh@suse.cz>
15667 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15668 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15669 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15670 of cgraph_get_create_node.
15671 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15673 2013-03-16 Jason Merrill <jason@redhat.com>
15676 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15677 with DW_AT_default_value.
15679 2013-03-16 Jakub Jelinek <jakub@redhat.com>
15681 * BASE-VER: Set to 4.9.0.
15683 2013-03-14 Andi Kleen <ak@linux.intel.com>
15686 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15687 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15688 Document _x* TSX intrinsics.
15690 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
15691 David Holsgrove <david.holsgrove@xilinx.com>
15693 * configure.ac: Add MicroBlaze TLS support detection.
15694 * configure: Regenerate.
15695 * config/microblaze/microblaze-protos.h
15696 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15697 symbol_mentioned_p, label_mentioned_p): Add prototypes.
15698 * config/microblaze/microblaze.c (microblaze_address_type): Add
15699 ADDRESS_TLS and tls_reloc address types.
15700 (microblaze_address_info): Add tls_reloc.
15701 (TARGET_HAVE_TLS): Define.
15702 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15703 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15704 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15705 load_tls_operand, microblaze_call_tls_get_addr,
15706 microblaze_legitimize_tls_address): New functions.
15707 (microblaze_classify_unspec): Handle UNSPEC_TLS.
15708 (get_base_reg): Use microblaze_tls_symbol_p.
15709 (microblaze_classify_address): Handle TLS.
15710 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15711 label_mentioned_p and microblaze_tls_referenced_p.
15712 (microblaze_legitimize_address): Handle TLS.
15713 (microblaze_address_insns): Handle ADDRESS_TLS.
15714 (pic_address_needs_scratch): Handle TLS.
15715 (print_operand_address): Handle TLS.
15716 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15717 (microblaze_expand_move): Handle TLS.
15718 (microblaze_legitimate_constant_p): Check
15719 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15720 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15721 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15722 (PIC_OFFSET_TABLE_REGNUM): Set.
15723 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15724 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15725 (addsi3, movsi_internal2, movdf_internal): Update constraints
15726 * config/microblaze/predicates.md (arith_plus_operand): Define
15727 (move_operand): Redefine as move_src_operand,
15728 check microblaze_tls_referenced_p.
15730 2013-03-14 Ian Bolton <ian.bolton@arm.com>
15732 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15733 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15735 2013-03-14 Ian Bolton <ian.bolton@arm.com>
15737 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15740 2013-03-14 Jakub Jelinek <jakub@redhat.com>
15742 PR tree-optimization/53265
15743 * common.opt (Waggressive-loop-optimizations): New option.
15744 * tree-ssa-loop-niter.c: Include tree-pass.h.
15745 (do_warn_aggressive_loop_optimizations): New function.
15746 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
15747 if number_of_latch_executions returned constant.
15748 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15749 early. If number_of_latch_executions returned constant, set
15750 nb_iterations_upper_bound back to it.
15751 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15753 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15754 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15756 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15757 (MULTILIB_OSDIRNAMES): Set.
15758 * genmultilib: If defaultosdirname doesn't start with :: , set
15759 defaultosdirname2 instead, clear it and emit two . multilib_raw
15760 entries instead of just one.
15762 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
15764 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15765 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15766 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15767 (SUBTARGET_OVERRIDE_OPTIONS): New.
15769 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
15772 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15773 (musermode): Convert to Var(TARGET_USERMODE).
15774 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15775 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15776 * config/sh/sh.c (sh_option_override): Use
15777 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15778 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15780 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15782 (udivsi3_i4_single, divsi3_i4_single): Use
15783 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15785 2013-03-13 Dave Korn <dave.korn.cygwin@....>
15787 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15790 2013-03-13 Richard Biener <rguenther@suse.de>
15792 PR tree-optimization/56608
15793 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15794 calls when vectorizing basic-blocks.
15796 2013-03-13 Jakub Jelinek <jakub@redhat.com>
15799 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15802 2013-03-12 Jakub Jelinek <jakub@redhat.com>
15804 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15806 2013-03-11 Jan Hubicka <jh@suse.cz>
15809 * lto-streamer-out.c (output_symbol_p): Skip references from
15810 constructors of external variables.
15812 2013-03-11 Jan Hubicka <jh@suse.cz>
15814 PR middle-end/56571
15815 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15817 * emit-rtl.c (verify_rtx_sharing): Likewise.
15818 (copy_insn_1): Likewise.
15819 * rtl.c (copy_rtx): Likewise.
15821 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
15824 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15825 output_operand_lossage message.
15827 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
15830 * arm.c (shift_op): Validate RTL pattern on the fly.
15831 (arm_print_operand, case 'S'): Don't use shift_operator to validate
15834 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
15837 * config/pa/pa.md (call_value): Check for calls to powf and direct to
15838 new call patterns that clobber %fr12.
15839 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15840 split and postreload patterns.
15841 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15842 registers %fr12 and %fr12R as call used.
15844 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
15846 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15847 (canon_address, record_store, replace_read, check_mem_read_rtx,
15848 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15849 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15850 rest_of_handle_dse): Likewise.
15852 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
15854 PR middle-end/56524
15855 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15857 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15858 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15859 (save_optabs_if_changed): Replace with...
15860 (init_tree_optimization_optabs): ...this.
15861 * optabs.c (save_optabs_if_changed): Rename to...
15862 (init_tree_optimization_optabs): ...this. Take the optimization node
15863 as argument. Do nothing if the base optabs are already correct.
15864 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15865 to recompute optabs.
15866 * function.h (function): Remove optabs field.
15867 * function.c (invoke_set_current_function_hook): Call
15868 init_tree_optimization_optabs. Use the result to initialize
15871 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
15873 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15874 if GTMA_HAS_NO_INSTRUMENTATION.
15875 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15876 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15877 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15878 * gimple-pretty-print.c (dump_gimple_transaction): Handle
15879 GTMA_HAS_NO_INSTRUMENTATION.
15881 2013-03-08 Jakub Jelinek <jakub@redhat.com>
15883 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15886 2013-03-08 Marek Polacek <polacek@redhat.com>
15887 Jakub Jelinek <jakub@redhat.com>
15889 PR tree-optimization/56478
15890 * predict.c (is_comparison_with_loop_invariant_p): Change the
15891 type of loop_step to tree.
15892 (predict_loops): Adjust.
15893 (predict_iv_comparison): Perform the computations on double_ints.
15895 2013-03-08 Richard Biener <rguenther@suse.de>
15897 PR tree-optimization/56570
15898 * tree-cfg.c (verify_expr_location_1): Verify locations for
15900 * tree-sra.c (create_access_replacement): Strip locations
15901 from DECL_DEBUG_EXPRs.
15903 2013-03-08 Richard Biener <rguenther@suse.de>
15905 * tree-inline.c (expand_call_inline): Do not associate
15906 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15907 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15909 2013-03-08 Richard Biener <rguenther@suse.de>
15911 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15912 or block changes with -Og. Fix for location / block encoding
15913 changes and PHI arguments with locations.
15915 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
15917 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15919 (struct output_info): Likewise.
15920 (register_overhead): Remove bad gcc_assert.
15921 (bitmap_find_bit): If there is only a single bitmap element, do not
15922 count a miss as a search.
15923 (print_statistics): Update for counter type changes.
15924 (dump_bitmap_statistics): Likewise. Print headers such that they
15925 are properly lined up with the printed counters.
15927 2013-03-07 Jakub Jelinek <jakub@redhat.com>
15929 PR tree-optimization/56559
15930 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15931 check that it has only a single use.
15933 2013-03-07 Richard Biener <rguenther@suse.de>
15935 * doc/invoke.texi (fwhole-program): Discourage use in combination
15938 2013-03-06 Jakub Jelinek <jakub@redhat.com>
15940 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15942 PR tree-optimization/56539
15943 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15944 instead of GSI_CONTINUE_LINKING as last argument to
15945 force_gimple_operand_gsi. Adjust function comment.
15947 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15950 PR middle-end/56548
15951 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15952 promoted mode, convert the result back to the original mode.
15954 2013-03-06 Richard Biener <rguenther@suse.de>
15956 PR middle-end/56294
15957 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15958 (insert_updated_phi_nodes_compare_uids): New function.
15959 (update_ssa): Sort symbols_to_rename after UID before
15960 traversing it to insert PHI nodes.
15962 2013-03-06 Richard Biener <rguenther@suse.de>
15964 PR middle-end/50494
15965 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15966 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15969 2013-02-13 Richard Biener <rguenther@suse.de>
15972 * varasm.c (output_constant_def_1): Get the decl representing
15973 the constant as argument.
15974 (output_constant_def): Wrap output_constant_def_1.
15975 (make_decl_rtl): Use output_constant_def_1 with the decl
15976 representing the constant.
15977 (build_constant_desc): Optionally re-use a decl already
15978 representing the constant.
15979 (tree_output_constant_def): Adjust.
15981 2013-03-06 Joey Ye <joey.ye@arm.com>
15984 * gcc.c (convert_white_space): New function.
15985 (main): Handles white space in function name.
15987 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
15990 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15991 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
15992 to SH_DIV_CALL_TABLE for TARGET_SH2.
15993 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15995 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15996 call-table options.
15998 2013-03-05 Sterling Augustine <saugustine@google.com>
15999 Cary Coutant <ccoutant@google.com>
16002 * dwarf2out.c (resolve_addr): Don't call
16003 remove_loc_list_addr_table_entries a second time for the same
16006 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16009 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
16010 (avoid_complex_debug_insns): New function.
16011 (expand_debug_locations): Call it.
16013 PR rtl-optimization/56484
16014 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
16015 lifetimes of hard registers on small register class machines.
16017 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
16019 * config/microblaze/microblaze-protos.h: Rename
16020 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
16021 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
16023 (microblaze_fast_interrupt_function_p): New function.
16024 (microblaze_is_interrupt_handler): Rename to
16025 microblaze_is_interrupt_variant and add fast_interrupt check.
16026 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
16027 (save_restore_insns): Likewise.
16028 (compute_frame_size): Likewise.
16029 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
16030 (microblaze_globalize_label): Likewise.
16031 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
16032 * config/microblaze/microblaze.md: Use wrapper
16033 microblaze_is_interrupt_variant.
16035 2013-03-05 Kai Tietz <ktietz@redhat.com>
16037 * sdbout.c (sdbout_one_type): Switch to current function's section
16038 supporting cold/hot.
16040 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
16042 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
16045 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16047 PR middle-end/56461
16048 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
16049 if VALGRIND_GET_VBITS is defined, temporarily make object
16050 memory all defined, and restore previous valgrind addressability
16051 and definability afterwards. Free this_object at the end.
16053 PR middle-end/56461
16054 * lra.c (lra): Call lra_clear_live_ranges if live_p,
16055 right before calling lra_create_live_ranges, also call it
16056 when clearing live_p. Only call lra_clear_live_ranges
16057 at the end if live_p.
16059 PR middle-end/56461
16060 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
16062 2013-03-05 Richard Biener <rguenther@suse.de>
16064 PR tree-optimization/56521
16065 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
16068 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
16071 * except.h (remove_unreachable_eh_regions): New prototype.
16072 * except.c (remove_eh_handler_splicer): New function, split out
16073 of remove_eh_handler.
16074 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
16075 warning about running it on many EH regions one at a time.
16076 (remove_unreachable_eh_regions_worker): New function, walk the
16077 EH tree in depth-first order and remove non-marked regions.
16078 (remove_unreachable_eh_regions): New function.
16079 * tree-eh.c (mark_reachable_handlers): New function, split out
16080 from remove_unreachable_handlers.
16081 (remove_unreachable_handlers): Use mark_reachable_handlers and
16082 remove_unreachable_eh_regions.
16083 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
16084 and remove_unreachable_eh_regions.
16086 2013-03-05 Richard Biener <rguenther@suse.de>
16088 PR middle-end/56525
16089 * loop-init.c (fix_loop_structure): Remove loops in two stages,
16090 not freeing them until the end.
16092 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
16094 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
16096 2013-03-05 Richard Biener <rguenther@suse.de>
16098 PR tree-optimization/56270
16099 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
16100 of loads after scheduling an SLP instance.
16102 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16104 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
16106 (check_gcc_parallelize): Run guality.exp as a separate job from
16107 vect.exp with unsorted.exp and $(dg_target_exps) separately from
16108 struct-layout-1.exp with stackalign.exp.
16110 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
16112 PR middle-end/56461
16113 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
16114 load_index sbitmap even if some bit in it isn't set.
16116 PR middle-end/56461
16117 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
16118 (discover_iteration_bound_by_body_walk): Change queues to
16119 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
16120 spelling in comment. Call safe_push on queues[bound_index] directly.
16121 Release queues[queue_index] in every iteration unconditionally.
16122 Release bounds vector.
16124 PR middle-end/56461
16125 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
16126 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
16127 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
16130 2013-03-05 Richard Biener <rguenther@suse.de>
16133 * tree-inline.c (remap_blocks_to_null): New function.
16134 (expand_call_inline): When expanding a call stmt without
16135 an associated block inline remap all callee blocks to NULL.
16137 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16139 PR rtl-optimization/56494
16140 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
16141 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
16142 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
16144 PR middle-end/56461
16145 * sel-sched-ir.c (free_sched_pools): Release
16146 succs_info_pool.stack[succs_info_pool.max_top] vectors too
16147 if succs_info_pool.max_top isn't -1.
16150 * opts.c (opts_obstack, opts_concat): Moved to...
16151 * opts-common.c (opts_obstack, opts_concat): ... here.
16153 2013-03-04 Jakub Jelinek <jakub@redhat.com>
16155 PR middle-end/56461
16156 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
16158 2013-03-04 Martin Jambor <mjambor@suse.cz>
16160 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
16161 all appropriate places.
16163 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
16165 PR tree-optimization/56424
16166 * ipa-split.c (split_function): Do not set the RSO flag if result is
16167 not by reference and its type is a register type.
16169 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
16171 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
16172 (microblaze_legitimate_pic_operand): Likewise
16173 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
16174 new function microblaze_legitimate_pic_operand
16175 * config/microblaze/microblaze-protos.h
16176 (microblaze_legitimate_pic_operand): Declare.
16178 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16180 * config/microblaze/predicates.md (call_insn_simple_operand):
16181 New predicate for supported rtx code types.
16182 * config/microblaze/microblaze.md (call_internal1): Use
16183 call_insn_simple_operand predicate.
16185 2013-03-04 Jakub Jelinek <jakub@redhat.com>
16187 PR middle-end/56461
16188 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
16189 partitions.ordered_remove.
16191 PR middle-end/56461
16192 * tree-vect-stmts.c (vectorizable_conversion): Don't call
16193 vec_oprnds0.create (1) for modifier == NONE.
16195 PR middle-end/56461
16196 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
16197 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
16198 vec_oprnds1 right before pushing anything to it for
16201 PR middle-end/56461
16202 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
16203 set nbbs to 0 instead of having separate code path.
16204 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
16205 instead of false as last argument if returning NULL.
