1 2013-09-04 Teresa Johnson <tejohnson@google.com>
3 * dumpfile.c (dump_finish): Don't close stderr/stdout.
5 2013-09-04 James Greenhalgh <james.greenhalgh@arm.com>
7 * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
9 2013-09-04 Jan Hubicka <jh@suse.cz>
11 * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
12 * ipa-devirt.c: Include diganostic.h
13 (odr_type_d): Add types and types_set.
14 (hash_type_name): Work for types with vtables during LTO.
15 (odr_hasher::remove): Fix comment; destroy types_set.
16 (add_type_duplicate): New function,
17 (get_odr_type): Use it.
18 (dump_type_inheritance_graph): Dump type duplicates.
19 * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
21 * tree.c (types_same_for_odr): Give exact answers on types with
24 2013-09-04 Dodji Seketeli <dodji@redhat.com>
26 * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
27 explaining their differences.
29 2013-09-04 Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
31 * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
33 2013-09-03 Jeff Law <law@redhat.com>
35 * tree-ssa-threadedge.c (thread_across_edge): Record entire path
36 when not threading through a joiner block. Pass joiner/no joiner
37 state to register_jump_thread.
38 * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
39 state from argument rather than implying on path length.
40 Dump the entire jump thread path into debugging dump.
41 * tree-flow.h (register_jump_thread): Update prototype.
43 2013-08-29 Xinliang David Li <davidxl@google.com>
45 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
46 Remove a trivial gcc_assert.
48 2013-08-29 Xinliang David Li <davidxl@google.com>
50 * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
51 * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
52 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
53 Delay base decl alignment adjustment.
54 * tree-vectorizer.c (vect_destroy_datarefs): New function.
55 * tree-vectorizer.h: New data structure.
56 (set_dr_misalignment): New function.
57 (dr_misalignment): Ditto.
58 * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
59 (vectorizable_load): Ditto.
60 (ensure_base_align): New function.
61 (vectorize_loops): Add dbg_cnt support.
62 (execute_vect_slp): Ditto.
63 * dbgcnt.def: New debug counter.
64 * Makefile: New dependency.
66 2013-09-03 Meador Inge <meadori@codesourcery.com>
70 2013-08-30 Meador Inge <meadori@codesourcery.com>
72 * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
74 2013-09-03 David Edelsohn <dje.gcc@gmail.com>
76 * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
79 2013-09-03 Richard Biener <rguenther@suse.de>
81 * tree-affine.c (add_elt_to_tree): Fix association issue,
82 avoid useless converts and make sure to always return a
83 properly typed result.
85 2013-09-03 Richard Biener <rguenther@suse.de>
88 * fold-const.c (negate_expr_p): Fix division case.
89 (negate_expr): Likewise.
91 2013-09-03 Richard Biener <rguenther@suse.de>
94 * tree-streamer-out.c: Include tm.h.
95 * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
97 2013-09-03 Jan Hubicka <jh@suse.cz>
99 * tree-profile.c (tree_profiling): Cleanup CFG when done.
101 2013-09-03 Alan Modra <amodra@gmail.com>
103 * config.gcc (powerpc*-*-linux*): Add support for little-endian
104 multilibs to big-endian target and vice versa.
105 * config/rs6000/t-linux64: Use := assignment on all vars.
106 (MULTILIB_EXTRA_OPTS): Remove fPIC.
107 (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
108 * config/rs6000/t-linux64le: New file.
109 * config/rs6000/t-linux64bele: New file.
110 * config/rs6000/t-linux64lebe: New file.
112 2013-09-02 Jan Hubicka <jh@suse.cz>
114 * ipa-inline-transform.c (inline_transform): Do not
115 optimize_inline_calls when not optimizing.
117 2013-09-02 Jan Hubicka <jh@suse.cz>
119 * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
120 duplicated nodes for assembler names.
121 * symtab.c (symtab_unregister_node): Do not attempt to unlink
122 hard registers from assembler name hash.
124 2013-09-02 Jan Hubicka <jh@suse.cz>
126 * ipa-split.c (execute_split_functions): Split externally visible
127 functions called once.
129 2013-09-02 Martin Jambor <mjambor@suse.cz>
132 * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
133 linked list. When finding the correct duplicate, also consider also
134 the caller in additon to its inlined_to node.
136 2013-09-02 James Greenhalgh <james.greenhalgh@arm.com>
138 * config/aarch64/aarch64-simd-builtins.def
139 (dup_lane_scalar): Remove.
140 * config/aarch64/aarch64-simd.md
141 (aarch64_simd_dup): Add 'w->w' alternative.
142 (aarch64_dup_lane<mode>): Allow for VALL.
143 (aarch64_dup_lane_scalar<mode>): Remove.
144 (aarch64_dup_lane_<vswap_width_name><mode>): New.
145 (aarch64_get_lane_signed<mode>): Add w->w altenative.
146 (aarch64_get_lane_unsigned<mode>): Likewise.
147 (aarch64_get_lane<mode>): Likewise.
148 * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
149 (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
150 * config/aarch64/iterators.md (VSWAP_WIDTH): New.
151 (VCON): Change container of V2SF.
152 (vswap_width_name): Likewise.
153 * config/aarch64/arm_neon.h
154 (__aarch64_vdup_lane_any): New.
155 (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
156 (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
157 (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
159 2013-09-02 Eric Botcazou <ebotcazou@adacore.com>
162 * expr.c (emit_move_complex): Do not move complex FP values as parts if
163 the source or the destination is a single hard register.
165 2013-09-02 Richard Biener <rguenther@suse.de>
168 * tree-scalar-evolution.c (instantiate_scev_name): Allow
171 2013-09-02 Richard Biener <rguenther@suse.de>
173 * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
174 arithmetic to sizetype.
176 2013-09-02 Bin Cheng <bin.cheng@arm.com>
178 * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
179 Find auto-increment use both before and after candidate.
181 2013-09-02 Marek Polacek <polacek@redhat.com>
183 * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
185 2013-09-01 Jan Hubicka <jh@suse.cz>
187 * Makefile.in: Add ipa-profile.o
188 (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
189 * cgraph.c (struct cgraph_propagate_frequency_data,
190 cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
191 ipa-profile.c; replace cgraph_ by ipa_ prefix.
192 * cgraph.h (cgraph_propagate_frequency): Remove.
193 * ipa-inline-analysis.c: Include ipa-utils.h; drop duplicated cfgloop.h.
194 (inline_update_callee_summaries): Update.
195 * ipa-profile.c: New file.
196 * ipa-utils.h (ipa_propagate_frequency): Declare.
197 * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
198 data-streamer.h, value-prof.h
199 (symtab_remove_unreachable_nodes): Update profile.
200 (struct histogram_entry, histogram, histogram_pool, histogram_hash,
201 account_time_size, cmp_counts, dump_histogram,
202 ipa_profile_generate_summary, ipa_profile_write_summary,
203 ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
204 pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
205 Move to ipa-profile.c
207 2013-09-01 John David Anglin <danglin@gcc.gnu.org>
209 * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
211 2013-09-01 Jan Hubicka <jh@suse.cz>
213 * common.opt (fdevirtualize-speculatively): New function.
214 * invoke.texi (fdevirtualize-speculatively): Document.
215 * ipa-devirt.c: Include ipa-inline.h
216 (likely_target_p): New function.
217 (ipa_devirt): New function.
218 (gate_ipa_devirt): New function.
219 (pass_data_ipa_devirt): New static var.
220 (pass_ipa_devirt): Likewise.
221 (make_pass_ipa_devirt): New function.
222 * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
223 (common_handle_option): Disable devirtualization when
224 value range profiling is available.
225 * passes.def (pass_ipa_devirt): Add.
226 * timever.def (TV_IPA_DEVIRT): New timevar.
227 * tree-pass.h (make_pass_ipa_devirt):
229 2013-09-01 Iain Sandoe <iain@codesourcery.com>
231 * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
232 include sanitize(undefined).
234 2013-08-31 Diego Novillo <dnovillo@google.com>
236 * Makefile.in (TREE_CORE_H): Define.
238 (GTFILES): Add tree-core.h.
239 * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
241 * tree-core.h: New file.
242 Move all data structures, enum, typedefs, global
243 declarations and constants from ...
246 2013-08-31 Jan Hubicka <jh@suse.cz>
248 * bulitins.c (expand_builtin): Do not early exit for gcov
249 instrumented functions.
251 2013-08-31 Marek Polacek <polacek@redhat.com>
253 * ubsan.c: Include tm_p.h.
255 2013-08-31 Jan Hubicka <jh@suse.cz>
257 * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
260 * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
261 * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
262 * tree-cfg.c (verify_gimple_label): ... here.
263 * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
264 (ipa_merge_profiles): New function.
265 * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
266 (lto_input_function_body): Likewise.
267 * ipa-utils.h (ipa_merge_profiles): Declare.
268 * lto-streamer.h (lto_input_function_body): Update prototype.
269 (emit_label_in_global_context_p): Remove.
270 * lto-symtab.c: Include ipa-utils.h
271 (lto_cgraph_replace_node): Use ipa_merge_profiles.
273 2013-08-31 Jan Hubicka <jh@suse.cz>
275 * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
277 2013-08-31 Jan Hubicka <jh@suse.cz>
279 * basic-block.h (apply_scale): Make scale parmeter gcov_type.
281 2013-08-31 Uros Bizjak <ubizjak@gmail.com>
283 * config/alpha/alpha.c (alpha_emit_conditional_move): Update
284 "cmp" RTX before signed_comparison_operator check to account
287 2013-08-30 Jan Hubicka <jh@suse.cz>
289 * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
290 (detect_type_change_1): Rename to ...
291 (detect_type_change): ... this one; early return on non-polymorphic
293 (detect_type_change_ssa): Add comp_type parameter; update
294 use of detect_type_change.
295 (compute_complex_assign_jump_func): Add param_type parameter;
296 update use of detect_type_change_ssa.
297 (compute_complex_ancestor_jump_func): Likewise.
298 (ipa_get_callee_param_type): New function.
299 (ipa_compute_jump_functions_for_edge): Compute parameter type;
300 update calls to the jump function computation functions.
302 2013-08-30 Teresa Johnson <tejohnson@google.com>
303 Steven Bosscher <steven@gcc.gnu.org>
305 * cfgrtl.c (fixup_new_cold_bb): New routine.
306 (commit_edge_insertions): Invoke fixup_partitions.
307 (find_partition_fixes): New routine.
308 (fixup_partitions): Ditto.
309 (verify_hot_cold_block_grouping): Update comments.
310 (rtl_verify_edges): Invoke find_partition_fixes.
311 (rtl_verify_bb_pointers): Update comments.
312 (rtl_verify_bb_layout): Ditto.
313 * basic-block.h (probably_never_executed_edge_p): Declare.
314 (fixup_partitions): Ditto.
315 * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
316 * bb-reorder.c (sanitize_hot_paths): New function.
317 (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
319 * predict.c (probably_never_executed_edge_p): New routine.
320 * cfg.c (check_bb_profile): Add partition insanity warnings.
322 2013-08-30 Meador Inge <meadori@codesourcery.com>
324 * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
326 2013-08-30 Marek Polacek <polacek@redhat.com>
328 * Makefile.in (ubsan.o): Add.
329 (c-family/c-ubsan.o): Add.
330 (builtins.o): Add ubsan.h dependency.
333 * common.opt: Add -fsanitize=undefined option.
334 (flag_sanitize): Add variable.
335 (fsanitize=): Add option. Add Driver.
336 (fsanitize=thread): Remove option.
337 (fsanitize=address): Likewise.
338 (static-libubsan): New option.
339 * doc/invoke.texi: Document the new flag and -static-libubsan.
340 * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
341 (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
342 * builtin-attrs.def (ATTR_COLD): Define.
343 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
344 * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
345 BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
346 * flag-types.h (sanitize_code): New enum.
347 * opts.c (common_handle_option): Parse command line arguments
348 of -fsanitize=. Add -fsanitize=unreachable option.
349 * varasm.c (get_variable_section): Adjust.
350 (assemble_noswitch_variable): Likewise.
351 (assemble_variable): Likewise.
352 (output_constant_def_contents): Likewise.
353 (categorize_decl_for_section): Likewise.
354 (place_block_symbol): Likewise.
355 (output_object_block): Likewise.
356 * builtins.def: Likewise.
357 * toplev.c (compile_file): Likewise.
358 (process_options): Likewise.
359 * cppbuiltin.c: Likewise.
360 * tsan.c (tsan_pass): Likewise.
361 (tsan_gate): Likewise.
362 (tsan_gate_O0): Likewise.
363 * cfgexpand.c (partition_stack_vars): Likewise.
364 (expand_stack_vars): Likewise.
365 (defer_stack_allocation): Likewise.
366 (expand_used_vars): Likewise.
367 * cfgcleanup.c (old_insns_match_p): Likewise.
368 * asan.c (asan_finish_file): Likewise.
369 (asan_instrument): Likewise.
370 (gate_asan): Likewise.
371 (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
372 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
373 (asan_global_struct): Use pointer_sized_int_node instead
374 calling build_nonstandard_integer_type.
375 (initialize_sanitizer_builtins): Likewise.
376 (asan_finish_file): Likewise.
377 * gcc.c: Document %{%:function(args):X}.
378 (static_spec_functions): Add sanitize.
379 (handle_spec_function): Add retval_nonnull argument and if non-NULL,
380 store funcval != NULL there.
381 (do_spec_1): Adjust handle_spec_function caller.
382 (handle_braces): Allow %:function(args) as condition.
383 (sanitize_spec_function): New function.
384 (ADD_STATIC_LIBUBSAN_LIBS): Define.
385 (LIBUBSAN_SPEC): Likewise.
386 (LIBUBSAN_EARLY_SPEC): Likewise.
387 (SANITIZER_SPEC): Handle libubsan.
388 (SANITIZER_EARLY_SPEC): Likewise.
389 * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
390 instead of fsanitize=address.
391 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
392 instead of fsanitize=address*.
393 * builtins.c: Include ubsan.h.
394 (fold_builtin_0): Instrument __builtin_unreachable.
395 * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
396 instead of flag_asan.
397 * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
398 (pointer_sized_int_node): Define.
399 * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
401 2013-08-30 Mike Stump <mikestump@comcast.net>
403 * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
406 2013-08-29 Jan Hubicka <jh@suse.cz>
408 * cgraph.c (cgraph_function_body_availability): Handle weakref
410 * passes.def: Remove pass_fixup_cfg.
411 * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
412 track when we need to remove functions.
413 (gate_ipa_inline): Execute inlining always; add comment why.
414 (pass_data_ipa_inline): Remove TODO_remove_functions.
415 * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
416 do not produce summaries.
417 * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
418 (symtab_nonoverwritable_alias): Assert we are not called on weakref.
419 * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
420 constant pool and vtable.
422 2013-08-30 Tejas Belagod <tejas.belagod@arm.com>
424 * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
425 New arm_neon.h's internal macros to specify 64-bit constants.
426 Avoid using stdint.h's macros.
428 2013-08-30 Joern Rennecke <joern.rennecke@embecosm.com>
430 * recog.c (verify_changes): Verify that changes[i].old is non-zero
431 before applying REG_P.
433 2013-08-30 Jakub Jelinek <jakub@redhat.com>
435 PR tree-optimization/58277
436 * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
437 after seeing too many stmts with vdef in between dombb and current
438 bb, invalidate everything.
440 2013-08-30 Richard Biener <rguenther@suse.de>
442 * fold-const.c (fold_single_bit_test): Fix overflow test.
444 2013-08-30 Eric Botcazou <ebotcazou@adacore.com>
446 * function.c (assign_parm_setup_reg): For a parameter passed by pointer
447 and which can live in a register, always retrieve the value on entry.
448 * var-tracking.c (add_stores): Treat the copy on entry for a parameter
449 passed by invisible reference specially.
450 (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
451 (vt_add_function_parameter): Correctly deal with a parameter passed by
454 2013-08-30 Jan Hubicka <jh@suse.cz>
456 * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
458 2013-08-30 Richard Biener <rguenther@suse.de>
460 PR tree-optimization/58228
461 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
462 allow invariant loads in nested loop vectorization.
464 2013-08-30 Richard Biener <rguenther@suse.de>
466 PR tree-optimization/58223
467 * tree-loop-distribution.c (has_anti_dependence): Rename to ...
468 (has_anti_or_output_dependence): ... this and adjust to also
469 look for output dependences.
470 (mark_nodes_having_upstream_mem_writes): Adjust.
471 (rdg_flag_uses): Likewise.
473 2013-08-30 Richard Biener <rguenther@suse.de>
475 PR tree-optimization/58010
476 * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
477 assert that we have a loop-closed PHI.
479 2013-08-29 Jan Hubicka <jh@suse.cz>
481 * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
482 * cgraph.c (cgraph_release_function_body): Free decl_in_state.
483 * lto-section-in.c (lto_free_function_in_decl_state): New function.
484 (lto_free_function_in_decl_state_for_node): New function.
486 2013-08-29 Xinliang David Li <davidxl@google.com>
488 * loop-unroll.c (report_unroll_peel): Minor message change.
489 * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
490 Emit alignment peeling message with default -fopt-info.
491 (vect_loop_versioning): Emit loop version info message.
492 * tree-vectorizer.c (vectorize_loops): Minor message change.
493 (execute_vect_slp): Ditto.
495 2013-08-29 Eric Botcazou <ebotcazou@adacore.com>
497 * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
498 of the clone from the DECL_NAME of the original function.
500 2013-08-29 Oleg Endo <olegendo@gcc.gnu.org>
502 * passes.c (register_pass): Add overload.
503 * tree-pass.h (register_pass): Forward declare it. Add comment.
505 2013-08-29 Jan Hubicka <jh@suse.cz>
507 * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
508 DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
509 * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
510 DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
511 (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
512 * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
513 Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
514 (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
516 2013-08-29 Teresa Johnson <tejohnson@google.com>
518 * dumpfile.c (dump_loc): Output column number.
519 * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
520 * doc/invoke.texi: Document optall -fopt-info flag.
521 * profile.c (read_profile_edge_counts): Use new dump framework.
522 (compute_branch_probabilities): Ditto.
523 * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
524 when pass not in any opt group.
525 * pass_manager.h (pass_manager::get_pass_profile): New method.
526 * value-prof.c (check_counter): Use new dump framework.
527 (check_ic_target): Ditto.
528 * coverage.c (get_coverage_counts): Ditto.
529 (coverage_init): Setup new dump framework.
531 2013-08-29 Richard Biener <rguenther@suse.de>
533 PR tree-optimization/58246
534 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
535 handle the dominance check inside a basic-block.
537 2013-08-29 Richard Biener <rguenther@suse.de>
540 * tree-ssa-copy.c (may_propagate_copy): Allow propagating
541 of default defs that appear in abnormal PHI nodes.
543 2013-08-29 Richard Biener <rguenther@suse.de>
545 PR tree-optimization/57685
546 * tree-vrp.c (register_edge_assert_for_1): Recurse only for
547 single-use operands to avoid exponential complexity.
549 2013-08-28 Dehao Chen <dehao@google.com>
551 * ipa-inline.c (edge_badness): Fix integer underflow.
553 2013-08-28 Uros Bizjak <ubizjak@gmail.com>
555 * gtm-builtins.def (_ITM_free): Declare leaf.
557 2013-08-28 Jakub Jelinek <jakub@redhat.com>
560 * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
561 (*tls_local_dynamic_base_64_largepic): Likewise.
562 (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
563 Remove predicate from call operand.
564 * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
565 return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
567 2013-08-28 Jeff Law <law@redhat.com>
569 * tree-ssa-threadedge.c (thread_around_empty_block): Remove
570 checks for the number of predecessors and successors allowed.
571 * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
572 which require copying a joiner block if there is a request which
573 is a subpath that requires no joiner block copying.
575 2013-08-28 Jan Hubicka <jh@suse.cz>
577 * lto-streamer-out.c (DFS_write_tree_body): Drop
578 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
579 (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
580 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
581 * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
582 Do not read DECL_ERROR_ISSUED.
583 (unpack_ts_decl_with_vis_value_fields): Do not read
585 (lto_input_ts_binfo_tree_pointers): Do not read
586 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
587 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
588 write DECL_ERROR_ISSUED..
589 (pack_ts_decl_with_vis_value_fields): Do not write
591 (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
592 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
593 * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
594 * tree.h (tree_decl_common): Update comment.
595 (DECL_ERROR_ISSUED): Remove.
597 2013-08-28 Jakub Jelinek <jakub@redhat.com>
600 * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
602 2013-08-28 Jan Hubicka <jh@suse.cz>
604 * builtins.def (free): Declare leaf.
606 2013-08-27 David Malcolm <dmalcolm@redhat.com>
609 * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
610 * configure: Regenerate.
612 2013-08-27 Martin Jambor <mjambor@suse.cz>
614 * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
615 (ipa_ancestor_jf_data): Likewise.
616 (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
617 (ipa_get_jf_pass_through_type_preserved): New function.
618 (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
619 (ipa_get_jf_ancestor_type_preserved): New function.
620 * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
621 (ipa_get_jf_ancestor_result): Likewise.
622 (propagate_vals_accross_pass_through): Use
623 ipa_get_jf_pass_through_result to do all the value mappings.
624 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
626 (ipa_set_jf_cst_copy): New function.
627 (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
628 (ipa_set_jf_arith_pass_through): Likewise.
629 (ipa_set_ancestor_jf): Likewise.
630 (compute_complex_assign_jump_func): Set type_preserved instead of
632 (ipa_compute_jump_functions_for_edge): Likewise.
633 (combine_known_type_and_ancestor_jfs): Honor type_preserved.
634 (update_jump_functions_after_inlining): Update type_preserved.
635 Explicitely create jump functions when combining one with pass_through.
636 (ipa_write_jump_function): Stream the type_preserved flags.
637 (ipa_read_jump_function): Likewise.
639 2013-08-27 Jakub Jelinek <jakub@redhat.com>
640 Aldy Hernandez <aldyh@redhat.com>
642 * Makefile.in (omp-low.o): Depend on $(TARGET_H).
643 * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
644 * function.h (struct function): Add has_force_vect_loops and
646 * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
647 * gimple.c (gimple_build_omp_critical): Add KIND argument and
649 * gimple.def: Update CLAUSES comments.
650 * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
651 (gimple_build_omp_for): Add argument to prototype.
652 (gimple_omp_for_kind): New.
653 (gimple_omp_for_set_kind): New.
654 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
655 GOVD_DATA_SHARE_CLASS.
656 (enum omp_region_type): Add ORT_SIMD.
657 (gimple_add_tmp_var): Handle ORT_SIMD.
658 (gimplify_var_or_parm_decl): Same.
659 (is_gimple_stmt): Same.
660 (omp_firstprivatize_variable): Same.
661 (omp_add_variable): Only use splay_tree_insert if lookup failed.
662 (omp_notice_variable): Handle ORT_SIMD.
663 (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
664 (omp_check_private): Handle ORT_SIMD.
665 (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
667 (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
668 Handle OMP_CLAUSE_LASTPRIVATE.
669 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
671 (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
672 (gimplify_expr): Handle OMP_SIMD.
673 * internal-fn.c (expand_GOMP_SIMD_LANE): New.
674 (expand_GOMP_SIMD_VF): New.
675 (expand_GOMP_SIMD_LAST_LANE): New.
676 * internal-fn.def (GOMP_SIMD_LANE): New.
678 (GOMP_SIMD_LAST_LANE): New.
679 * omp-low.c: Include target.h.
680 (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
682 (check_omp_nesting_restrictions): Same.
684 (lower_rec_simd_input_clauses): New.
685 (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
687 (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
688 GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
689 (expand_omp_build_assign): New.
690 (expand_omp_for_init_counts): New.
691 (expand_omp_for_init_vars): New.
692 (extract_omp_for_update_vars): New.
693 (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
694 and rewrite accordingly.
695 (expand_omp_simd): New.
696 (expand_omp_for): Use expand_omp_simd.
697 (lower_omp_for_lastprivate): Unshare vinit when appropriate.
698 (lower_omp_for): Do not lower the body.
699 * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
701 * tree-flow.h (find_omp_clause): Remove prototype.
702 * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
703 forcing vectorization of the loop, or if flag_tree_vectorize.
704 (gate_tree_if_conversion): Similarly.
705 * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
706 gimple_build_omp_for.
707 (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
708 * tree-parloops (create_parallel_loop): Pass kind argument to
709 gimple_build_omp_for.
710 * tree-pretty-print.c (dump_omp_clause): Add cases for
711 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
712 OMP_CLAUSE__SIMDUID_.
713 (dump_generic_node): Handle OMP_SIMD.
714 * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
715 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
716 unroll OMP_SIMD loops here.
717 * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
718 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
720 (vect_analyze_data_refs): Handle simd loops.
721 * tree-vect-loop.c (vectorizable_live_operation): Handle
723 * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
724 (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
725 (vectorizable_load): Same.
726 * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
727 (struct simduid_to_vf): New.
728 (simduid_to_vf::hash): New.
729 (simduid_to-vf::equal): New.
730 (struct simd_array_to_simduid): New.
731 (simd_array_to_simduid::hash): New.
732 (simd_array_to_simduid::equal): New.
733 (adjust_simduid_builtins): New.
734 (struct note_simd_array_uses_struct): New.
735 (note_simd_array_uses_cb): New.
736 (note_simd_array_uses): New.
737 (vectorize_loops): Handle simd hints and adjust simd builtins
739 * tree-vectorizer.h (struct _stmt_vec_info): Add
740 simd_lane_access_p field.
741 (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
742 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
743 OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
744 (omp_clause_code_name): Same.
745 (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
746 OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
747 * tree.def (OMP_SIMD): New entry.
748 * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
749 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
750 (OMP_CLAUSE_DECL): Adjust range for new clauses.
751 (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
752 (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
753 (OMP_CLAUSE_LINEAR_STEP): New.
754 (OMP_CLAUSE_SAFELEN_EXPR): New.
755 (OMP_CLAUSE__SIMDUID__DECL): New.
756 (find_omp_clause): New prototype.
758 2013-08-27 H.J. Lu <hongjiu.lu@intel.com>
760 * config/i386/driver-i386.c (host_detect_local_cpu): Update
761 Haswell processor detection.
763 2013-08-27 Christian Widmer <shadow@umbrox.de>
766 * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
767 of Ivy Bridge and Haswell processors. Assume core-avx2 for unknown
768 AVX2 capable processors.
770 2013-08-27 Tejas Belagod <tejas.belagod@arm.com>
772 * config/aarch64/arm_neon.h: Replace all inline asm implementations
773 of vget_low_* with implementations in terms of other intrinsics.
775 2013-08-27 Marc Glisse <marc.glisse@inria.fr>
778 * doc/extend.texi (__builtin_isinf_sign): Restrict the return
779 values to -1, 0 and 1.
781 2013-08-27 Vidya Praveen <vidyapraveen@arm.com>
783 * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
784 UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
785 (<optab><mode>3_insn): Remove.
786 (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
787 (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
788 (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
789 (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
790 (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
791 (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
792 (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
793 (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
794 (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
795 (ror<mode>3_insn): Likewise.
796 * config/aarch64/predicates.md (aarch64_simd_register): New.