16207 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
16209 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
16210 the attribute is now called "target" instead of "option".
16211 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
16212 * doc/tm.texi.in (Target Attributes): Likewise document the correct
16213 attribute/pragma name for TARGET_OPTION_VALID_P and
16214 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
16215 * doc/tm.texi: Regenerated.
16217 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
16219 * config/microblaze/microblaze.c:
16220 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
16221 * config/microblaze/microblaze.h: Add -mxl-reorder to
16223 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
16224 instructions emitted if TARGET_REORDER.
16225 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
16226 or 0 for -m/-mno case, but initialises as 2 to detect default use case
16229 2013-03-01 Xinliang David Li <davidxl@google.com>
16231 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
16234 2013-03-01 Jakub Jelinek <jakub@redhat.com>
16236 PR middle-end/56461
16237 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
16238 vector even when returning true. Fix up function comment formatting.
16240 PR middle-end/56461
16241 * ira-build.c (ira_loop_nodes_count): New variable.
16242 (create_loop_tree_nodes): Initialize it.
16243 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
16245 PR middle-end/56461
16246 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
16247 method on dr_chain and result_chain.
16248 * tree-vect-stmts.c (vectorizable_store): Only call
16249 result_chain.create if j == 0.
16251 PR middle-end/56461
16252 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
16253 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
16254 before overwriting it.
16256 2013-03-01 Tobias Burnus <burnus@net-b.de>
16258 * doc/extended.texi (C Extensions): Change order in @menu
16260 (Other MIPS Built-in Functions): Move last MIPS entry before
16261 "picoChip Built-in Functions".
16262 (SH Built-in Functions): Move after RX Built-in Functions.
16263 * doc/gcc.texi (Introduction): Change order in @menu
16265 * doc/md.texi (Constraints): Ditto.
16266 * gty.texi (Type Information): Ditto.
16267 (User-provided marking routines for template types): Make
16269 * doc/invoke.texi (AArch64 Options): Move before
16270 "Adapteva Epiphany Options".
16272 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
16273 Jakub Jelinek <jakub@redhat.com>
16276 * asan.c (gate_asan): Lookup no_sanitize_address instead of
16277 no_address_safety_analysis attribute.
16278 * doc/extend.texi (no_address_safety_attribute): Rename to
16279 no_sanitize_address attribute, mention no_address_safety_analysis
16280 attribute as deprecated alias.
16282 2013-02-28 Jakub Jelinek <jakub@redhat.com>
16284 PR middle-end/56461
16285 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
16286 type to vec<vec<tree> > *.
16287 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
16288 to be vec<tree> instead of vec<tree> *, set vec_defs
16289 to vNULL and call vec_defs.create (number_of_vects), adjust other
16291 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
16292 vectorizable_condition): Adjust vect_get_slp_defs callers.
16294 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16296 * config/aarch64/aarch64.c
16297 (aarch64_float_const_representable): Remove unused variable.
16299 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16301 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
16303 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16305 * config/aarch64/aarch64-builtins.c
16306 (aarch64_init_simd_builtins): Make static.
16308 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16310 * config/aarch64/aarch64.c
16311 (aarch64_simd_make_constant): Make static.
16313 2013-02-28 Martin Jambor <mjambor@suse.cz>
16315 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
16316 with no initialization to the RHS of debug statements.
16318 2013-02-28 Martin Jambor <mjambor@suse.cz>
16320 PR tree-optimization/56294
16321 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
16323 (get_access_replacement): Do not call create_access_replacement.
16324 Assert a replacement exists.
16325 (get_repl_default_def_ssa_name): Create the replacement declaration
16328 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16330 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
16331 final_end_function.
16333 2013-02-28 Marek Polacek <polacek@redhat.com>
16335 PR rtl-optimization/56466
16336 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
16337 if we're changing a loop.
16338 (peel_loops_completely): Likewise.
16340 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
16343 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
16345 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16348 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
16349 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
16350 INTX_FTYPE_FX, FX_FTYPE_INTX.
16351 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
16353 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16355 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
16356 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
16357 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
16358 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
16359 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
16360 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
16361 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
16362 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
16363 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
16364 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
16365 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
16366 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
16367 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
16368 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
16369 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
16370 (avrxmega6): Increase max flash segments from 5 to 6.
16371 * config/avr/t-multilib: Regenerate.
16372 * config/avr/avr-tables.opt: Regenerate.
16373 * doc/avr-mmcu.texi: Regenerate.
16375 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16377 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
16378 (avr_device_to_arch): Rename to avr_device_to_ld.
16379 (avr_device_to_as): New prototype.
16380 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
16381 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
16382 * config/avr/driver-avr.c (avr_device_to_as): New.
16383 (avr_device_to_arch): Rename to avr_device_to_ld.
16385 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16387 PR middle-end/56461
16388 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
16389 method on dr_chain and result_chain.
16391 PR middle-end/56461
16392 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
16393 pointer_set_destroy on not_executed_last_iteration.
16395 PR middle-end/56461
16396 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
16398 PR middle-end/56461
16399 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
16400 FOR_EACH_DEFINED_FUNCTION when freeing state.
16402 PR middle-end/56461
16403 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
16405 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
16408 PR middle-end/56461
16409 * ipa-cp.c (decide_whether_version_node): Call vec_free on
16410 known_aggs[i].items and release known_aggs vector.
16412 PR middle-end/56461
16413 * ipa-reference.c (propagate): Free node_info even for alias nodes.
16415 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16417 * config/microblaze/microblaze.c (microblaze_emit_compare):
16418 Use xor for EQ/NE comparisions.
16419 * config/microblaze/microblaze.md (cstoresf4): Add constraints
16420 (cbranchsf4): Adjust operator to comparison_operator.
16422 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16424 PR middle-end/56461
16425 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
16427 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
16428 vec_safe_push, always update *slot.
16429 (redirect_edge_var_map_clear): Use vec_free.
16430 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
16431 (free_var_map_entry): Use vec_free.
16432 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
16433 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
16435 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
16437 PR middle-end/45472
16438 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
16439 when the may_trap_p bit of the exprs being merged differs.
16440 Reorder tests for speculativeness in the logical and operator.
16442 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16444 * incpath.c (add_standard_paths): Use reconcat instead of concat
16445 where appropriate and avoid leaking memory.
16447 * opts.h: Include obstack.h.
16448 (opts_concat): New prototype.
16449 (opts_obstack): New declaration.
16450 * opts.c (opts_concat): New function.
16451 (opts_obstack): New variable.
16452 (init_options_struct): Call gcc_init_obstack on opts_obstack.
16453 (finish_options): Use opts_concat instead of concat
16454 and XOBNEWVEC instead of XNEWVEC.
16455 * opts-common.c (generate_canonical_option, decode_cmdline_option,
16456 generate_option): Likewise.
16457 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
16458 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
16461 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
16462 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
16464 2013-02-26 Jakub Jelinek <jakub@redhat.com>
16466 PR middle-end/56461
16467 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
16469 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
16471 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
16472 (arm_block_move_unaligned_straight): Likewise.
16473 (arm_adjust_block_mem): Likewise.
16475 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
16478 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
16479 temp, cond and label.
16480 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
16483 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16484 * config/c6x/c6x.h (dbx_register_map): Update declaration.
16487 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16488 of prologue/epilogue functions.
16491 * config/tilegx/tilegx.c (tilegx_expand_prologue):
16492 Remove unused variable cfa_offset.
16493 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16496 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16497 type promotion to unsigned.
16500 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
16501 for HOST_WIDE_INT of 32 bit / same size as int.
16502 (arm_block_move_unaligned_straight): Likewise.
16503 (arm_adjust_block_mem): Likewise.
16506 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
16509 2013-02-26 Marek Polacek <polacek@redhat.com>
16511 PR tree-optimization/56426
16512 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
16514 2013-02-26 Richard Biener <rguenther@suse.de>
16517 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
16518 unused variable loops.
16520 2013-02-26 Jakub Jelinek <jakub@redhat.com>
16522 PR tree-optimization/56448
16523 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16524 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16525 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16526 later operands of the references, or even first operand for
16527 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16529 PR tree-optimization/56443
16530 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16531 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16532 to type_for_mode langhook.
16534 2013-02-25 Matt Turner <mattst88@gmail.com>
16536 * doc/invoke.texi: Document r4700.
16538 2013-02-25 Richard Biener <rguenther@suse.de>
16540 PR tree-optimization/56175
16541 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16543 (simplify_bitwise_binary): ... here. Also guard the conversion
16544 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16546 2013-02-25 Catherine Moore <clm@codesourcery.com>
16549 2013-02-24 Catherine Moore <clm@codesourcery.com>
16550 Maciej W. Rozycki <macro@codesourcery.com>
16551 Tom de Vries <tom@codesourcery.com>
16552 Nathan Sidwell <nathan@codesourcery.com>
16553 Iain Sandoe <iain@codesourcery.com>
16554 Nathan Froyd <froydnj@codesourcery.com>
16555 Chao-ying Fu <fu@mips.com>
16557 * doc/extend.texi: (micromips, nomicromips, nocompression):
16558 Document new function attributes.
16559 * doc/invoke.texi (minterlink-compressed, mmicromips,
16560 m14k, m14ke, m14kec): Document new options.
16561 (minterlink-mips16): Update documentation.
16562 * doc/md.texi (ZC, ZD): Document new constraints.
16563 * configure.ac (gcc_cv_as_micromips): Check if linker
16564 supports the .set micromips directive.
16565 * configure: Regenerate.
16566 * config.in: Regenerate.
16567 * config/mips/mips-tables.opt: Regenerate.
16568 * config/mips/micromips.md: New file.
16569 * constraints.md (ZC, AD): New constraints.
16570 * config/mips/predicates.md (movep_src_register): New predicate.
16571 (movep_src_operand): New predicate.
16572 (non_volatile_mem_operand): New predicate.
16573 * config/mips/mips.md (multimem): New type.
16574 (length): Differentiate between 17-bit and 18-bit branch offsets.
16575 (MOVEP1, MOVEP2): New mode iterator.
16576 (mov_<load>l): Use ZC constraint.
16577 (mov_<load>r): Likewise.
16578 (mov_<store>l): Likewise.
16579 (mov_<store>r): Likewise.
16580 (*branch_equality<mode>_inverted): Add microMIPS support.
16581 (*branch_equality<mode>): Likewise.
16582 (*jump_absolute): Likewise.
16583 (indirect_jump_<mode>): Likewise.
16584 (tablejump_<mode>): Likewise.
16585 (<optab>_internal): Likewise.
16586 (sibcall_internal): Likewise.
16587 (sibcall_value_internal): Likewise.
16588 (prefetch): Use constraint ZD.
16589 * config/mips/mips.opt (minterlink-compressed): New option.
16590 (minterlink-mips16): Now an alias for minterlink-compressed.
16591 (mmicromips): New option.
16592 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16593 (compare_and_swap_12): Likewise.
16594 (sync_add<mode>): Likewise.
16595 (sync_<optab>_12): Likewise.
16596 (sync_old_<optab>_12): Likewise.
16597 (sync_new_<optab>_12): Likewise.
16598 (sync_nand_12): Likewise.
16599 (sync_old_nand_12): Likewise.
16600 (sync_new_nand_12): Likewise.
16601 (sync_sub<mode>): Likewise.
16602 (sync_old_add<mode>): Likewise.
16603 (sync_old_sub<mode>): Likewise.
16604 (sync_new_add<mode>): Likewise.
16605 (sync_new_sub<mode>): Likewise.
16606 (sync_<optab><mode>): Likewise.
16607 (sync_old_<optab><mode>): Likewise.
16608 (sync_new_<optab><mode>): Likewise.
16609 (sync_nand<mode>): Likewise.
16610 (sync_old_nand<mode>): Likewise.
16611 (sync_new_nand<mode>): Likewise.
16612 (sync_lock_test_and_set<mode>): Likewise.
16613 (test_and_set_12): Likewise.
16614 (atomic_compare_and_swap<mode>): Likewise.
16615 (atomic_exchange<mode>_llsc): Likewise.
16616 (atomic_fetch_add<mode>_llsc): Likewise.
16617 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16618 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16619 (umips_save_restore_pattern_p): Likewise.
16620 (umips_load_store_pair_p): Likewise.
16621 (umips_output_load_store_pair): Likewise.
16622 (umips_movep_target_p): Likewise.
16623 (umips_12bit_offset_address_p): Likewise.
16624 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16625 (mips_base_mips16): Rename this...
16626 (mips_base_compression_flags): ...to this. Update all uses.
16627 (mips_attribute_table): Add micromips, nomicromips and nocompression.
16628 (mips_mips16_decl_p): Delete.
16629 (mips_nomips16_decl_p): Delete.
16630 (mips_get_compress_on_flags): New function.
16631 (mips_get_compress_off_flags): New function.
16632 (mips_get_compress_mode): New function.
16633 (mips_get_compress_on_name): New function.
16634 (mips_get_compress_off_name): New function.
16635 (mips_insert_attributes): Support multiple compression types.
16636 (mips_merge_decl_attributes): Likewise.
16637 (umips_12bit_offset_address_p): New function.
16638 (mips_start_function_definition): Emit .set micromips directive.
16639 (mips_call_may_need_jalx_p): New function.
16640 (mips_function_ok_for_sibcall): Add microMIPS support.
16641 (mips_print_operand_punctuation): Support short delay slots and
16643 (umips_swm_mask, umips_swm_encoding): New.
16644 (umips_build_save_restore): New function.
16645 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16646 (was_mips16_p): Remove.
16647 (old_compression_mode): New.
16648 (mips_set_compression_mode): New function.
16649 (mips_set_current_function): Add microMIPS support.
16650 (mips_option_override): Likewise.
16651 (umips_save_restore_pattern_p): New function.
16652 (umips_output_save_restore): New function.
16653 (umips_load_store_pair_p_1): New function.
16654 (umips_load_store_pair_p): New function.
16655 (umips_output_load_store_pair_1): New function.
16656 (umips_output_load_store_pair): New function.
16657 (umips_movep_target_p) New function.
16658 (mips_prepare_pch_save): Add microMIPS support.
16659 * config/mips/mips.h (TARGET_COMPRESSION): New.
16660 (TARGET_CPU_CPP_BUILTINS): Update macro
16661 to use new compression flags and to support microMIPS.
16662 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16663 (MIPS_ARCH_FLOAT_SPEC): Likewise.
16664 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16665 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16666 (ASM_SPEC): Support mmicromips and mno-micromips.
16667 (M16STORE_REG_P): New macro.
16668 (MIPS_CALL): Support TARGET_MICROMIPS.
16669 (MICROMIPS_J): New macro.
16670 (mips_base_mips16): Rename this...
16671 (mips_base_compression_flags): ...to this.
16672 (UMIPS_12BIT_OFFSET_P): New macro.
16673 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16674 (MULTILIB_DIRNAMES): Likewise.
16676 2013-02-25 Tom de Vries <tom@codesourcery.com>
16678 PR rtl-optimization/56131
16679 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16680 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16681 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16683 2013-02-25 Tobias Burnus <burnus@net-b.de>
16685 * doc/invoke.texi (-fsanitize=): Move from optimization
16686 to debugging options.
16688 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
16690 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16692 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
16693 Alexander Monakov <amonakov@ispras.ru>
16695 PR middle-end/56077
16696 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16697 flush pending lists also on non-jumps. Adjust comment.
16699 2013-02-24 Catherine Moore <clm@codesourcery.com>
16700 Maciej W. Rozycki <macro@codesourcery.com>
16701 Tom de Vries <tom@codesourcery.com>
16702 Nathan Sidwell <nathan@codesourcery.com>
16703 Iain Sandoe <iain@codesourcery.com>
16704 Nathan Froyd <froydnj@codesourcery.com>
16705 Chao-ying Fu <fu@mips.com>
16707 * doc/extend.texi: (micromips, nomicromips, nocompression):
16708 Document new function attributes.
16709 * doc/invoke.texi (minterlink-compressed, mmicromips,
16710 m14k, m14ke, m14kec): Document new options.
16711 (minterlink-mips16): Update documentation.
16712 * doc/md.texi (ZC, ZD): Document new constraints.
16713 * configure.ac (gcc_cv_as_micromips): Check if linker
16714 supports the .set micromips directive.
16715 * configure: Regenerate.
16716 * config.in: Regenerate.
16717 * config/mips/mips-tables.opt: Regenerate.
16718 * config/mips/micromips.md: New file.
16719 * constraints.md (ZC, AD): New constraints.
16720 * config/mips/predicates.md (movep_src_register): New predicate.
16721 (movep_src_operand): New predicate.
16722 (non_volatile_mem_operand): New predicate.
16723 * config/mips/mips.md (multimem): New type.
16724 (length): Differentiate between 17-bit and 18-bit branch offsets.
16725 (MOVEP1, MOVEP2): New mode iterator.
16726 (mov_<load>l): Use ZC constraint.
16727 (mov_<load>r): Likewise.
16728 (mov_<store>l): Likewise.
16729 (mov_<store>r): Likewise.
16730 (*branch_equality<mode>_inverted): Add microMIPS support.
16731 (*branch_equality<mode>): Likewise.
16732 (*jump_absolute): Likewise.
16733 (indirect_jump_<mode>): Likewise.
16734 (tablejump_<mode>): Likewise.
16735 (<optab>_internal): Likewise.
16736 (sibcall_internal): Likewise.
16737 (sibcall_value_internal): Likewise.
16738 (prefetch): Use constraint ZD.
16739 * config/mips/mips.opt (minterlink-compressed): New option.
16740 (minterlink-mips16): Now an alias for minterlink-compressed.
16741 (mmicromips): New option.
16742 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16743 (compare_and_swap_12): Likewise.
16744 (sync_add<mode>): Likewise.
16745 (sync_<optab>_12): Likewise.
16746 (sync_old_<optab>_12): Likewise.
16747 (sync_new_<optab>_12): Likewise.
16748 (sync_nand_12): Likewise.
16749 (sync_old_nand_12): Likewise.
16750 (sync_new_nand_12): Likewise.
16751 (sync_sub<mode>): Likewise.
16752 (sync_old_add<mode>): Likewise.
16753 (sync_old_sub<mode>): Likewise.
16754 (sync_new_add<mode>): Likewise.
16755 (sync_new_sub<mode>): Likewise.
16756 (sync_<optab><mode>): Likewise.
16757 (sync_old_<optab><mode>): Likewise.
16758 (sync_new_<optab><mode>): Likewise.
16759 (sync_nand<mode>): Likewise.
16760 (sync_old_nand<mode>): Likewise.
16761 (sync_new_nand<mode>): Likewise.
16762 (sync_lock_test_and_set<mode>): Likewise.
16763 (test_and_set_12): Likewise.
16764 (atomic_compare_and_swap<mode>): Likewise.
16765 (atomic_exchange<mode>_llsc): Likewise.
16766 (atomic_fetch_add<mode>_llsc): Likewise.
16767 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16768 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16769 (umips_save_restore_pattern_p): Likewise.
16770 (umips_load_store_pair_p): Likewise.
16771 (umips_output_load_store_pair): Likewise.
16772 (umips_movep_target_p): Likewise.
16773 (umips_12bit_offset_address_p): Likewise.
16774 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16775 (mips_base_mips16): Rename this...
16776 (mips_base_compression_flags): ...to this. Update all uses.
16777 (mips_attribute_table): Add micromips, nomicromips and nocompression.
16778 (mips_mips16_decl_p): Delete.
16779 (mips_nomips16_decl_p): Delete.
16780 (mips_get_compress_on_flags): New function.
16781 (mips_get_compress_off_flags): New function.
16782 (mips_get_compress_mode): New function.
16783 (mips_get_compress_on_name): New function.
16784 (mips_get_compress_off_name): New function.
16785 (mips_insert_attributes): Support multiple compression types.
16786 (mips_merge_decl_attributes): Likewise.
16787 (umips_12bit_offset_address_p): New function.
16788 (mips_start_function_definition): Emit .set micromips directive.
16789 (mips_call_may_need_jalx_p): New function.
16790 (mips_function_ok_for_sibcall): Add microMIPS support.
16791 (mips_print_operand_punctuation): Support short delay slots and
16793 (umips_swm_mask, umips_swm_encoding): New.
16794 (umips_build_save_restore): New function.
16795 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16796 (was_mips16_p): Remove.
16797 (old_compression_mode): New.
16798 (mips_set_compression_mode): New function.
16799 (mips_set_current_function): Add microMIPS support.
16800 (mips_option_override): Likewise.
16801 (umips_save_restore_pattern_p): New function.
16802 (umips_output_save_restore): New function.
16803 (umips_load_store_pair_p_1): New function.
16804 (umips_load_store_pair_p): New function.
16805 (umips_output_load_store_pair_1): New function.
16806 (umips_output_load_store_pair): New function.
16807 (umips_movep_target_p) New function.
16808 (mips_prepare_pch_save): Add microMIPS support.
16809 * config/mips/mips.h (TARGET_COMPRESSION): New.
16810 (TARGET_CPU_CPP_BUILTINS): Update macro
16811 to use new compression flags and to support microMIPS.
16812 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16813 (MIPS_ARCH_FLOAT_SPEC): Likewise.
16814 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16815 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16816 (ASM_SPEC): Support mmicromips and mno-micromips.
16817 (M16STORE_REG_P): New macro.
16818 (MIPS_CALL): Support TARGET_MICROMIPS.
16819 (MICROMIPS_J): New macro.
16820 (mips_base_mips16): Rename this...
16821 (mips_base_compression_flags): ...to this.
16822 (UMIPS_12BIT_OFFSET_P): New macro.
16823 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16824 (MULTILIB_DIRNAMES): Likewise.
16826 2013-02-24 Jakub Jelinek <jakub@redhat.com>
16829 * target-globals.c (save_target_globals): For init_reg_sets and
16830 target_reinit remporarily set this_fn_optabs to this_target_optabs.
16832 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
16834 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16835 * config/aarch64/t-aarch64
16836 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16838 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
16840 PR inline-asm/56148
16841 * lra-constraints.c (process_alt_operands): Reload operand
16842 conflicting with earlier clobber only if no more other conflicting
16845 2013-02-22 Jakub Jelinek <jakub@redhat.com>
16848 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16849 if not linking a shared library.
16851 2013-02-22 Seth LaForge <sethml@google.com>
16853 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16855 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
16857 * config/arm/arm.md (split for extendsidi): Update condition.
16858 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16859 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16860 (qhs_zextenddi_cstr): Likewise.
16862 2013-02-21 Jakub Jelinek <jakub@redhat.com>
16864 PR middle-end/56420
16865 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16866 avoid signed wrapping.
16867 (expand_mult): Handle properly multiplication by
16868 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
16869 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
16870 in the compiler if coeff is HOST_WIDE_INT_MIN.
16871 (expand_divmod): Don't make ext_op1 static, change it's type to
16872 uhwi. Avoid undefined behavior in -INTVAL (op1).
16874 PR rtl-optimization/50339
16875 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16877 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16878 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16879 into splitting_ashiftrt field.
16880 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16882 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16885 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
16887 PR middle-end/56108
16888 * trans-mem.c (execute_tm_mark): Do not expand transactions that
16889 are sure to go irrevocable.
16891 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
16893 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16894 scalars are valid operands.
16896 2013-02-21 Martin Jambor <mjambor@suse.cz>
16898 PR tree-optimization/56310
16899 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16900 only matching indices and non-negative final offsets.
16901 (intersect_aggregates_with_edge): Pass src_idx to
16902 agg_replacements_to_vector. Pass src_idx insstead of index to
16903 intersect_with_agg_replacements.
16905 2013-02-21 Martin Jambor <mjambor@suse.cz>
16907 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16908 instead of hard-wired defaults.
16910 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
16912 * doc/invoke.texi (MIPS Options): Update documentation of the
16913 floating-point multiply-accumulate instruction restrictions.
16915 2013-02-21 Kostya Serebryany <kcc@google.com>
16917 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16918 asan_shadow_offset on x86_64 linux.
16920 2013-02-21 Richard Biener <rguenther@suse.de>
16922 PR tree-optimization/56415
16924 2013-02-11 Richard Biener <rguenther@suse.de>
16926 PR tree-optimization/56273
16927 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16930 2013-02-21 Jakub Jelinek <jakub@redhat.com>
16933 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16936 PR inline-asm/56405
16937 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16938 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16940 2013-02-20 Jan Hubicka <jh@suse.cz>
16942 PR tree-optimization/56265
16943 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16944 when target is referenced for first time.
16946 2013-02-20 Richard Biener <rguenther@suse.de>
16948 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16949 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16950 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16951 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16952 not return anything.
16953 (rename_ssa_copies): Do not remove unused locals.
16954 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16955 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16956 * passes.c (execute_function_todo): Do not schedule unused locals
16957 removal if cleanup_tree_cfg did something.
16958 * tree-ssa-live.c (remove_unused_locals): Dump statistics
16959 about the number of removed locals.
16961 2013-02-20 Richard Biener <rguenther@suse.de>
16963 PR tree-optimization/56398
16964 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16966 2013-02-20 Martin Jambor <mjambor@suse.cz>
16968 PR tree-optimization/55334
16969 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16970 restricted pointers to arrays.
16972 2013-02-20 Richard Biener <rguenther@suse.de>
16973 Jakub Jelinek <jakub@redhat.com>
16975 PR tree-optimization/56396
16976 * tree-ssa-ccp.c (n_const_val): New static variable.
16977 (get_value): Return NULL for SSA names we don't have a lattice
16979 (ccp_initialize): Initialize n_const_val.
16980 * tree-ssa-copy.c (n_copy_of): New static variable.
16981 (init_copy_prop): Initialize n_copy_of.
16982 (get_value): Return NULL_TREE for SSA names we don't have a
16985 2013-02-20 Martin Jambor <mjambor@suse.cz>
16987 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16989 2013-02-20 Richard Biener <rguenther@suse.de>
16991 * genpreds.c (write_lookup_constraint): Do not compare first
16992 letter of the constraint again.
16994 2013-02-20 Richard Biener <rguenther@suse.de>
16996 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16998 (get_use_iv_cost): Terminate hashtable walk when coming across
17001 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
17003 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
17004 reassociation for avx2 targets.
17006 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17008 * config/microblaze/microblaze.c: microblaze_has_clz = 0
17009 Add version check for v8.10.a to enable microblaze_has_clz
17010 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
17011 version and TARGET_PATTERN_COMPARE check
17012 * config/microblaze/microblaze.md: New clzsi2 instruction
17014 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17016 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
17017 function before branching.
17019 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
17021 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
17023 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
17025 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
17027 PR middle-end/55889
17028 * sel-sched.c: Include ira.h.
17029 (implicit_clobber_conflict_p): New function.
17030 (moveup_expr): Use it.
17031 * Makefile.in (sel-sched.o): Depend on ira.h.
17033 2013-02-19 Richard Biener <rguenther@suse.de>
17035 PR tree-optimization/56384
17036 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
17037 (vn_hash_type): Split out from ...
17038 (vn_hash_constant_with_type): ... here.
17039 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
17040 (vn_phi_eq): Compare types from vn_phi_s structure.
17041 (vn_phi_lookup): Populate vn_phi_s type.
17042 (vn_phi_insert): Likewise.
17044 2013-02-19 Jakub Jelinek <jakub@redhat.com>
17046 PR tree-optimization/56350
17047 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
17048 if haven't found reduction or nested cycle operand, rather than
17049 asserting we must find it.
17051 PR tree-optimization/56381
17052 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
17055 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
17056 Jakub Jelinek <jakub@redhat.com>
17059 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
17060 (swap_optab_enable): Same.
17061 (init_all_optabs): Use argument instead of global.
17062 * tree.h (struct tree_optimization_option): New field target_optabs.
17063 * expr.h (init_all_optabs): Add argument to prototype.
17064 (TREE_OPTIMIZATION_OPTABS): New.
17065 (save_optabs_if_changed): Protoize.
17066 * optabs.h: Declare this_fn_optabs.
17067 * optabs.c (save_optabs_if_changed): New.
17068 Declare this_fn_optabs.
17069 (init_optabs): Add argument to init_all_optabs() call.
17070 * function.c (invoke_set_current_function_hook): Handle per
17072 * function.h (struct function): New field optabs.
17073 * config/mips/mips.c (mips_set_mips16_mode): Handle when
17074 optimization_current_node has changed.
17075 * target-globals.h (save_target_globals_default_opts): Protoize.
17076 * target-globals.c (save_target_globals_default_opts): New.
17078 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17081 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
17082 registers %fr12 and %fr12R as call used.
17085 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
17086 and HImode, require all displacements to be an integer multiple of
17088 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
17089 only allow QImode and HImode when reload is in progress and strict is
17090 true. Likewise for symbolic addresses. Use base14_operand to check
17091 displacements in REG+BASE addresses.
17093 2013-02-18 Richard Biener <rguenther@suse.de>
17095 PR tree-optimization/56366
17096 * tree-vect-loop.c (get_initial_def_for_induction): Properly
17097 handle sign-conversion of outer-loop initial induction value.
17099 2013-02-18 Richard Biener <rguenther@suse.de>
17101 PR middle-end/56349
17102 * cfghooks.c (merge_blocks): If we merge a latch into another
17103 block adjust references to it.
17104 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
17105 (verify_loop_structure): Verify that a recorded latch is in fact
17108 2013-02-18 Richard Biener <rguenther@suse.de>
17110 PR tree-optimization/56321
17111 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
17112 order SSA name release and virtual operand unlinking.
17114 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17116 * config/microblaze/microblaze.md (save_stack_block): Define.
17117 (restore_stack_block): Likewise.
17119 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17121 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
17122 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
17123 * config/microblaze/microblaze.c (microblaze_option_override):
17124 Bail out early for PIC modes when target does not support PIC.
17126 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17128 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
17129 Replace with a microblaze version.
17130 (microblaze_trampoline_init): Adapt for microblaze.