798 2013-08-27 Richard Biener <rguenther@suse.de>
800 PR tree-optimization/57521
801 * tree-if-conv.c (if_convertible_bb_p): Verify that at least
802 one edge is non-critical.
803 (find_phi_replacement_condition): Make sure to use a non-critical
804 edge. Cleanup and remove old bug workarounds.
805 (bb_postdominates_preds): Remove.
806 (if_convertible_loop_p_1): Do not compute post-dominators.
807 (combine_blocks): Do not free post-dominators.
808 (main_tree_if_conversion): Likewise.
809 (pass_data_if_conversion): Add TODO_verify_ssa.
811 2013-08-27 DJ Delorie <dj@redhat.com>
813 * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
815 2013-08-27 Yufeng Zhang <yufeng.zhang@arm.com>
817 * function.c (assign_parm_find_data_types): Set passed_mode and
818 nominal_mode to the TYPE_MODE of nominal_type for the built
819 pointer type in case of the struct-pass-by-reference.
821 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
823 * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
824 (UINT16_TYPE): Change default to "unsigned int".
826 * config/avr/avr.opt (mfract-convert-truncate): New option.
827 * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
828 is set, round negative fractional integers according to n1169
829 when converting to integer types.
831 2013-08-26 Jan Hubicka <jh@suse.cz>
833 * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
834 methods can not be called indirectly when their address is not taken.
836 2013-08-26 Jan Hubicka <jh@suse.cz>
838 * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
841 2013-08-26 Jan Hubicka <jh@suse.cz>
843 * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
846 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
848 * reload.c (find_valid_class): Allow classes that do not include
849 FIRST_PSEUDO_REGISTER - 1.
851 2013-08-26 Jan Hubicka <jh@suse.cz>
853 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
854 fix edge count/frequency when speculation failed; fix type check
857 2013-08-26 Jan Hubicka <jh@suse.cz>
859 * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
861 2013-08-26 Jan Hubicka <jh@suse.cz>
863 * ipa-inline-transform.c (inline_transform): Be ready for basic block
864 to be changed by edge redirection.
866 2013-08-26 Jan Hubicka <jh@suse.cz>
868 * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
869 formating; add sanity check.
870 (cgraph_resolve_speculation): Add FIXME about scaling profiles.
871 (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
872 * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
873 (resolve_noninline_speculation): Update callee keys, too.
875 2013-08-26 Jan Hubicka <jh@suse.cz>
877 * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
878 (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
880 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
882 * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
885 2013-08-26 Uros Bizjak <ubizjak@gmail.com>
887 * config/i386/i386.c (ix86_debug_options): Remove prototype.
888 (x86_64_elf_select_section): Ditto.
889 (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
891 (ix86_pass_by_reference): Ditto.
892 (output_set_got): Ditto.
893 (ix86_unary_operator_ok): Ditto.
894 (ix86_expand_builtin): Ditto.
896 2013-08-23 Jan Hubicka <jh@suse.cz>
898 * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
900 2013-08-23 Jan Hubicka <jh@suse.cz>
902 * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
903 (tree_decl_with_vis): Add FINAL field.
905 2013-08-23 Jeff Law <law@redhat.com>
907 * tree-ssa-pre.c (do_regular_insertion): Include the expression in
908 the debugging dump when the expression is fully redundant.
910 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
912 * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
913 * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
914 * pretty-print.c (pp_formatted_text_data): Likewise.
915 (pp_write_text_to_stream): Likewise.
916 (pp_write_text_as_dot_label_to_stream): Likewise.
917 (pp_append_r): Likewise.
918 (pp_format): Likewise.
919 (pp_flush): Likewise.
920 (pp_clear_output_area): Likewise.
921 (pp_append_text): Likewise.
922 (pp_formatted_text): Likewise.
923 (pp_remaining_character_count_for_line): Likewise.
924 (pp_newline): Likewise.
925 (pp_character): Likewise.
926 (output_buffer::~output_buffer): Define.
927 (pretty_printer::~pretty_printer): Destruct output buffer.
928 * pretty-print.h (output_buffer::~output_buffer): Declare.
929 (pretty_printer::~pretty_printer): Declare virtual.
931 2013-08-24 Marc Glisse <marc.glisse@inria.fr>
934 * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
935 HOST_WIDE_INT_M1U): New macros.
936 * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
937 fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
938 unsigned -1 for lshift.
939 * cse.c (cse_insn): Likewise.
940 * double-int.c (rshift_double, lshift_double): Likewise.
941 * builtins.c (fold_builtin_bitop): Likewise.
942 * combine.c (force_to_mode): Likewise.
943 * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
944 * simplify-rtx.c (simplify_const_unary_operation,
945 simplify_const_binary_operation): Likewise.
946 * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
947 check_va_list_escapes): Likewise.
948 * rtlanal.c (nonzero_bits1): Likewise.
949 * expmed.c (expand_smod_pow2): Likewise.
950 * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
952 2013-08-23 Jan Hubicka <jh@suse.cz>
954 * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
955 as having address taken.
957 2013-08-23 Jan Hubicka <jh@suse.cz>
959 * ipa-utils.h (method_class_type): Declare.
960 * ipa-devirt.c (method_class_type): Export.
962 * cgraphunit.c (analyze_functions): Do basic devirtualization;
963 do not walk base classes of anonymous types.
965 2013-08-23 Kaz Kojima <kkojima@gcc.gnu.org>
967 PR rtl-optimization/58220
969 * final.c (reemit_insn_block_notes): Use NEXT_INSN to
970 handle SEQUENCE insns properly.
972 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
974 * pretty-print.h (pp_newline_and_flush): Declare. Remove macro
976 (pp_newline_and_indent): Likewise.
977 (pp_separate_with): Likewise.
978 * pretty-print.c (pp_newline_and_flush): Define.
979 (pp_newline_and_indent): Likewise.
980 (pp_separate_with): Likewise.
982 2013-08-23 Jakub Jelinek <jakub@redhat.com>
985 * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
986 * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
988 2013-08-23 Kirill Yukhin <kirill.yukhin@intel.com>
990 * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
991 * gcc/config/i386/i386.md (*movti_internal): Use
992 predicate to determine if EVEX is needed.
993 (*movsi_internal): Ditto.
994 (*movdf_internal): Ditto.
995 (*movsf_internal): Ditto.
996 * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
998 2013-08-23 Jakub Jelinek <jakub@redhat.com>
1000 PR tree-optimization/58209
1001 * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
1002 (find_tail_calls): Give up for pointer result types if m is non-NULL.
1003 (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
1004 emit POINTER_PLUS_EXPR.
1005 (create_tailcall_accumulator): For pointer result type accumulate in
1008 2013-08-22 Paolo Carlini <paolo.carlini@oracle.com>
1010 * configure.ac: Add backslashes missing from the last change.
1011 * configure: Regenerate.
1013 2013-08-22 Jan Hubicka <jh@susue.cz>
1015 * ipa.c (function_and_variable_visibility): First remember function
1016 was global and then make it local.
1018 2013-08-22 Julian Brown <julian@codesourcery.com>
1020 * configure.ac: Add aarch64 to list of arches which use "nop" in
1022 * configure: Regenerate.
1024 2013-08-22 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1026 * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
1027 gnu_libc_has_function.
1028 * config/s390/tpf.h: Likewise.
1030 2013-08-22 Jan Hubicka <jh@susue.cz>
1032 * timevar.c (validate_phases): Add cast.
1034 2013-08-22 Jan Hubicka <jh@susue.cz>
1036 * timevar.c (validate_phases): Use size_t for memory.
1037 * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
1039 2013-08-22 Gabriel Dos Reis <gdr@integrable-solutions.net>
1041 * pretty-print.h (output_buffer::output_buffer): Declare.
1042 (pretty_printer::pretty_printer): Likewise.
1043 (pp_construct): Remove.
1044 * pretty-print.c (output_buffer::output_buffer): Define.
1045 (pretty_printer::pretty_printer): Rename from pp_construct. Simplify.
1046 * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
1047 (print_gimple_expr): Likewise.
1048 (print_gimple_seq): Likewise.
1049 (gimple_dump_bb): Likewise.
1050 * sched-vis.c (dump_value_slim): Likewise.
1051 (dump_insn_slim): Likewise.
1052 (dump_rtl_slim): Likewise.
1053 (str_pattern_slim): Likewise.
1054 * tree-mudflap.c (mf_varname_tree): Likewise.
1055 * graph.c (print_graph_cfg): Likewise.
1056 (start_graph_dump): Likewise.
1057 * tree-pretty-print.c (maybe_init_pretty_print): Likewise. Use
1059 * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
1060 pretty printer initialization.
1061 * coretypes.h (diagnostic_context): Remove superflous type alias
1063 (pretty_printer): Likewise. Declare directly as a class.
1064 (pretty_print_info): Remove declaration as class.
1065 * asan.c (asan_emit_stack_protection): Remove call to pp_construct
1066 and pp_clear_output_area.
1067 (asan_add_global): Likewise.
1069 2013-08-22 Jan Hubicka <jh@suse.cz>
1071 * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
1072 * ipa-utils.h (update_type_inheritance_graph): Declare.
1073 (possible_polymorphic_call_target_p): Declare.
1074 (possible_polymorphic_call_target_p): New.
1075 * ipa-devirt.c: Update toplevel comments.
1076 (cached_polymorphic_call_targets): Move up.
1077 (odr_type_d): Move ID down.
1078 (polymorphic_type_binfo_p): Update comment.
1079 (odr_hasher::remove): Likewise;
1080 (get_odr_type): Set anonymous_namespace.
1081 (dump_odr_type): Dump it.
1082 (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
1083 (maybe_record_node): Record node in cached_polymorphic_call_targets.
1084 (record_binfo): Add comment.
1085 (free_polymorphic_call_targets_hash): Do not ICE when cache is not
1087 (devirt_node_removal_hook): Do not iCE when cache is freed.
1088 (possible_polymorphic_call_target_p): New predicate.
1089 (update_type_inheritance_graph): New function.
1091 2013-08-22 Alexander Ivchenko <alexander.ivchenko@intel.com>
1092 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
1093 Sergey Lega <sergey.s.lega@intel.com>
1094 Anna Tikhonova <anna.tikhonova@intel.com>
1095 Ilya Tocar <ilya.tocar@intel.com>
1096 Andrey Turetskiy <andrey.turetskiy@intel.com>
1097 Ilya Verbin <ilya.verbin@intel.com>
1098 Kirill Yukhin <kirill.yukhin@intel.com>
1099 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
1101 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
1102 (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
1103 (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
1104 (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
1105 (OPTION_MASK_ISA_AVX2_UNSET): Update.
1106 (OPTION_MASK_ISA_AVX512F_UNSET): New.
1107 (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
1108 (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
1109 (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
1110 (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
1111 OPT_mavx512pf, OPT_mavx512er cases.
1112 * config/i386/constraints.md (v): New constraint.
1113 (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
1114 * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
1115 (bit_AVX512CD): New.
1116 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1117 AVX512F, AVX512ER, AVX512PF, AVX512CD features.
1118 * config/i386/i386-c.c (ix86_target_macros_internal):
1119 Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
1121 * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
1122 (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
1123 * config/i386/i386.c (regclass_map, dbx_register_map)
1124 (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
1125 (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
1126 (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
1128 (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
1129 PTA_AVX512PF, PTA_AVX512CD. Handle -mavx512f, -mavx512er, -mavx512cd,
1130 -mavx512pf options. Fix formatting.
1131 (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
1132 targets. Squash EVEX_SSE_REGS if AVX512F is disabled.
1133 (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
1134 -mavx512cd, -mavx512pf options.
1135 (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
1136 (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
1137 (ix86_preferred_output_reload_class): Replace SSE_REGS with
1139 (ix86_hard_regno_mode_ok): Support 512-bit registers.
1140 (ix86_set_reg_reg_cost): Ditto.
1141 (x86_order_regs_for_local_alloc): Ditto.
1142 (MAX_VECT_LEN): Extend to 64-byte.
1143 (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
1144 * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
1145 (TARGET_AVX512ER, TARGET_AVX512CD): New.
1146 (BIGGEST_ALIGNMENT): Extend to 512-bits.
1147 (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
1148 (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
1149 (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
1150 (SSE_REG_MODE_P): Support new modes.
1151 (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
1152 (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
1153 (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
1154 (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
1155 (REG_CLASS_CONTENTS): Add new registers.
1156 (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
1157 (EXT_REX_SSE_REGNO_P): New.
1158 (HI_REGISTER_NAMES): Add new registers.
1159 * config/i386/i386.md: Define constants for new registers.
1160 (mode): Add new 512-bit modes.
1161 (prefix): Support evex prefix.
1162 (isa): Support avx512f, noavx512f, fma_avx512f.
1163 (ssemodesuffix): Add new 512-bit modes.
1165 (*movxi_internal_avx512f): Ditto.
1166 (*movdi_internal): Replace constraint "x" with the new constraint "v".
1168 (*movsi_internal): Likewise.
1169 (*movdf_internal): Likewise.
1170 (*movsf_internal): Likewise.
1171 (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
1172 (<code><mode>3): Likewise.
1173 * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
1175 * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
1176 with the new constraint "v".
1177 * config/i386/sse.md (*mov<mode>_internal): Support new registers and
1179 (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
1180 with the new constraint "v".
1181 (<sse2>_loaddqu<avxsizesuffix>): Likewise.
1182 (<sse2>_storedqu<avxsizesuffix>): Likewise.
1183 (*<plusminus_insn><mode>3): Likewise.
1184 (<sse>_vm<plusminus_insn><mode>3): Likewise.
1185 (*mul<mode>3): Likewise.
1186 (<sse>_vmmul<mode>3): Likewise.
1187 (<sse>_div<mode>3): Likewise.
1188 (<sse>_vmdiv<mode>3): Likewise.
1189 (<sse>_sqrt<mode>2): Likewise.
1190 (<sse>_vmsqrt<mode>2): Likewise.
1191 (*<code><mode>3_finite): Likewise.
1192 (*<code><mode>3) <smaxmin>: Likewise.
1193 (<sse>_vm<code><mode>3): Likewise.
1194 (*<code><mode>3) <any_logic>: Likewise.
1195 (*fma_fmadd_<mode>): Likewise.
1196 (*fma_fmsub_<mode>): Likewise.
1197 (*fma_fnmadd_<mode>): Likewise.
1198 (*fma_fnmsub_<mode>): Likewise.
1199 (*fma_fmaddsub_<mode>): Likewise.
1200 (*fma_fmsubadd_<mode>): Likewise.
1201 (*fmai_fmadd_<mode>): Likewise.
1202 (*fmai_fmsub_<mode>): Likewise.
1203 (*fmai_fnmadd_<mode>): Likewise.
1204 (*fmai_fnmsub_<mode>): Likewise.
1205 (sse_cvtsi2ss): Likewise.
1206 (sse_cvtsi2ssq): Likewise.
1207 (sse_cvtss2si): Likewise.
1208 (sse_cvtss2si_2): Likewise.
1209 (sse_cvtss2siq): Likewise.
1210 (sse_cvtss2siq_2): Likewise.
1211 (sse_cvttss2si): Likewise.
1212 (sse_cvtss2siq_2): Likewise.
1213 (float<sseintvecmodelower><mode>2): Likewise.
1214 (sse2_cvtsd2si_2): Likewise.
1215 (sse2_cvtsd2siq_2): Likewise.
1216 (*<plusminus_insn><mode>3): Likewise.
1217 (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
1218 (*<sse4_1_avx2>_mul<mode>3): Likewise.
1219 (ashr<mode>3): Likewise.
1220 (<shift_insn><mode>3): Likewise.
1221 (avx2_<code><mode>3): Likewise.
1222 (*avx2_<code><mode>3): Likewise.
1223 (*andnot<mode>3): Likewise.
1224 (*<code><mode>3) <any_logic>: Likewise.
1225 (abs<mode>2): Likewise.
1226 (avx2_permvar<mode>): Likewise.
1227 (avx2_perm<mode>_1): Likewise.
1228 (*avx_vpermilp<mode>): Likewise.
1229 (avx_vpermilvar<mode>3): Likewise.
1230 (avx2_ashrv<mode>): Likewise.
1231 (avx2_<shift_insn>v<mode>): Likewise.
1232 * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
1234 * doc/rtl.texi: Document XImode.
1236 2013-08-21 Jeff Law <law@redhat.com>
1238 * tree-flow.h (register_jump_thread): Pass vector of edges
1239 instead of each important edge.
1240 * tree-ssa-threadedge.c (thread_across_edge): Build the jump
1241 thread path into a vector and pass that to register_jump_thread.
1242 * tree-ssa-threadupdate.c (register_jump_thread): Conver the
1243 passed in edge vector to the current 3-edge form.
1246 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
1248 * dce.c (fini_dce): Call df_analyze again just in case
1249 delete_unmarked_insns removed anything.
1251 2013-08-21 Joern Rennecke <joern.rennecke@embecosm.com>
1253 * reload.h (struct reg_equivs): Rename to ..
1254 (struct reg_equivs_s): .. this.
1256 2013-08-20 Martin Liska <marxin.liska@gmail.com>
1258 * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
1260 2013-08-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1262 * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
1264 2013-08-21 Jeff Law <law@redhat.com>
1266 * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
1267 simplify assignments too. If the RHS collapses to a singleton
1268 range, then return the value for the range.
1270 2013-08-21 Kirill Yukhin <kirill.yukhin@intel.com>
1272 * config/i386/sse.md (V16): Rename to...
1274 (mov<mode>): Update iterator name.
1275 (*mov<mode>_internal): Ditto.
1276 (push<mode>1): Ditto.
1277 (movmisalign<mode>): Ditto.
1279 2013-08-20 Jan Hubicka <jh@suse.cz>
1282 * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
1283 entry for direct edges.
1284 (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
1286 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1288 Revert my last two changes, r201865 and r201864:
1291 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1293 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1294 instances can own GC refs.
1296 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1297 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1298 (gcc::context::gt_pch_nx): Likewise.
1299 (gcc::context::gt_pch_nx): Likewise.
1300 * ggc.h (gt_ggc_mx <T>): New.
1301 (gt_pch_nx_with_op <T>): New.
1302 (gt_pch_nx <T>): New.
1303 * passes.c (opt_pass::gt_ggc_mx): New.
1304 (opt_pass::gt_pch_nx): New.
1305 (opt_pass::gt_pch_nx_with_op): New.
1306 (pass_manager::gt_ggc_mx): New.
1307 (pass_manager::gt_pch_nx): New.
1308 (pass_manager::gt_pch_nx_with_op): New.
1309 (pass_manager::operator new): Use
1310 ggc_internal_cleared_alloc_stat rather than xcalloc.
1311 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1312 (pass_manager::gt_ggc_mx): New.
1313 (pass_manager::gt_pch_nx): New.
1314 (pass_manager::gt_pch_nx_with_op): New.
1315 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1316 (opt_pass::operator new): New.
1317 (opt_pass::gt_ggc_mx): New.
1318 (opt_pass::gt_pch_nx): New.
1319 (opt_pass::gt_pch_nx_with_op): New.
1322 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1324 * Makefile.in (GTFILES): Add context.h.
1325 * context.c (gcc::context::operator new): New.
1326 (gcc::context::gt_ggc_mx): New.
1327 (gcc::context::gt_pch_nx): New.
1328 (gcc::context::gt_pch_nx): New.
1329 * context.h (gcc::context): Add GTY((user)) marking.
1330 (gcc::context::operator new): New.
1331 (gcc::context::gt_ggc_mx): New.
1332 (gcc::context::gt_pch_nx): New.
1333 (gcc::context::gt_pch_nx): New.
1334 (g): Add GTY marking.
1335 (gt_ggc_mx (gcc::context *)): New.
1336 (gt_pch_nx (gcc::context *)): New.
1337 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1338 void *cookie)): New.
1339 * gengtype.c (open_base_files) <ifiles>: Add context.h.
1341 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
1343 * dce.c (fini_dce): Call df_analyze again just in case
1344 delete_unmarked_insns removed anything.
1346 2013-08-20 Teresa Johnson <tejohnson@google.com>
1348 PR rtl-optimizations/57451
1349 * final.c (reemit_insn_block_notes): Prevent lexical blocks
1350 from crossing split section boundaries.
1352 2013-08-20 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
1354 * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
1355 * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
1356 with MULTLIB_DEFAULTS.
1358 2013-08-20 Nick Clifton <nickc@redhat.com>
1360 * target.def (narrow_volatile_bitfield): Note that the default
1361 value is false, not !TARGET_STRICT_ALIGN.
1362 * doc/tm.texi: Regenerate.
1364 2013-08-20 Pavel Chupin <pavel.v.chupin@intel.com>
1366 Fix LIB_SPEC for systems without libpthread.
1368 * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
1369 * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
1371 * config/i386/linux-common.h: Likewise.
1372 * config/mips/linux-common.h: Likewise.
1374 2013-08-20 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
1376 * tree-ssa-ccp.c (get_default_value): Remove redundant condition
1379 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1381 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
1382 instances can own GC refs.
1384 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
1385 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
1386 (gcc::context::gt_pch_nx): Likewise.
1387 (gcc::context::gt_pch_nx): Likewise.
1388 * ggc.h (gt_ggc_mx <T>): New.
1389 (gt_pch_nx_with_op <T>): New.
1390 (gt_pch_nx <T>): New.
1391 * passes.c (opt_pass::gt_ggc_mx): New.
1392 (opt_pass::gt_pch_nx): New.
1393 (opt_pass::gt_pch_nx_with_op): New.
1394 (pass_manager::gt_ggc_mx): New.
1395 (pass_manager::gt_pch_nx): New.
1396 (pass_manager::gt_pch_nx_with_op): New.
1397 (pass_manager::operator new): Use
1398 ggc_internal_cleared_alloc_stat rather than xcalloc.
1399 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
1400 (pass_manager::gt_ggc_mx): New.
1401 (pass_manager::gt_pch_nx): New.
1402 (pass_manager::gt_pch_nx_with_op): New.
1403 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
1404 (opt_pass::operator new): New.
1405 (opt_pass::gt_ggc_mx): New.
1406 (opt_pass::gt_pch_nx): New.
1407 (opt_pass::gt_pch_nx_with_op): New.
1409 2013-08-20 David Malcolm <dmalcolm@redhat.com>
1411 * Makefile.in (GTFILES): Add context.h.
1412 * context.c (gcc::context::operator new): New.
1413 (gcc::context::gt_ggc_mx): New.
1414 (gcc::context::gt_pch_nx): New.
1415 (gcc::context::gt_pch_nx): New.
1416 * context.h (gcc::context): Add GTY((user)) marking.
1417 (gcc::context::operator new): New.
1418 (gcc::context::gt_ggc_mx): New.
1419 (gcc::context::gt_pch_nx): New.
1420 (gcc::context::gt_pch_nx): New.
1421 (g): Add GTY marking.
1422 (gt_ggc_mx (gcc::context *)): New.
1423 (gt_pch_nx (gcc::context *)): New.
1424 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
1425 void *cookie)): New.
1426 * gengtype.c (open_base_files) <ifiles>: Add context.h.
1428 2013-08-20 Alan Modra <amodra@gmail.com>
1431 * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
1432 (rs6000_emit_epilogue): Likewise.
1434 2013-08-19 Dehao Chen <dehao@google.com>
1436 * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
1438 2013-08-19 Peter Bergner <bergner@vnet.ibm.com>
1439 Jakub Jelinek <jakub@redhat.com>
1441 * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
1442 (BUILT_IN_FABSD64): Likewise.
1443 (BUILT_IN_FABSD128): Likewise.
1444 * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
1445 (fold_builtin_1): Likewise.
1446 * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
1447 destination and source operands.
1448 (*abstd2_fpr): Likewise.
1449 (*nabstd2_fpr): Likewise.
1451 2013-08-19 Richard Sandiford <rdsandiford@googlemail.com>
1453 * config/mips/mips.c (mips_adjust_insn_length): Add checks for
1456 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
1458 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
1461 2013-08-19 Alexander Ivchenko <alexander.ivchenko@intel.com>
1463 * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
1464 * builtins.c (default_libc_has_function): New.
1465 (gnu_libc_has_function): Ditto.
1466 (no_c99_libc_has_function): Ditto.
1467 (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
1468 instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
1469 (fold_builtin_sincos): Likewise.
1470 (fold_builtin_cexp): Likewise.
1471 * builtins.def (DEF_C94_BUILTIN): Likewise.
1472 (DEF_C99_BUILTIN): Likewise.
1473 (DEF_C99_C90RES_BUILTIN): Likewise.
1474 (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
1475 definitions to using this define.
1476 * config/darwin-protos.h (darwin_libc_has_function): New.
1477 * config/darwin.c (darwin_libc_has_function): Ditto.
1478 * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
1479 TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
1480 * config/darwin.h: Ditto.
1481 * config/elfos.h: Ditto.
1482 * config/freebsd.h: Ditto.
1483 * config/i386/cygming.h: Ditto.
1484 * config/i386/djgpp.h: Ditto.
1485 * config/i386/i386-interix.h: Ditto.
1486 * config/microblaze/microblaze.h: Ditto.
1487 * config/mmix/mmix.h: Ditto.
1488 * config/gnu-user.h: Ditto.
1489 * config/ia64/hpux.h: Ditto.
1490 * config/pa/pa-hpux.h: Ditto.
1491 * config/pdp11/pdp11.h: Ditto.
1492 * config/picochip/picochip.h: Ditto.
1493 * config/linux.h: Ditto.
1494 * config/netbsd.h: Ditto.
1495 * config/openbsd.h: Ditto.
1496 * config/rs6000/aix43.h: Ditto.
1497 * config/rs6000/aix51.h: Ditto.
1498 * config/rs6000/aix52.h: Ditto.
1499 * config/rs6000/aix53.h: Ditto.
1500 * config/rs6000/aix61.h: Ditto.
1501 * config/rs6000/darwin.h: Ditto.
1502 * config/rs6000/linux.h: Ditto.
1503 * config/rs6000/linux64.h: Ditto.
1504 * config/s390/tpf.h: Ditto.
1505 * config/sol2-10.h: Ditto.
1506 * config/sol2.h: Ditto.
1507 * config/vms/vms.h: Ditto.
1508 * config/vxworks.h: Ditto.
1509 * config/linux-android.c (linux_android_libc_has_function):
1510 New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
1511 * config/linux-protos.h (linux_android_libc_has_function):
1513 * config/i386/i386.c (ix86_libc_has_function): New.
1514 * config/i386/i386-protos.h
1515 (ix86_libc_has_function): New declaration.
1516 * config/i386/i386.md
1517 ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
1518 ("isinf<mode>2): Likewise.
1519 * convert.c (convert_to_integer): Using new target hook
1520 TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
1521 TARGET_C99_FUNCTIONS.
1522 * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
1523 * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
1524 * coretypes.h (function_class): New enum for different
1525 classes of functions.
1526 * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
1527 * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
1528 (TARGET_HAS_SINCOS): Likewise.
1529 (TARGET_LIBC_HAS_FUNCTION): New.
1530 * doc/tm.texi: Regenerated.
1531 * targhooks.h (default_libc_has_function): New declaration.
1532 (no_c99_libc_has_function): Ditto.
1533 (gnu_libc_has_function): Ditto.
1534 * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
1535 and TARGET_HAS_SINCOS.
1537 2013-08-18 Jan Hubicka <jh@suse.cz>
1539 * Makeifle-in (ipa-devirt.o): New.