17131 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
17134 2013-02-16 Jakub Jelinek <jakub@redhat.com>
17135 Dodji Seketeli <dodji@redhat.com>
17138 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
17139 (instrument_mem_region_access): Do not forget to always put
17140 instrumentation of the of 'base' and 'base + len' in a "if (len !=
17141 0) statement, even for cases where either 'base' or 'base + len'
17142 are not instrumented -- because they have been previously
17143 instrumented. Simplify the logic by putting all the statements
17144 instrument 'base + len' inside a sequence, and then insert that
17145 sequence right before the current insertion point. Then, to
17146 instrument 'base + len', just get an iterator on that statement.
17147 And do not forget to update the pointer to iterator the function
17148 received as argument.
17150 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
17152 PR rtl-optimization/56348
17153 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
17155 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
17157 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
17158 (clean_graph_dump_file): Pass base to start_graph_dump.
17160 2013-02-14 Richard Henderson <rth@redhat.com>
17163 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
17165 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
17167 * collect2-aix.h: Define F_LOADONLY.
17169 2013-02-14 Richard Biener <rguenther@suse.de>
17172 * varasm.c (output_constant_def_1): Get the decl representing
17173 the constant as argument.
17174 (output_constant_def): Wrap output_constant_def_1.
17175 (make_decl_rtl): Use output_constant_def_1 with the decl
17176 representing the constant.
17177 (build_constant_desc): Optionally re-use a decl already
17178 representing the constant.
17179 (tree_output_constant_def): Adjust.
17181 2013-02-14 Dodji Seketeli <dodji@redhat.com>
17184 * asan.c (instrument_builtin_call): Really put the length of the
17185 second source argument into src1_len.
17187 2013-02-13 Jakub Jelinek <jakub@redhat.com>
17189 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
17190 argument. If it is false, don't create edge from then_bb to
17192 (insert_if_then_before_iter): Pass true to it.
17193 (build_check_stmt): Pass false to it.
17194 (transform_statements): Flush hash table only on extended basic
17195 block boundaries, rather than at the beginning of every bb.
17196 Don't flush hash table on nonfreeing_call_p calls.
17197 * tree-flow.h (nonfreeing_call_p): New prototype.
17198 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
17200 2013-02-13 David S. Miller <davem@davemloft.net>
17202 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
17204 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
17207 * ira.c (max_regno_before_ira): Move from ...
17209 (fix_reg_equiv_init): Use max_regno_before_ira instead of
17212 2013-02-13 Jakub Jelinek <jakub@redhat.com>
17214 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
17216 2013-02-13 Richard Biener <rguenther@suse.de>
17219 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
17220 globals in MEM_REFs.
17222 2013-02-13 Richard Biener <rguenther@suse.de>
17224 * loop-init.c (loop_optimizer_init): Clear loop state when
17225 re-initializing preserved loops.
17226 * loop-unswitch.c (unswitch_single_loop): Return whether
17227 we unswitched the loop. Do not verify loop state here.
17228 (unswitch_loops): When we unswitched a loop discover new loops.
17230 2013-02-13 Kostya Serebryany <kcc@google.com>
17232 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
17234 * sanitizer.def: Rename __asan_init to __asan_init_v1.
17236 2013-02-12 Dodji Seketeli <dodji@redhat.com>
17238 Avoid instrumenting duplicated memory access in the same basic block
17239 * Makefile.in (asan.o): Add new dependency on hash-table.h
17240 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
17241 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
17242 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
17243 (free_mem_ref_resources, has_mem_ref_been_instrumented)
17244 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
17245 (get_mem_ref_of_assignment): New functions.
17246 (get_mem_refs_of_builtin_call): Extract from
17247 instrument_builtin_call and tweak a little bit to make it fit with
17249 (instrument_builtin_call): Use the new
17250 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
17251 of is_gimple_builtin_call.
17252 (instrument_derefs, instrument_mem_region_access): Insert the
17253 instrumented memory reference into the hash table.
17254 (maybe_instrument_assignment): Renamed instrument_assignment into
17255 this, and change it to advance the iterator when instrumentation
17256 actually happened and return true in that case. This makes it
17257 homogeneous with maybe_instrument_assignment, and thus give a
17258 chance to callers to be more 'regular'.
17259 (transform_statements): Clear the memory reference hash table
17260 whenever we enter a new BB, when we cross a function call, or when
17261 we are done transforming statements. Use
17262 maybe_instrument_assignment instead of instrumentation. No more
17263 need to special case maybe_instrument_assignment and advance the
17264 iterator after calling it; it's now handled just like
17265 maybe_instrument_call. Update comment.
17267 2013-02-13 Richard Biener <rguenther@suse.de>
17269 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
17270 Fix loop discovery code.
17272 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
17274 PR inline-asm/56148
17275 * lra-constraints.c (process_alt_operands): Match early clobber
17276 operand with itself. Check conflicts with earlyclobber only if
17277 the operand is not reloaded. Prefer to reload conflicting operand
17278 if earlyclobber and matching operands are the same.
17280 2013-02-12 Richard Biener <rguenther@suse.de>
17283 * lto-streamer-out.c (write_symbol): Do not output symbols
17284 for hard register variables.
17286 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
17289 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
17290 (umulsidi3_insn, mulsidi3_insn): New insns.
17292 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
17294 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
17295 (struct tune_params): Add vec_costs field.
17296 * config/arm/arm.c (arm_builtin_vectorization_cost)
17297 (arm_add_stmt_cost): New functions.
17298 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
17299 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
17300 (arm_default_vec_cost): New struct of type cpu_vec_costs.
17301 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
17302 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
17303 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
17304 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
17306 2013-02-12 Richard Biener <rguenther@suse.de>
17309 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
17310 decls again if possible.
17312 2013-02-12 Richard Biener <rguenther@suse.de>
17314 PR middle-end/56288
17315 * tree-ssa.c (verify_ssa_name): Fix check, move
17316 SSA_NAME_IN_FREE_LIST check up.
17318 2013-02-12 Jakub Jelinek <jakub@redhat.com>
17319 Steven Bosscher <steven@gcc.gnu.org>
17321 PR rtl-optimization/56151
17322 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
17323 equal to op0 or op1, and last_insn pattern is CODE operation
17324 with MEM dest and one of the operands matches that MEM.
17326 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
17328 * doc/extend.texi: Document Function Multiversioning and "default"
17329 parameter string to target attribute.
17330 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
17331 target attribute parameter is "default".
17332 (ix86_compare_version_priority): Remove checks for target attribute.
17333 (ix86_mangle_function_version_assembler_name): Change error to sorry.
17334 Remove check for target attribute equal to NULL. Add assert.
17335 (ix86_generate_version_dispatcher_body): Change error to sorry.
17337 2013-02-11 Iain Sandoe <iain@codesourcery.com>
17338 Jack Howarth <howarth@bromo.med.uc.edu>
17339 Patrick Marlier <patrick.marlier@gmail.com>
17342 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
17343 define ENDFILE_SPEC as TM_DESTRUCTOR.
17344 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
17346 2013-02-11 Alexander Potapenko <glider@google.com>
17347 Jack Howarth <howarth@bromo.med.uc.edu>
17348 Jakub Jelinek <jakub@redhat.com>
17351 * config/darwin.c (cdtor_record): Rename ctor_record.
17352 (sort_cdtor_records): Rename sort_ctor_records.
17353 (finalize_dtors): New routine to sort destructors by
17354 priority before use in assemble_integer.
17355 (machopic_asm_out_destructor): Use finalize_dtors if needed.
17357 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
17359 PR rtl-optimization/56275
17360 * simplify-rtx.c (avoid_constant_pool_reference): Check that
17361 offset is non-negative and less than cmode size before
17362 calling simplify_subreg.
17364 2013-02-11 Richard Biener <rguenther@suse.de>
17366 PR tree-optimization/56264
17367 * cfgloop.h (fix_loop_structure): Adjust prototype.
17368 * loop-init.c (fix_loop_structure): Return the number of
17369 newly discovered loops.
17370 * tree-cfgcleanup.c (repair_loop_structures): When new loops
17371 are discovered, do a full loop-closed SSA rewrite.
17373 2013-02-11 Richard Biener <rguenther@suse.de>
17375 PR tree-optimization/56273
17376 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17378 (check_array_ref): Fix missing newline in dumps.
17379 (search_for_addr_array): Likewise.
17381 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
17383 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
17385 2013-02-09 Jakub Jelinek <jakub@redhat.com>
17388 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
17390 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
17392 PR rtl-optimization/56246
17393 * lra-constraints.c (simplify_operand_subreg): Try to reuse
17395 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
17396 constraints are satisfied.
17398 2013-02-08 Jeff Law <law@redhat.com>
17401 * emit-rtl.c (reg_is_parm_p): New function.
17402 * regs.h (reg_is_parm_p): New prototype.
17403 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
17404 callee-clobbered registers.
17406 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
17409 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
17410 If there is no implicit builtin declaration, just return NULL.
17412 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17414 * config/i386/sse.md (FMAMODEM): New mode iterator.
17415 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
17416 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
17418 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17420 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
17421 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
17422 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
17424 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17426 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
17427 (microblaze*-*-elf): Likewise.
17428 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
17430 * config/microblaze/microblaze-c.c: Add builtin defines for
17431 _LITTLE_ENDIAN and _BIG_ENDIAN.
17432 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
17433 add to TARGET_DEFAULT flags.
17434 Expand ASM_SPEC and LINK_SPEC.
17435 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
17436 * config/microblaze/microblaze.md: Update extendsidi2 and
17437 movdi_internal instructions to use low-order / high-order reg
17439 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
17440 options and inversemask / mask of LITTLE_ENDIAN.
17441 * config/microblaze/t-microblaze: Expand multilib options to
17442 include mlittle-endian (le) and update exceptions patterns.
17444 2013-02-08 Jakub Jelinek <jakub@redhat.com>
17446 PR rtl-optimization/56195
17447 * lra-constraints.c (get_reload_reg): Don't reuse regs
17448 if they have smaller mode than requested, if they have
17449 wider mode than requested, try to return a SUBREG.
17451 PR tree-optimization/56250
17452 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
17453 if type is unsigned and code isn't MULT_EXPR.
17455 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
17457 PR tree-optimization/56064
17458 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
17459 bits according to mode.
17460 * fixed-value.h (fixed_from_double_int)
17461 (const_fixed_from_double_int): Adjust comments.
17463 2013-02-08 Richard Biener <rguenther@suse.de>
17466 * lto-streamer.h (struct data_in): Remove current_file, current_line
17467 and current_col members.
17468 * lto-streamer-out.c (lto_output_location): Stream changed bits
17469 en-block for efficiency.
17470 * lto-streamer-in.c (clear_line_info): Remove.
17471 (lto_input_location): Cache current file, line and column
17472 globally via local statics. Read changed bits en-block.
17473 (input_function): Do not call clear_line_info.
17474 (lto_read_body): Likewise.
17475 (lto_input_toplevel_asms): Likewise.
17477 2013-02-08 Michael Matz <matz@suse.de>
17479 PR tree-optimization/52448
17480 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
17481 (nt_call_phase): New static.
17482 (add_or_mark_expr): Only mark accesses with newer phase than any
17484 (nonfreeing_call_p): New.
17485 (nt_init_block): Update nt_call_phase, mark blocks as visited.
17486 (nt_fini_block): Keep blocks marked as visited.
17487 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
17489 2013-02-08 Richard Biener <rguenther@suse.de>
17491 * ira.c (ira): Free broken dominator information.
17493 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17495 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
17497 2013-02-08 Marek Polacek <polacek@redhat.com>
17499 * cfgloop.c (verify_loop_structure): Add more checking of headers.
17501 2013-02-08 Richard Biener <rguenther@suse.de>
17503 PR middle-end/56181
17504 * cfgloop.h (flow_loops_find): Adjust.
17505 (bb_loop_header_p): Declare.
17506 * cfgloop.c (bb_loop_header_p): New function split out from ...
17507 (flow_loops_find): ... here. Adjust function signature,
17508 support incremental loop structure update.
17509 (verify_loop_structure): Cleanup. Verify a loop is a loop.
17510 * cfgloopmanip.c (fix_loop_structure): Move ...
17511 * loop-init.c (fix_loop_structure): ... here.
17512 (apply_loop_flags): Split out from ...
17513 (loop_optimizer_init): ... here.
17514 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
17515 in incremental mode, only remove dead loops here.
17517 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
17520 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
17521 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
17522 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17523 (*round<mode>3.libgcc): New insns for fixed-modes.
17524 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17525 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17526 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17527 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17528 implementations. Define to __builtin_avr_absFX,
17529 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17530 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17531 __builtin_avr_countlsFX, respectively.
17532 * config/avr/avr-c.c (target.h): Include it.
17533 (enum avr_builtin_id): New enum.
17534 (avr_resolve_overloaded_builtin): New static function.
17535 (avr_register_target_pragmas): Use it to set
17536 targetm.resolve_overloaded_builtin.
17537 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17538 tree nodes used by DEF_BUILTIN.
17539 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17540 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17541 <AVR_BUILTIN_xxBITS>: Same.
17543 2013-02-08 Richard Biener <rguenther@suse.de>
17545 * cfgloop.c (verify_loop_structure): Properly handle
17546 a loop exiting to another loop header.
17547 * ira-int.h (ira_loops): Remove.
17548 * ira.c (ira_loops): Remove.
17549 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17550 (do_reload): Use loop_optimizer_finalize.
17551 * ira-build.c (create_loop_tree_nodes): Use get_loops and
17552 number_of_loops to access the loop tree.
17553 (more_one_region_p): Likewise.
17554 (finish_loop_tree_nodes): Likewise.
17555 (rebuild_regno_allocno_maps): Likewise.
17556 (mark_loops_for_removal): Likewise.
17557 (mark_all_loops_for_removal): Likewise.
17558 (remove_unnecessary_regions): Likewise.
17559 (ira_build): Likewise.
17560 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17562 2013-02-08 Richard Biener <rguenther@suse.de>
17564 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17565 * ipa-pure-const.c (analyze_function): Avoid calling
17566 mark_irreducible_loops twice.
17567 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17569 2013-02-07 David S. Miller <davem@davemloft.net>
17571 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17573 * var-tracking.c (vt_add_function_parameter): Test the presence of
17574 HAVE_window_save properly and do not remap argument registers when
17575 we have a leaf function.
17577 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
17580 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17582 * config/i386/i386.c (ix86_print_operand): Ditto.
17584 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
17586 * lra-constraints.c (process_alt_operands): Fix recently added comment.
17588 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
17590 PR rtl-optimization/56225
17591 * lra-constraints.c (process_alt_operands): Check that reload hard
17592 reg can hold value for strict_low_part.
17594 2013-02-07 Jakub Jelinek <jakub@redhat.com>
17597 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17598 dwarf2out_end_function.
17599 (in_first_function_p, maybe_at_text_label_p,
17600 first_loclabel_num_not_at_text_label): New variables.
17601 (dwarf2out_var_location): In the first function find out
17602 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17603 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17606 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
17608 PR rtl-optimization/56178
17609 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17610 SUBREG of a register. Tidy up related block of code.
17611 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17612 note if the source is a register or a SUBREG of a register.
17614 2013-02-07 Jakub Jelinek <jakub@redhat.com>
17617 * config/rs6000/rs6000.md (ptrm): New mode attr.
17618 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17619 call_value_indirect_aix<pttrsize>,
17620 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17623 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
17625 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17626 if -bnortl. Convert to strcmp and strncmp.
17628 2013-02-07 Alan Modra <amodra@gmail.com>
17631 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17632 addresses won't wrap when offsetting.
17633 (rs6000_secondary_reload): Provide secondary reloads needed for
17634 wrapping LO_SUM addresses.
17636 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
17638 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17639 MACH, just __MACH__.
17641 2013-02-06 Richard Biener <rguenther@suse.de>
17643 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17644 instead of calling fix_loop_structure.
17646 2013-02-06 Jakub Jelinek <jakub@redhat.com>
17648 PR middle-end/56217
17649 * omp-low.c (use_pointer_for_field): Return false if
17650 lower_send_shared_vars doesn't generate any copy-out code.
17652 2013-02-06 Tom de Vries <tom@codesourcery.com>
17654 PR rtl-optimization/56131
17655 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17656 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
17657 of the label is NULL. Add comment.
17659 2013-02-05 Jakub Jelinek <jakub@redhat.com>
17661 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17664 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17665 (STATIC_LIBTSAN_LIBS): Likewise.
17666 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17667 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17668 is defined, don't add anything else beyond that.
17669 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17670 (LINK_COMMAND_SPEC): Use them.
17672 PR tree-optimization/56205
17673 * tree-stdarg.c (check_all_va_list_escapes): Return true if
17674 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17675 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17677 2013-02-05 Richard Biener <rguenther@suse.de>
17679 PR tree-optimization/53342
17680 PR tree-optimization/53185
17681 * tree-vectorizer.h (vect_check_strided_load): Remove.
17682 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17683 not disallow peeling for vectorized strided loads.
17684 (vect_check_strided_load): Make static and simplify.
17685 (vect_analyze_data_refs): Adjust.
17686 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17687 correctly when vectorizing strided loads.
17689 2013-02-05 Richard Biener <rguenther@suse.de>
17691 * doc/install.texi: Refer to ISL, not PPL.
17693 2013-02-05 Jan Hubicka <jh@suse.cz>
17695 PR tree-optimization/55789
17696 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17698 2013-02-05 Jan Hubicka <jh@suse.cz>
17700 PR tree-optimization/55789
17701 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17702 the dead call anyway.
17704 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
17707 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17709 2013-02-04 Alexander Potapenko <glider@google.com>
17710 Jack Howarth <howarth@bromo.med.uc.edu>
17711 Jakub Jelinek <jakub@redhat.com>
17714 * config/darwin.c (sort_ctor_records): Stabilized qsort
17715 on constructor priority by using original position.
17716 (finalize_ctors): New routine to sort constructors by
17717 priority before use in assemble_integer.
17718 (machopic_asm_out_constructor): Use finalize_ctors if needed.
17720 2013-02-04 Jakub Jelinek <jakub@redhat.com>
17723 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17724 about visibility on artificial decls.
17725 * config/sol2.c (solaris_assemble_visibility): Likewise.
17727 2013-02-04 Kai Tietz <ktietz@redhat.com>
17730 * config/i386/i386.c (function_value_ms_64): Add additional valtype
17731 argument and improve checking of return-argument types for 16-byte
17733 (ix86_function_value_1): Add additional valtype argument on call
17734 of function_value_64.
17735 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17736 handling infunction_value_64 function.
17738 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
17740 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17742 2013-02-04 Richard Biener <rguenther@suse.de>
17744 PR tree-optimization/56188
17745 * tree-ssa-structalias.c (label_visit): Consider case with
17746 initially non-empty points-to set.
17747 (perform_var_substitution): Dump node mapping and clean up.
17749 2013-02-04 Richard Guenther <rguenther@suse.de>
17752 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17753 node prevail as last resort.
17754 (lto_symtab_merge_decls): Remove guard on LTRANS here.
17755 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17757 2013-02-04 Richard Biener <rguenther@suse.de>
17759 PR tree-optimization/56113
17760 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17762 (equiv_class_lookup_or_add): ... this.
17763 (label_visit): Adjust and fix error in previous patch.
17764 (perform_var_substitution): Adjust.
17766 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
17768 * config/sh/divtab.c: Fix formatting and comments throughout the file.
17769 * config/sh/sh4-300.md: Likewise.
17770 * config/sh/sh4a.md: Likewise.
17771 * config/sh/constraints.md: Likewise.
17772 * config/sh/sh.md: Likewise.
17773 * config/sh/netbsd-elf.h: Likewise.
17774 * config/sh/predicates.md: Likewise.
17775 * config/sh/sh-protos.h: Likewise.
17776 * config/sh/ushmedia.h: Likewise.
17777 * config/sh/linux.h: Likewise.
17778 * config/sh/sh.c: Likewise.
17779 * config/sh/superh.h: Likewise.
17780 * config/sh/elf.h: Likewise.
17781 * config/sh/sh4.md: Likewise.
17782 * config/sh/sh.h: Likewise.
17784 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17786 * config/pa/constraints.md: Adjust unused letters. Change "T"
17787 constraint to match_test floating_point_store_memory_operand().
17788 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17789 (base14_operand): New.
17790 (floating_point_store_memory_operand): New.
17791 (integer_store_memory_operand): Revise to use base14_operand and
17792 reg_plus_base_memory_operand.
17793 (move_dest_operand): Allow symbolic_memory_operands.
17794 (symbolic_memory_operand): Check for LO_SOM.
17795 (symbolic_operand): Change default case to break.
17796 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17797 CONST_DOUBLE values to be reloaded by putting them into memory when
17798 the destination is a floating point register.
17799 (movdf): Remove code to handle CONST_DOUBLE.
17801 (reload_indf_r1): New.
17802 (reload_insf_r1): New.
17803 Consistently use "Q" and "T" constraints with integer and floating
17804 point move instructions, respectively.
17805 (movdi): Remove FAIL.
17806 Change predicate for source operand unamed DImode move from
17807 general_operand to move_src_operand.
17808 (umulsidi3): Change predicate for destination operand to
17810 Likewise for similar unamed patterns.
17811 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17812 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17813 (hppa_legitimize_address): Simplify mask calculation.
17814 (pa_emit_move_sequence): Revised handling of secondary reloads from
17815 REG+D addresses for floating point loads and stores. Directly handle
17816 loading CONST0_RTX (mode) to a floating point register.
17817 (pa_secondary_reload): Handle reloading DF and SFmode constant values
17818 to floating point registers. Don't restrict secondary reloads to
17819 floating point registers to integer modes. Revise some comments and
17821 (TARGET_LEGITIMATE_ADDRESS_P): Define.
17822 (pa_legitimate_address_p): New.
17823 (pa_legitimize_reload_address): New.
17824 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17825 (STRICT_REG_OK_FOR_BASE_P): New.
17826 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
17827 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17829 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
17830 Andrew Dixie <andrewd@gentrack.com>
17832 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17835 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
17837 * expmed.c (extract_bit_field_1): Pass the full width of the
17838 structure to get_best_reg_extraction_insn.
17840 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
17843 * configure.ac (use_cxa_atexit): Add AIX.
17844 * configure: Regenerate.
17846 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17848 2013-02-01 Jakub Jelinek <jakub@redhat.com>
17851 * final.c (need_profile_function): New variable.
17852 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17853 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17854 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17855 notes, targetm.asm_out.function_prologue doesn't emit anything,
17856 HAVE_prologue and profiler should be emitted before prologue,
17857 set need_profile_function instead of emitting it.
17858 (final_scan_insn): If need_profile_function, emit
17859 profile_function on the first NOTE_INSN_BASIC_BLOCK or
17860 NOTE_INSN_FUNCTION_BEG note.
17862 2013-02-01 Richard Henderson <rth@redhat.com>
17864 * config/rs6000/rs6000.md (smulditi3): New.
17867 * config/alpha/alpha.md (umulditi3): New.
17869 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
17871 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17872 (ASM_OUTPUT_ALIGNED_LOCAL): New.
17874 2013-02-01 Richard Biener <rguenther@suse.de>
17876 PR tree-optimization/56113
17877 * tree-ssa-structalias.c (label_visit): Reduce work for
17878 single-predecessor nodes.
17880 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
17882 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17883 range isn't testing for zero.
17885 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
17887 PR middle-end/56113
17888 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17890 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
17891 Nick Clifton <nickc@redhat.com>
17893 * config/v850/constraints.md (Q): Define as a memory constraint.
17894 * config/v850/predicates.md (label_ref_operand): New predicate.
17895 (e3v5_shift_operand): New predicate.
17896 (ior_operator): New predicate.
17897 * config/v850/t-v850: Add e3v5 multilib.
17898 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17899 (v850_gen_movdi): Prototype.
17900 * config/v850/v850.c: Add support for e3v5 architecture.
17901 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17903 (construct_save_jarl): Add e3v5 long JARL support.
17904 (v850_adjust_insn_length): New function. Adjust length of call
17905 insns when using e3v5 instructions.
17906 (v850_gen_movdi): New function: Generate instructions to move a
17908 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17909 (CPP_SPEC): Define __v850e3v5__ as appropriate.
17910 (TARGET_USE_FPU): Enable for e3v5.
17911 (CONST_OK_FOR_W): New macro.
17912 (ADJUST_INSN_LENGTH): Define.
17913 * config/v850/v850.md (UNSPEC_LOOP): Define.
17914 (attr cpu): Add v850e3v5.
17915 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17916 (movdi): New pattern.
17917 (movdi_internal): New pattern.
17918 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17919 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17920 (cstoresf4): Likewise.
17921 (cstoredf4): Likewise.
17922 (insv): New pattern.
17923 (rotlso3_a): New pattern.
17924 (rotlsi3_b): New pattern
17925 (rotlsi3_v850e3v5): New pattern.
17926 (doloop_begin): New pattern.
17927 (fix_loop_counter): New pattern.
17928 (doloop_end): New pattern.
17929 (branch_normal): Add e3v5 long branch support.
17930 (branch_invert): Likewise.
17931 (branch_z_normal): Likewise.
17932 (branch_z_invert): Likewise.
17933 (branch_nz_normal): Likewise.
17934 (branch_nz_invert): Likewise.
17935 (call_internal_short): Add e3v5 register-indirect JARL support.
17936 (call_internal_long): Likewise.
17937 (call_value_internal_short): Likewise.
17938 (call_value_internal_long): Likewise.
17939 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17940 (mloop): New option.
17941 * config.gcc: Add support for configuring v840e3v5 target.
17942 * doc/invoke.texi: Document new v850 specific command line options.
17944 2013-01-31 Paul Koning <ni1d@arrl.net>
17948 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17949 children if parent is a class.
17950 (prune_unused_types_prune): Don't add DW_AT_declaration.
17952 2013-01-31 Richard Biener <rguenther@suse.de>
17954 PR tree-optimization/56157
17955 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17956 match up operand with SLP child.
17958 2013-01-31 Jason Merrill <jason@redhat.com>
17961 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17962 parameters the first time.
17963 (gen_scheduled_generic_parms_dies): Check completeness here.
17965 2013-01-31 Richard Biener <rguenther@suse.de>
17967 PR middle-end/53073
17968 * common.opt (faggressive-loop-optimizations): New flag,
17969 enabled by default.
17970 * doc/invoke.texi (faggressive-loop-optimizations): Document.
17971 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17972 infer_loop_bounds_from_undefined by it.
17974 2013-01-31 Richard Biener <rguenther@suse.de>
17976 PR tree-optimization/56150
17977 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17978 visit virtual operands.
17979 (find_uses_to_rename_bb): Likewise.
17981 2013-01-31 Richard Biener <rguenther@suse.de>
17983 PR tree-optimization/56150
17984 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17985 mixed store non-store stmts.
17987 2013-01-30 Jakub Jelinek <jakub@redhat.com>
17990 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17991 LIBASAN_EARLY_SPEC is defined.
17992 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17993 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17995 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17998 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17999 invalid args instead of ICEing on it.
18000 (ix86_valid_target_attribute_tree): Return error_mark_node if
18001 ix86_valid_target_attribute_inner_p failed.
18002 (ix86_valid_target_attribute_p): Return false only if
18003 ix86_valid_target_attribute_tree returned error_mark_node. Allow
18004 target("default") attribute.
18005 (sorted_attr_string): Change argument from const char * to tree,
18006 merge in all target attribute arguments rather than just one.
18007 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
18008 instead of free. Avoid using strcat.
18009 (ix86_mangle_function_version_assembler_name): Mangle
18010 target("default") as if no target attribute is present. Adjust
18011 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
18012 instead of xmalloc and XDELETEVEC instead of free.
18013 (ix86_function_versions): Don't return true if one of the decls
18014 doesn't have target attribute. If they don't and one of the decls
18015 is DECL_FUNCTION_VERSIONED, report an error. Adjust
18016 sorted_attr_string caller. Use XDELETEVEC instead of free.
18017 (ix86_supports_function_versions): Remove.
18018 (make_name): Fix up formatting.
18019 (make_dispatcher_decl): Remove resolver_name and its initialization.
18020 Avoid leaking memory.
18021 (is_function_default_version): Return true if there is
18022 target("default") attribute rather than no target attribute at all.
18023 (make_resolver_func): Avoid leaking memory.
18024 (ix86_generate_version_dispatcher_body): Likewise.
18025 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
18026 * target.def (supports_function_versions): Remove.
18027 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
18028 * doc/tm.texi: Regenerated.
18030 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
18032 PR rtl-optimization/56144
18033 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
18034 for values with side effects.
18036 2013-01-30 Richard Biener <rguenther@suse.de>
18038 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
18039 (sparseset_pop): Likewise.
18040 * cfganal.c (compute_idf): Likewise. Increase work-stack size
18041 to be able to use quick_push in the worker loop.
18043 2013-01-30 Marek Polacek <polacek@redhat.com>
18045 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
18047 2013-01-30 Richard Biener <rguenther@suse.de>
18050 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
18052 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
18054 PR tree-optimization/56064
18055 * fixed-value.c (fixed_from_double_int): New function.
18056 * fixed-value.h (fixed_from_double_int): New prototype.
18057 (const_fixed_from_double_int): New static inline function.
18058 * fold-const.c (native_interpret_fixed): New static function.
18059 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
18060 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
18061 (native_encode_fixed): New static function.
18062 (native_encode_expr) <FIXED_CST>: Use it.
18063 (native_interpret_int): Move double_int worker code to...
18064 * double-int.c (double_int::from_buffer): ...this new static method.
18065 * double-int.h (double_int::from_buffer): Prototype it.
18067 2013-01-30 Richard Biener <rguenther@suse.de>
18069 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
18070 New pointer-map and obstack.
18071 (init_alias_vars): Allocate pointer-map and obstack.