1540 (GTFILES): Add ipa-utils.h and ipa-devirt.c
1541 * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
1542 (analyze_functions): Look into possible targets of polymorphic call.
1543 * dumpfile.c (dump_files): Add type-inheritance dump.
1544 * dumpfile.h (TDI_inheritance): New.
1545 * ipa-devirt.c: New file.
1546 * ipa-utils.h (odr_type_d): Forward declare.
1547 (odr_type): New type.
1548 (build_type_inheritance_graph): Declare.
1549 (possible_polymorphic_call_targets): Declare and introduce inline
1550 variant when only edge is pased.
1551 (dump_possible_polymorphic_call_targets): Likewise.
1552 * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
1553 * tree.c (type_in_anonymous_namespace_p): Break out from ...
1554 (types_same_for_odr): ... here.
1555 * tree.h (type_in_anonymous_namespace_p): Declare.
1557 2013-08-18 Jakub Jelinek <jakub@redhat.com>
1559 PR tree-optimization/58006
1560 * tree-parloops.c (take_address_of): Don't ICE if get_name
1562 (eliminate_local_variables_stmt): Remove clobber stmts.
1564 2013-08-18 Eric Botcazou <ebotcazou@adacore.com>
1566 * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
1567 error message is issued for an alias to undefined symbol.
1569 2013-08-18 Jan Hubicka <jh@suse.cz>
1571 * cgraph.c (cgraph_create_indirect_edge): Discover
1572 polymorphic calls and record basic info into indirect_info.
1573 * gimple-fold.c (gimple_fold_call): When doing BINFO based
1574 devirtualization, ignore objc function calls.
1575 * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
1576 call with no parm index info.
1577 * ipa-prop.c (ipa_analyze_call_uses): Likewise.
1578 * tree.c (virtual_method_call_p): New function.
1579 * tree.h (virtual_method_call_p): Declare.
1581 2013-08-16 Jan Hubicka <jh@suse.cz>
1584 * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
1586 2013-08-16 David Edelsohn <dje.gcc@gmail.com>
1588 * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
1591 2013-08-16 David Malcolm <dmalcolm@redhat.com>
1593 * gengtype.c (type_for_name): Add special-case support for
1594 locating types within the "gcc::" namespace.
1595 (open_base_files): Emit a "using namespace gcc" directive.
1597 2013-08-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1600 * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1601 memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1603 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1604 array instead of each individual operand as a separate argument.
1605 (emit_fusion_gpr_load): Likewise.
1606 (expand_fusion_gpr_load): Add new function declaration.
1608 * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1609 signature to have the operands passed as an array, instead of as
1610 separate arguments. Allow ZERO_EXTEND to be in the memory
1611 address, and also SIGN_EXTEND if -mpower8-fusion-sign. Do not
1612 depend on the register live/dead flags when peepholes are run.
1613 (expand_fusion_gpr_load): New function to be called from the
1614 peephole2 pass, to change the register that addis sets to be the
1616 (emit_fusion_gpr_load): Change the calling signature to have the
1617 operands passed as an array, instead of as separate arguments.
1618 Allow ZERO_EXTEND to be in the memory address, and also
1619 SIGN_EXTEND if -mpower8-fusion-sign.
1621 * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1623 (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1624 adjust the register addis loads up in the peephole2 pass. Do not
1625 depend on the register live/dead state when the peephole pass is done.
1627 2013-08-16 David Malcolm <dmalcolm@redhat.com>
1629 * gengtype.c (create_user_defined_type): Ensure that the kind
1630 is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1631 declaration is seen before the GTY((user)) marking.
1633 2013-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
1636 * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1638 2013-08-16 Jan Hubicka <jh@suse.cz>
1640 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1641 arugment expected_type.
1642 (gimple_fold_call): Use it.
1643 * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1644 * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1645 * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1646 (try_make_edge_direct_virtual_call): Likewise.
1647 * tree.c (obj_type_ref_class): New.
1648 * tree.h (obj_type_ref_class): Use it.
1650 2013-08-16 Gabriel Dos Reis <gdr@integrable-solutions.net>
1652 * sched-vis.c (rtl_slim_pp_initialized): Remove.
1653 (rtl_slim_pp): Likewise.
1654 (init_rtl_slim_pretty_print): Likewise.
1655 (dump_value_slim): Don't call it. Use local pretty printer.
1656 (dump_insn_slim): Likewise.
1657 (dump_rtl_slim): Likewise.
1658 (str_pattern_slim): Likewise.
1659 * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1662 2013-08-16 Jakub Jelinek <jakub@redhat.com>
1664 PR tree-optimization/58164
1665 * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1666 walk gimple_goto_dest of GIMPLE_GOTO.
1668 PR tree-optimization/58165
1669 * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1670 bi_call must be the last stmt in a bb, don't split_block, instead
1671 use fallthru edge from it and give up if there is none.
1672 Release conds vector when returning early.
1674 2013-08-14 Xinliang David Li <davidxl@google.com>
1676 * config/i386/i386.c (ix86_option_override_internal):
1677 Remove unused variable and field.
1679 2013-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1682 * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1683 * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1684 * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1685 and Linux, correct BLKmode alignment when 128-bit alignment is
1686 required and compatibility flag is not set.
1687 (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1688 for zero-size arguments when compatibility flag is not set.
1690 2013-08-14 Jakub Jelinek <jakub@redhat.com>
1692 PR tree-optimization/58145
1693 * tree-sra.c (build_ref_for_offset): If prev_base has
1694 TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1696 2013-08-14 Xinliang David Li <davidxl@google.com>
1698 * config/i386/i386.c (ix86_option_override_internal):
1699 Fix uninitialized variable error.
1701 2013-08-14 Xinliang David Li <davidxl@google.com>
1703 * config/i386/i386.opt: Define two new options.
1704 * config/i386/x86-tune.def: Add arch selector field in macros.
1705 * config/i386/i386.h: Adjust macro definition.
1706 * config/i386/i386.c (ix86_option_override_internal):
1708 (parse_mtune_ctrl_str): New function.
1709 (set_ix86_tune_features): New function.
1710 (ix86_function_specific_restore): Call the new helper function.
1712 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
1714 PR rtl-optimization/57662
1715 * sel-sched.c (code_motion_process_successors): When the current insn
1716 is removed after the recursive traversal, break from the loop.
1717 Add comments and debug printouts.
1719 2013-08-14 Jakub Jelinek <jakub@redhat.com>
1720 Alexandre Oliva <aoliva@redhat.com>
1723 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1724 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1725 there also UNSPEC_PLTOFF.
1727 2013-08-14 Marek Polacek <polacek@redhat.com>
1729 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1730 NUM_CONDITIONS bit positions.
1732 2013-08-13 Cary Coutant <ccoutant@google.com>
1734 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1735 (attr_checksum): Hash vector contents instead of pointer.
1736 (attr_checksum_ordered): Likewise.
1738 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
1740 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1741 when Pmode != word_mode. Add length_address attribute.
1742 (sse3_monitor_<mode>): Merge from sse3_monitor and
1743 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
1744 Pmode != word_mode. Update insn length attribute.
1745 * config/i386/i386.c (ix86_option_override_internal): Update
1746 ix86_gen_monitor selection for merged sse3_monitor insn.
1748 2013-08-13 Julian Brown <julian@codesourcery.com>
1750 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1751 perform invalid legitimization on greater-than-word-size modes for
1754 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
1756 * ira.c (setup_class_translate_array): Use aclass instead of cl
1757 for classes not fully covered by allocno classes.
1759 2013-08-13 Jakub Jelinek <jakub@redhat.com>
1761 PR tree-optimization/57661
1762 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1763 * tree-inline.c (tree_function_versioning): Initialize it.
1764 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1765 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1766 that is not being copied.
1769 * asan.c (instrument_strlen_call): Fix typo in comment.
1770 Use char * type even for the lhs of POINTER_PLUS_EXPR.
1772 2013-08-13 Steve Ellcey <sellcey@mips.com>
1774 * config/mips/mips.md (prefetch): Use lw instead of ld on
1775 loongson in 32bit mode.
1777 2013-08-13 Nick Clifton <nickc@redhat.com>
1779 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1781 2013-08-13 Jan Hubicka <jh@suse.cz>
1783 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1784 introduced edge; fix typo in sanity check.
1785 (cgraph_resolve_speculation): Export; improve diagnostic.
1786 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1787 speculation at type mismatch.
1788 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1789 (cgraph_resolve_speculation): Declare.
1790 (symtab_can_be_discarded): New function.
1791 * value-prof.c (gimple_ic_transform): Remove actual transform code.
1792 * ipa-inline-transform.c (speculation_removed): New global var.
1793 (clone_inlined_nodes): See if speculation can be removed.
1794 (inline_call): If speculations was removed, we growths may not match.
1795 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1796 (speculation_useful_p): New function.
1797 (resolve_noninline_speculation): New function.
1798 (inline_small_functions): Resolve useless speculations.
1799 * ipa-inline.h (speculation_useful_p): Declare
1800 * ipa.c (can_replace_by_local_alias): Simplify.
1801 (ipa_profile): Produce speculative calls in non-lto, too;
1802 add simple cost model; produce local aliases.
1804 2013-08-13 David Malcolm <dmalcolm@redhat.com>
1806 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1809 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
1811 * config/i386/i386.c (ix86_function_versions): Use error + inform.
1813 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
1815 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1816 iterator instead of X87MODEF.
1818 2013-08-12 Perez Read <netfirewall@gmail.com>
1821 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
1822 operand 0 for intel asm alternative.
1823 (*movabs<mode>_2): Ditto for operand 1.
1825 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
1827 * config/aarch64/arm_none.h
1828 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1830 2013-08-12 Nick Clifton <nickc@redhat.com>
1832 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1834 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
1836 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1837 expand for QI/HImode operand to produce more effictive code for
1838 unsigned char(short) --> float(double) conversion.
1840 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
1842 * doc/invoke.texi: Mention that -ftls-model does not force the final
1845 2013-08-12 Marek Polacek <polacek@redhat.com>
1846 Marc Glisse <marc.glisse@inria.fr>
1848 PR tree-optimization/57980
1849 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1850 when creating -1 constant.
1852 2013-08-10 Jan Hubicka <jh@suse.cz>
1854 Workaround binutils PR14342.
1855 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1856 (gimple_init_edge_profiler): Likewise.
1857 (gimple_gen_ic_func_profiler): Likewise.
1859 2013-08-09 Jan Hubicka <jh@suse.cz>
1861 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1863 2013-08-09 Xinliang David Li <davidxl@google.com>
1865 * config/i386/stringop.def: New file.
1866 * config/i386/stringop.opt: New file.
1867 * config/i386/i386-opts.h: Include stringopt.def.
1868 * config/i386/i386.opt: Include stringopt.opt.
1869 * config/i386/i386.c (ix86_option_override_internal):
1870 Override default size based stringop inline strategies with options.
1871 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1874 2013-08-09 Jan Hubicka <jh@suse.cz>
1876 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1878 2013-08-09 Jan Hubicka <jh@suse.cz>
1880 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1882 (dump_cgraph_node): Dump profile-id.
1883 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1884 and common_target_probability.
1885 * lto-cgraph.c (lto_output_edge): Stream common targets.
1886 (lto_output_node): Stream profile ids.
1887 (input_node): Stream profile ids.
1888 (input_edge): Stream common targets.
1889 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1890 * ipa.c: Include value-prof.h
1891 (ipa_profile_generate_summary): Turn indirect call statement histograms
1892 into common targets.
1893 (ipa_profile): Turn common targets into speculative edges.
1895 2013-08-09 Jan Hubicka <jh@suse.cz>
1897 * cgraph.h (cgraph_node): Add profile_id.
1898 * value-prof.c (cgraph_node_map): Turn into pointer_map.
1899 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1900 (del_node_map): Update.
1901 (find_func_by_funcdef_no): Replace by ...
1902 (find_func_by_profile_id): ... this one.
1903 (gimple_ic_transform): Do not remove useful histograms when
1904 speculation is not done; dump info when indirect call removal
1906 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1907 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1908 (__gcov_indirect_call_profiler_v2): .. this one.
1909 * profile.h (init_node_map): Update.
1910 * coverage.c (coverage_compute_profile_id): New function.
1911 * coverage.h (coverage_compute_profile_id): Declare.
1912 * tree-profile.c (init_ic_make_global_vars): Make
1913 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
1914 (gimple_init_edge_profiler): Update prototype of
1915 __gcov_indirect_call_profiler.
1916 (gimple_gen_ic_func_profiler): Simplify.
1917 (tree_profiling): Use init_node_map
1919 2013-08-09 Jan Hubicka <jh@suse.cz>
1921 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1922 non-speculative refs.
1923 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1924 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1925 (cgraph_set_call_stmt): Likewise.
1926 (cgraph_create_edge_1): Fix release checking compilatoin;
1928 (cgraph_free_edge): Free indirect info.
1929 (cgraph_turn_edge_to_speculative): New function.
1930 (cgraph_speculative_call_info): New function.
1931 (cgraph_make_edge_direct): Return direct edge; handle speculation.
1932 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1933 (dump_cgraph_node): Dump speculation.
1934 (verify_edge_count_and_frequency): Accept speculative edges.
1935 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1936 (verify_cgraph_node): Handle speculation.
1937 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1938 (cgraph_set_call_stmt): Update prototype.
1939 (cgraph_make_edge_direct): Update prototype.
1940 (cgraph_speculative_call_info): Declare.
1941 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1942 to change; update call of ipa_find_references.
1943 * ipa-ref.c (ipa_record_reference): Fix return value; clear
1944 lto_stmt_uid and speculative flags.
1945 (ipa_dump_references): Dump speculation.
1946 (ipa_clone_references): Clone speculative flag.
1947 (ipa_clone_referring): Likewise.
1948 (ipa_clone_ref): New function.
1949 (ipa_find_reference): Look into lto_stmt_uids
1950 (ipa_clear_stmts_in_references): Do not clear speculative calls.
1951 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1952 (ipa_find_reference): Update declaration.
1953 (ipa_clone_ref): Declare.
1954 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1955 stream speculative flag.
1956 (lto_output_ref): Stream statements uids and speculation.
1957 (input_ref): Likewise.
1958 (input_edge): Stream speuclation.
1959 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1960 (cgraph_set_call_stmt_including_clones): Handle speculation.
1961 * ipa-inline.c (heap_edge_removal_hook): New function.
1962 (inline_small_functions): Register it.
1963 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1964 also initialize refs.
1965 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1967 (try_make_edge_direct_simple_call): Likewise.
1968 (try_make_edge_direct_simple_call): Likewise.
1969 (update_indirect_edges_after_inlining): Likewise.
1970 (remove_described_reference): Look proper lto_stmt_uid.
1971 (propagate_controlled_uses): Likewise.
1972 (propagate_controlled_uses): Liekwise.
1973 * tree-inline.c (copy_bb): Copy speculative edges.
1974 (redirect_all_calls): New function.
1975 (copy_cfg_body): Do redirection after loop info is updated.
1976 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1978 2013-08-09 Jan Hubicka <jh@suse.cz>
1980 * lto-streamer-out.c (output_function): Renumber PHIs.
1981 * lto-streamer-in.c (input_function): Likewise.
1983 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
1985 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1986 (get_lane_unsigned): Likewise.
1987 (dup_lane_scalar): Likewise.
1988 (get_lane): enable for VALL.
1989 * config/aarch64/aarch64-simd.md
1990 (aarch64_dup_lane_scalar<mode>): Remove.
1991 (aarch64_get_lane_signed<mode>): Likewise.
1992 (aarch64_get_lane_unsigned<mode>): Likewise.
1993 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1994 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1995 (aarch64_get_lane<mode>): Enable for all vector modes.
1996 (aarch64_get_lanedi): Remove misleading constraints.
1997 * config/aarch64/arm_neon.h
1998 (__aarch64_vget_lane_any): Define.
1999 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
2000 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
2001 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
2002 * config/aarch64/iterators.md (VDQQH): New.
2006 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
2008 * configure.ac: Add GAS check for LEON instructions on SPARC.
2009 * configure: Regenerate.
2010 * config.in: Likewise.
2011 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
2013 * config/sparc/sparc.opt (LEON, LEON3): New masks.
2014 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
2016 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
2017 (AS_LEON_FLAG): New macro.
2018 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
2019 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
2020 Deal with LEON and LEON3 for the memory model.
2021 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
2022 (atomic_compare_and_swap<mode>_1): Likewise.
2023 (*atomic_compare_and_swap<mode>_1): Likewise.
2025 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
2027 * config/arm/neon.md (vcond): Fix floating-point vector
2028 comparisons against 0.
2030 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
2032 * lra-constraints.c (emit_spill_move): Remove assert.
2033 (process_alt_operands): Add more debugging
2034 output. Increase reject for spilling into memory. Decrease
2035 reject for reloading scratch.
2036 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
2038 2013-08-08 Steve Ellcey <sellcey@mips.com>
2040 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
2041 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
2042 micromips incompatible. Add nan2008.
2043 (MULTILIB_DIRNAMES): Add nan2008.
2044 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
2045 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
2046 and micromips incompatible. Add nan2008.
2047 (MULTILIB_DIRNAMES): Add nan2008.
2048 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
2050 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
2052 PR rtl-optimization/58079
2053 * combine.c (combine_simplify_rtx): Avoid using SUBST if
2054 simplify_comparison has widened a comparison with an integer.
2056 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2058 * config/arm/neon.md (movmisalign<mode>): Disable when we
2059 don't allow unaligned accesses.
2060 (*movmisalign<mode>_neon_store): Likewise.
2061 (*movmisalign<mode>_neon_load): Likewise.
2062 (*movmisalign<mode>_neon_store): Likewise.
2063 (*movmisalign<mode>_neon_load): Likewise.
2065 2013-08-08 Jan Hubicka <jh@suse.cz>
2067 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
2068 (make_pass_rebuild_cgraph_edges): Also clear references.
2069 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
2070 * ipa-inline-transform.c (inline_transform): Remove all references
2072 * cgraphunit.c (expand_function): Remove all references after
2074 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
2075 (ipa_find_reference): Rewrite to iterator.
2076 (remove_stmt_references): Likewise.
2077 (ipa_clear_stmts_in_references): New function.
2078 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
2079 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
2081 * ipa-split.c (split_function): Remove references in split function.
2083 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
2086 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
2087 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
2089 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
2092 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
2093 suggested mode for the assignment isn't compatible with the
2096 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2099 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
2101 2013-08-07 Xinliang David Li <davidxl@google.com>
2103 * config/i386/i386.opt: New option -mtune-ctrl=.
2104 * config/i386/x86-tune.def: New file.
2105 * config/i386/i386.h: include x86-tune.def.
2106 * config/i386/i386.c (ix86_option_override_internal):
2107 Parsing -mtune-ctrl= option and set tune features.
2109 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
2112 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
2113 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
2116 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
2118 * rtl.h (update_alignments): Declare.
2119 * final.c (grow_label_align): New function extracted from...
2120 (shorten_branches): ...here. Call it.
2121 (update_alignments): New function.
2122 * reorg.c (sibling_labels): New variable.
2123 (get_label_before): Add SIBLING parameter. If it is non-zero, push
2124 the new label along with it onto the sibling_labels vector.
2125 (fill_simple_delay_slots): Adjust call to get_label_before.
2126 (fill_slots_from_thread): Likewise.
2127 (relax_delay_slots): Likewise.
2128 (make_return_insns): Likewise.
2129 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
2131 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
2133 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
2134 document its semantics.
2135 (diagnostic_report_diagnostic): Adjust accordingly.
2137 2013-08-07 David Malcolm <dmalcolm@redhat.com>
2139 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
2140 (sparc_option_override): ...and port to new C++ pass API.
2141 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
2143 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
2145 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
2147 2013-08-06 Caroline Tice <cmtice@google.com>
2149 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
2150 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
2151 * tree-pass.h: Add pass_vtable_verify.
2152 * varasm.c (assemble_variable): Add code to properly set the comdat
2153 section and name for the .vtable_map_vars section.
2154 (assemble_vtyv_preinit_initializer): New function.
2155 (default_sectin_type_flags): Make sure .vtable_map_vars section has
2157 * output.h: Add function decl for assemble_vtv_preinit_initializer.
2158 * vtable-verify.c: New file.
2159 * vtable-verify.h: New file.
2160 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
2161 initialiation levels.
2162 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
2163 * passes.def: Insert pass_vtable_verify.
2164 * aclocal.m4: Reorder includes.
2165 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
2166 -fvtv-counts options.
2167 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
2168 as appropriate, if -fvtable-verify=... is used.
2169 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
2170 -fvtable-verify=... is used.
2171 * Makefile.in (OBJS): Add vtable-verify.o to list.
2172 (vtable-verify.o): Add new build rule.
2173 (GTFILES): Add vtable-verify.c to list.
2174 * common.opt (fvtable-verify=): New flag.
2175 (vtv_priority): Values for fvtable-verify= flag.
2176 (fvtv-counts): New flag.
2177 (fvtv-debug): New flag.
2178 * tree.h (save_vtable_map_decl): New extern function decl.
2180 2013-08-07 David Malcolm <dmalcolm@redhat.com>
2182 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
2183 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
2184 (pass_data_rl78_devirt): ...new pass_data instance and...
2185 (make_pass_rl78_devirt): ...new function.
2186 (rl78_asm_file_start): Port pass registration to new C++ API.
2188 2013-08-07 David Malcolm <dmalcolm@redhat.com>
2190 * coretypes.h (rtl_opt_pass): Add.
2191 (gcc::context): Add.
2192 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
2193 (epiphany_init): Port to new C++ pass API.
2194 (epiphany_optimize_mode_switching): Likewise.
2195 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
2196 (pass_manager::get_pass_mode_switching): New.
2197 (pass_manager::get_pass_peephole2): New.
2198 * mode-switching.c (pass_mode_switching): Add clone method.
2199 * recog.c (pass_peephole2): Add clone method.
2200 (pass_split_all_insns): Add clone method.
2202 2013-08-06 David Malcolm <dmalcolm@redhat.com>
2204 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
2205 (mips_option_override): ...here, porting to new C++ API for passes.
2207 2013-08-06 Jan Hubicka <jh@suse.cz>
2209 * cgraph.c (cgraph_get_body): New function based on lto.c
2211 * cgraph.h (cgraph_get_body): Declare.
2212 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
2214 * cgraphunit.c (expand_function): Get body prior expanding.
2215 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
2216 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
2218 * passes.c (do_per_function_toporder): Get body.
2219 * tree-inline.c (expand_call_inline): Get body prior inlining it.
2220 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
2222 2013-08-06 Martin Jambor <mjambor@suse.cz>
2225 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
2226 is not re-finalized. Rename second parameter to no_collect.
2228 2013-08-06 Martin Jambor <mjambor@suse.cz>
2231 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
2232 MEM_REF has proper alignment information.
2234 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
2237 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
2239 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
2240 rtx (*) (rtx, ...) with insn_gen_fn.
2241 * genoutput.c (output_insn_data): Cast gen_? function pointers to
2242 insn_gen_fn::stored_funcptr. Add initializer braces.
2244 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2246 Rewrite how instances of passes are cloned to remove assumptions
2247 about their sizes (thus allowing pass subclasses to have
2248 additional data fields, albeit non-GC-managed ones at this point).
2250 * passes.c (make_pass_instance): Now that passes have clone
2251 methods, rewrite this function to eliminate XNEW and memcpy
2252 calls that used hardcoded sizes. Since this function no longer
2253 creates pass instances, rename it to...
2254 (add_pass_instance): ...this. Document the old way that passes were
2255 numbered and flagged, and rework this function to continue using it.
2256 (next_pass_1): Add an initial_pass argument for use by
2258 (position_pass): When adding multiple instances of a pass, use
2259 the pass's clone method, rather than relying on the XNEW/memcpy
2260 within the former make_pass_instance (now add_pass_instance).
2261 (pass_manager::pass_manager): When invoking next_pass_1, also supply
2262 the initial instance of the current pass within the pass manager.
2264 2013-08-05 David Malcolm <dmalcolm@redhat.com>
2266 This is the automated part of the conversion of passes from C
2267 structs to C++ classes.
2269 Patch autogenerated by refactor_passes.py from
2270 https://github.com/davidmalcolm/gcc-refactoring-scripts
2271 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
2273 * asan.c (pass_asan): Convert from a global struct to a subclass of
2274 gimple_opt_pass along with...
2275 (pass_data_asan): ...new pass_data instance and...
2276 (make_pass_asan): ...new function.
2277 (pass_asan_O0): Convert from a global struct to a subclass of
2278 gimple_opt_pass along with...
2279 (pass_data_asan_O0): ...new pass_data instance and...
2280 (make_pass_asan_O0): ...new function.
2281 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
2282 subclass of rtl_opt_pass along with...
2283 (pass_data_inc_dec): ...new pass_data instance and...
2284 (make_pass_inc_dec): ...new function.
2285 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
2286 a subclass of rtl_opt_pass along with...
2287 (pass_data_reorder_blocks): ...new pass_data instance and...
2288 (make_pass_reorder_blocks): ...new function.
2289 (pass_duplicate_computed_gotos): Convert from a global struct to a
2290 subclass of rtl_opt_pass along with...
2291 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
2292 (make_pass_duplicate_computed_gotos): ...new function.
2293 (pass_partition_blocks): Convert from a global struct to a subclass of
2294 rtl_opt_pass along with...
2295 (pass_data_partition_blocks): ...new pass_data instance and...
2296 (make_pass_partition_blocks): ...new function.
2297 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
2298 struct to a subclass of rtl_opt_pass along with...
2299 (pass_data_branch_target_load_optimize1): ...new pass_data instance
2301 (make_pass_branch_target_load_optimize1): ...new function.
2302 (pass_branch_target_load_optimize2): Convert from a global struct to a
2303 subclass of rtl_opt_pass along with...
2304 (pass_data_branch_target_load_optimize2): ...new pass_data instance
2306 (make_pass_branch_target_load_optimize2): ...new function.
2307 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
2308 of rtl_opt_pass along with...
2309 (pass_data_jump): ...new pass_data instance and...
2310 (make_pass_jump): ...new function.
2311 (pass_jump2): Convert from a global struct to a subclass of
2312 rtl_opt_pass along with...
2313 (pass_data_jump2): ...new pass_data instance and...
2314 (make_pass_jump2): ...new function.
2315 * cfgexpand.c (pass_expand): Convert from a global struct to a
2316 subclass of rtl_opt_pass along with...
2317 (pass_data_expand): ...new pass_data instance and...
2318 (make_pass_expand): ...new function.
2319 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
2320 of rtl_opt_pass along with...
2321 (pass_data_free_cfg): ...new pass_data instance and...
2322 (make_pass_free_cfg): ...new function.
2323 (pass_into_cfg_layout_mode): Convert from a global struct to a
2324 subclass of rtl_opt_pass along with...
2325 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
2326 (make_pass_into_cfg_layout_mode): ...new function.
2327 (pass_outof_cfg_layout_mode): Convert from a global struct to a
2328 subclass of rtl_opt_pass along with...
2329 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
2330 (make_pass_outof_cfg_layout_mode): ...new function.
2331 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
2332 struct to a subclass of gimple_opt_pass along with...