18072 (delete_points_to_sets): Free them.
18073 (find_what_var_points_to): Cache result.
18074 (find_what_p_points_to): Adjust for changed interface of
18075 find_what_var_points_to.
18076 (compute_points_to_sets): Likewise.
18077 (ipa_pta_execute): Likewise.
18079 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18081 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
18082 * configure: Regenerate.
18083 * config.in: Regenerate.
18084 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
18085 #nobits/#progbits if supported.
18087 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
18090 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
18091 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
18092 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
18094 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18096 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
18097 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
18099 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18101 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
18102 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
18104 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18106 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
18108 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
18109 * config/arm/cortex-a7.md: New bypasses using
18110 arm_mac_accumulator_is_result.
18112 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18114 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
18115 (cortex_a7_neon_mla): Likewise.
18116 (cortex_a7_fpfmad): New reservation.
18117 (cortex_a7_fpmacs): Use ffmas and update required units.
18118 (cortex_a7_fpmuld): Update required units and latency.
18119 (cortex_a7_fpmacd): Likewise.
18120 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
18121 (cortex_a7_neon). Likewise.
18122 (bypass) Update participating units.
18124 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18126 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
18127 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
18129 * config/arm/vfp11.md (vfp_farith): Use ffmas.
18130 (vfp_fmul): Use ffmad.
18131 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
18132 (cortex_r4_fmacd): Use ffmad.
18133 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
18134 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
18135 (cortex_a9_fmacd): Use ffmad.
18136 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
18137 (cortex_a8_vfp_macd): Use ffmad.
18138 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
18139 (cortex_a5_fpmacd): Use ffmad.
18140 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
18141 (cortex_a15_vfp_macd): Use ffmad.
18142 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
18144 2013-01-29 Jason Merrill <jason@redhat.com>
18147 * varasm.c (default_assemble_visibility): Don't warn about
18148 visibility on artificial decls.
18150 2013-01-29 Richard Biener <rguenther@suse.de>
18152 PR tree-optimization/56113
18153 * tree-ssa-structalias.c (equiv_class_lookup): Also return
18155 (label_visit): Free duplicate bitmaps and record the leader instead.
18156 (perform_var_substitution): Adjust.
18158 2013-01-29 Richard Biener <rguenther@suse.de>
18160 PR tree-optimization/55270
18161 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
18162 the CFG, schedule loops for fixup.
18164 2013-01-29 Nick Clifton <nickc@redhat.com>
18166 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
18169 2013-01-28 Leif Ekblad <leif@rdos.net>
18171 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
18172 * config/i386/i386.h (TARGET_RDOS): New macro.
18173 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
18174 * config/i386/i386.c (ix86_option_override_internal): For 64bit
18175 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
18176 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
18177 DEFAULT_LARGE_SECTION_THRESHOLD.
18178 * config/i386/i386.md (R14_REG, R15_REG): New constants.
18179 * config/i386/rdos.h: New file.
18180 * config/i386/rdos64.h: New file.
18182 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
18185 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
18188 2013-01-28 Jakub Jelinek <jakub@redhat.com>
18190 PR rtl-optimization/56117
18191 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
18192 call cselib_lookup_from_insn on the MEM before calling
18193 add_insn_mem_dependence.
18195 2013-01-28 Richard Biener <rguenther@suse.de>
18197 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
18198 to a stmt that didn't have one.
18199 (copy_phis_for_bb): Likewise for PHI arguments.
18200 (copy_debug_stmt): Likewise for debug stmts.
18202 2013-01-28 Richard Biener <rguenther@suse.de>
18204 PR tree-optimization/56034
18205 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
18206 (partition_builtin_p): Adjust.
18207 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
18208 it is the last partition.
18209 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
18210 up the vertex for the definition.
18211 (classify_partition): Classify whether a partition is a
18212 PKIND_REDUCTION, thus has uses outside of the loop.
18213 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
18214 Merge all PKIND_REDUCTION partitions into the last partition.
18215 (tree_loop_distribution): Seed partitions from reductions as well.
18217 2013-01-28 Jakub Jelinek <jakub@redhat.com>
18219 PR tree-optimization/56125
18220 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
18221 pow(x,c) into sqrt(x) * powi(x, n/2) or
18222 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
18223 optimizing for size.
18224 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
18225 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
18228 PR tree-optimization/56094
18229 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
18230 to UNKNOWN_LOCATION while gimplifying expr.
18232 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
18235 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
18236 operand 0 in movabs insn template for -masm=intel asm alternative.
18237 (*movabs<mode>_2): Ditto for operand 1.
18239 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
18242 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
18245 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
18247 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
18250 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
18252 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
18253 Undef to avoid warning.
18255 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
18257 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
18258 * configure: Regenerate.
18260 2013-01-25 Jakub Jelinek <jakub@redhat.com>
18262 PR tree-optimization/56098
18263 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
18264 for stmts with volatile ops.
18265 (cond_store_replacement): Don't optimize if assign has volatile ops.
18266 (cond_if_else_store_replacement_1): Don't optimize if either
18267 then_assign or else_assign have volatile ops.
18268 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
18271 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18273 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
18275 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18277 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
18278 missing ':' in asm example.
18280 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
18282 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
18283 entries into lane and laneq entries.
18284 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
18285 Remove AdvSIMD scalar modes.
18286 (aarch64_sq<r>dmulh_laneq<mode>): New.
18287 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
18289 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
18290 builtin implementations to relfect changes in RTL in aarch64-simd.md.
18291 * config/aarch64/iterators.md (VCOND): New.
18294 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18297 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
18298 Add NULL LIBNAME argument to existing definitions.
18299 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
18300 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
18301 * config/avr/avr.c (DEF_BUILTIN): Same.
18302 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
18303 (avr_expand_builtin): Expand to a vanilla call if a libgcc
18304 implementation is available (DECL_ASSEMBLER_NAME is set).
18305 (avr_fold_absfx): New static function.
18306 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
18307 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
18308 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
18309 AVR_BUILTIN_ABSLLK.
18310 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
18311 (abshk, absk, abslk, absllk): Provide as static inline functions.
18313 2013-01-25 Marek Polacek <polacek@redhat.com>
18315 PR tree-optimization/56035
18316 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
18318 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
18320 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
18321 (*movtf_internal_rex64): Add (!o,C) alternative
18322 (*movxf_internal_rex64): Ditto.
18323 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
18325 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
18327 * doc/invoke.texi: fix typo.
18328 * doc/objc.texi: fix typo.
18330 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
18332 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
18333 for the first two alternatives.
18335 2013-01-24 Diego Novillo <dnovillo@google.com>
18337 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
18338 (ggc-zone.o): Remove.
18339 * configure.ac: Remove option --with-gc.
18340 * configure: Re-generate.
18341 * doc/install.texi: Remove documentation for --with-gc.
18342 * gengtype.c (write_enum_defn): Remove. Update all users.
18343 (write_Types_process_field): Remove generation of gt_e_* argument.
18344 (output_type_enum): Remove. Update all users.
18345 (write_enum_defn): Remove. Update all users.
18346 (enum alloc_zone): Remove. Update all users.
18347 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
18348 * ggc-common.c (ggc_splay_alloc): Remove first argument.
18349 Update all callers.
18350 (struct ptr_data): Remove field TYPE. Update all users.
18351 (gt_pch_note_object): Remove argument TYPE. Update all users.
18352 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
18354 * ggc-none.c (ggc_alloc_typed_stat): Remove.
18355 (struct alloc_zone): Remove.
18356 (ggc_internal_alloc_zone_stat): Remove.
18357 (ggc_internal_cleared_alloc_zone_stat): Remove.
18358 * ggc-page.c (ggc_alloc_typed_stat): Remove.
18359 (ggc_pch_count_object): Remove last argument. Update all users.
18360 (ggc_pch_alloc_object): Remove last argument. Update all users.
18361 (struct alloc_zone): Remove.
18362 * ggc-zone.c: Remove.
18363 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
18364 (struct alloc_zone): Remove.
18365 (ggc_alloc_typed_stat): Remove.
18366 (ggc_alloc_typed): Remove.
18367 (ggc_splay_alloc): Remove first argument.
18368 (rtl_zone): Remove. Update all users.
18369 (tree_zone): Remove. Update all users.
18370 (tree_id_zone): Remove. Update all users.
18371 (ggc_internal_zone_alloc_stat): Remove. Update all users.
18372 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
18373 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
18374 * tree-ssanames.c: Remove references to zone allocator in comments.
18376 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
18378 * config/avr/avr.c (avr_out_fract): Make register numbers that
18379 might be outside of source operand signed.
18381 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
18383 * config/i386/constraints.md (Yf): New constraint.
18384 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
18385 of f constraint to conditionaly disable x87 register preferences.
18386 (*movdf_internal): Ditto.
18387 (*movsf_internal): Ditto.
18389 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
18391 PR inline-asm/55934
18392 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
18393 that have operands with impossible constraints.
18394 Add a FIXME for a speed-up opportunity.
18395 * lra-constraints.c (process_alt_operands): Verify that a class
18396 selected from constraints on asms is valid for the operand mode.
18397 (curr_insn_transform): Remove incorrect comment.
18399 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
18401 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
18402 TOC operand is a valid symbol ref in the constant pool.
18404 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
18406 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
18408 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
18411 * config/avr/stdfix.h: New file.
18412 * t-avr (stdfix-gcc.h): New rule to build it.
18413 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
18415 2013-01-23 Kostya Serebryany <kcc@google.com>
18417 * config/darwin.h: remove dependency on
18418 CoreFoundation (asan on Mac OS).
18420 2013-01-23 Jakub Jelinek <jakub@redhat.com>
18423 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
18424 instead of cmpdi_operand for first comparison operand.
18425 Don't assert that comparison operands aren't both constants.
18427 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
18429 * doc/install.texi (Downloading the Source): Update references to
18430 downloading separate components.
18432 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
18434 * doc/extend.texi (__int128): Improve grammar.
18436 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
18439 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
18440 alternative to (o,r).
18441 (*movdi_internal_rex64): Remove (!o,n) alternative.
18442 (DImode immediate->memory splitter): Remove.
18443 (DImode immediate->memory peephole2): Remove.
18444 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
18445 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
18446 alternative to (!o,*r).
18447 (*movtf_internal_sse): New pattern.
18448 (*movxf_internal_rex64): New pattern.
18449 (*movxf_internal): Disable for TARGET_64BIT.
18450 (*movdf_internal_rex64): Remove (!o,F) alternative.
18452 2013-01-22 Jakub Jelinek <jakub@redhat.com>
18454 PR middle-end/56074
18455 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
18456 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
18457 * tree-vect-loop-manip.c (find_loop_location): Also ignore
18458 stmt locations where LOCATION_LOCUS of the stmt location is
18459 UNKNOWN_LOCATION or BUILTINS_LOCATION.
18462 * config/i386/i386.md (UNSPEC_STOS): New.
18463 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
18464 *strsetqi_1): Add UNSPEC_STOS.
18466 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
18469 * doc/invoke.texi: Remove left over -Wsynth example.
18471 2013-01-21 Jakub Jelinek <jakub@redhat.com>
18473 PR tree-optimization/56051
18474 * fold-const.c (fold_binary_loc): Don't fold
18475 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
18476 a narrowing conversion, or widening conversion from signed
18479 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
18481 PR rtl-optimization/56023
18482 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
18483 dependent on debug instruction.
18485 2013-01-21 Martin Jambor <mjambor@suse.cz>
18487 PR middle-end/56022
18488 * function.c (allocate_struct_function): Call
18489 invoke_set_current_function_hook earlier.
18491 2013-01-21 Jakub Jelinek <jakub@redhat.com>
18493 * reload1.c (init_reload): Only initialize reload_obstack
18494 during the first call.
18496 2013-01-21 Marek Polacek <polacek@redhat.com>
18498 * cfgloop.c (verify_loop_structure): Fix up grammar.
18500 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
18502 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
18503 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
18505 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18508 * config/arm/marvell-pj4.md: Update copyright year.
18509 Fix up use of alu to alu_reg and simple_alu_imm.
18511 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
18513 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
18515 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
18518 * lra-constraints.c (curr_insn_transform): Don't reuse original
18519 insn for secondary memory move when memory mode should be different.
18521 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18523 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18524 atomic_storedi_1): New patterns.
18526 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
18528 btver2 pipeline descriptions.
18529 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18531 * config/i386/i386.md (btver2_decode): New type attributes.
18532 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18534 * config/i386/btver2.md: New file describing btver2 pipelines.
18536 2013-01-19 Andrew Pinski <apinski@cavium.com>
18538 PR tree-optimization/52631
18539 * tree-ssa-sccvn (visit_use): Before looking up the original
18540 statement, try looking up the simplified expression.
18542 2013-01-19 Anthony Green <green@moxielogic.com>
18544 * config/moxie/moxie.c (moxie_expand_prologue): Set
18545 current_function_static_stack_size.
18547 2013-01-18 Jakub Jelinek <jakub@redhat.com>
18549 PR tree-optimization/56029
18550 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18551 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18553 2013-01-18 Sharad Singhai <singhai@google.com>
18555 PR tree-optimization/55995
18556 * dumpfile.c (dump_loc): Print location only if available.
18557 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18559 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
18562 * lra-constraints.c (curr_insn_transform): Reuse original insn for
18563 secondary memory move.
18564 (inherit_reload_reg): Use rclass instead of cl for
18565 check_secondary_memory_needed_p.
18567 2013-01-18 Jakub Jelinek <jakub@redhat.com>
18569 PR middle-end/56015
18570 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18571 the case where writing real complex part of target modifies op1.
18573 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
18575 * config/aarch64/aarch64-simd.md
18576 (aarch64_vcond_internal<mode>): Handle unordered cases.
18577 * config/aarch64/iterators.md (v_cmp_result): New.
18579 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
18580 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18582 * config/arm/marvell-pj4.md: New file.
18583 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18584 * config/arm/arm.md (generic_sched): Add marvell_pj4.
18585 (generic_vfp): Likewise.
18586 * config/arm/arm-cores.def: Add marvell-pj4.
18587 * config/arm/arm-tune.md: Regenerate.
18588 * config/arm/arm-tables.opt: Regenerate.
18589 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18590 * doc/invoke.texi: Document marvell-pj4.
18592 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
18594 * config/aarch64/arm_neon.h: Map scalar types to standard types.
18596 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18601 * var-tracking.c (negative_power_of_two_p): New.
18602 (global_get_addr_cache, local_get_addr_cache): New.
18603 (get_addr_from_global_cache, get_addr_from_local_cache): New.
18604 (vt_canonicalize_addr): Rewrite using the above. Adjust the
18606 (vt_stack_offset_p): Remove.
18607 (vt_canon_true_dep): Always canonicalize loc's address.
18608 (clobber_overlapping_mems): Make sure we have a MEM.
18609 (local_get_addr_clear_given_value): New.
18610 (val_reset): Clear local cached entries.
18611 (compute_bb_dataflow): Create and release the local cache.