2333 (pass_data_build_cgraph_edges): ...new pass_data instance and...
2334 (make_pass_build_cgraph_edges): ...new function.
2335 (pass_rebuild_cgraph_edges): Convert from a global struct to a
2336 subclass of gimple_opt_pass along with...
2337 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
2338 (make_pass_rebuild_cgraph_edges): ...new function.
2339 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
2340 subclass of gimple_opt_pass along with...
2341 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
2343 (make_pass_remove_cgraph_callee_edges): ...new function.
2344 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
2345 struct to a subclass of rtl_opt_pass along with...
2346 (pass_data_stack_adjustments): ...new pass_data instance and...
2347 (make_pass_stack_adjustments): ...new function.
2348 * combine.c (pass_combine): Convert from a global struct to a subclass
2349 of rtl_opt_pass along with...
2350 (pass_data_combine): ...new pass_data instance and...
2351 (make_pass_combine): ...new function.
2352 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
2353 global struct to a subclass of rtl_opt_pass along with...
2354 (pass_data_compare_elim_after_reload): ...new pass_data instance
2356 (make_pass_compare_elim_after_reload): ...new function.
2357 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
2358 of rtl_opt_pass along with...
2359 (pass_data_rtl_cprop): ...new pass_data instance and...
2360 (make_pass_rtl_cprop): ...new function.
2361 * cse.c (pass_cse): Convert from a global struct to a subclass of
2362 rtl_opt_pass along with...
2363 (pass_data_cse): ...new pass_data instance and...
2364 (make_pass_cse): ...new function.
2365 (pass_cse2): Convert from a global struct to a subclass of
2366 rtl_opt_pass along with...
2367 (pass_data_cse2): ...new pass_data instance and...
2368 (make_pass_cse2): ...new function.
2369 (pass_cse_after_global_opts): Convert from a global struct to a
2370 subclass of rtl_opt_pass along with...
2371 (pass_data_cse_after_global_opts): ...new pass_data instance and...
2372 (make_pass_cse_after_global_opts): ...new function.
2373 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
2374 of rtl_opt_pass along with...
2375 (pass_data_ud_rtl_dce): ...new pass_data instance and...
2376 (make_pass_ud_rtl_dce): ...new function.
2377 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
2378 rtl_opt_pass along with...
2379 (pass_data_fast_rtl_dce): ...new pass_data instance and...
2380 (make_pass_fast_rtl_dce): ...new function.
2381 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
2382 a subclass of rtl_opt_pass along with...
2383 (pass_data_df_initialize_opt): ...new pass_data instance and...
2384 (make_pass_df_initialize_opt): ...new function.
2385 (pass_df_initialize_no_opt): Convert from a global struct to a
2386 subclass of rtl_opt_pass along with...
2387 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
2388 (make_pass_df_initialize_no_opt): ...new function.
2389 (pass_df_finish): Convert from a global struct to a subclass of
2390 rtl_opt_pass along with...
2391 (pass_data_df_finish): ...new pass_data instance and...
2392 (make_pass_df_finish): ...new function.
2393 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
2394 rtl_opt_pass along with...
2395 (pass_data_rtl_dse1): ...new pass_data instance and...
2396 (make_pass_rtl_dse1): ...new function.
2397 (pass_rtl_dse2): Convert from a global struct to a subclass of
2398 rtl_opt_pass along with...
2399 (pass_data_rtl_dse2): ...new pass_data instance and...
2400 (make_pass_rtl_dse2): ...new function.
2401 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
2402 subclass of rtl_opt_pass along with...
2403 (pass_data_dwarf2_frame): ...new pass_data instance and...
2404 (make_pass_dwarf2_frame): ...new function.
2405 * except.c (pass_set_nothrow_function_flags): Convert from a global
2406 struct to a subclass of rtl_opt_pass along with...
2407 (pass_data_set_nothrow_function_flags): ...new pass_data instance
2409 (make_pass_set_nothrow_function_flags): ...new function.
2410 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
2411 subclass of rtl_opt_pass along with...
2412 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
2414 (make_pass_convert_to_eh_region_ranges): ...new function.
2415 * final.c (pass_compute_alignments): Convert from a global struct to a
2416 subclass of rtl_opt_pass along with...
2417 (pass_data_compute_alignments): ...new pass_data instance and...
2418 (make_pass_compute_alignments): ...new function.
2419 (pass_final): Convert from a global struct to a subclass of
2420 rtl_opt_pass along with...
2421 (pass_data_final): ...new pass_data instance and...
2422 (make_pass_final): ...new function.
2423 (pass_shorten_branches): Convert from a global struct to a subclass of
2424 rtl_opt_pass along with...
2425 (pass_data_shorten_branches): ...new pass_data instance and...
2426 (make_pass_shorten_branches): ...new function.
2427 (pass_clean_state): Convert from a global struct to a subclass of
2428 rtl_opt_pass along with...
2429 (pass_data_clean_state): ...new pass_data instance and...
2430 (make_pass_clean_state): ...new function.
2431 * function.c (pass_instantiate_virtual_regs): Convert from a global
2432 struct to a subclass of rtl_opt_pass along with...
2433 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
2434 (make_pass_instantiate_virtual_regs): ...new function.
2435 (pass_leaf_regs): Convert from a global struct to a subclass of
2436 rtl_opt_pass along with...
2437 (pass_data_leaf_regs): ...new pass_data instance and...
2438 (make_pass_leaf_regs): ...new function.
2439 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
2440 subclass of rtl_opt_pass along with...
2441 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
2443 (make_pass_thread_prologue_and_epilogue): ...new function.
2444 (pass_match_asm_constraints): Convert from a global struct to a
2445 subclass of rtl_opt_pass along with...
2446 (pass_data_match_asm_constraints): ...new pass_data instance and...
2447 (make_pass_match_asm_constraints): ...new function.
2448 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
2449 subclass of rtl_opt_pass along with...
2450 (pass_data_rtl_fwprop): ...new pass_data instance and...
2451 (make_pass_rtl_fwprop): ...new function.
2452 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
2453 rtl_opt_pass along with...
2454 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
2455 (make_pass_rtl_fwprop_addr): ...new function.
2456 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
2457 rtl_opt_pass along with...
2458 (pass_data_rtl_pre): ...new pass_data instance and...
2459 (make_pass_rtl_pre): ...new function.
2460 (pass_rtl_hoist): Convert from a global struct to a subclass of
2461 rtl_opt_pass along with...
2462 (pass_data_rtl_hoist): ...new pass_data instance and...
2463 (make_pass_rtl_hoist): ...new function.
2464 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
2465 subclass of gimple_opt_pass along with...
2466 (pass_data_lower_cf): ...new pass_data instance and...
2467 (make_pass_lower_cf): ...new function.
2468 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
2469 from a global struct to a subclass of gimple_opt_pass along with...
2470 (pass_data_strength_reduction): ...new pass_data instance and...
2471 (make_pass_strength_reduction): ...new function.
2472 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
2473 of rtl_opt_pass along with...
2474 (pass_data_rtl_ifcvt): ...new pass_data instance and...
2475 (make_pass_rtl_ifcvt): ...new function.
2476 (pass_if_after_combine): Convert from a global struct to a subclass of
2477 rtl_opt_pass along with...
2478 (pass_data_if_after_combine): ...new pass_data instance and...
2479 (make_pass_if_after_combine): ...new function.
2480 (pass_if_after_reload): Convert from a global struct to a subclass of
2481 rtl_opt_pass along with...
2482 (pass_data_if_after_reload): ...new pass_data instance and...
2483 (make_pass_if_after_reload): ...new function.
2484 * init-regs.c (pass_initialize_regs): Convert from a global struct to
2485 a subclass of rtl_opt_pass along with...
2486 (pass_data_initialize_regs): ...new pass_data instance and...
2487 (make_pass_initialize_regs): ...new function.
2488 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
2489 of ipa_opt_pass_d along with...
2490 (pass_data_ipa_cp): ...new pass_data instance and...
2491 (make_pass_ipa_cp): ...new function.
2492 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
2493 global struct to a subclass of gimple_opt_pass along with...
2494 (pass_data_inline_parameters): ...new pass_data instance and...
2495 (make_pass_inline_parameters): ...new function.
2496 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
2497 subclass of gimple_opt_pass along with...
2498 (pass_data_early_inline): ...new pass_data instance and...
2499 (make_pass_early_inline): ...new function.
2500 (pass_ipa_inline): Convert from a global struct to a subclass of
2501 ipa_opt_pass_d along with...
2502 (pass_data_ipa_inline): ...new pass_data instance and...
2503 (make_pass_ipa_inline): ...new function.
2504 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
2505 struct to a subclass of gimple_opt_pass along with...
2506 (pass_data_local_pure_const): ...new pass_data instance and...
2507 (make_pass_local_pure_const): ...new function.
2508 (pass_ipa_pure_const): Convert from a global struct to a subclass of
2509 ipa_opt_pass_d along with...
2510 (pass_data_ipa_pure_const): ...new pass_data instance and...
2511 (make_pass_ipa_pure_const): ...new function.
2512 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
2513 to a subclass of ipa_opt_pass_d along with...
2514 (pass_data_ipa_reference): ...new pass_data instance and...
2515 (make_pass_ipa_reference): ...new function.
2516 * ipa-split.c (pass_split_functions): Convert from a global struct to
2517 a subclass of gimple_opt_pass along with...
2518 (pass_data_split_functions): ...new pass_data instance and...
2519 (make_pass_split_functions): ...new function.
2520 (pass_feedback_split_functions): Convert from a global struct to a
2521 subclass of gimple_opt_pass along with...
2522 (pass_data_feedback_split_functions): ...new pass_data instance and...
2523 (make_pass_feedback_split_functions): ...new function.
2524 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
2525 global struct to a subclass of simple_ipa_opt_pass along with...
2526 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
2528 (make_pass_ipa_function_and_variable_visibility): ...new function.
2529 (pass_ipa_free_inline_summary): Convert from a global struct to a
2530 subclass of simple_ipa_opt_pass along with...
2531 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
2532 (make_pass_ipa_free_inline_summary): ...new function.
2533 (pass_ipa_whole_program_visibility): Convert from a global struct to a
2534 subclass of ipa_opt_pass_d along with...
2535 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
2537 (make_pass_ipa_whole_program_visibility): ...new function.
2538 (pass_ipa_profile): Convert from a global struct to a subclass of
2539 ipa_opt_pass_d along with...
2540 (pass_data_ipa_profile): ...new pass_data instance and...
2541 (make_pass_ipa_profile): ...new function.
2542 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
2543 ipa_opt_pass_d along with...
2544 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
2545 (make_pass_ipa_cdtor_merge): ...new function.
2546 * ira.c (pass_ira): Convert from a global struct to a subclass of
2547 rtl_opt_pass along with...
2548 (pass_data_ira): ...new pass_data instance and...
2549 (make_pass_ira): ...new function.
2550 (pass_reload): Convert from a global struct to a subclass of
2551 rtl_opt_pass along with...
2552 (pass_data_reload): ...new pass_data instance and...
2553 (make_pass_reload): ...new function.
2554 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
2555 subclass of rtl_opt_pass along with...
2556 (pass_data_cleanup_barriers): ...new pass_data instance and...
2557 (make_pass_cleanup_barriers): ...new function.
2558 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
2559 of rtl_opt_pass along with...
2560 (pass_data_loop2): ...new pass_data instance and...
2561 (make_pass_loop2): ...new function.
2562 (pass_rtl_loop_init): Convert from a global struct to a subclass of
2563 rtl_opt_pass along with...
2564 (pass_data_rtl_loop_init): ...new pass_data instance and...
2565 (make_pass_rtl_loop_init): ...new function.
2566 (pass_rtl_loop_done): Convert from a global struct to a subclass of
2567 rtl_opt_pass along with...
2568 (pass_data_rtl_loop_done): ...new pass_data instance and...
2569 (make_pass_rtl_loop_done): ...new function.
2570 (pass_rtl_move_loop_invariants): Convert from a global struct to a
2571 subclass of rtl_opt_pass along with...
2572 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
2573 (make_pass_rtl_move_loop_invariants): ...new function.
2574 (pass_rtl_unswitch): Convert from a global struct to a subclass of
2575 rtl_opt_pass along with...
2576 (pass_data_rtl_unswitch): ...new pass_data instance and...
2577 (make_pass_rtl_unswitch): ...new function.
2578 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
2579 subclass of rtl_opt_pass along with...
2580 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
2582 (make_pass_rtl_unroll_and_peel_loops): ...new function.
2583 (pass_rtl_doloop): Convert from a global struct to a subclass of
2584 rtl_opt_pass along with...
2585 (pass_data_rtl_doloop): ...new pass_data instance and...
2586 (make_pass_rtl_doloop): ...new function.
2587 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
2588 a subclass of rtl_opt_pass along with...
2589 (pass_data_lower_subreg): ...new pass_data instance and...
2590 (make_pass_lower_subreg): ...new function.
2591 (pass_lower_subreg2): Convert from a global struct to a subclass of
2592 rtl_opt_pass along with...
2593 (pass_data_lower_subreg2): ...new pass_data instance and...
2594 (make_pass_lower_subreg2): ...new function.
2595 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
2596 struct to a subclass of ipa_opt_pass_d along with...
2597 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
2598 (make_pass_ipa_lto_gimple_out): ...new function.
2599 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
2600 of ipa_opt_pass_d along with...
2601 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2602 (make_pass_ipa_lto_finish_out): ...new function.
2603 * mode-switching.c (pass_mode_switching): Convert from a global struct
2604 to a subclass of rtl_opt_pass along with...
2605 (pass_data_mode_switching): ...new pass_data instance and...
2606 (make_pass_mode_switching): ...new function.
2607 * modulo-sched.c (pass_sms): Convert from a global struct to a
2608 subclass of rtl_opt_pass along with...
2609 (pass_data_sms): ...new pass_data instance and...
2610 (make_pass_sms): ...new function.
2611 * omp-low.c (pass_expand_omp): Convert from a global struct to a
2612 subclass of gimple_opt_pass along with...
2613 (pass_data_expand_omp): ...new pass_data instance and...
2614 (make_pass_expand_omp): ...new function.
2615 (pass_lower_omp): Convert from a global struct to a subclass of
2616 gimple_opt_pass along with...
2617 (pass_data_lower_omp): ...new pass_data instance and...
2618 (make_pass_lower_omp): ...new function.
2619 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2620 of gimple_opt_pass along with...
2621 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2622 (make_pass_diagnose_omp_blocks): ...new function.
2623 * passes.c (pass_early_local_passes): Convert from a global struct to
2624 a subclass of simple_ipa_opt_pass along with...
2625 (pass_data_early_local_passes): ...new pass_data instance and...
2626 (make_pass_early_local_passes): ...new function.
2627 (pass_all_early_optimizations): Convert from a global struct to a
2628 subclass of gimple_opt_pass along with...
2629 (pass_data_all_early_optimizations): ...new pass_data instance and...
2630 (make_pass_all_early_optimizations): ...new function.
2631 (pass_all_optimizations): Convert from a global struct to a subclass
2632 of gimple_opt_pass along with...
2633 (pass_data_all_optimizations): ...new pass_data instance and...
2634 (make_pass_all_optimizations): ...new function.
2635 (pass_all_optimizations_g): Convert from a global struct to a subclass
2636 of gimple_opt_pass along with...
2637 (pass_data_all_optimizations_g): ...new pass_data instance and...
2638 (make_pass_all_optimizations_g): ...new function.
2639 (pass_rest_of_compilation): Convert from a global struct to a subclass
2640 of rtl_opt_pass along with...
2641 (pass_data_rest_of_compilation): ...new pass_data instance and...
2642 (make_pass_rest_of_compilation): ...new function.
2643 (pass_postreload): Convert from a global struct to a subclass of
2644 rtl_opt_pass along with...
2645 (pass_data_postreload): ...new pass_data instance and...
2646 (make_pass_postreload): ...new function.
2647 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2648 subclass of rtl_opt_pass along with...
2649 (pass_data_gcse2): ...new pass_data instance and...
2650 (make_pass_gcse2): ...new function.
2651 * postreload.c (pass_postreload_cse): Convert from a global struct to
2652 a subclass of rtl_opt_pass along with...
2653 (pass_data_postreload_cse): ...new pass_data instance and...
2654 (make_pass_postreload_cse): ...new function.
2655 * predict.c (pass_profile): Convert from a global struct to a subclass
2656 of gimple_opt_pass along with...
2657 (pass_data_profile): ...new pass_data instance and...
2658 (make_pass_profile): ...new function.
2659 (pass_strip_predict_hints): Convert from a global struct to a subclass
2660 of gimple_opt_pass along with...
2661 (pass_data_strip_predict_hints): ...new pass_data instance and...
2662 (make_pass_strip_predict_hints): ...new function.
2663 * recog.c (pass_peephole2): Convert from a global struct to a subclass
2664 of rtl_opt_pass along with...
2665 (pass_data_peephole2): ...new pass_data instance and...
2666 (make_pass_peephole2): ...new function.
2667 (pass_split_all_insns): Convert from a global struct to a subclass of
2668 rtl_opt_pass along with...
2669 (pass_data_split_all_insns): ...new pass_data instance and...
2670 (make_pass_split_all_insns): ...new function.
2671 (pass_split_after_reload): Convert from a global struct to a subclass
2672 of rtl_opt_pass along with...
2673 (pass_data_split_after_reload): ...new pass_data instance and...
2674 (make_pass_split_after_reload): ...new function.
2675 (pass_split_before_regstack): Convert from a global struct to a
2676 subclass of rtl_opt_pass along with...
2677 (pass_data_split_before_regstack): ...new pass_data instance and...
2678 (make_pass_split_before_regstack): ...new function.
2679 (pass_split_before_sched2): Convert from a global struct to a subclass
2680 of rtl_opt_pass along with...
2681 (pass_data_split_before_sched2): ...new pass_data instance and...
2682 (make_pass_split_before_sched2): ...new function.
2683 (pass_split_for_shorten_branches): Convert from a global struct to a
2684 subclass of rtl_opt_pass along with...
2685 (pass_data_split_for_shorten_branches): ...new pass_data instance
2687 (make_pass_split_for_shorten_branches): ...new function.
2688 * ree.c (pass_ree): Convert from a global struct to a subclass of
2689 rtl_opt_pass along with...
2690 (pass_data_ree): ...new pass_data instance and...
2691 (make_pass_ree): ...new function.
2692 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2693 subclass of rtl_opt_pass along with...
2694 (pass_data_stack_regs): ...new pass_data instance and...
2695 (make_pass_stack_regs): ...new function.
2696 (pass_stack_regs_run): Convert from a global struct to a subclass of
2697 rtl_opt_pass along with...
2698 (pass_data_stack_regs_run): ...new pass_data instance and...
2699 (make_pass_stack_regs_run): ...new function.
2700 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2701 subclass of rtl_opt_pass along with...
2702 (pass_data_cprop_hardreg): ...new pass_data instance and...
2703 (make_pass_cprop_hardreg): ...new function.
2704 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2705 subclass of rtl_opt_pass along with...
2706 (pass_data_reginfo_init): ...new pass_data instance and...
2707 (make_pass_reginfo_init): ...new function.
2708 * regmove.c (pass_regmove): Convert from a global struct to a subclass
2709 of rtl_opt_pass along with...
2710 (pass_data_regmove): ...new pass_data instance and...
2711 (make_pass_regmove): ...new function.
2712 * regrename.c (pass_regrename): Convert from a global struct to a
2713 subclass of rtl_opt_pass along with...
2714 (pass_data_regrename): ...new pass_data instance and...
2715 (make_pass_regrename): ...new function.
2716 * reorg.c (pass_delay_slots): Convert from a global struct to a
2717 subclass of rtl_opt_pass along with...
2718 (pass_data_delay_slots): ...new pass_data instance and...
2719 (make_pass_delay_slots): ...new function.
2720 (pass_machine_reorg): Convert from a global struct to a subclass of
2721 rtl_opt_pass along with...
2722 (pass_data_machine_reorg): ...new pass_data instance and...
2723 (make_pass_machine_reorg): ...new function.
2724 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2725 of rtl_opt_pass along with...
2726 (pass_data_sched): ...new pass_data instance and...
2727 (make_pass_sched): ...new function.
2728 (pass_sched2): Convert from a global struct to a subclass of
2729 rtl_opt_pass along with...
2730 (pass_data_sched2): ...new pass_data instance and...
2731 (make_pass_sched2): ...new function.
2732 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2733 to a subclass of rtl_opt_pass along with...
2734 (pass_data_stack_ptr_mod): ...new pass_data instance and...
2735 (make_pass_stack_ptr_mod): ...new function.
2736 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2737 to a subclass of rtl_opt_pass along with...
2738 (pass_data_rtl_store_motion): ...new pass_data instance and...
2739 (make_pass_rtl_store_motion): ...new function.
2740 * tracer.c (pass_tracer): Convert from a global struct to a subclass
2741 of gimple_opt_pass along with...
2742 (pass_data_tracer): ...new pass_data instance and...
2743 (make_pass_tracer): ...new function.
2744 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2745 to a subclass of gimple_opt_pass along with...
2746 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2747 (make_pass_diagnose_tm_blocks): ...new function.
2748 (pass_lower_tm): Convert from a global struct to a subclass of
2749 gimple_opt_pass along with...
2750 (pass_data_lower_tm): ...new pass_data instance and...
2751 (make_pass_lower_tm): ...new function.
2752 (pass_tm_init): Convert from a global struct to a subclass of
2753 gimple_opt_pass along with...
2754 (pass_data_tm_init): ...new pass_data instance and...
2755 (make_pass_tm_init): ...new function.
2756 (pass_tm_mark): Convert from a global struct to a subclass of
2757 gimple_opt_pass along with...
2758 (pass_data_tm_mark): ...new pass_data instance and...
2759 (make_pass_tm_mark): ...new function.
2760 (pass_tm_edges): Convert from a global struct to a subclass of
2761 gimple_opt_pass along with...
2762 (pass_data_tm_edges): ...new pass_data instance and...
2763 (make_pass_tm_edges): ...new function.
2764 (pass_tm_memopt): Convert from a global struct to a subclass of
2765 gimple_opt_pass along with...
2766 (pass_data_tm_memopt): ...new pass_data instance and...
2767 (make_pass_tm_memopt): ...new function.
2768 (pass_ipa_tm): Convert from a global struct to a subclass of
2769 simple_ipa_opt_pass along with...
2770 (pass_data_ipa_tm): ...new pass_data instance and...
2771 (make_pass_ipa_tm): ...new function.
2772 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2773 subclass of gimple_opt_pass along with...
2774 (pass_data_call_cdce): ...new pass_data instance and...
2775 (make_pass_call_cdce): ...new function.
2776 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2777 subclass of gimple_opt_pass along with...
2778 (pass_data_build_cfg): ...new pass_data instance and...
2779 (make_pass_build_cfg): ...new function.
2780 (pass_split_crit_edges): Convert from a global struct to a subclass of
2781 gimple_opt_pass along with...
2782 (pass_data_split_crit_edges): ...new pass_data instance and...
2783 (make_pass_split_crit_edges): ...new function.
2784 (pass_warn_function_return): Convert from a global struct to a
2785 subclass of gimple_opt_pass along with...
2786 (pass_data_warn_function_return): ...new pass_data instance and...
2787 (make_pass_warn_function_return): ...new function.
2788 (pass_warn_function_noreturn): Convert from a global struct to a
2789 subclass of gimple_opt_pass along with...
2790 (pass_data_warn_function_noreturn): ...new pass_data instance and...
2791 (make_pass_warn_function_noreturn): ...new function.
2792 (pass_warn_unused_result): Convert from a global struct to a subclass
2793 of gimple_opt_pass along with...
2794 (pass_data_warn_unused_result): ...new pass_data instance and...
2795 (make_pass_warn_unused_result): ...new function.
2796 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2797 a subclass of gimple_opt_pass along with...
2798 (pass_data_merge_phi): ...new pass_data instance and...
2799 (make_pass_merge_phi): ...new function.
2800 * tree-complex.c (pass_lower_complex): Convert from a global struct to
2801 a subclass of gimple_opt_pass along with...
2802 (pass_data_lower_complex): ...new pass_data instance and...
2803 (make_pass_lower_complex): ...new function.
2804 (pass_lower_complex_O0): Convert from a global struct to a subclass of
2805 gimple_opt_pass along with...
2806 (pass_data_lower_complex_O0): ...new pass_data instance and...
2807 (make_pass_lower_complex_O0): ...new function.
2808 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2809 subclass of gimple_opt_pass along with...
2810 (pass_data_lower_eh): ...new pass_data instance and...
2811 (make_pass_lower_eh): ...new function.
2812 (pass_refactor_eh): Convert from a global struct to a subclass of
2813 gimple_opt_pass along with...
2814 (pass_data_refactor_eh): ...new pass_data instance and...
2815 (make_pass_refactor_eh): ...new function.
2816 (pass_lower_resx): Convert from a global struct to a subclass of
2817 gimple_opt_pass along with...
2818 (pass_data_lower_resx): ...new pass_data instance and...
2819 (make_pass_lower_resx): ...new function.
2820 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2821 of gimple_opt_pass along with...
2822 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2823 (make_pass_lower_eh_dispatch): ...new function.
2824 (pass_cleanup_eh): Convert from a global struct to a subclass of
2825 gimple_opt_pass along with...
2826 (pass_data_cleanup_eh): ...new pass_data instance and...
2827 (make_pass_cleanup_eh): ...new function.
2828 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2829 to a subclass of simple_ipa_opt_pass along with...
2830 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2831 (make_pass_ipa_lower_emutls): ...new function.
2832 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2833 a subclass of gimple_opt_pass along with...
2834 (pass_data_if_conversion): ...new pass_data instance and...
2835 (make_pass_if_conversion): ...new function.
2836 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2837 subclass of gimple_opt_pass along with...
2838 (pass_data_build_ssa): ...new pass_data instance and...
2839 (make_pass_build_ssa): ...new function.
2840 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2841 global struct to a subclass of gimple_opt_pass along with...
2842 (pass_data_loop_distribution): ...new pass_data instance and...
2843 (make_pass_loop_distribution): ...new function.
2844 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2845 subclass of gimple_opt_pass along with...
2846 (pass_data_mudflap_1): ...new pass_data instance and...
2847 (make_pass_mudflap_1): ...new function.
2848 (pass_mudflap_2): Convert from a global struct to a subclass of
2849 gimple_opt_pass along with...
2850 (pass_data_mudflap_2): ...new pass_data instance and...
2851 (make_pass_mudflap_2): ...new function.
2852 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2853 subclass of gimple_opt_pass along with...
2854 (pass_data_mudflap_1): ...new pass_data instance and...
2855 (make_pass_mudflap_1): ...new function.
2856 (pass_mudflap_2): Convert from a global struct to a subclass of
2857 gimple_opt_pass along with...
2858 (pass_data_mudflap_2): ...new pass_data instance and...
2859 (make_pass_mudflap_2): ...new function.
2860 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2861 gimple_opt_pass along with...
2862 (pass_data_nrv): ...new pass_data instance and...
2863 (make_pass_nrv): ...new function.
2864 (pass_return_slot): Convert from a global struct to a subclass of
2865 gimple_opt_pass along with...
2866 (pass_data_return_slot): ...new pass_data instance and...
2867 (make_pass_return_slot): ...new function.
2868 * tree-object-size.c (pass_object_sizes): Convert from a global struct
2869 to a subclass of gimple_opt_pass along with...
2870 (pass_data_object_sizes): ...new pass_data instance and...
2871 (make_pass_object_sizes): ...new function.
2872 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2873 global struct to a subclass of gimple_opt_pass along with...
2874 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2876 (make_pass_cleanup_cfg_post_optimizing): ...new function.
2877 (pass_fixup_cfg): Convert from a global struct to a subclass of
2878 gimple_opt_pass along with...
2879 (pass_data_fixup_cfg): ...new pass_data instance and...
2880 (make_pass_fixup_cfg): ...new function.
2881 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2882 (make_pass_mudflap_1): ...new function.
2883 (pass_mudflap_2): Replace declaration with that of...
2884 (make_pass_mudflap_2): ...new function.
2885 (pass_asan): Replace declaration with that of...
2886 (make_pass_asan): ...new function.
2887 (pass_asan_O0): Replace declaration with that of...
2888 (make_pass_asan_O0): ...new function.
2889 (pass_tsan): Replace declaration with that of...
2890 (make_pass_tsan): ...new function.
2891 (pass_tsan_O0): Replace declaration with that of...
2892 (make_pass_tsan_O0): ...new function.
2893 (pass_lower_cf): Replace declaration with that of...
2894 (make_pass_lower_cf): ...new function.
2895 (pass_refactor_eh): Replace declaration with that of...
2896 (make_pass_refactor_eh): ...new function.
2897 (pass_lower_eh): Replace declaration with that of...
2898 (make_pass_lower_eh): ...new function.
2899 (pass_lower_eh_dispatch): Replace declaration with that of...
2900 (make_pass_lower_eh_dispatch): ...new function.
2901 (pass_lower_resx): Replace declaration with that of...
2902 (make_pass_lower_resx): ...new function.
2903 (pass_build_cfg): Replace declaration with that of...
2904 (make_pass_build_cfg): ...new function.
2905 (pass_early_tree_profile): Replace declaration with that of...
2906 (make_pass_early_tree_profile): ...new function.
2907 (pass_cleanup_eh): Replace declaration with that of...
2908 (make_pass_cleanup_eh): ...new function.
2909 (pass_sra): Replace declaration with that of...
2910 (make_pass_sra): ...new function.
2911 (pass_sra_early): Replace declaration with that of...
2912 (make_pass_sra_early): ...new function.
2913 (pass_early_ipa_sra): Replace declaration with that of...
2914 (make_pass_early_ipa_sra): ...new function.
2915 (pass_tail_recursion): Replace declaration with that of...
2916 (make_pass_tail_recursion): ...new function.
2917 (pass_tail_calls): Replace declaration with that of...
2918 (make_pass_tail_calls): ...new function.
2919 (pass_tree_loop): Replace declaration with that of...
2920 (make_pass_tree_loop): ...new function.
2921 (pass_tree_loop_init): Replace declaration with that of...
2922 (make_pass_tree_loop_init): ...new function.
2923 (pass_lim): Replace declaration with that of...
2924 (make_pass_lim): ...new function.
2925 (pass_tree_unswitch): Replace declaration with that of...
2926 (make_pass_tree_unswitch): ...new function.
2927 (pass_predcom): Replace declaration with that of...
2928 (make_pass_predcom): ...new function.
2929 (pass_iv_canon): Replace declaration with that of...
2930 (make_pass_iv_canon): ...new function.
2931 (pass_scev_cprop): Replace declaration with that of...
2932 (make_pass_scev_cprop): ...new function.
2933 (pass_empty_loop): Replace declaration with that of...
2934 (make_pass_empty_loop): ...new function.
2935 (pass_record_bounds): Replace declaration with that of...
2936 (make_pass_record_bounds): ...new function.
2937 (pass_graphite): Replace declaration with that of...
2938 (make_pass_graphite): ...new function.
2939 (pass_graphite_transforms): Replace declaration with that of...
2940 (make_pass_graphite_transforms): ...new function.
2941 (pass_if_conversion): Replace declaration with that of...
2942 (make_pass_if_conversion): ...new function.
2943 (pass_loop_distribution): Replace declaration with that of...
2944 (make_pass_loop_distribution): ...new function.
2945 (pass_vectorize): Replace declaration with that of...
2946 (make_pass_vectorize): ...new function.
2947 (pass_slp_vectorize): Replace declaration with that of...
2948 (make_pass_slp_vectorize): ...new function.
2949 (pass_complete_unroll): Replace declaration with that of...
2950 (make_pass_complete_unroll): ...new function.
2951 (pass_complete_unrolli): Replace declaration with that of...
2952 (make_pass_complete_unrolli): ...new function.
2953 (pass_parallelize_loops): Replace declaration with that of...
2954 (make_pass_parallelize_loops): ...new function.
2955 (pass_loop_prefetch): Replace declaration with that of...
2956 (make_pass_loop_prefetch): ...new function.
2957 (pass_iv_optimize): Replace declaration with that of...
2958 (make_pass_iv_optimize): ...new function.
2959 (pass_tree_loop_done): Replace declaration with that of...
2960 (make_pass_tree_loop_done): ...new function.
2961 (pass_ch): Replace declaration with that of...
2962 (make_pass_ch): ...new function.
2963 (pass_ccp): Replace declaration with that of...
2964 (make_pass_ccp): ...new function.
2965 (pass_phi_only_cprop): Replace declaration with that of...
2966 (make_pass_phi_only_cprop): ...new function.
2967 (pass_build_ssa): Replace declaration with that of...
2968 (make_pass_build_ssa): ...new function.
2969 (pass_build_alias): Replace declaration with that of...
2970 (make_pass_build_alias): ...new function.
2971 (pass_build_ealias): Replace declaration with that of...
2972 (make_pass_build_ealias): ...new function.
2973 (pass_dominator): Replace declaration with that of...
2974 (make_pass_dominator): ...new function.
2975 (pass_dce): Replace declaration with that of...
2976 (make_pass_dce): ...new function.
2977 (pass_dce_loop): Replace declaration with that of...
2978 (make_pass_dce_loop): ...new function.
2979 (pass_cd_dce): Replace declaration with that of...
2980 (make_pass_cd_dce): ...new function.
2981 (pass_call_cdce): Replace declaration with that of...
2982 (make_pass_call_cdce): ...new function.
2983 (pass_merge_phi): Replace declaration with that of...
2984 (make_pass_merge_phi): ...new function.
2985 (pass_split_crit_edges): Replace declaration with that of...
2986 (make_pass_split_crit_edges): ...new function.
2987 (pass_pre): Replace declaration with that of...
2988 (make_pass_pre): ...new function.
2989 (pass_profile): Replace declaration with that of...
2990 (make_pass_profile): ...new function.
2991 (pass_strip_predict_hints): Replace declaration with that of...
2992 (make_pass_strip_predict_hints): ...new function.
2993 (pass_lower_complex_O0): Replace declaration with that of...
2994 (make_pass_lower_complex_O0): ...new function.
2995 (pass_lower_complex): Replace declaration with that of...
2996 (make_pass_lower_complex): ...new function.
2997 (pass_lower_vector): Replace declaration with that of...
2998 (make_pass_lower_vector): ...new function.
2999 (pass_lower_vector_ssa): Replace declaration with that of...
3000 (make_pass_lower_vector_ssa): ...new function.
3001 (pass_lower_omp): Replace declaration with that of...
3002 (make_pass_lower_omp): ...new function.
3003 (pass_diagnose_omp_blocks): Replace declaration with that of...
3004 (make_pass_diagnose_omp_blocks): ...new function.
3005 (pass_expand_omp): Replace declaration with that of...
3006 (make_pass_expand_omp): ...new function.
3007 (pass_expand_omp_ssa): Replace declaration with that of...
3008 (make_pass_expand_omp_ssa): ...new function.
3009 (pass_object_sizes): Replace declaration with that of...
3010 (make_pass_object_sizes): ...new function.
3011 (pass_strlen): Replace declaration with that of...
3012 (make_pass_strlen): ...new function.
3013 (pass_fold_builtins): Replace declaration with that of...
3014 (make_pass_fold_builtins): ...new function.
3015 (pass_stdarg): Replace declaration with that of...
3016 (make_pass_stdarg): ...new function.
3017 (pass_early_warn_uninitialized): Replace declaration with that of...
3018 (make_pass_early_warn_uninitialized): ...new function.
3019 (pass_late_warn_uninitialized): Replace declaration with that of...
3020 (make_pass_late_warn_uninitialized): ...new function.
3021 (pass_cse_reciprocals): Replace declaration with that of...
3022 (make_pass_cse_reciprocals): ...new function.
3023 (pass_cse_sincos): Replace declaration with that of...
3024 (make_pass_cse_sincos): ...new function.
3025 (pass_optimize_bswap): Replace declaration with that of...
3026 (make_pass_optimize_bswap): ...new function.
3027 (pass_optimize_widening_mul): Replace declaration with that of...
3028 (make_pass_optimize_widening_mul): ...new function.
3029 (pass_warn_function_return): Replace declaration with that of...
3030 (make_pass_warn_function_return): ...new function.
3031 (pass_warn_function_noreturn): Replace declaration with that of...
3032 (make_pass_warn_function_noreturn): ...new function.
3033 (pass_cselim): Replace declaration with that of...
3034 (make_pass_cselim): ...new function.
3035 (pass_phiopt): Replace declaration with that of...
3036 (make_pass_phiopt): ...new function.
3037 (pass_forwprop): Replace declaration with that of...
3038 (make_pass_forwprop): ...new function.
3039 (pass_phiprop): Replace declaration with that of...
3040 (make_pass_phiprop): ...new function.
3041 (pass_tree_ifcombine): Replace declaration with that of...
3042 (make_pass_tree_ifcombine): ...new function.
3043 (pass_dse): Replace declaration with that of...
3044 (make_pass_dse): ...new function.
3045 (pass_nrv): Replace declaration with that of...
3046 (make_pass_nrv): ...new function.
3047 (pass_rename_ssa_copies): Replace declaration with that of...
3048 (make_pass_rename_ssa_copies): ...new function.
3049 (pass_sink_code): Replace declaration with that of...
3050 (make_pass_sink_code): ...new function.
3051 (pass_fre): Replace declaration with that of...
3052 (make_pass_fre): ...new function.
3053 (pass_check_data_deps): Replace declaration with that of...
3054 (make_pass_check_data_deps): ...new function.
3055 (pass_copy_prop): Replace declaration with that of...
3056 (make_pass_copy_prop): ...new function.
3057 (pass_vrp): Replace declaration with that of...
3058 (make_pass_vrp): ...new function.
3059 (pass_uncprop): Replace declaration with that of...
3060 (make_pass_uncprop): ...new function.
3061 (pass_return_slot): Replace declaration with that of...
3062 (make_pass_return_slot): ...new function.
3063 (pass_reassoc): Replace declaration with that of...
3064 (make_pass_reassoc): ...new function.
3065 (pass_rebuild_cgraph_edges): Replace declaration with that of...
3066 (make_pass_rebuild_cgraph_edges): ...new function.
3067 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
3068 (make_pass_remove_cgraph_callee_edges): ...new function.
3069 (pass_build_cgraph_edges): Replace declaration with that of...
3070 (make_pass_build_cgraph_edges): ...new function.
3071 (pass_local_pure_const): Replace declaration with that of...
3072 (make_pass_local_pure_const): ...new function.
3073 (pass_tracer): Replace declaration with that of...
3074 (make_pass_tracer): ...new function.
3075 (pass_warn_unused_result): Replace declaration with that of...
3076 (make_pass_warn_unused_result): ...new function.
3077 (pass_diagnose_tm_blocks): Replace declaration with that of...
3078 (make_pass_diagnose_tm_blocks): ...new function.
3079 (pass_lower_tm): Replace declaration with that of...
3080 (make_pass_lower_tm): ...new function.
3081 (pass_tm_init): Replace declaration with that of...
3082 (make_pass_tm_init): ...new function.
3083 (pass_tm_mark): Replace declaration with that of...
3084 (make_pass_tm_mark): ...new function.
3085 (pass_tm_memopt): Replace declaration with that of...
3086 (make_pass_tm_memopt): ...new function.
3087 (pass_tm_edges): Replace declaration with that of...
3088 (make_pass_tm_edges): ...new function.
3089 (pass_split_functions): Replace declaration with that of...
3090 (make_pass_split_functions): ...new function.
3091 (pass_feedback_split_functions): Replace declaration with that of...
3092 (make_pass_feedback_split_functions): ...new function.
3093 (pass_strength_reduction): Replace declaration with that of...
3094 (make_pass_strength_reduction): ...new function.
3095 (pass_ipa_lower_emutls): Replace declaration with that of...
3096 (make_pass_ipa_lower_emutls): ...new function.
3097 (pass_ipa_function_and_variable_visibility): Replace declaration with
3099 (make_pass_ipa_function_and_variable_visibility): ...new function.
3100 (pass_ipa_tree_profile): Replace declaration with that of...
3101 (make_pass_ipa_tree_profile): ...new function.
3102 (pass_early_local_passes): Replace declaration with that of...
3103 (make_pass_early_local_passes): ...new function.
3104 (pass_ipa_whole_program_visibility): Replace declaration with that
3106 (make_pass_ipa_whole_program_visibility): ...new function.
3107 (pass_ipa_lto_gimple_out): Replace declaration with that of...
3108 (make_pass_ipa_lto_gimple_out): ...new function.
3109 (pass_ipa_increase_alignment): Replace declaration with that of...
3110 (make_pass_ipa_increase_alignment): ...new function.
3111 (pass_ipa_inline): Replace declaration with that of...
3112 (make_pass_ipa_inline): ...new function.
3113 (pass_ipa_free_lang_data): Replace declaration with that of...
3114 (make_pass_ipa_free_lang_data): ...new function.
3115 (pass_ipa_free_inline_summary): Replace declaration with that of...
3116 (make_pass_ipa_free_inline_summary): ...new function.
3117 (pass_ipa_cp): Replace declaration with that of...
3118 (make_pass_ipa_cp): ...new function.
3119 (pass_ipa_reference): Replace declaration with that of...
3120 (make_pass_ipa_reference): ...new function.
3121 (pass_ipa_pure_const): Replace declaration with that of...
3122 (make_pass_ipa_pure_const): ...new function.
3123 (pass_ipa_pta): Replace declaration with that of...
3124 (make_pass_ipa_pta): ...new function.
3125 (pass_ipa_lto_finish_out): Replace declaration with that of...
3126 (make_pass_ipa_lto_finish_out): ...new function.
3127 (pass_ipa_tm): Replace declaration with that of...
3128 (make_pass_ipa_tm): ...new function.
3129 (pass_ipa_profile): Replace declaration with that of...
3130 (make_pass_ipa_profile): ...new function.
3131 (pass_ipa_cdtor_merge): Replace declaration with that of...
3132 (make_pass_ipa_cdtor_merge): ...new function.
3133 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
3135 (make_pass_cleanup_cfg_post_optimizing): ...new function.
3136 (pass_init_datastructures): Replace declaration with that of...
3137 (make_pass_init_datastructures): ...new function.
3138 (pass_fixup_cfg): Replace declaration with that of...
3139 (make_pass_fixup_cfg): ...new function.
3140 (pass_expand): Replace declaration with that of...
3141 (make_pass_expand): ...new function.
3142 (pass_instantiate_virtual_regs): Replace declaration with that of...
3143 (make_pass_instantiate_virtual_regs): ...new function.
3144 (pass_rtl_fwprop): Replace declaration with that of...
3145 (make_pass_rtl_fwprop): ...new function.
3146 (pass_rtl_fwprop_addr): Replace declaration with that of...
3147 (make_pass_rtl_fwprop_addr): ...new function.
3148 (pass_jump): Replace declaration with that of...
3149 (make_pass_jump): ...new function.
3150 (pass_jump2): Replace declaration with that of...
3151 (make_pass_jump2): ...new function.
3152 (pass_lower_subreg): Replace declaration with that of...
3153 (make_pass_lower_subreg): ...new function.
3154 (pass_cse): Replace declaration with that of...
3155 (make_pass_cse): ...new function.
3156 (pass_fast_rtl_dce): Replace declaration with that of...
3157 (make_pass_fast_rtl_dce): ...new function.
3158 (pass_ud_rtl_dce): Replace declaration with that of...
3159 (make_pass_ud_rtl_dce): ...new function.
3160 (pass_rtl_dce): Replace declaration with that of...
3161 (make_pass_rtl_dce): ...new function.
3162 (pass_rtl_dse1): Replace declaration with that of...
3163 (make_pass_rtl_dse1): ...new function.
3164 (pass_rtl_dse2): Replace declaration with that of...
3165 (make_pass_rtl_dse2): ...new function.
3166 (pass_rtl_dse3): Replace declaration with that of...
3167 (make_pass_rtl_dse3): ...new function.
3168 (pass_rtl_cprop): Replace declaration with that of...
3169 (make_pass_rtl_cprop): ...new function.
3170 (pass_rtl_pre): Replace declaration with that of...
3171 (make_pass_rtl_pre): ...new function.
3172 (pass_rtl_hoist): Replace declaration with that of...
3173 (make_pass_rtl_hoist): ...new function.
3174 (pass_rtl_store_motion): Replace declaration with that of...
3175 (make_pass_rtl_store_motion): ...new function.
3176 (pass_cse_after_global_opts): Replace declaration with that of...
3177 (make_pass_cse_after_global_opts): ...new function.
3178 (pass_rtl_ifcvt): Replace declaration with that of...
3179 (make_pass_rtl_ifcvt): ...new function.
3180 (pass_into_cfg_layout_mode): Replace declaration with that of...
3181 (make_pass_into_cfg_layout_mode): ...new function.
3182 (pass_outof_cfg_layout_mode): Replace declaration with that of...
3183 (make_pass_outof_cfg_layout_mode): ...new function.
3184 (pass_loop2): Replace declaration with that of...
3185 (make_pass_loop2): ...new function.
3186 (pass_rtl_loop_init): Replace declaration with that of...
3187 (make_pass_rtl_loop_init): ...new function.
3188 (pass_rtl_move_loop_invariants): Replace declaration with that of...
3189 (make_pass_rtl_move_loop_invariants): ...new function.
3190 (pass_rtl_unswitch): Replace declaration with that of...
3191 (make_pass_rtl_unswitch): ...new function.
3192 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
3193 (make_pass_rtl_unroll_and_peel_loops): ...new function.
3194 (pass_rtl_doloop): Replace declaration with that of...
3195 (make_pass_rtl_doloop): ...new function.
3196 (pass_rtl_loop_done): Replace declaration with that of...
3197 (make_pass_rtl_loop_done): ...new function.
3198 (pass_web): Replace declaration with that of...
3199 (make_pass_web): ...new function.
3200 (pass_cse2): Replace declaration with that of...
3201 (make_pass_cse2): ...new function.
3202 (pass_df_initialize_opt): Replace declaration with that of...
3203 (make_pass_df_initialize_opt): ...new function.
3204 (pass_df_initialize_no_opt): Replace declaration with that of...
3205 (make_pass_df_initialize_no_opt): ...new function.
3206 (pass_reginfo_init): Replace declaration with that of...
3207 (make_pass_reginfo_init): ...new function.
3208 (pass_inc_dec): Replace declaration with that of...
3209 (make_pass_inc_dec): ...new function.
3210 (pass_stack_ptr_mod): Replace declaration with that of...
3211 (make_pass_stack_ptr_mod): ...new function.
3212 (pass_initialize_regs): Replace declaration with that of...
3213 (make_pass_initialize_regs): ...new function.
3214 (pass_combine): Replace declaration with that of...
3215 (make_pass_combine): ...new function.
3216 (pass_if_after_combine): Replace declaration with that of...
3217 (make_pass_if_after_combine): ...new function.
3218 (pass_ree): Replace declaration with that of...
3219 (make_pass_ree): ...new function.
3220 (pass_partition_blocks): Replace declaration with that of...
3221 (make_pass_partition_blocks): ...new function.
3222 (pass_match_asm_constraints): Replace declaration with that of...
3223 (make_pass_match_asm_constraints): ...new function.
3224 (pass_regmove): Replace declaration with that of...
3225 (make_pass_regmove): ...new function.
3226 (pass_split_all_insns): Replace declaration with that of...
3227 (make_pass_split_all_insns): ...new function.
3228 (pass_fast_rtl_byte_dce): Replace declaration with that of...
3229 (make_pass_fast_rtl_byte_dce): ...new function.
3230 (pass_lower_subreg2): Replace declaration with that of...
3231 (make_pass_lower_subreg2): ...new function.
3232 (pass_mode_switching): Replace declaration with that of...
3233 (make_pass_mode_switching): ...new function.
3234 (pass_sms): Replace declaration with that of...
3235 (make_pass_sms): ...new function.
3236 (pass_sched): Replace declaration with that of...
3237 (make_pass_sched): ...new function.
3238 (pass_ira): Replace declaration with that of...
3239 (make_pass_ira): ...new function.
3240 (pass_reload): Replace declaration with that of...
3241 (make_pass_reload): ...new function.
3242 (pass_clean_state): Replace declaration with that of...
3243 (make_pass_clean_state): ...new function.
3244 (pass_branch_prob): Replace declaration with that of...
3245 (make_pass_branch_prob): ...new function.
3246 (pass_value_profile_transformations): Replace declaration with that
3248 (make_pass_value_profile_transformations): ...new function.
3249 (pass_postreload_cse): Replace declaration with that of...
3250 (make_pass_postreload_cse): ...new function.
3251 (pass_gcse2): Replace declaration with that of...
3252 (make_pass_gcse2): ...new function.
3253 (pass_split_after_reload): Replace declaration with that of...
3254 (make_pass_split_after_reload): ...new function.
3255 (pass_branch_target_load_optimize1): Replace declaration with that
3257 (make_pass_branch_target_load_optimize1): ...new function.
3258 (pass_thread_prologue_and_epilogue): Replace declaration with that
3260 (make_pass_thread_prologue_and_epilogue): ...new function.
3261 (pass_stack_adjustments): Replace declaration with that of...
3262 (make_pass_stack_adjustments): ...new function.
3263 (pass_peephole2): Replace declaration with that of...
3264 (make_pass_peephole2): ...new function.
3265 (pass_if_after_reload): Replace declaration with that of...
3266 (make_pass_if_after_reload): ...new function.
3267 (pass_regrename): Replace declaration with that of...
3268 (make_pass_regrename): ...new function.
3269 (pass_cprop_hardreg): Replace declaration with that of...
3270 (make_pass_cprop_hardreg): ...new function.
3271 (pass_reorder_blocks): Replace declaration with that of...
3272 (make_pass_reorder_blocks): ...new function.
3273 (pass_branch_target_load_optimize2): Replace declaration with that
3275 (make_pass_branch_target_load_optimize2): ...new function.
3276 (pass_leaf_regs): Replace declaration with that of...
3277 (make_pass_leaf_regs): ...new function.
3278 (pass_split_before_sched2): Replace declaration with that of...
3279 (make_pass_split_before_sched2): ...new function.
3280 (pass_compare_elim_after_reload): Replace declaration with that of...
3281 (make_pass_compare_elim_after_reload): ...new function.
3282 (pass_sched2): Replace declaration with that of...
3283 (make_pass_sched2): ...new function.
3284 (pass_stack_regs): Replace declaration with that of...
3285 (make_pass_stack_regs): ...new function.
3286 (pass_stack_regs_run): Replace declaration with that of...
3287 (make_pass_stack_regs_run): ...new function.
3288 (pass_df_finish): Replace declaration with that of...
3289 (make_pass_df_finish): ...new function.
3290 (pass_compute_alignments): Replace declaration with that of...
3291 (make_pass_compute_alignments): ...new function.
3292 (pass_duplicate_computed_gotos): Replace declaration with that of...
3293 (make_pass_duplicate_computed_gotos): ...new function.
3294 (pass_variable_tracking): Replace declaration with that of...
3295 (make_pass_variable_tracking): ...new function.
3296 (pass_free_cfg): Replace declaration with that of...
3297 (make_pass_free_cfg): ...new function.
3298 (pass_machine_reorg): Replace declaration with that of...
3299 (make_pass_machine_reorg): ...new function.
3300 (pass_cleanup_barriers): Replace declaration with that of...
3301 (make_pass_cleanup_barriers): ...new function.
3302 (pass_delay_slots): Replace declaration with that of...
3303 (make_pass_delay_slots): ...new function.
3304 (pass_split_for_shorten_branches): Replace declaration with that of...
3305 (make_pass_split_for_shorten_branches): ...new function.
3306 (pass_split_before_regstack): Replace declaration with that of...
3307 (make_pass_split_before_regstack): ...new function.
3308 (pass_convert_to_eh_region_ranges): Replace declaration with that
3310 (make_pass_convert_to_eh_region_ranges): ...new function.
3311 (pass_shorten_branches): Replace declaration with that of...
3312 (make_pass_shorten_branches): ...new function.
3313 (pass_set_nothrow_function_flags): Replace declaration with that of...
3314 (make_pass_set_nothrow_function_flags): ...new function.
3315 (pass_dwarf2_frame): Replace declaration with that of...
3316 (make_pass_dwarf2_frame): ...new function.
3317 (pass_final): Replace declaration with that of...
3318 (make_pass_final): ...new function.
3319 (pass_rtl_seqabstr): Replace declaration with that of...
3320 (make_pass_rtl_seqabstr): ...new function.
3321 (pass_release_ssa_names): Replace declaration with that of...
3322 (make_pass_release_ssa_names): ...new function.
3323 (pass_early_inline): Replace declaration with that of...
3324 (make_pass_early_inline): ...new function.
3325 (pass_inline_parameters): Replace declaration with that of...
3326 (make_pass_inline_parameters): ...new function.
3327 (pass_update_address_taken): Replace declaration with that of...
3328 (make_pass_update_address_taken): ...new function.
3329 (pass_convert_switch): Replace declaration with that of...
3330 (make_pass_convert_switch): ...new function.
3331 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
3332 to a subclass of simple_ipa_opt_pass along with...
3333 (pass_data_ipa_tree_profile): ...new pass_data instance and...
3334 (make_pass_ipa_tree_profile): ...new function.
3335 * tree-sra.c (pass_sra_early): Convert from a global struct to a
3336 subclass of gimple_opt_pass along with...
3337 (pass_data_sra_early): ...new pass_data instance and...
3338 (make_pass_sra_early): ...new function.
3339 (pass_sra): Convert from a global struct to a subclass of
3340 gimple_opt_pass along with...
3341 (pass_data_sra): ...new pass_data instance and...
3342 (make_pass_sra): ...new function.
3343 (pass_early_ipa_sra): Convert from a global struct to a subclass of
3344 gimple_opt_pass along with...
3345 (pass_data_early_ipa_sra): ...new pass_data instance and...
3346 (make_pass_early_ipa_sra): ...new function.
3347 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
3348 subclass of gimple_opt_pass along with...
3349 (pass_data_ccp): ...new pass_data instance and...
3350 (make_pass_ccp): ...new function.
3351 (pass_fold_builtins): Convert from a global struct to a subclass of
3352 gimple_opt_pass along with...
3353 (pass_data_fold_builtins): ...new pass_data instance and...