18612 Disable duplicate MEMs clobbering.
18613 (emit_notes_in_bb): Clobber MEMs likewise.
18614 (vt_emit_notes): Create and release the local cache.
18615 (vt_initialize, vt_finalize): Create and release the global
18616 cache, respectively.
18617 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18619 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18621 PR libmudflap/53359
18622 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18623 not found in the symtab.
18625 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18628 PR rtl-optimization/55547
18629 PR rtl-optimization/53827
18632 * alias.c (offset_overlap_p): New, factored out of...
18633 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
18634 the conservative special case for symbolic constants. Don't
18635 adjust zero sizes on alignment.
18637 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
18639 PR rtl-optimization/52573
18640 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18641 REG_UNUSED for the same register.
18643 2013-01-17 Richard Biener <rguenther@suse.de>
18644 Marek Polacek <polacek@redhat.com>
18646 PR rtl-optimization/55833
18647 * loop-unswitch.c (unswitch_loops): Move loop verification...
18648 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
18649 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18650 Set it to true when we're removing a loop from hierarchy tree in
18651 an irreducible region.
18652 (fix_bb_placements): Adjust caller.
18653 (fix_loop_placements): Likewise.
18655 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
18657 * config/avr/builtins.def (DEF_BUILTIN): Factor out
18658 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18659 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18660 Remove ID. Adjust comments.
18661 * config/avr/avr-c.c (avr_builtin_name): Remove.
18662 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18663 * config/avr/avr.c (avr_tolower): New static function.
18664 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
18665 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18666 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18669 2013-01-17 Jan Hubicka <jh@suse.cz>
18671 PR tree-optimization/55273
18672 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18674 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
18677 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18678 store through atomic_store<mode>_1.
18679 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18681 2013-01-17 Martin Jambor <mjambor@suse.cz>
18683 PR tree-optimizations/55264
18684 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18685 for virtual methods.
18686 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18687 virtual methods before inlining is over.
18688 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18690 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18693 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
18695 PR rtl-optimization/56005
18696 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18697 pending reads for prefetch.
18699 2013-01-16 Ian Bolton <ian.bolton@arm.com>
18701 * config/aarch64/aarch64.md
18702 (*cstoresi_neg_uxtw): New pattern.
18703 (*cmovsi_insn_uxtw): New pattern.
18704 (*<optab>si3_uxtw): New pattern.
18705 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18706 (*<optab>si3_insn_uxtw): New pattern.
18707 (*bswapsi2_uxtw): New pattern.
18709 2013-01-16 Richard Biener <rguenther@suse.de>
18711 * tree-inline.c (tree_function_versioning): Remove set but
18712 never used variable.
18714 2013-01-16 Richard Biener <rguenther@suse.de>
18716 PR tree-optimization/55964
18717 * tree-flow.h (rename_variables_in_loop): Remove.
18718 (rename_variables_in_bb): Likewise.
18719 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18720 (copy_loop_before): Adjust and delete update-ssa status.
18721 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18722 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
18723 (rename_variables_in_loop): Remove.
18724 (slpeel_update_phis_for_duplicate_loop): Likewise.
18725 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18726 use available cfg machinery instead of duplicating it.
18727 Update PHI nodes and perform poor-mans SSA update here.
18728 (slpeel_tree_peel_loop_to_edge): Adjust.
18730 2013-01-16 Richard Biener <rguenther@suse.de>
18732 PR tree-optimization/54767
18733 PR tree-optimization/53465
18734 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18735 (vrp_visit_phi_node): For PHI arguments coming via backedges
18736 drop all symbolical range information.
18737 (execute_vrp): Compute backedges.
18739 2013-01-16 Richard Biener <rguenther@suse.de>
18741 * doc/install.texi: Update CLooG and ISL requirements to
18744 2013-01-16 Christian Bruel <christian.bruel@st.com>
18747 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18748 (broken_move): Handle UNSPECV_SP_SWITCH_B.
18749 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18751 2013-01-16 DJ Delorie <dj@redhat.com>
18753 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18754 (UNSPECV_SP_SWITCH_E): New.
18755 (sp_switch_1): Change to an unspec.
18756 (sp_switch_2): Change to an unspec. Don't use post-inc when we
18759 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
18761 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18762 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18763 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18764 (expand_mem_thread_fence): Ditto.
18765 (expand_mem_signal_fence): Ditto.
18766 (expand_atomic_load): Ditto.
18767 (expand_atomic_store): Ditto.
18769 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
18771 PR rtl-optimization/55547
18772 PR rtl-optimization/53827
18775 * alias.c (memrefs_conflict_p): Set sizes to negative after
18778 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18781 * function.c (thread_prologue_and_epilogue_insns): Always
18782 add crtl->drap_reg to set_up_by_prologue.set, even if
18783 stack_realign_drap is false.
18785 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
18787 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18788 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18789 *call): Fix indention.
18791 2013-01-15 Tom de Vries <tom@codesourcery.com>
18794 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18797 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
18799 PR rtl-optimization/55153
18800 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18802 2013-01-15 Martin Jambor <mjambor@suse.cz>
18804 PR tree-optimization/55920
18805 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18806 accesses as grp_to_be_debug_replaced.
18808 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18810 PR tree-optimization/55920
18811 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18812 there is non-useless type conversion needed from debug rhs to lhs,
18813 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18815 2013-01-15 Joseph Myers <joseph@codesourcery.com>
18816 Mikael Pettersson <mikpe@it.uu.se>
18819 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18821 (ASM_OUTPUT_CASE_LABEL): Remove.
18822 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18823 * final.c (shorten_branches): Update alignment of labels before
18824 jump tables if CASE_VECTOR_SHORTEN_MODE.
18826 2013-01-15 Richard Biener <rguenther@suse.de>
18829 * system.h: Do not include gmp.h for building host tools.
18831 2013-01-15 Richard Biener <rguenther@suse.de>
18833 PR middle-end/55882
18834 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18835 account for bitpos when computing alignment.
18837 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
18839 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18840 (ix86_target_macros_internal): Likewise.
18842 * config/i386/i386.c (m_CORE2I7): Removed.
18843 (m_CORE_HASWELL): New macro.
18844 (m_CORE_ALL): Likewise.
18845 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18846 (initial_ix86_arch_features): Likewise.
18847 (processor_target_table): Initializations for Core avx2.
18848 (cpu_names): New names "core-avx2".
18849 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18850 PROCESSOR_CORE_HASWELL.
18851 (ix86_issue_rate): New case.
18852 (ia32_multipass_dfa_lookahead): Likewise.
18853 (ix86_sched_init_global): Likewise.
18855 * config/i386/i386.h (TARGET_HASWELL): New macro.
18856 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18857 (processor_type): New PROCESSOR_HASWELL.
18859 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18861 PR tree-optimization/55955
18862 * tree-vect-loop.c (vectorizable_reduction): Give up early on
18863 *SHIFT_EXPR and *ROTATE_EXPR codes.
18865 PR tree-optimization/48766
18866 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18867 -ftrapv disable -fwrapv.
18869 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18872 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18873 etc. to 1 and not to __flash.
18874 Use LL suffix for __INT24_MAX__ with -mint8.
18875 Use ULL suffix for __UINT24_MAX__ with -mint8.
18877 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18879 * config/avr/avr-arch.h
18880 (struct base_arch_s): Use typedef avr_arch_t instead.
18881 (struct arch_info_s): Use typedef avr_arch_info_t instead.
18882 (struct mcu_type_s): Use typedef avr_mcu_t instead.
18883 * config/avr/avr.c: Same.
18884 * config/avr/avr-devices.c: Same.
18885 * config/avr/driver-avr.c: Same.
18886 * config/avr/gen-avr-mmcu-texi.c: Same.
18887 * config/avr/avr-mcus.def: Adjust comment.
18889 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
18891 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18892 * config/aarch64/iterators.md (VALLDI): New.
18894 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
18895 Andi Kleen <ak@linux.intel.com>
18898 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18899 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18902 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18904 * config/avr/avr-stdint.h: Remove trailing blanks.
18905 * config/avr/avr-log.h: Same.
18906 * config/avr/avr-arch.h: Same.
18907 * config/avr/avr-devices.c: Same.
18908 * config/avr/avr-dimode.md: Same.
18909 * config/avr/predicates.md: Same.
18910 * config/avr/avr-c.c: Same. And fix typo.
18912 * config/avr/avr-protos.h: Same. And:
18913 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18914 (init_cumulative_args): Rename to avr_init_cumulative_args.
18915 (expand_prologue): Rename to avr_expand_prologue.
18916 (expand_epilogue): Rename to avr_expand_epilogue.
18917 (adjust_insn_length): Rename to avr_adjust_insn_length.
18918 (notice_update_cc): Rename to avr_notice_update_cc.
18919 (final_prescan_insn): Rename to avr_final_prescan_insn.
18920 * config/avr/avr.c: Same.
18921 * config/avr/avr.h: Same.
18922 * config/avr/avr.md: Remove trailing blanks.
18923 (prologue): Use avr_expand_prologue.
18924 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18926 2013-01-14 Richard Biener <rguenther@suse.de>
18928 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18929 verify_location, collect_subblocks): New functions.
18930 (verify_gimple_in_cfg): Verify that locations only reference
18931 BLOCKs in the functions BLOCK tree.
18933 2013-01-14 Richard Biener <rguenther@suse.de>
18935 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18937 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18939 (insert_out_of_ssa_copy_on_edge): Likewise.
18940 (rewrite_close_phi_out_of_ssa): Likewise.
18941 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18943 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18944 propagated constants.
18945 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18948 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18950 * config/avr/avr-modes.def: Add GPL copyright notice.
18952 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
18954 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18955 MEMMODEL_MASK to determine memory model.
18956 (atomic_store<mode>): Ditto from operands[2].
18957 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18959 2013-01-13 Jakub Jelinek <jakub@redhat.com>
18962 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18963 (fold_gimple_assign): Don't call unshare_expr here.
18964 (fold_ctor_reference): Call unshare_expr.
18966 2013-01-13 Terry Guo <terry.guo@arm.com>
18968 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18969 * doc/fragments.texi: Document MULTILIB_REUSE.
18970 * gcc.c (multilib_reuse): New internal spec.
18971 (set_multilib_dir): Also search multilib from multilib_reuse.
18972 * genmultilib (tmpmultilib3): Refactor code.
18973 (tmpmultilib4): Ditto.
18974 (multilib_reuse): New multilib argument.
18976 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
18978 * Makefile.in: Update copyright.
18980 2013-01-12 Tom de Vries <tom@codesourcery.com>
18982 PR middle-end/55890
18983 * calls.c (expand_call): Check if arg_nr is valid.
18985 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18987 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18988 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18989 documentation. Add missing '__' in front of
18990 __builtin_ia32_packssdw256.
18992 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
18995 * config/s390/s390.c (s390_preferred_reload_class): Do not return
18996 NO_REGS for larl operands.
18997 (s390_reload_larl_operand): Use s390_load_address instead of
19000 2013-01-11 Richard Biener <rguenther@suse.de>
19002 * tree-cfg.c (verify_node_sharing_1): Split out from ...
19003 (verify_node_sharing): ... here.
19004 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
19006 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
19008 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
19009 Substitute TREECHECKING.
19010 * configure: Regenerate.
19011 * Makefile.in (TREECHECKING): New.
19013 2013-01-11 Richard Guenther <rguenther@suse.de>
19015 PR tree-optimization/44061
19016 * tree-vrp.c (extract_range_basic): Compute zero as
19017 value-range for __builtin_constant_p of function parameters.
19019 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
19021 Update copyright years.
19023 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
19025 PR rtl-optimization/55672
19026 * lra-eliminations.c (mark_not_eliminable): Permit addition with
19027 const to be eliminable.
19029 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
19031 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
19032 * configure: Regenerate.
19034 2013-01-10 Richard Biener <rguenther@suse.de>
19036 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
19038 2013-01-10 Richard Biener <rguenther@suse.de>
19041 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
19042 locations for virtual PHI arguments.
19043 (rewrite_update_phi_arguments): Likewise.
19045 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
19047 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
19050 2013-01-10 Jakub Jelinek <jakub@redhat.com>
19052 PR tree-optimization/55921
19053 * tree-complex.c (expand_complex_asm): New function.
19054 (expand_complex_operations_1): Call it for GIMPLE_ASM.
19056 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
19059 * config/s390/s390.c (s390_symref_operand_p)
19060 (s390_loadrelative_operand_p): Merge the two functions.
19061 (s390_check_qrst_address, print_operand_address): Add parameters
19062 to s390_loadrelative_operand_p invokation.
19063 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
19064 (s390_reload_larl_operand, s390_secondary_reload): Use
19065 s390_loadrelative_operand_p instead of s390_symref_operand_p.
19066 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
19068 2013-01-09 Mike Stump <mikestump@comcast.net>
19070 * dse.c (record_store): Remove unnecessary assert.
19072 2013-01-09 Jan Hubicka <jh@suse.cz>
19074 PR tree-optimization/55569
19075 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
19076 * cfgloop.h (scale_loop_profile): Likewise.
19078 2013-01-09 Jan Hubicka <jh@suse.cz>
19081 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
19083 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
19085 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
19087 PR middle-end/55114
19088 * expr.h (maybe_emit_group_store): Declare.
19089 * expr.c (maybe_emit_group_store): New function.
19090 * builtins.c (expand_builtin_int_roundingfn): Call it.
19091 (expand_builtin_int_roundingfn_2): Likewise.
19093 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
19095 PR rtl-optimization/55829
19096 * lra-constraints.c (match_reload): Add code for absent output.
19097 (curr_insn_transform): Add code for reloads of matched inputs
19100 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
19102 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
19103 attribute of movddup insn to DF.
19104 (*vec_interleave_lowv2df): Ditto.
19105 (vec_dupv2df): Ditto.
19107 2013-01-09 Jan Hubicka <jh@suse.cz>
19109 PR tree-optimiation/55875
19110 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
19111 EVERY_ITERATION parameter.
19112 (number_of_iterations_exit): Check if exit is executed every iteration.
19113 (idx_infer_loop_bounds): Similarly here.
19114 (n_of_executions_at_most): Simplify
19115 to only test for cases where statement is dominated by the
19116 particular bound; handle correctly the "postdominance" test.
19117 (scev_probably_wraps_p): Use max loop iterations info
19118 as a global bound first.
19120 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
19121 Nick Clifton <nickc@redhat.com>
19123 * config/v850/v850.md (cbranchsf4): New pattern.
19124 (cstoresf4): New pattern.
19125 (cbranchdf4): New pattern.
19126 (cstoredf4): New pattern.
19127 (movsicc): Disallow floating point comparisons.
19128 (cmpsf_le_insn): Fix order of operators.
19129 (cmpsf_lt_insn): Likewise.
19130 (cmpsf_eq_insn): Likewise.
19131 (cmpdf_le_insn): Likewise.
19132 (cmpdf_lt_insn): Likewise.
19133 (cmpdf_eq_insn): Likewise.
19134 (cmpsf_ge_insn): Use LE comparison.
19135 (cmpdf_ge_insn): Likewise.