3354 (make_pass_fold_builtins): ...new function.
3355 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
3356 subclass of gimple_opt_pass along with...
3357 (pass_data_copy_prop): ...new pass_data instance and...
3358 (make_pass_copy_prop): ...new function.
3359 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
3360 global struct to a subclass of gimple_opt_pass along with...
3361 (pass_data_rename_ssa_copies): ...new pass_data instance and...
3362 (make_pass_rename_ssa_copies): ...new function.
3363 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
3364 subclass of gimple_opt_pass along with...
3365 (pass_data_dce): ...new pass_data instance and...
3366 (make_pass_dce): ...new function.
3367 (pass_dce_loop): Convert from a global struct to a subclass of
3368 gimple_opt_pass along with...
3369 (pass_data_dce_loop): ...new pass_data instance and...
3370 (make_pass_dce_loop): ...new function.
3371 (pass_cd_dce): Convert from a global struct to a subclass of
3372 gimple_opt_pass along with...
3373 (pass_data_cd_dce): ...new pass_data instance and...
3374 (make_pass_cd_dce): ...new function.
3375 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
3376 subclass of gimple_opt_pass along with...
3377 (pass_data_dominator): ...new pass_data instance and...
3378 (make_pass_dominator): ...new function.
3379 (pass_phi_only_cprop): Convert from a global struct to a subclass of
3380 gimple_opt_pass along with...
3381 (pass_data_phi_only_cprop): ...new pass_data instance and...
3382 (make_pass_phi_only_cprop): ...new function.
3383 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
3384 subclass of gimple_opt_pass along with...
3385 (pass_data_dse): ...new pass_data instance and...
3386 (make_pass_dse): ...new function.
3387 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
3388 a subclass of gimple_opt_pass along with...
3389 (pass_data_forwprop): ...new pass_data instance and...
3390 (make_pass_forwprop): ...new function.
3391 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
3392 struct to a subclass of gimple_opt_pass along with...
3393 (pass_data_tree_ifcombine): ...new pass_data instance and...
3394 (make_pass_tree_ifcombine): ...new function.
3395 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
3396 subclass of gimple_opt_pass along with...
3397 (pass_data_ch): ...new pass_data instance and...
3398 (make_pass_ch): ...new function.
3399 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
3400 subclass of gimple_opt_pass along with...
3401 (pass_data_tree_loop): ...new pass_data instance and...
3402 (make_pass_tree_loop): ...new function.
3403 (pass_tree_loop_init): Convert from a global struct to a subclass of
3404 gimple_opt_pass along with...
3405 (pass_data_tree_loop_init): ...new pass_data instance and...
3406 (make_pass_tree_loop_init): ...new function.
3407 (pass_lim): Convert from a global struct to a subclass of
3408 gimple_opt_pass along with...
3409 (pass_data_lim): ...new pass_data instance and...
3410 (make_pass_lim): ...new function.
3411 (pass_tree_unswitch): Convert from a global struct to a subclass of
3412 gimple_opt_pass along with...
3413 (pass_data_tree_unswitch): ...new pass_data instance and...
3414 (make_pass_tree_unswitch): ...new function.
3415 (pass_predcom): Convert from a global struct to a subclass of
3416 gimple_opt_pass along with...
3417 (pass_data_predcom): ...new pass_data instance and...
3418 (make_pass_predcom): ...new function.
3419 (pass_vectorize): Convert from a global struct to a subclass of
3420 gimple_opt_pass along with...
3421 (pass_data_vectorize): ...new pass_data instance and...
3422 (make_pass_vectorize): ...new function.
3423 (pass_graphite): Convert from a global struct to a subclass of
3424 gimple_opt_pass along with...
3425 (pass_data_graphite): ...new pass_data instance and...
3426 (make_pass_graphite): ...new function.
3427 (pass_graphite_transforms): Convert from a global struct to a subclass
3428 of gimple_opt_pass along with...
3429 (pass_data_graphite_transforms): ...new pass_data instance and...
3430 (make_pass_graphite_transforms): ...new function.
3431 (pass_check_data_deps): Convert from a global struct to a subclass of
3432 gimple_opt_pass along with...
3433 (pass_data_check_data_deps): ...new pass_data instance and...
3434 (make_pass_check_data_deps): ...new function.
3435 (pass_iv_canon): Convert from a global struct to a subclass of
3436 gimple_opt_pass along with...
3437 (pass_data_iv_canon): ...new pass_data instance and...
3438 (make_pass_iv_canon): ...new function.
3439 (pass_scev_cprop): Convert from a global struct to a subclass of
3440 gimple_opt_pass along with...
3441 (pass_data_scev_cprop): ...new pass_data instance and...
3442 (make_pass_scev_cprop): ...new function.
3443 (pass_record_bounds): Convert from a global struct to a subclass of
3444 gimple_opt_pass along with...
3445 (pass_data_record_bounds): ...new pass_data instance and...
3446 (make_pass_record_bounds): ...new function.
3447 (pass_complete_unroll): Convert from a global struct to a subclass of
3448 gimple_opt_pass along with...
3449 (pass_data_complete_unroll): ...new pass_data instance and...
3450 (make_pass_complete_unroll): ...new function.
3451 (pass_complete_unrolli): Convert from a global struct to a subclass of
3452 gimple_opt_pass along with...
3453 (pass_data_complete_unrolli): ...new pass_data instance and...
3454 (make_pass_complete_unrolli): ...new function.
3455 (pass_parallelize_loops): Convert from a global struct to a subclass
3456 of gimple_opt_pass along with...
3457 (pass_data_parallelize_loops): ...new pass_data instance and...
3458 (make_pass_parallelize_loops): ...new function.
3459 (pass_loop_prefetch): Convert from a global struct to a subclass of
3460 gimple_opt_pass along with...
3461 (pass_data_loop_prefetch): ...new pass_data instance and...
3462 (make_pass_loop_prefetch): ...new function.
3463 (pass_iv_optimize): Convert from a global struct to a subclass of
3464 gimple_opt_pass along with...
3465 (pass_data_iv_optimize): ...new pass_data instance and...
3466 (make_pass_iv_optimize): ...new function.
3467 (pass_tree_loop_done): Convert from a global struct to a subclass of
3468 gimple_opt_pass along with...
3469 (pass_data_tree_loop_done): ...new pass_data instance and...
3470 (make_pass_tree_loop_done): ...new function.
3471 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
3472 struct to a subclass of gimple_opt_pass along with...
3473 (pass_data_cse_reciprocals): ...new pass_data instance and...
3474 (make_pass_cse_reciprocals): ...new function.
3475 (pass_cse_sincos): Convert from a global struct to a subclass of
3476 gimple_opt_pass along with...
3477 (pass_data_cse_sincos): ...new pass_data instance and...
3478 (make_pass_cse_sincos): ...new function.
3479 (pass_optimize_bswap): Convert from a global struct to a subclass of
3480 gimple_opt_pass along with...
3481 (pass_data_optimize_bswap): ...new pass_data instance and...
3482 (make_pass_optimize_bswap): ...new function.
3483 (pass_optimize_widening_mul): Convert from a global struct to a
3484 subclass of gimple_opt_pass along with...
3485 (pass_data_optimize_widening_mul): ...new pass_data instance and...
3486 (make_pass_optimize_widening_mul): ...new function.
3487 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
3488 subclass of gimple_opt_pass along with...
3489 (pass_data_phiopt): ...new pass_data instance and...
3490 (make_pass_phiopt): ...new function.
3491 (pass_cselim): Convert from a global struct to a subclass of
3492 gimple_opt_pass along with...
3493 (pass_data_cselim): ...new pass_data instance and...
3494 (make_pass_cselim): ...new function.
3495 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
3496 subclass of gimple_opt_pass along with...
3497 (pass_data_phiprop): ...new pass_data instance and...
3498 (make_pass_phiprop): ...new function.
3499 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
3500 subclass of gimple_opt_pass along with...
3501 (pass_data_pre): ...new pass_data instance and...
3502 (make_pass_pre): ...new function.
3503 (pass_fre): Convert from a global struct to a subclass of
3504 gimple_opt_pass along with...
3505 (pass_data_fre): ...new pass_data instance and...
3506 (make_pass_fre): ...new function.
3507 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
3508 subclass of gimple_opt_pass along with...
3509 (pass_data_reassoc): ...new pass_data instance and...
3510 (make_pass_reassoc): ...new function.
3511 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
3512 subclass of gimple_opt_pass along with...
3513 (pass_data_sink_code): ...new pass_data instance and...
3514 (make_pass_sink_code): ...new function.
3515 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
3516 subclass of gimple_opt_pass along with...
3517 (pass_data_strlen): ...new pass_data instance and...
3518 (make_pass_strlen): ...new function.
3519 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
3520 struct to a subclass of gimple_opt_pass along with...
3521 (pass_data_build_alias): ...new pass_data instance and...
3522 (make_pass_build_alias): ...new function.
3523 (pass_build_ealias): Convert from a global struct to a subclass of
3524 gimple_opt_pass along with...
3525 (pass_data_build_ealias): ...new pass_data instance and...
3526 (make_pass_build_ealias): ...new function.
3527 (pass_ipa_pta): Convert from a global struct to a subclass of
3528 simple_ipa_opt_pass along with...
3529 (pass_data_ipa_pta): ...new pass_data instance and...
3530 (make_pass_ipa_pta): ...new function.
3531 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
3532 subclass of gimple_opt_pass along with...
3533 (pass_data_uncprop): ...new pass_data instance and...
3534 (make_pass_uncprop): ...new function.
3535 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
3536 global struct to a subclass of gimple_opt_pass along with...
3537 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
3538 (make_pass_late_warn_uninitialized): ...new function.
3539 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
3540 to a subclass of gimple_opt_pass along with...
3541 (pass_data_init_datastructures): ...new pass_data instance and...
3542 (make_pass_init_datastructures): ...new function.
3543 (pass_early_warn_uninitialized): Convert from a global struct to a
3544 subclass of gimple_opt_pass along with...
3545 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
3546 (make_pass_early_warn_uninitialized): ...new function.
3547 (pass_update_address_taken): Convert from a global struct to a
3548 subclass of gimple_opt_pass along with...
3549 (pass_data_update_address_taken): ...new pass_data instance and...
3550 (make_pass_update_address_taken): ...new function.
3551 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
3552 struct to a subclass of gimple_opt_pass along with...
3553 (pass_data_release_ssa_names): ...new pass_data instance and...
3554 (make_pass_release_ssa_names): ...new function.
3555 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
3556 subclass of gimple_opt_pass along with...
3557 (pass_data_stdarg): ...new pass_data instance and...
3558 (make_pass_stdarg): ...new function.
3559 * tree-switch-conversion.c (pass_convert_switch): Convert from a
3560 global struct to a subclass of gimple_opt_pass along with...
3561 (pass_data_convert_switch): ...new pass_data instance and...
3562 (make_pass_convert_switch): ...new function.
3563 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
3564 to a subclass of gimple_opt_pass along with...
3565 (pass_data_tail_recursion): ...new pass_data instance and...
3566 (make_pass_tail_recursion): ...new function.
3567 (pass_tail_calls): Convert from a global struct to a subclass of
3568 gimple_opt_pass along with...
3569 (pass_data_tail_calls): ...new pass_data instance and...
3570 (make_pass_tail_calls): ...new function.
3571 * tree-vect-generic.c (pass_lower_vector): Convert from a global
3572 struct to a subclass of gimple_opt_pass along with...
3573 (pass_data_lower_vector): ...new pass_data instance and...
3574 (make_pass_lower_vector): ...new function.
3575 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
3576 gimple_opt_pass along with...
3577 (pass_data_lower_vector_ssa): ...new pass_data instance and...
3578 (make_pass_lower_vector_ssa): ...new function.
3579 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
3580 to a subclass of gimple_opt_pass along with...
3581 (pass_data_slp_vectorize): ...new pass_data instance and...
3582 (make_pass_slp_vectorize): ...new function.
3583 (pass_ipa_increase_alignment): Convert from a global struct to a
3584 subclass of simple_ipa_opt_pass along with...
3585 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
3586 (make_pass_ipa_increase_alignment): ...new function.
3587 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
3588 gimple_opt_pass along with...
3589 (pass_data_vrp): ...new pass_data instance and...
3590 (make_pass_vrp): ...new function.
3591 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
3592 subclass of simple_ipa_opt_pass along with...
3593 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
3594 (make_pass_ipa_free_lang_data): ...new function.
3595 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
3596 gimple_opt_pass along with...
3597 (pass_data_tsan): ...new pass_data instance and...
3598 (make_pass_tsan): ...new function.
3599 (pass_tsan_O0): Convert from a global struct to a subclass of
3600 gimple_opt_pass along with...
3601 (pass_data_tsan_O0): ...new pass_data instance and...
3602 (make_pass_tsan_O0): ...new function.
3603 * var-tracking.c (pass_variable_tracking): Convert from a global
3604 struct to a subclass of rtl_opt_pass along with...
3605 (pass_data_variable_tracking): ...new pass_data instance and...
3606 (make_pass_variable_tracking): ...new function.
3607 * web.c (pass_web): Convert from a global struct to a subclass of
3608 rtl_opt_pass along with...
3609 (pass_data_web): ...new pass_data instance and...
3610 (make_pass_web): ...new function.
3611 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3612 declaration with that of...
3613 (make_pass_mode_switch_use): ...new function.
3614 (pass_resolve_sw_modes): Replace declaration with that of...
3615 (make_pass_resolve_sw_modes): ...new function.
3616 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3617 from a global struct to a subclass of rtl_opt_pass along with...
3618 (pass_data_mode_switch_use): ...new pass_data instance and...
3619 (make_pass_mode_switch_use): ...new function.
3620 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3621 from a global struct to a subclass of rtl_opt_pass along with...
3622 (pass_data_resolve_sw_modes): ...new pass_data instance and...
3623 (make_pass_resolve_sw_modes): ...new function.
3624 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3625 struct to a subclass of rtl_opt_pass along with...
3626 (pass_data_insert_vzeroupper): ...new pass_data instance and...
3627 (make_pass_insert_vzeroupper): ...new function.
3628 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3629 global struct to a subclass of rtl_opt_pass along with...
3630 (pass_data_work_around_errata): ...new pass_data instance and...
3631 (make_pass_work_around_errata): ...new function.
3632 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3633 struct to a subclass of rtl_opt_pass along with...
3634 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3635 (make_pass_mips_machine_reorg2): ...new function.
3637 2013-08-05 David Malcolm <dmalcolm@redhat.com>
3639 * passes.c (pass_manager::operator new): New.
3641 2013-08-05 David Malcolm <dmalcolm@redhat.com>
3643 Handwritten part of conversion of passes to C++ classes.
3645 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3646 (toplev.o): Add dep on PASS_MANAGER_H.
3647 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3648 of early local pases to reflect this moving from a global to a
3649 member of gcc::pass_manager.
3650 (cgraph_add_new_function): Likewise.
3651 * lto-cgraph.c (lto_output_node): Update for conversion of
3652 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3653 * passes.c (opt_pass::clone): New.
3654 (opt_pass::gate): New.
3655 (opt_pass::execute): New.
3656 (opt_pass::opt_pass): New.
3657 (pass_manager::execute_early_local_passes): New.
3658 (pass_manager::execute_pass_mode_switching): new.
3659 (finish_optimization_passes): Convert to...
3660 (pass_manager::finish_optimization_passes): ...this.
3661 (finish_optimization_passes): Update for conversion of passes to
3663 (register_dump_files_1): Use has_gate since we cannot portably
3664 check a vtable entry against NULL.
3665 (dump_one_pass): Likewise.
3666 (ipa_write_summaries_2): Likewise.
3667 (ipa_write_optimization_summaries_1): Likewise.
3668 (ipa_read_summaries_1): Likewise.
3669 (ipa_read_optimization_summaries_1): Likewise.
3670 (execute_ipa_stmt_fixups): Likewise.
3671 (pass_manager::pass_manager): Rewrite pass-creation, invoking
3672 pass-creation functions rather than wiring up globals, and
3673 storing the results in fields of pass_manager generated using
3675 (pass_manager::dump_profile_report): Update for conversion of
3676 passes to C++ classes.
3677 (pass_manager::execute_ipa_summary_passes): Likewise.
3678 (execute_one_ipa_transform_pass): Likewise.
3679 (execute_one_pass): Use has_gate and has_execute since we cannot
3680 portably check a vtable entry against NULL.
3681 * pass_manager.h (pass_manager::finish_optimization_passes): New.
3682 (pass_manager): Use pass-instances.def to add fields for the
3683 various pass instances.
3684 * toplev.c (finalize): Update for move of
3685 finish_optimization_passes to a method of gcc::pass_manager.
3686 * toplev.h (finish_optimization_passes): Move to method of class
3688 * tree-pass.h (struct pass_data): New.
3689 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3690 (opt_pass::gate): Convert to virtual function.
3691 (opt_pass::~opt_pass): New.
3692 (opt_pass::clone): New.
3693 (opt_pass::execute): Convert to virtual function.
3694 (opt_pass::opt_pass): New.
3695 (opt_pass::ctxt_): new.
3696 (gimple_opt_pass): Convert to subclass of opt_pass.
3697 (gimple_opt_pass::gimple_opt_pass): New.
3698 (rtl_opt_pass): Convert to subclass of opt_pass.
3699 (rtl_opt_pass::rtl_opt_pass): New.
3700 (ipa_opt_pass_d): Convert to subclass of opt_pass.
3701 (ipa_opt_pass_d::ipa_opt_pass_d): New.
3702 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3703 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3704 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3705 invocation of pass_mode_switching to reflect this moving from a
3706 global to a member of gcc::pass_manager.
3707 (ix86_option_override): Rework how pass_insert_vzeroupper is
3708 added to the pass_manager to reflect autogenerated changes.
3709 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3711 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
3713 PR rtl-optimization/57708
3714 * recog.c (peep2_find_free_register): Validate all regs in a
3717 2013-08-05 Jan Hubicka <jh@suse.cz>
3720 * cgraph.c (verify_cgraph_node): Accept local flags from other
3722 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3723 (function_and_variable_visibility): Likewise.
3724 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3726 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3728 * graph.c (init_graph_slim_pretty_print): Remove.
3729 (print_graph_cfg): Do not call it. Use local pretty printer.
3730 (start_graph_dump): Likewise.
3732 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3734 * gimple-pretty-print.c (buffer): Remove.
3735 (initialized): Likewise.
3736 (maybe_init_pretty_print): Likewise.
3737 (print_gimple_stmt): Do not call it. Use non-static local
3738 pretty_printer variable.
3739 (print_gimple_expr): Likewise.
3740 (print_gimple_seq): Likewise.
3741 (gimple_dump_bb): Likewise.
3743 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
3745 * asan.c (asan_pp): Remove.
3746 (asan_pp_initialized): Likewise.
3747 (asan_pp_initialize): Likewise.
3748 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
3749 (asan_emit_stack_protection): Tidy. Use local pretty printer.
3750 (asan_add_global): Likewise.
3752 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
3754 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
3755 * diagnostic.h (diagnostic_flush_buffer): Adjust.
3756 * pretty-print.c (pp_formatted_text_data): Likewise.
3757 (pp_indent): Rename from pp_base_indent.
3758 (pp_format): Rename from pp_base_format.
3759 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3760 (pp_format_verbatim): Rename from pp_base_format_verbatim.
3761 (pp_flush): Rename from pp_base_flush.
3762 (pp_set_line_maximum_length): Rename from
3763 pp_base_set_line_maximum_length.
3764 (pp_clear_output_area): Rename from pp_base_clear_output_area.
3765 (pp_set_prefix): Rename from pp_base_set_prefix.
3766 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3767 (pp_emit_prefix): Rename from pp_base_emit_prefix.
3768 (pp_append_text): Rename from pp_base_append_text.
3769 (pp_formatted_text): Rename from pp_base_formatted_text.
3770 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3771 (pp_remaining_character_count_for_line): Rename from
3772 pp_base_remaining_character_count_for_line.
3773 (pp_newline): Rename from pp_base_newline.
3774 (pp_character): Rename from pp_base_character.
3775 (pp_string): Rename from pp_base_string.
3776 (pp_maybe_space): Rename from pp_base_maybe_space.
3777 * asan.c (asan_pp_string): Adjust.
3778 (asan_emit_stack_protection): Likewise.
3779 (asan_add_global): Likewise.
3780 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3781 * tree-mudflap.c (mf_varname_tree): Likewise.
3782 * tree-pretty-print.c (pp_tree_identifier): Rename from
3783 pp_base_tree_identifier.
3784 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3785 Declare as function.
3787 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3789 * pretty-print.h (pp_bar_bar): New.
3790 (pp_ampersand_ampersand): Likewise.
3791 (pp_less_equal): Likewise.
3792 (pp_greater_equal): Likewise.
3793 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3794 printer functions instead of pp_string or operators and punctuators.
3795 (dump_gimple_call): Likewise.
3796 (dump_gimple_omp_for): Likewise.
3797 (dump_gimple_transaction): Likewise.
3798 (dump_gimple_phi): Likewise.
3799 (pp_gimple_stmt_1): Likewise.
3800 * sched-vis.c (print_insn): Likewise.
3801 * tree-mudflap.c (mf_varname_tree): Likewise.
3802 * tree-pretty-print.c (dump_block_node): Likewise.
3803 (dump_generic_node): Likewise.
3805 2013-08-02 Jan Hubicka <jh@suse.cz>
3807 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3809 * lto-streamer-out.c (tree_is_indexable): Results decls and
3810 parm decls are not indexable.
3811 (DFS_write_tree_body): Do not follow args and results.
3812 (hash_tree): Likewise.
3813 (output_functions): Rearrange so struct function is needed
3814 only when real body is output; be able to also ouptut abstract
3815 functions; output DECL_ARGUMENTS and DECL_RESULT.
3816 (lto_output): When not in WPA, ale store abstract functions.
3817 (write_symbol): Do not care about RESULT_DECL.
3818 (output_symbol_p): Handle correctly sbtract decls.
3819 * lto-streamer-in.c (input_function): Rearrange so struct
3820 function can be NULL at entry; allow streaming of
3821 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3822 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3823 sanity check during LTO.
3824 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3825 RESULT_DECl and DECL_ARGUMENTS.
3826 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3829 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3831 * pretty-print.h (pp_underscore): New.
3833 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3834 printer functions instead of pp_character.
3835 (dump_binary_rhs): Likewise.
3836 (dump_ternary_rhs): Likewise.
3837 (dump_gimple_call_args): Likewise.
3838 (pp_points_to_solution): Likewise.
3839 (dump_gimple_call): Likewise.
3840 (dump_gimple_switch): Likewise.
3841 (dump_gimple_cond): Likewise.
3842 (dump_gimple_bind): Likewise.
3843 (dump_gimple_try): Likewise.
3844 (dump_gimple_omp_for): Likewise.
3845 (dump_gimple_omp_continue): Likewise.
3846 (dump_gimple_omp_single): Likewise.
3847 (dump_gimple_omp_sections): Likewise.
3848 (dump_gimple_omp_block): Likewise.
3849 (dump_gimple_omp_critical): Likewise.
3850 (dump_gimple_transaction): Likewise.
3851 (dump_gimple_asm): Likewise.
3852 (dump_gimple_phi): Likewise.
3853 (dump_gimple_omp_parallel): Likewise.
3854 (dump_gimple_omp_task): Likewise.
3855 (dump_gimple_omp_atomic_load): Likewise.
3856 (dump_gimple_omp_atomic_store): Likewise.
3857 (dump_gimple_mem_ops): Likewise.
3858 (pp_gimple_stmt_1): Likewise.
3859 (pp_cfg_jump): Likewise.
3860 (dump_implicit_edges): Likewise.
3861 (gimple_dump_bb_for_graph): Likewise.
3862 * graph.c (draw_cfg_node): Likewise.
3863 * langhooks.c (lhd_print_error_function): Likewise.
3864 * sched-vis.c (print_exp): Likewise.
3865 (print_value): Likewise.
3866 (print_pattern): Likewise.
3867 (print_insn): Likewise.
3868 (rtl_dump_bb_for_graph): Likewise.
3869 * tree-pretty-print.c (dump_function_declaration): Likewise.
3870 (dump_array_domain): Likewise.
3871 (dump_omp_clause): Likewise.
3872 (dump_location): Likewise.
3873 (dump_generic_node): Likewise.
3874 (print_struct_decl): Likewise.
3875 * diagnostic.c (diagnostic_show_locus): Use pp_space.
3877 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
3879 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3880 candidate table when replacing a candidate statement.
3881 (replace_rhs_if_not_dup): Likewise.
3882 (replace_one_candidate): Likewise.
3884 2013-08-02 Jan Hubicka <jh@suse.cz>
3885 Martin Liska <marxin.liska@gmail.com>
3887 * cgraphunit.c (add_new_function): Fix logic when adding from
3889 (assemble_thunk): Rename to ...
3890 (expand_thunk); .. this one; export; get it working with
3891 general functions; make produced gimple valid.
3892 * cgraph.h (expand_thunk): Declare.
3894 2013-08-02 Jan Hubicka <jh@suse.cz>
3896 * ipa-cp.c (gather_context_independent_values): Use
3897 ipa_get_param_move_cost.
3898 (get_replacement_map): Remove PARAM; move parameter folding
3900 (create_specialized_node): Update.
3901 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3902 assert that we have gimple body; update move_cost.
3903 (count_formal_params): Assert that we have gimple body.
3904 (ipa_dump_param): New function.
3905 (ipa_alloc_node_params): Break out from ...
3906 (ipa_initialize_node_params): ... here.
3907 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3908 (ipa_write_node_info): Stream move costs.
3909 (ipa_read_node_info): Read move costs.
3910 (ipa_update_after_lto_read): Do not recompute node params.
3911 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3912 (ipa_get_param): Check we are not in WPA.
3913 (ipa_get_param_move_cost): New.
3914 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3915 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3916 parm numbers to be present.
3918 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
3920 PR rtl-optimization/58048
3921 * lra-constraints.c (process_alt_operands): Don't check asm
3922 operand on register.
3924 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
3926 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3927 the implied StoreLoad barrier for atomic operations if before.
3929 2013-08-02 Jan Hubicka <jh@suse.cz>
3930 Martin Liska <marxin.liska@gmail.com>
3932 * cgraph.c (cgraph_function_body_availability): Do not check
3934 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3935 symtab_node_availability): Declare.
3936 * ipa.c (can_replace_by_local_alias): New.
3937 (function_and_variable_visibility): Use it.
3938 * symtab.c (symtab_for_node_and_aliases,
3939 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3941 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
3943 PR rtl-optimization/57963
3944 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3945 (lra_constraints): Use them.
3947 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
3949 * config/arm/types.md (define_attr "type"): Add "load_acq"
3951 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3953 (cortex_a53_store1): Likewise.
3955 2013-08-01 Jan Hubicka <jh@suse.cz>
3957 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3958 partitions are not needed.
3960 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
3962 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3963 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3964 MAYBE_NON_Q_CLASS_P where appropriate.
3966 2013-08-01 Jan Hubicka <jh@suse.cz>
3968 * cgraph.h (release_function_body): Declare.
3969 * tree.c (free_lang_data_in_decl): Free, parameters and return values
3970 of unused delcarations.
3972 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3974 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3975 RTL form when subtracting a constant.
3977 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3979 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3980 Generate canonical plus rtx with negated immediate instead of minus
3982 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3984 2013-08-01 Jan Hubicka <jh@suse.cz>
3986 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3987 (cgraph_release_function_body): Likewise.
3988 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3989 * cgraph.h (cgrpah_node): Rename abstract_and_needed
3990 to used_as_abstract_origin.
3991 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3992 symbols used as abstract origins.
3993 * cgraphunit.c (analyze_functions): Update.
3994 * ipa.c (symtab_remove_unreachable_nodes): Recompute
3995 used_as_abstract_origin.
3996 * tree-inline.c (tree_function_versioning): Update
3997 used_as_abstract_origin; be ready for DECL_RESULT and
3998 DECL_ARGUMENTS to be NULL.
4000 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
4001 for abstract functions.
4002 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
4005 2013-08-01 Jan Hubicka <jh@suse.cz>
4007 * profile.c (compute_value_histograms): Fix thinko.
4009 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
4011 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
4012 aarch-common-protos.h to extra_headers.
4013 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
4014 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
4015 * config/aarch64/t-aarch64 (aarch-common.o): Define.
4017 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
4019 * config/aarch64/aarch64.md (define_attr "type"): Delete.
4020 Include "../arm/types.md". Define "type" attribute for all patterns.
4021 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
4024 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4026 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
4027 to support power8 load fusion.
4028 (fusion_gpr_mem_load): Likewise.
4030 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
4032 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
4033 declarations for power8 load fusion.
4034 (emit_fusion_gpr_load): Likewise.
4036 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
4037 tuning for power8, turn on fusion mode by default. Turn on sign
4038 extending fusion mode if normal fusion mode is on, and we are at
4040 (fusion_gpr_load_p): New function, return true if we can fuse an
4041 addis instruction with a dependent load to a GPR.
4042 (emit_fusion_gpr_load): Emit the instructions for power8 load
4045 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
4046 (VSX load fusion peepholes): New peepholes to fuse together an
4047 addi instruction with a VSX load instruction.
4049 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
4050 peepholes to fuse an addis instruction with a load to a GPR base
4051 register. If we are supporting sign extending fusions, convert
4052 sign extending loads to zero extending loads and add an explicit
4055 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
4057 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
4058 aarch-common-protos.h to extra_headers.
4059 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
4060 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
4061 (arm_early_store_addr_dep): Likewise.
4062 (arm_no_early_alu_shift_dep): Likewise.
4063 (arm_no_early_alu_shift_value_dep): Likewise.
4064 (arm_no_early_mul_dep): Likewise.
4065 (arm_no_early_store_addr_dep): Likewise.
4066 (arm_mac_accumulator_is_mul_result): Likewise.
4067 (arm_mac_accumulator_is_result): Likewise.
4068 * config/arm/aarch-common.c: ... here. New file.
4069 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
4071 (arm_early_store_addr_dep): Likewise.
4072 (arm_no_early_alu_shift_dep): Likewise.
4073 (arm_no_early_alu_shift_value_dep): Likewise.
4074 (arm_no_early_mul_dep): Likewise.
4075 (arm_no_early_store_addr_dep): Likewise.
4076 (arm_mac_accumulator_is_mul_result): Likewise.
4077 (arm_mac_accumulator_is_result): Likewise.
4078 * config/arm/aarch-common-protos.h: ... here. New file.
4079 * config/arm/t-arm (aarch-common.o): Define.
4081 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
4083 * config/arm/arm.md: Include new file "types.md".
4084 (define_attr "type"): Move from here to ...
4085 (define_attr "mul32"): Likewise.
4086 (define_attr "mul64"): Likewise.
4087 * config/arm/types.md: ... here. New file.
4089 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
4091 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
4092 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
4094 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4096 * gen-pass-instances.awk: Fix offset of substr().
4098 2013-07-31 David Malcolm <dmalcolm@redhat.com>
4100 * Makefile.in (pass-instances.def): New.
4101 (passes.o): Replace dependency on passes.def with one on
4104 * gen-pass-instances.awk: New.
4106 * passes.c (pass_manager::pass_manager): Use pass-instances.def
4107 rather than passes.def, updating local definition of NEXT_PASS
4108 macro to add an extra NUM parameter (currently unused).
4110 2013-07-30 David Malcolm <dmalcolm@redhat.com>
4112 * Makefile.in (PASS_MANAGER_H): New.
4113 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
4114 (passes.o): Likewise.
4115 (statistics.o): Likewise.
4116 (cgraphunit.o): Likewise.
4117 (context.o): Depend on PASS_MANAGER_H.
4119 * pass_manager.h: New.
4121 * cgraphunit.c (cgraph_add_new_function): Update for moves
4122 of globals to fields of pass_manager.
4123 (analyze_function): Likewise.
4124 (expand_function): Likewise.
4125 (ipa_passes): Likewise.
4126 (compile): Likewise.
4128 * context.c (context::context): New.
4129 * context.h (context::context): New.
4130 (context::get_passes): New.
4131 (context::passes_): New.
4133 * lto-cgraph.c (input_node): Update for moves of globals to
4134 fields of pass_manager.
4136 * passes.c (all_passes): Remove, in favor of a field of the
4137 same name within the new class pass_manager.
4138 (all_small_ipa_passes): Likewise.
4139 (all_lowering_passes): Likewise.
4140 (all_regular_ipa_passes): Likewise.
4141 (all_late_ipa_passes): Likewise.
4142 (all_lto_gen_passes): Likewise.
4143 (passes_by_id): Likewise.
4144 (passes_by_id_size): Likewise.
4145 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4146 the new class pass_manager.
4147 (set_pass_for_id): Convert to...
4148 (pass_manager::set_pass_for_id): ...method.
4149 (get_pass_for_id): Convert to...
4150 (pass_manager::get_pass_for_id): ...method.
4151 (register_one_dump_file): Move body of implementation into...
4152 (pass_manager::register_one_dump_file): ...here.
4153 (register_dump_files_1): Convert to...
4154 (pass_manager::register_dump_files_1): ...method.
4155 (register_dump_files): Convert to...
4156 (pass_manager::register_dump_files): ...method.
4157 (create_pass_tab): Update for moves of globals to fields of
4159 (dump_passes): Move body of implementation into...
4160 (pass_manager::dump_passes): ...here.
4161 (register_pass): Move body of implementation into...
4162 (pass_manager::register_pass): ...here.
4163 (init_optimization_passes): Convert into...
4164 (pass_manager::pass_manager): ...constructor for new
4165 pass_manager class, and initialize the pass_lists array.
4166 (check_profile_consistency): Update for moves of globals to
4167 fields of pass_manager.
4168 (dump_profile_report): Move body of implementation into...
4169 (pass_manager::dump_profile_report): ...here.
4170 (ipa_write_summaries_1): Update for moves of pass lists from
4171 being globals to fields of pass_manager.
4172 (ipa_write_optimization_summaries): Likewise.
4173 (ipa_read_summaries): Likewise.
4174 (ipa_read_optimization_summaries): Likewise.
4175 (execute_all_ipa_stmt_fixups): Likewise.
4177 * statistics.c (statistics_fini): Update for moves of globals to
4178 fields of pass_manager.
4180 * toplev.c (general_init): Replace call to
4181 init_optimization_passes with construction of the pass_manager
4184 * tree-pass.h (all_passes): Remove, in favor of a field of the
4185 same name within the new class pass_manager.
4186 (all_small_ipa_passes): Likewise.
4187 (all_lowering_passes): Likewise.
4188 (all_regular_ipa_passes): Likewise.
4189 (all_lto_gen_passes): Likewise.
4190 (all_late_ipa_passes): Likewise.
4191 (passes_by_id): Likewise.
4192 (passes_by_id_size): Likewise.
4193 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
4194 the new class pass_manager.
4195 (get_pass_for_id): Remove.
4197 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
4199 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
4202 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
4204 * arm.md (mulhi3): New expand pattern.
4206 2013-07-30 Jan Hubicka <jh@suse.cz>
4207 Martin Liska <marxin.liska@gmail.com>
4209 * profile.c (compute_value_histograms): Do not ICE when
4210 there is mismatch only on some counters.
4212 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
4214 PR rtl-optimization/57637
4215 * function.c (move_insn_for_shrink_wrap): Also check the
4216 GEN set of the LIVE problem for the liveness analysis
4217 if it exists, otherwise give up.
4219 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4221 PR tree-optimization/57993
4222 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
4223 replaced statement in the candidate table.
4224 (phi_add_costs): Return infinite cost when the hidden basis does
4225 not dominate all phis on which the candidate is dependent.
4226 (replace_one_candidate): Record replaced statement in the
4229 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
4231 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
4232 (ashlv2si3): New expander.
4233 (*ashlv2si3_i): New define_insn_and_split.
4234 * predicates.md (float_operation): Allow patterns with three
4237 PR rtl-optimization/58021
4238 * mode-switching.c (create_pre_exit): Always split off preceding
4239 insns if we are not at the basic block head.
4241 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
4243 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
4244 (UCLIBC_DYNAMIC_LINKER): New macro.
4245 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
4247 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
4248 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
4250 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
4251 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
4252 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
4253 for SF and DF modes. Use ieee_quad_format for TF mode.
4254 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
4255 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
4256 (mips_option_override): Handle `-mnan=legacy'.
4257 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
4258 `-mabs=2008' and `-mnan=2008'.
4259 (OPTION_DEFAULT_SPECS): Add "nan" default.
4260 (ASM_SPEC): Handle `-mnan='.
4261 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
4262 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
4263 comment accordingly.
4264 (neg<mode>2): Likewise.
4265 * config/mips/mips.opt (mabs, mnan): New options.
4266 * doc/install.texi (Configuration): Document `--with-nan=' option.
4267 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
4269 (MIPS Options): Document them.
4270 * config.gcc <mips*-*-*>: Handle `--with-nan='.
4271 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
4272 * configure: Regenerate.
4273 * config.in: Regenerate.
4275 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
4277 * config/i386/i386.md (float post-reload splitters): Do not check
4278 for subregs of SSE registers.
4280 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
4281 H.J. Lu <hongjiu.lu@intel.com>
4285 * config/i386/i386.md (post-reload splitter
4286 to avoid partial SSE reg dependency stalls): New pattern.
4288 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
4290 * config/s390/s390.md ("movcc"): Swap load and store instructions.
4292 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
4294 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
4295 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
4297 2013-07-26 Cary Coutant <ccoutant@google.com>
4299 * dwarf2out.c (die_checksum_ordered): Don't include template
4300 instantiations in signature.
4301 (is_template_parameter): New function.
4302 (is_template_instantiation): New function.
4303 (generate_skeleton_bottom_up): Don't include template instantiations
4305 (generate_skeleton): Likewise.
4306 (break_out_comdat_types): Move recursive call to break out nested
4308 (prune_unused_types_mark_generic_parms_dies): Call
4309 is_template_parameter.
4311 2013-07-26 Ian Bolton <ian.bolton@arm.com>
4313 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
4314 uses vector registers.
4315 * config/aarch64/iterators.md: Add attributes rtn and vas.
4317 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4318 Richard Earnshaw <richard.earnshaw@arm.com>
4320 * combine.c (simplify_comparison): Re-canonicalize operands
4322 * config/arm/arm.md (movcond_addsi): New splitter.
4324 2013-07-25 Sterling Augustine <saugustine@google.com>
4326 * dwarf2out.c (size_of_pubnames): Move code to...
4327 (include_pubname_in_output): ...here. New.
4328 (want_pubnames): Rearrange.
4329 (output_pubnames): Call include_pubname_in_output. Move assertion.
4331 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
4333 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
4335 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
4338 * doc/extend.texi: Remove obsolete builtins. Fix
4339 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
4341 2013-07-25 Jan Hubicka <jh@suse.cz>
4343 * cgraph.c (release_function_body): Break out from ...
4344 (cgraph_release_function_body): ... this one; also release DECL_RESULT
4346 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
4347 old_tree in the map.
4348 (create_specialized_node): Update.
4349 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
4351 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
4352 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
4353 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
4354 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
4357 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4359 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
4360 addsi3_carryin_alt2_<optab>): Correct output template.
4362 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4364 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
4365 Adjust for arm_restrict_it.
4366 Remove trailing whitespace.
4368 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
4370 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
4371 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
4373 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
4375 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
4377 PR rtl-optimization/57960
4378 * lra-constraints.c (process_alt_operands): Use the right mode
4379 when checking strict_low.
4381 2013-07-25 Jan Hubicka <jh@suse.cz>
4383 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
4384 * cgraph.c (cgraph_remove_node): Do not release function body
4385 when in cgraph streaming.
4386 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
4387 in other partitions are not considered reachable; fix handling of
4390 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4392 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
4394 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4399 * config/arm/arm.md ("*sibcall_insn): Replace use of
4400 Ss with US. Adjust output for v5 and v4t.
4401 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
4403 * config/arm/constraints.md ("Ss"): Rename to US.
4405 2013-07-25 Terry Guo <terry.guo@arm.com>
4407 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
4408 shift_add/shift_sub0/shift_sub1 RTXs.
4410 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
4411 Anton Blanchard <anton@au1.ibm.com>
4413 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
4414 (altivec_vpks<VI_char>ss): Likewise.
4415 (altivec_vpks<VI_char>us): Likewise.
4416 (altivec_vpku<VI_char>us): Likewise.
4417 (altivec_vpku<VI_char>um): Likewise.
4419 2013-07-24 David Malcolm <dmalcolm@redhat.com>
4421 Introduce context class.
4423 * Makefile.in (CONTEXT_H): New.
4424 (OBJS): Add context.o.
4425 (toplev.o): Add CONTEXT_H to dependencies.
4428 * toplev.c (general_init): Create the singleton gcc::context instance.
4434 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
4436 PR rtl-optimization/57968
4437 * mode-switching.c (create_pre_exit): Allow instructions that
4438 don't set a return register to need a non-exit mode.
4440 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4441 Anton Blanchard <anton@au1.ibm.com>
4443 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
4444 operands to vperm for little endian.
4445 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
4446 of lvsl to create the control mask for a vperm for little endian.
4448 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4449 Anton Blanchard <anton@au1.ibm.com>
4451 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
4452 two operands for little-endian.
4454 2013-07-23 Steve Ellcey <sellcey@mips.com>
4456 * config/mips/mips.c (mips_case_values_threshold): New.
4457 (TARGET_CASE_VALUES_THRESHOLD): Define.
4459 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4460 Anton Blanchard <anton@au1.ibm.com>
4462 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
4463 selection of field for vector splat in little endian mode.
4465 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4467 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
4468 expanders to rs6000.md.
4469 (ior<mode>3): Likewise.
4470 (and<mode>3): Likewise.
4471 (one_cmpl<mode>2): Likewise.
4472 (nor<mode>3): Likewise.
4473 (andc<mode>3): Likewise.
4474 (eqv<mode>3): Likewise.
4475 (nand<mode>3): Likewise.
4476 (orc<mode>3): Likewise.
4478 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
4481 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
4482 to split multi-word logical operations.
4483 (rs6000_split_logical_di): Likewise.
4484 (rs6000_split_logical): Likewise.
4486 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
4487 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
4488 and allow TImode operations in 32-bit.
4489 (vsx_and<mode>3_64bit): Likewise.
4490 (vsx_ior<mode>3_32bit): Likewise.
4491 (vsx_ior<mode>3_64bit): Likewise.
4492 (vsx_xor<mode>3_32bit): Likewise.
4493 (vsx_xor<mode>3_64bit): Likewise.
4494 (vsx_one_cmpl<mode>2_32bit): Likewise.
4495 (vsx_one_cmpl<mode>2_64bit): Likewise.
4496 (vsx_nor<mode>3_32bit): Likewise.
4497 (vsx_nor<mode>3_64bit): Likewise.
4498 (vsx_andc<mode>3_32bit): Likewise.
4499 (vsx_andc<mode>3_64bit): Likewise.
4500 (vsx_eqv<mode>3_32bit): Likewise.
4501 (vsx_eqv<mode>3_64bit): Likewise.
4502 (vsx_nand<mode>3_32bit): Likewise.
4503 (vsx_nand<mode>3_64bit): Likewise.
4504 (vsx_orc<mode>3_32bit): Likewise.
4505 (vsx_orc<mode>3_64bit): Likewise.
4507 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
4508 logical types in GPRs.
4510 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
4511 logical insns to rs6000.md, and allow TImode operations in
4513 (altivec_ior<mode>3): Likewise.
4514 (altivec_xor<mode>3): Likewise.
4515 (altivec_one_cmpl<mode>2): Likewise.
4516 (altivec_nor<mode>3): Likewise.
4517 (altivec_andc<mode>3): Likewise.
4519 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
4520 attributes for moving the 128-bit logical operations into
4522 (BOOL_REGS_OUTPUT): Likewise.
4523 (BOOL_REGS_OP1): Likewise.
4524 (BOOL_REGS_OP2): Likewise.
4525 (BOOL_REGS_UNARY): Likewise.
4526 (BOOL_REGS_AND_CR0): Likewise.
4527 (one_cmpl<mode>2): Add support for DI logical operations on
4528 32-bit, splitting the operations to 32-bit.
4532 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
4533 changes to combine the 32/64-bit code, allow logical operations on
4534 TI mode in 32-bit, and to use similar match_operator patterns like
4535 scalar mode uses. Combine the Altivec and VSX code for logical
4536 operations, and move it here.
4537 (ior<mode>3, 128-bit types): Likewise.
4538 (xor<mode>3, 128-bit types): Likewise.
4539 (one_cmpl<mode>3, 128-bit types): Likewise.
4540 (nor<mode>3, 128-bit types): Likewise.
4541 (andc<mode>3, 128-bit types): Likewise.
4542 (eqv<mode>3, 128-bit types): Likewise.
4543 (nand<mode>3, 128-bit types): Likewise.
4544 (orc<mode>3, 128-bit types): Likewise.
4545 (and<mode>3_internal): Likewise.
4546 (bool<mode>3_internal): Likewise.
4547 (boolc<mode>3_internal1): Likewise.
4548 (boolc<mode>3_internal2): Likewise.
4549 (boolcc<mode>3_internal1): Likewise.
4550 (boolcc<mode>3_internal2): Likewise.
4551 (eqv<mode>3_internal1): Likewise.
4552 (eqv<mode>3_internal2): Likewise.
4553 (one_cmpl1<mode>3_internal): Likewise.
4555 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
4557 * config/microblaze/microblaze.c (microblaze_expand_prologue):
4558 Rename flag_stack_usage to flag_stack_usage_info.
4560 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
4562 * config/microblaze/sync.md: New file.
4563 * config/microblaze/microblaze.md: Include sync.md
4564 * config/microblaze/microblaze.c: Add print_operand 'y'.
4565 * config/microblaze/constraints.md: Add memory_contraint
4566 'Q' which is a single register.
4568 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
4570 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
4572 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
4574 * reload.c (find_reloads): Exit loop once we find this operand
4575 cannot be reloaded somehow for this alternative.
4577 * reload.c (find_reloads): Exit loop once we find a hard register.
4579 * rtlanal.c (computed_jump_p): Exit loop once we find label
4582 * i386.c (ix86_pad_returns): Exit loop after setting replace.
4584 * cfgloopmanip.c (remove_path): Exit loop after setting
4587 * gensupport.c (subst_dup): Avoid loop if code is not
4588 MATCH_DUP nor MATCH_OP_DUP.
4590 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
4592 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
4594 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4596 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
4597 true for SP_REGNUM if mode == ptr_mode.
4598 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
4599 with value R0_REGNUM + 31.
4601 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4603 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4604 pad pointer-typed argument downward.
4606 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4608 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4609 and __ILP32__ when the ILP32 model is in use.
4611 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4613 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4614 (aarch64_load_symref_appropriately): In the case of
4615 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4616 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4617 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4618 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4619 if 'mode' doesn't equal to 'ptr_mode'.
4620 (aarch64_output_mi_thunk): Add an assertion on the alignment of
4621 'vcall_offset'; change to call aarch64_emit_move differently depending
4622 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4623 to calculate the upper bound of 'vcall_offset'.
4624 (aarch64_cannot_force_const_mem): Change to also return true if
4626 (aarch64_legitimize_reload_address): In the case of large
4627 displacements, add new local variable 'xmode' and an assertion
4628 based on it; change to use 'xmode' to generate the new rtx and
4630 (aarch64_asm_trampoline_template): Change to generate the template
4631 differently depending on TARGET_ILP32 or not; change to use
4632 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4633 (aarch64_trampoline_size): Removed.
4634 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4635 and replace immediate literals with it. Change to use 'ptr_mode'
4636 instead of 'DImode' and call convert_memory_address if the mode
4637 of 'fnaddr' doesn't equal to 'ptr_mode'.
4638 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4640 (aarch64_elf_asm_destructor): Likewise.
4641 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4642 on TARGET_ILP32 instead of aarch64_trampoline_size.
4643 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4644 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4645 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4646 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
4647 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4648 (storewb_pair<GPI:mode>_<P:mode>): ... this.
4649 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4650 depending on the value of 'mode'.
4651 (add_losym_<mode>): New.
4652 (ldr_got_small_<mode>): New, based on ldr_got_small.
4653 (ldr_got_small): Remove.
4654 (ldr_got_small_sidi): New.
4655 * config/aarch64/iterators.md (P): New.
4656 (PTR): Change to 'ptr_mode' in the condition.
4658 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
4660 * config.gcc (aarch64*-*-*): Support --with-abi.
4661 (aarch64*-*-elf): Support --with-multilib-list.
4662 (aarch64*-*-linux*): Likewise.
4663 (supported_defaults): Add abi to aarch64*-*-*.
4664 * configure.ac: Mention AArch64 for --with-multilib-list.
4665 * configure: Re-generated.
4666 * config/aarch64/biarchilp32.h: New file.
4667 * config/aarch64/biarchlp64.h: New file.
4668 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4670 (MULTILIB_DEFAULTS): Ditto.
4671 (DRIVER_SELF_SPECS): Ditto.
4672 (ASM_SPEC): Update to also substitute -mabi.
4673 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4674 file whose name depends on -mabi= and -mbig-endian.
4675 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4677 (POINTER_SIZE): New define.
4678 (POINTERS_EXTEND_UNSIGNED): Ditto.
4679 (enum aarch64_abi_type): New enumeration tag.
4680 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4681 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4682 (TARGET_ILP32): New define.
4683 * config/aarch64/aarch64.opt (mabi): New.
4685 (ilp32, lp64): New values for -mabi.
4686 * config/aarch64/t-aarch64 (comma): New define.
4687 (MULTILIB_OPTIONS): Ditto.
4688 (MULTILIB_DIRNAMES): Ditto.
4689 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4690 * doc/invoke.texi: Document -mabi for AArch64.
4692 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
4694 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4696 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4697 Anton Blanchard <anton@au1.ibm.com>
4699 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4700 endianness when selecting field to splat.
4702 2013-07-22 Eric Christopher <echristo@gmail.com>
4704 * dwarf2out.c (die_odr_checksum): New function to use
4705 CHECKSUM_ macros and ULEB128 for DIE tag.
4706 (generate_type_signature): Use.
4708 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
4710 * config.gcc (sparc*-*-*): Accept leon3 processor.
4711 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4712 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4713 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4714 * config/sparc/sparc.opt (enum processor_type): Add leon3.
4715 (mfix-ut699): Adjust comment.
4716 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4717 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4718 (CPP_CPU_SPEC): Likewise.
4719 (ASM_CPU_SPEC): Likewise.
4720 * config/sparc/sparc.c (leon3_cost): New constant.
4721 (sparc_option_override): Add leon3 support.
4722 (mem_ref): New function.
4723 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4724 (sparc_do_work_around_errata): Look into the instruction in the delay
4725 slot and adjust accordingly. Add fix for the data cache nullify issues
4726 of the UT699. Change insertion position for the NOP.
4727 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4728 (leon3_load): New reservation.
4729 (leon_store): Bump latency to 2.
4730 (grfpu): New automaton.
4731 (grfpu_alu): New unit.
4732 (grfpu_ds): Likewise.
4733 (leon_fp_alu): Adjust.
4734 (leon_fp_mult): Delete.
4735 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4736 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4737 * config/sparc/sparc.md (cpu): Add leon3.
4738 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4740 (atomic_test_and_set): Likewise.
4743 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
4745 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
4746 default for R5900 targets.
4747 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4748 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4749 * config/mips/mips.c (mips_option_override): Report an error for
4750 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
4751 for -march=r5900 -mhard-float.
4753 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
4755 * df-problems.c (can_move_insns_across): Exit loop once we
4756 find a non-fixed, non-global register.
4758 * ipa-pure-const.c (propagate_nothrow): Exit loop after
4761 * omega.c (omega_eliminate_red): Break after setting red_found.
4762 (omega_problem_has_red_equations): Similarly after setting found.
4763 (omega_query_variable): Similarly after setting coupled.
4765 2013-07-22 Marek Polacek <polacek@redhat.com>
4767 * gimplify.c: Don't include gimple.h twice.
4769 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4771 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4772 instead of TARGET_THUMB1.
4773 (Pz): New constraint.
4774 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4776 (compare_negsi_si): Likewise.
4777 (compare_addsi2_op0): Likewise.
4778 (compare_addsi2_op1): Likewise.
4779 (addsi3_carryin_<optab>): Likewise.
4780 (addsi3_carryin_alt2_<optab>): Likewise.
4781 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4782 for arm_restrict_it.
4783 (subsi3_carryin): Likewise.
4784 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4785 (minmax_arithsi): Disable for arm_restrict_it.
4786 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4787 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4788 (satsi_<SAT:code>_shift): Likewise.
4789 (arm_shiftsi3): Add alternative for 16-bit encoding.
4790 (arm32_movhf): Disable for arm_restrict_it.
4791 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4792 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4794 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
4796 * config/arm/arm.md (attribute "insn"): Delete.
4797 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4798 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4799 (not_shiftsi): Update for attribute change.
4800 (not_shiftsi_compare0): Likewise.
4801 (not_shiftsi_compare0_scratch): Likewise.
4802 (arm_one_cmplsi2): Likewise.
4803 (thumb1_one_cmplsi2): Likewise.
4804 (notsi_compare0): Likewise.
4805 (notsi_compare0_scratch): Likewise.
4806 (thumb1_movdi_insn): Likewise.
4807 (arm_movsi_insn): Likewise.
4808 (movhi_insn_arch4): Likewise.
4809 (movhi_bytes): Likewise.
4810 (arm_movqi_insn): Likewise.
4811 (thumb1_movqi_insn): Likewise.
4812 (arm32_movhf): Likewise.
4813 (thumb1_movhf): Likewise.
4814 (arm_movsf_soft_insn): Likewise.
4815 (thumb1_movsf_insn): Likewise.
4816 (thumb_movdf_insn): Likewise.
4817 (movsicc_insn): Likewise.
4818 (movsfcc_soft_insn): Likewise.
4819 (and_scc): Likewise.
4820 (cond_move): Likewise.
4821 (if_move_not): Likewise.
4822 (if_not_move): Likewise.
4823 (if_shift_move): Likewise.
4824 (if_move_shift): Likewise.
4825 (if_shift_shift): Likewise.
4826 (if_not_arith): Likewise.
4827 (if_arith_not): Likewise.
4828 (cond_move_not): Likewise.