19136 (cmpsf_gt_insn): Use LT comparison.
19137 (cmpdf_gt_insn): Likewise.
19138 (cmpsf_ne_insn): Delete pattern.
19139 (cmpdf_ne_insn): Delete pattern.
19140 * config/v850/v850.c (v850_gen_float_compare): Use
19141 gen_cmpdf_eq_insn for NE comparison.
19142 (v850_float_z_comparison_operator)
19143 (v850_float_nz_comparison_operator): Move from here ...
19144 * config/v850/predicates.md: ... to here. Move GT and GE
19145 comparisons into v850_float_z_comparison_operator.
19146 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
19148 (v850_float_nz_comparison_operator): Likewise.
19150 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19152 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
19153 with calls to gen_insvsi/gen_insvdi.
19155 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
19157 * config/i386/i386.c (initial_ix86_tune_features): Set up
19158 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
19160 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
19161 Jakub Jelinek <jakub@redhat.com>
19163 PR tree-optimization/48189
19164 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
19165 If nitercst is 0, don't predict the exit edge.
19167 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
19169 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
19170 in asm_fprintf with reg_names.
19171 (aarch64_print_operand_address): Likewise.
19172 (aarch64_return_addr): Likewise.
19173 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
19175 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19177 * config/pa/pa.h (VAL_U6_BITS_P): Define.
19178 (INT_U6_BITS): Likewise.
19179 * config/pa/predicates.md (uint6_operand): New predicate.
19180 (shift5_operand, shift6_operand): Likewise.
19181 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
19183 (lshrdi3): Use shift6_operand.
19184 (shrpsi4, shrpdi4): New insn patterns.
19185 (extzv): Delete expander.
19186 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
19187 predicates in unamed zero extract patterns. Tighten common constraint.
19188 (extv): Delete expander.
19189 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
19190 predicates in unamed sign extract patterns. Tighten common constraint.
19191 (insv): Delete expander.
19192 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
19193 predicates in unamed insert patterns. Tighten common constraint.
19194 Change uint32_operand predicate to uint6_operand predicate in unamed
19195 DImode pattern to insert constant values of type 1...1xxxx.
19197 2013-01-04 Jan Hubicka <jh@suse.cz>
19199 PR tree-optimization/55823
19200 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
19203 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19204 Uros Bizjak <ubizjak@gmail.com>
19206 PR rtl-optimization/55845
19207 * df-problems.c (can_move_insns_across): Stop scanning at
19208 volatile_insn_p source instruction or give up if
19209 across_from .. across_to range contains any volatile_insn_p
19212 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
19214 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
19215 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
19217 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
19218 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
19220 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19223 * asan.c (asan_clear_shadow): New function.
19224 (asan_emit_stack_protection): Use it.
19226 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
19228 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
19229 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
19230 with tab instead of space.
19232 2013-01-08 Nick Clifton <nickc@redhat.com>
19234 * config/rl78/rl78.c (rl78_expand_prologue): Always select
19235 register bank 0 at the start of an interrupt handler.
19236 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
19239 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
19241 * config/aarch64/aarch64-simd.md
19242 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
19243 (aarch64_simd_bsl): Likewise.
19244 (aarch64_vcond_internal<mode>): Likewise.
19245 (vcond<mode><mode>): Likewise.
19246 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
19247 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
19249 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
19251 * config/aarch64/aarch64-builtins.c
19252 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
19254 2013-01-08 Martin Jambor <mjambor@suse.cz>
19257 * tree-sra.c (analyze_access_subtree): Return true also after
19258 potentially creating a debug-only replacement.
19260 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19262 PR middle-end/55890
19263 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
19265 PR tree-optimization/54120
19266 * tree-vrp.c (range_fits_type_p): Don't allow
19267 src_precision < precision from signed vr to unsigned_p
19268 if vr->min or vr->max is negative.
19269 (simplify_float_conversion_using_ranges): Test can_float_p
19270 against CODE_FOR_nothing.
19272 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19273 Richard Biener <rguenther@suse.de>
19275 PR middle-end/55851
19276 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
19277 types instead of just INTEGER_TYPE types.
19279 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
19281 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
19284 2013-01-07 Steve Ellcey <sellcey@mips.com>
19287 * config/mips/mips.opt: Change mad to mmad to match documentation.
19289 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19292 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
19293 .progmemx.data now.
19295 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19298 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
19299 (avr_addrspace_t): Add .section_name field.
19300 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
19302 (avr_addrspace): Same. Initialize .section_name. Remove last
19303 NULL entry. Put __memx into .progmemx.data.
19304 (progmem_section_prefix): Remove.
19305 (avr_asm_init_sections): No need to initialize progmem_section.
19306 (avr_asm_named_section): Use avr_addrspace[].section_name to get
19307 section name prefix.
19308 (avr_asm_select_section): Ditto. And use get_unnamed_section to
19309 retrieve the progmem section.
19310 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
19311 boundary to run over avr_addrspace[].
19312 (avr_register_target_pragmas): Ditto.
19314 2013-01-06 Jakub Jelinek <jakub@redhat.com>
19316 * varasm.c (output_constant_def_contents): For asan_protect_global
19317 protected strings, adjust DECL_ALIGN if needed, before testing for
19319 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
19320 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
19322 (output_object_block): For asan protected decls, emit asan padding
19323 after their contents.
19324 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
19325 (asan_finish_file): Test it here instead.
19327 2013-01-07 Nick Clifton <nickc@redhat.com>
19328 Matthias Klose <doko@debian.org>
19329 Doug Kwan <dougkwan@google.com>
19330 H.J. Lu <hongjiu.lu@intel.com>
19333 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
19335 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
19337 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
19339 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
19341 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
19343 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19346 * doc/install.texi (Cross-Compiler-Specific Options): Document
19349 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
19351 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
19352 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
19353 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
19354 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
19355 vqmovun_high_s64): Fix source operand number and update copyright.
19357 2013-01-07 Richard Biener <rguenther@suse.de>
19359 PR middle-end/55890
19360 * gimple.h (gimple_call_builtin_p): New overload.
19361 * gimple.c (validate_call): New function.
19362 (gimple_call_builtin_p): Likewise.
19363 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19364 Use gimple_call_builtin_p.
19365 (find_func_clobbers): Likewise.
19366 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
19367 (strlen_optimize_stmt): Likewise.
19369 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
19371 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
19372 (vld1q_dup_*): Likewise.
19373 (vld1_*): Likewise.
19374 (vld1q_*): Likewise.
19375 (vld1_lane_*): Likewise.
19376 (vld1q_lane_*): Likewise.
19378 2013-01-07 Richard Biener <rguenther@suse.de>
19380 * lto-streamer.h (LTO_minor_version): Bump to 2.
19382 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
19384 * config/aarch64/aarch64-protos.h
19385 (aarch64_const_double_zero_rtx_p): Rename to...
19386 (aarch64_float_const_zero_rtx_p): ...this.
19387 (aarch64_float_const_representable_p): New.
19388 (aarch64_output_simd_mov_immediate): Likewise.
19389 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
19390 move immediate case.
19391 * config/aarch64/aarch64.c
19392 (aarch64_const_double_zero_rtx_p): Rename to...
19393 (aarch64_float_const_zero_rtx_p): ...this.
19394 (aarch64_print_operand): Allow printing of new constants.
19395 (aarch64_valid_floating_const): New.
19396 (aarch64_legitimate_constant_p): Check for valid floating-point
19398 (aarch64_simd_valid_immediate): Likewise.
19399 (aarch64_vect_float_const_representable_p): New.
19400 (aarch64_float_const_representable_p): Likewise.
19401 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
19402 (aarch64_output_simd_mov_immediate): New.
19403 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
19404 (*movdf_aarch64): Likewise.
19405 * config/aarch64/constraints.md (Ufc): New.
19406 (Y): call aarch64_float_const_zero_rtx.
19407 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
19409 2013-01-07 Richard Biener <rguenther@suse.de>
19411 PR tree-optimization/55888
19412 PR tree-optimization/55862
19413 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
19414 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
19415 not if it is contained therein.
19417 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19419 * config/avr/t-avr: Typo.
19421 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19424 * config/avr/t-avr: Don't automatically rebuild
19425 $(srcdir)/config/avr/t-multilib
19426 $(srcdir)/config/avr/avr-tables.opt
19427 $(srcdir)/doc/avr-mmcu.texi
19428 (avr-mcus): New phony target to build them on request.
19429 (s-avr-mlib, s-avr-mmcu-texi): Remove.
19430 * avr/avr-mcus.def: Adjust comments.
19432 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
19434 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
19436 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
19438 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
19440 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
19442 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
19444 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
19446 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
19447 to generate profiling.
19448 * config/rs6000/aix64.h (LIB_SPEC): Same.
19450 2013-01-04 Andrew Pinski <apinski@cavium.com>
19452 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
19454 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
19456 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
19458 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
19460 (ix86_expand_move): Ditto.
19461 (ix86_zero_extend_to_Pmode): Ditto.
19462 (ix86_expand_call): Ditto.
19463 (ix86_expand_special_args_builtin): Ditto.
19464 (ix86_expand_builtin): Ditto.
19466 2013-01-04 Richard Biener <rguenther@suse.de>
19468 PR tree-optimization/55862
19469 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
19470 translating them through PHI nodes.
19472 2013-01-04 Martin Jambor <mjambor@suse.cz>
19474 PR tree-optimization/55755
19475 * tree-sra.c (sra_modify_assign): Do not check that an access has no
19476 children when trying to avoid producing a VIEW_CONVERT_EXPR.
19478 2013-01-04 Marek Polacek <polacek@redhat.com>
19480 PR middle-end/55859
19481 * opts.c (default_options_optimization): Clarify error message.
19483 2013-01-04 Richard Biener <rguenther@suse.de>
19485 PR middle-end/55863
19486 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
19489 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19492 * config/pa/pa.md (movsi): Revert previous change.
19493 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
19496 2013-01-03 Richard Henderson <rth@redhat.com>
19498 * config/i386/i386.c (ix86_expand_move): Always assign to op1
19499 after eliminating TLS symbols.
19501 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
19504 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
19505 * graphite-poly.c (debug_gmp_value): Likewise.
19507 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
19510 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
19511 selected code model, define __code_mode_small__, __code_model_medium__,
19512 __code_model_large__, __code_model_32__ or __code_model_kernel__.
19513 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
19514 xchg temporary register with %k. Declare temporary register as
19516 [__x86_64__]: For medium and large code models, preserve %rbx register.
19518 2013-01-03 Richard Biener <rguenther@suse.de>
19520 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
19521 (dump_subscript): Adjust.
19522 (finalize_ddr_dependent): Do not dump redundant info.
19523 (analyze_siv_subscript): Adjust.
19524 (subscript_dependence_tester): Likewise.
19525 (compute_affine_dependence): Likewise.
19527 2013-01-03 Richard Biener <rguenther@suse.de>
19530 2013-01-03 Richard Biener <rguenther@suse.de>
19532 PR tree-optimization/55857
19533 * tree-vect-stmts.c (vectorizable_load): Do not setup
19534 re-alignment for invariant loads.
19536 2013-01-02 Richard Biener <rguenther@suse.de>
19538 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19539 invariant load do not generate a vector load from the scalar location.
19541 2013-01-03 Richard Biener <rguenther@suse.de>
19543 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19544 for not vectorizing.
19545 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19546 not build INDIRECT_REFs, call get_name once only.
19547 (vect_create_data_ref_ptr): Likewise. Dump base object kind
19548 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19550 2013-01-03 Richard Biener <rguenther@suse.de>
19552 PR tree-optimization/55857
19553 * tree-vect-stmts.c (vectorizable_load): Do not setup
19554 re-alignment for invariant loads.
19556 2013-01-03 Richard Biener <rguenther@suse.de>
19559 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19560 prefer a built-in decl.
19562 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19564 * gcc.c (process_command): Update copyright notice dates.
19565 * gcov.c (print_version): Likewise.
19566 * gcov-dump.c (print_version): Likewise.
19568 PR rtl-optimization/55838
19569 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19570 iv0.step, iv1.step and step.
19572 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19573 Marc Glisse <marc.glisse@inria.fr>
19575 PR tree-optimization/55832
19576 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19577 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19578 integer_{one,zero}_node.
19580 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19583 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19584 * var-tracking.c (reverse_op): Don't add reverse ops to
19585 VALUEs that have already
19586 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19588 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19590 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19592 2013-01-02 Teresa Johnson <tejohnson@google.com>
19594 * dumpfile.c (dump_loc): Print filename with location.
19595 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19596 new location_t parameter to emit complete unroll message with
19597 new dump framework.
19598 (canonicalize_loop_induction_variables): Compute loops location
19599 and pass to try_unroll_loop_completely.
19600 * loop-unroll.c (report_unroll_peel): New function.
19601 (peel_loops_completely): Use new dump format with location
19602 for main dumpfile message, and invoke report_unroll_peel on success.
19603 (decide_unrolling_and_peeling): Ditto.
19604 (decide_peel_once_rolling): Remove old dumpfile message subsumed
19605 by report_unroll_peel.
19606 (decide_peel_completely): Ditto.
19607 (decide_unroll_constant_iterations): Ditto.
19608 (decide_unroll_runtime_iterations): Ditto.
19609 (decide_peel_simple): Ditto.
19610 (decide_unroll_stupid): Ditto.
19611 * cfgloop.c (get_loop_location): New function.
19612 * cfgloop.h (get_loop_location): Declare.
19614 2013-01-02 Sriraman Tallam <tmsriram@google.com>
19616 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19619 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19621 PR middle-end/55198
19622 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19623 BLKmode objects when EXPAND_MEMORY is specified.
19625 2013-01-02 Sriraman Tallam <tmsriram@google.com>
19627 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19629 (fold_builtin_cpu): Do not share cpu model decls across statements.
19631 2013-01-02 Jason Merrill <jason@redhat.com>
19634 * tree.c (build_array_type_1): Revert earlier change.
19636 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
19638 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19640 * config/aarch64/aarch64-tune.md: Re-generate.
19642 2013-01-02 Richard Biener <rguenther@suse.de>
19644 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19645 invariant load do not generate a vector load from the scalar location.
19647 2013-01-02 Richard Biener <rguenther@suse.de>
19650 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19651 * configure: Regenerate.
19653 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
19655 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19656 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19657 (expand_builtin_int_roundingfn_2): Keep the original target around
19658 for the fallback case.
19660 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
19662 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19663 to be clear for sign changes.
19665 2013-01-01 Jan Hubicka <jh@suse.cz>
19667 * ipa-inline-analysis.c: Fix formatting.
19669 2013-01-01 Jakub Jelinek <jakub@redhat.com>
19671 PR tree-optimization/55831
19672 * tree-vect-loop.c (get_initial_def_for_induction): Use
19673 gsi_after_labels instead of gsi_start_bb.
19675 Copyright (C) 2013 Free Software Foundation, Inc.
19677 Copying and distribution of this file, with or without modification,
19678 are permitted in any medium without royalty provided the copyright
19679 notice and this notice are preserved.