4829 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4830 (neon_mov<mode>): Likewise.
4831 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4832 (thumb2_movsi_vfp): Likewise.
4833 (movsf_vfp): Likewise.
4834 (thumb2_movsf_vfp): Likewise.
4835 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4837 (cortexa7_older_only): Likewise.
4838 (cortexa7_younger): Likewise.
4839 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4840 (1020alu_shift_op): Likewise.
4841 (1020alu_shift_reg_op): Likewise.
4842 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4843 (alu_shift_op): Likewise.
4844 (alu_shift_reg_op): Likewise.
4845 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4846 (11_alu_shift_op): Likewise.
4847 (11_alu_shift_reg_op): Likewise.
4848 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4849 (9_alu_shift_reg_op): Likewise.
4850 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4852 (cortex_a15_alu_shift): Likewise.
4853 (cortex_a15_alu_shift_reg): Likewise.
4854 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4855 (cortex_a5_alu_shift): Likewise.
4856 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4858 (cortex_a53_alu_shift): Likewise.
4859 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4861 (cortex_a7_alu_reg): Likewise.
4862 (cortex_a7_alu_shift): Likewise.
4863 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4864 (cortex_a8_alu_shift): Likewise.
4865 (cortex_a8_alu_shift_reg): Likewise.
4866 (cortex_a8_mov): Likewise.
4867 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4868 (cortex_a9_dp_shift): Likewise.
4869 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4870 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4871 (cortex_r4_mov): Likewise.
4872 (cortex_r4_alu_shift): Likewise.
4873 (cortex_r4_alu_shift_reg): Likewise.
4874 * config/arm/fa526.md (526_alu_op): Update for attribute change.
4875 (526_alu_shift_op): Likewise.
4876 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4877 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4878 (626te_alu_shift_op): Likewise.
4879 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4880 (726te_alu_op): Likewise.
4881 (726te_alu_shift_op): Likewise.
4882 (726te_alu_shift_reg_op): Likewise.
4883 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4884 (mp626_alu_shift_op): Likewise.
4885 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4886 (pj4_alu_e1_conds): Likewise.
4887 (pj4_alu): Likewise.
4888 (pj4_alu_conds): Likewise.
4889 (pj4_shift): Likewise.
4890 (pj4_shift_conds): Likewise.
4891 (pj4_alu_shift): Likewise.
4892 (pj4_alu_shift_conds): Likewise.
4894 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4896 * config/arm/predicates.md (shiftable_operator_strict_it):
4898 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4899 Disable cond_exec version for arm_restrict_it.
4900 (thumb2_smaxsi3): Convert to generate cond_exec.
4901 (thumb2_sminsi3): Likewise.
4902 (thumb32_umaxsi3): Likewise.
4903 (thumb2_uminsi3): Likewise.
4904 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4905 (thumb2_neg_abssi2): Likewise.
4906 (thumb2_mov_scc): Add alternative for 16-bit encoding.
4907 (thumb2_movsicc_insn): Adjust alternatives.
4908 (thumb2_mov_negscc): Disable for arm_restrict_it.
4909 (thumb2_mov_negscc_strict_it): New pattern.
4910 (thumb2_mov_notscc_strict_it): New pattern.
4911 (thumb2_mov_notscc): Disable for arm_restrict_it.
4912 (thumb2_ior_scc): Likewise.
4913 (thumb2_ior_scc_strict_it): New pattern.
4914 (thumb2_cond_move): Adjust for arm_restrict_it.
4915 (thumb2_cond_arith): Disable for arm_restrict_it.
4916 (thumb2_cond_arith_strict_it): New pattern.
4917 (thumb2_cond_sub): Adjust for arm_restrict_it.
4918 (thumb2_movcond): Likewise.
4919 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4920 (thumb2_zero_extendhisi2_v6): Likewise.
4921 (thumb2_zero_extendqisi2_v6): Likewise.
4922 (orsi_notsi_si): Likewise.
4923 (orsi_not_shiftsi_si): Likewise.
4925 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
4927 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4928 instruction sequence is 1 byte shorter.
4930 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
4932 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4933 it is not needed after split.
4935 2013-07-20 Iain Sandoe <iain@codesourcery.com>
4938 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4939 second label for nonlocal goto receivers. Don't output pic base labels
4940 unless we're producing PIC; mark that action unreachable().
4941 (ix86_save_reg): If the function contains a nonlocal label, save the
4943 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4944 * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4945 (update_pic_label_number_if_needed): New.
4946 (machopic_output_function_base_name): Adjust for nonlocal receiver
4948 (machopic_should_output_picbase_label): New.
4949 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4950 (nonlocal_goto_receiver): New insn and split.
4952 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
4954 * config/aarch64/aarch64-builtins.c
4955 (aarch64_fold_builtin): Fold abs in all modes.
4956 * config/aarch64/aarch64-simd-builtins.def
4957 (abs): Enable for all modes.
4958 * config/aarch64/arm_neon.h
4959 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4960 (vabs_f64): Add missing intrinsic.
4962 2013-07-19 Ian Bolton <ian.bolton@arm.com>
4964 * config/aarch64/arm_neon.h (vabs_s64): New function
4966 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
4969 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4970 * config/avr/avr.md (adjust_len): Add `round'.
4971 * config/avr/avr-protos.h (avr_out_round): New prototype.
4972 (avr_out_plus): Add `out_label' argument.
4973 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4974 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4975 Handle the case where `insn' is just a pattern.
4976 (avr_out_bitop): Handle the case where `insn' is just a pattern.
4977 (avr_out_round): New function.
4978 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4980 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
4982 * config/microblaze/microblaze.c (microblaze_expand_prologue):
4983 Add check for flag_stack_usage to handle -fstack-usage support
4985 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
4987 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4988 interaction for new Power8 flags and VSX.
4990 2013-07-18 Sriraman Tallam <tmsriram@google.com>
4993 * tree-inline.c (expand_call_inline): Emit errors during
4994 early_inlining only if optimization is not turned on.
4996 2013-07-18 David Malcolm <dmalcolm@redhat.com>
5000 * passes.c (init_optimization_passes): Move the construction of
5001 the pass hierarchy into a new passes.def file.
5003 * Makefile.in (passes.o): Add dependency on passes.def.
5005 2013-07-18 David Malcolm <dmalcolm@redhat.com>
5007 * passes.c (init_optimization_passes): Introduce macros for
5008 constructing the tree of passes (INSERT_PASSES_AFTER,
5009 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
5010 TERMINATE_PASS_LIST).
5012 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
5013 Wei Mi <wmi@google.com>
5015 PR rtl-optimization/57878
5016 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
5018 (reload_pseudo_compare_func): Check nregs first for reload
5021 2013-07-18 David Malcolm <dmalcolm@redhat.com>
5023 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
5025 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
5027 * read-rtl.c (validate_const_int): Once an invalid character is
5028 seen, quit the loop.
5030 * gengtype.c (write_roots): Similarly once we find the "deletable"
5031 or "if_marked" option.
5033 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
5035 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
5036 "xtab" and "sat". Move value "clz" from here to ...
5037 (attriubte "type"): ... here.
5038 (satsi_<SAT:code>): Delete "insn" attribute.
5039 (satsi_<SAT:code>_shift): Likewise.
5040 (arm_zero_extendqisi2addsi): Likewise.
5041 (arm_extendqisi2addsi): Likewise.
5042 (clzsi2): Update for attribute changes.
5043 (rbitsi2): Likewise.
5044 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
5046 (arm_usatsihi): Likewise.
5047 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
5049 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
5051 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
5052 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
5053 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
5054 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
5055 in alphabetical order.
5056 (attribute "core_cycles"): Update for attribute changes.
5057 (arm_addsi3): Likewise.
5058 (addsi3_compare0): Likewise.
5059 (addsi3_compare0_scratch): Likewise.
5060 (addsi3_compare_op1): Likewise.
5061 (addsi3_compare_op2): Likewise.
5062 (compare_addsi2_op0): Likewise.
5063 (compare_addsi2_op1): Likewise.
5064 (addsi3_carryin_shift_<optab>): Likewise.
5065 (subsi3_carryin_shift): Likewise.
5066 (rsbsi3_carryin_shift): Likewise.
5067 (arm_subsi3_insn): Likewise.
5068 (subsi3_compare0): Likewise.
5069 (subsi3_compare): Likewise.
5070 (arm_andsi3_insn): Likewise.
5071 (thumb1_andsi3_insn): Likewise.
5072 (andsi3_compare0): Likewise.
5073 (andsi3_compare0_scratch): Likewise.
5074 (zeroextractsi_compare0_scratch
5075 (andsi_not_shiftsi_si): Likewise.
5076 (iorsi3_insn): Likewise.
5077 (iorsi3_compare0): Likewise.
5078 (iorsi3_compare0_scratch): Likewise.
5079 (arm_xorsi3): Likewise.
5080 (thumb1_xorsi3_insn): Likewise.
5081 (xorsi3_compare0): Likewise.
5082 (xorsi3_compare0_scratch): Likewise.
5083 (satsi_<SAT:code>_shift): Likewise.
5085 (arm_shiftsi3): Likewise.
5086 (shiftsi3_compare0): Likewise.
5087 (not_shiftsi): Likewise.
5088 (not_shiftsi_compare0): Likewise.
5089 (not_shiftsi_compare0_scratch): Likewise.
5090 (arm_one_cmplsi2): Likewise.
5091 (thumb_one_complsi2): Likewise.
5092 (notsi_compare0): Likewise.
5093 (notsi_compare0_scratch): Likewise.
5094 (thumb1_zero_extendhisi2): Likewise.
5095 (arm_zero_extendhisi2): Likewise.
5096 (arm_zero_extendhisi2_v6): Likewise.
5097 (arm_zero_extendhisi2addsi): Likewise.
5098 (thumb1_zero_extendqisi2): Likewise.
5099 (thumb1_zero_extendqisi2_v6): Likewise.
5100 (arm_zero_extendqisi2): Likewise.
5101 (arm_zero_extendqisi2_v6): Likewise.
5102 (arm_zero_extendqisi2addsi): Likewise.
5103 (thumb1_extendhisi2): Likewise.
5104 (arm_extendhisi2): Likewise.
5105 (arm_extendhisi2_v6): Likewise.
5106 (arm_extendqisi): Likewise.
5107 (arm_extendqisi_v6): Likewise.
5108 (arm_extendqisi2addsi): Likewise.
5109 (thumb1_extendqisi2): Likewise.
5110 (thumb1_movdi_insn): Likewise.
5111 (arm_movsi_insn): Likewise.
5112 (movsi_compare0): Likewise.
5113 (movhi_insn_arch4): Likewise.
5114 (movhi_bytes): Likewise.
5115 (arm_movqi_insn): Likewise.
5116 (thumb1_movqi_insn): Likewise.
5117 (arm32_movhf): Likewise.
5118 (thumb1_movhf): Likewise.
5119 (arm_movsf_soft_insn): Likewise.
5120 (thumb1_movsf_insn): Likewise.
5121 (movdf_soft_insn): Likewise.
5122 (thumb_movdf_insn): Likewise.
5123 (arm_cmpsi_insn): Likewise.
5124 (cmpsi_shiftsi): Likewise.
5125 (cmpsi_shiftsi_swp): Likewise.
5126 (arm_cmpsi_negshiftsi_si): Likewise.
5127 (movsicc_insn): Likewise.
5128 (movsfcc_soft_insn): Likewise.
5129 (arith_shiftsi): Likewise.
5130 (arith_shiftsi_compare0
5131 (arith_shiftsi_compare0_scratch
5132 (sub_shiftsi): Likewise.
5133 (sub_shiftsi_compare0
5134 (sub_shiftsi_compare0_scratch
5135 (and_scc): Likewise.
5136 (cond_move): Likewise.
5137 (if_plus_move): Likewise.
5138 (if_move_plus): Likewise.
5139 (if_move_not): Likewise.
5140 (if_not_move): Likewise.
5141 (if_shift_move): Likewise.
5142 (if_move_shift): Likewise.
5143 (if_shift_shift): Likewise.
5144 (if_not_arith): Likewise.
5145 (if_arith_not): Likewise.
5146 (cond_move_not): Likewise.
5147 (thumb1_ashlsi3): Set type attribute.
5148 (thumb1_ashrsi3): Likewise.
5149 (thumb1_lshrsi3): Likewise.
5150 (thumb1_rotrsi3): Likewise.
5151 (shiftsi3_compare0_scratch): Likewise.
5152 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
5153 (neon_mov<mode>): Likewise.
5154 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
5156 (thumb2_movsi_insn): Likewise.
5157 (thumb2_cmpsi_neg_shiftsi): Likewise.
5158 (thumb2_extendqisi_v6): Likewise.
5159 (thumb2_zero_extendhisi2_v6): Likewise.
5160 (thumb2_zero_extendqisi2_v6): Likewise.
5161 (thumb2_shiftsi3_short): Likewise.
5162 (thumb2_addsi3_compare0_scratch): Likewise.
5163 (orsi_not_shiftsi_si): Likewise.
5164 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
5165 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
5167 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
5168 (1020alu_shift_op): Likewise.
5169 (1020alu_shift_reg_op): Likewise.
5170 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
5171 (alu_shift_op): Likewise.
5172 (alu_shift_reg_op): Likewise.
5173 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
5174 (11_alu_shift_op): Likewise.
5175 (11_alu_shift_reg_op): Likewise.
5176 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
5177 (9_alu_shift_reg_op): Likewise.
5178 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
5180 (cortex_a15_alu_shift): Likewise.
5181 (cortex_a15_alu_shift_reg): Likewise.
5182 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
5184 (cortex_a5_alu_shift): Likewise.
5185 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
5187 (cortex_a53_alu_shift): Likewise.
5188 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
5190 (cortex_a7_alu_reg): Likewise.
5191 (cortex_a7_alu_shift): Likewise.
5192 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
5194 (cortex_a8_alu_shift): Likewise.
5195 (cortex_a8_alu_shift_reg): Likewise.
5196 (cortex_a8_mov): Likewise.
5197 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
5198 (cortex_a9_dp_shift): Likewise.
5199 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
5201 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
5203 (cortex_r4_mov): Likewise.
5204 (cortex_r4_alu_shift): Likewise.
5205 (cortex_r4_alu_shift_reg): Likewise.
5206 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
5207 (526_alu_shift_op): Likewise.
5208 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
5209 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
5210 (626te_alu_shift_op): Likewise.
5211 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
5212 (726te_alu_op): Likewise.
5213 (726te_alu_shift_op): Likewise.
5214 (726te_alu_shift_reg_op): Likewise.
5215 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
5216 (mp626_alu_shift_op): Likewise.
5217 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
5218 (pj4_alu_e1_conds): Likewise.
5219 (pj4_alu): Likewise.
5220 (pj4_alu_conds): Likewise.
5221 (pj4_shift): Likewise.
5222 (pj4_shift_conds): Likewise.
5223 (pj4_alu_shift): Likewise.
5224 (pj4_alu_shift_conds): Likewise.
5225 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
5227 (cortexa7_older_only): Likewise.
5228 (cortexa7_younger): Likewise.
5230 2013-07-18 David Malcolm <dmalcolm@redhat.com>
5232 * ipa-pure-const.c (generate_summary): Rename to...
5233 (pure_const_generate_summary): ... this.
5235 2013-07-17 Iain Sandoe <iain@codesourcery.com>
5237 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
5239 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
5242 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
5245 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5247 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
5248 enabled without -march=zEC12.
5249 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
5252 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5254 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
5255 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
5256 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
5257 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
5258 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
5259 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
5260 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
5262 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
5263 and ISA_HAS_NMADD3_NMSUB3.
5264 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
5265 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
5266 (nmsub4<mode>, nmsub3<mode>): Likewise.
5267 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
5269 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5271 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
5272 TARGET_MIPS5400 checking.
5274 2013-07-16 Jakub Jelinek <jakub@redhat.com>
5275 Peter Bergner <bergner@vnet.ibm.com>
5277 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
5278 registers in the comment.
5279 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
5280 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
5281 rather than FIRST_PSEUDO_REGISTERS.
5283 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
5285 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
5286 enable extra ISA flags with TARGET_HTM.
5288 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
5290 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
5293 2013-07-15 Cong Hou <congh@google.com>
5295 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
5296 in compare function for sorting.
5298 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
5300 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
5301 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
5302 * config/rs6000/rs6000.opt: Add -mhtm option.
5303 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
5304 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
5305 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
5306 __HTM__ if the HTM instructions are available.
5307 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
5308 htm_spr_reg_operand): New define_predicates.
5309 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
5310 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
5312 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
5313 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
5314 HTM builtin functions.
5315 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
5316 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
5317 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
5318 (rs6000_builtin_mask_calculate): Likewise.
5319 (rs6000_option_override_internal): Likewise.
5320 (bdesc_htm): Add new HTM builtin support.
5321 (htm_spr_num): New function.
5322 (htm_spr_regno): Likewise.
5323 (rs6000_htm_spr_icode): Likewise.
5324 (htm_expand_builtin): Likewise.
5325 (htm_init_builtins): Likewise.
5326 (rs6000_expand_builtin): Add support for HTM builtin functions.
5327 (rs6000_init_builtins): Likewise.
5328 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
5330 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
5331 (TARGET_HTM, MASK_HTM): Define macros.
5332 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
5333 (FIXED_REGISTERS): Likewise.
5334 (CALL_USED_REGISTERS): Likewise.
5335 (CALL_REALLY_USED_REGISTERS): Likewise.
5336 (REG_ALLOC_ORDER): Likewise.
5337 (enum reg_class): Likewise.
5338 (REG_CLASS_NAMES): Likewise.
5339 (REG_CLASS_CONTENTS): Likewise.
5340 (REGISTER_NAMES): Likewise.
5341 (ADDITIONAL_REGISTER_NAMES): Likewise.
5342 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
5343 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
5344 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
5345 * config/rs6000/htm.md: New file.
5346 * config/rs6000/htmintrin.h: New file.
5347 * config/rs6000/htmxlintrin.h: New file.
5349 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
5351 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5352 Define SYMBOL_TINY_GOT, update comment.
5353 * config/aarch64/aarch64.c
5354 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
5355 (aarch64_expand_mov_immediate): Likewise.
5356 (aarch64_print_operand): Likewise.
5357 (aarch64_classify_symbol): Likewise.
5358 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
5359 (ldr_got_tiny): Define.
5361 2013-07-13 Tobias Grosser <tobias@grosser.es>
5363 PR tree-optimization/54094
5364 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
5365 scheduling dimension for the parallelism check from the polyhedral
5366 information in the AST.
5367 * graphite-dependences.c (carries_deps): Do not assume the schedule is
5370 2013-07-13 Jason Merrill <jason@redhat.com>
5372 * print-tree.c (debug_vec_tree): Use debug_raw.
5373 (debug_raw (vec<tree, va_gc> &)): New.
5374 (debug_raw (vec<tree, va_gc> *)): New.
5375 * tree.h: Declare them.
5377 2013-07-13 Bin Cheng <bin.cheng@arm.com>
5379 * ifcvt.c (ifcvt_after_combine): New static variable.
5380 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
5382 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
5383 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
5384 rest_of_handle_if_after_reload): Pass new argument for if_convert.
5386 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
5388 * config/mips/mips.c (mips_expand_call): Remove empty statement.
5390 2013-07-12 Michael Matz <matz@suse.de>
5393 * convert.c (convert_to_real): Reject non-float inner types.
5395 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
5397 * config/aarch64/aarch64-protos.h
5398 (aarch64_simd_immediate_valid_for_move): Remove.
5399 * config/aarch64/aarch64.c (simd_immediate_info): New member.
5400 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
5402 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
5404 2013-07-11 Steve Ellcey <sellcey@mips.com>
5406 * config/mips/mips.c (mips_conditional_register_usage): Do not
5407 use t[0-7] registers in MIPS16 mode when optimizing for size.
5409 2013-07-11 Sriraman Tallam <tmsriram@google.com>
5411 * config/i386/i386.c (dispatch_function_versions): Fix array
5412 indexing of function_version_info to match actual_versions.
5414 2013-07-11 Teresa Johnson <tejohnson@google.com>
5416 * vec.h (struct va_gc): Move release out-of-line.
5417 (va_gc::release): Call ggc_free on released vec.
5419 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
5421 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
5422 Require GOT register as additional operand in UNSPEC.
5423 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
5424 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5425 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
5426 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
5427 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
5428 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
5429 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
5431 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
5434 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
5435 name seen by assembler/linker rather if available.
5437 2013-07-11 Andreas Schwab <schwab@suse.de>
5439 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
5441 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
5443 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
5445 2013-07-10 Joseph Myers <joseph@codesourcery.com>
5447 * doc/tm.texi.in: Move hook documentation to ....
5448 * target.def: ... here.
5450 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
5452 * doc/tm.texi: Regenerate.
5454 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
5457 * doc/invoke.texi: Document -Wconditionally-supported.
5459 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5462 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5465 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5468 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
5469 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
5470 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
5471 Remove duplicate devices.
5472 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
5473 * config/avr/t-multilib: Regenerate.
5474 * config/avr/avr-tables.opt: Regenerate.
5475 * doc/avr-mmcu.texi: Regenerate.
5477 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5480 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
5482 2013-07-10 Graham Stott <graham.stott@btinternet.com>
5484 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
5485 the cost of MULT when optimizing for size.
5487 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5489 * config/cr16/cr16-protos.h: Don't include target.h.
5491 2013-07-09 Joseph Myers <joseph@codesourcery.com>
5493 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
5494 adjust register size for TDmode and TFmode for VSX registers.
5496 2013-07-08 Kai Tietz <ktietz@redhat.com>
5499 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
5500 hook_bool_const_tree_true.
5502 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5504 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
5505 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
5506 * config/s390/s390.md: Define FPR*_REGNUM constants.
5507 Fix FPR2_REGNUM constant (18 -> 17).
5508 ("*trunc<BFP:mode><DFP_ALL:mode>2")
5509 ("*trunc<DFP_ALL:mode><BFP:mode>2")
5510 ("trunc<BFP:mode><DFP_ALL:mode>2")
5511 ("trunc<DFP_ALL:mode><BFP:mode>2")
5512 ("*extend<BFP:mode><DFP_ALL:mode>2")
5513 ("*extend<DFP_ALL:mode><BFP:mode>2")
5514 ("extend<BFP:mode><DFP_ALL:mode>2")
5515 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
5518 2013-07-08 Graham Stott <graham.stott@btinternet.com>
5520 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
5522 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5524 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
5525 and cfun_fpr_bit_p to cfun_fpr_save_p.
5526 (s390_frame_area, s390_register_info, s390_frame_info)
5527 (s390_emit_prologue, s390_emit_epilogue)
5528 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
5530 * config/s390/s390.h: Define *_REGNUM macros for floating point
5533 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
5535 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
5537 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
5539 PR rtl-optimization/57786
5540 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
5541 and break out of the loop when it is set to false.
5543 2013-07-08 Jakub Jelinek <jakub@redhat.com>
5546 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
5547 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
5548 (const_int 63)) 0)).
5549 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
5550 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
5551 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
5553 PR rtl-optimization/57829
5554 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
5555 mask bits outside of mode are just sign-extension from mode to HWI.
5557 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
5559 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
5560 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
5561 adjust_address instead of change_address to keep info about alignment.
5562 (emit_strmov): Remove.
5563 (emit_memmov): New function.
5564 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
5565 (expand_movmem_epilogue): Likewise and return updated rtx for
5567 (expand_constant_movmem_prologue): Likewise and return updated rtx for
5568 destination and source.
5569 (decide_alignment): Refactor, handle vector_loop.
5570 (ix86_expand_movmem): Likewise.
5571 (ix86_expand_setmem): Likewise.
5572 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
5574 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
5576 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
5577 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
5579 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
5581 * config/i386/sse.md (sse_movlhps): Change alternative 3
5582 of operand 2 to "m".
5584 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
5587 * config/i386/sse.md (iptr): New mode attribute.
5588 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
5589 (<sse>_vm<plusminus_insn><mode>3): Ditto.
5590 (<sse>_vmmul<mode>3): Ditto.
5591 (<sse>_vmdiv<mode>3): Ditto.
5592 (sse_vmrcpv4sf2): Ditto.
5593 (<sse>_vmsqrt<mode>2): Ditto.
5594 (sse_vmrsqrtv4sf2): Ditto.
5595 (<sse>_vm<code><mode>3): Ditto.
5596 (avx_vmcmp<mode>3): Ditto.
5597 (<sse>_vmmaskcmp<mode>3): Ditto.
5598 (<sse>_comi): Ditto.
5599 (<sse>_ucomi): Ditto.
5600 (*xop_vmfrcz_<mode>): Ditto.
5601 (*fmai_fmadd_<mode>): Ditto.
5602 (*fmai_fmsub_<mode>): Ditto.
5603 (*fmai_fnmadd_<mode>): Ditto.
5604 (*fmai_fnmsub_<mode>): Ditto.
5605 (*fma4i_vmfmadd_<mode>): Ditto.
5606 (*fma4i_vmfmsub_<mode>): Ditto.
5607 (*fma4i_vmfnmadd_<mode>): Ditto.
5608 (*fma4i_vmfnmsub_<mode>): Ditto.
5609 (*xop_vmfrcz_<mode>): Ditto.
5610 (sse_cvtps2pi): Ditto.
5611 (sse_cvttps2pi): Ditto.
5612 (sse_cvtss2si): Ditto.
5613 (sse_cvtss2si_2): Ditto.
5614 (sse_cvtss2siq_2): Ditto.
5615 (sse_cvttss2si): Ditto.
5616 (sse_cvttss2siq): Ditto.
5617 (sse_cvtsd2si): Ditto.
5618 (sse_cvtsd2si_2): Ditto.
5619 (sse_cvtsd2siq_2): Ditto.
5620 (sse_cvttsd2si): Ditto.
5621 (sse_cvttsd2siq): Ditto.
5622 (sse_cvtsd2ss): Ditto.
5623 (sse_cvtss2sd): Ditto.
5624 (avx2_pbroadcast<mode>): Ditto.
5625 (avx2_pbroadcast<mode>_1): Ditto.
5626 (*avx_vperm_broadcast_v4sf): Ditto.
5628 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5629 (sse_movlhps): Ditto.
5630 (sse_storehps): Ditto.
5631 (sse_loadhps): Ditto.
5632 (sse_storelps): Ditto.
5633 (sse_loadlps): Ditto.
5634 (*vec_concatv4sf): Ditto.
5635 (*vec_interleave_highv2df): Ditto.
5636 (*vec_interleave_lowv2df): Ditto.
5637 (*vec_extractv2df_1_sse): Ditto.
5638 (*vec_extractv2df_0_sse): Ditto.
5639 (sse2_storelpd): Ditto.
5640 (sse2_loadlpd): Ditto.
5641 (sse2_movsd): Ditto.
5642 (*vec_concatv4si): Ditto.
5643 (vec_concatv2di): Ditto.
5645 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5646 for Intel asm dialect.
5647 (mmx_punpcklwd): Ditto.
5648 (mmx_punpckldq): Ditto.
5650 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5651 for intel assembler dialect.
5653 2013-07-06 Jakub Jelinek <jakub@redhat.com>
5656 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5657 for BUILT_IN_C{LZ,LRSB}*.
5658 * tree.h (CASE_INT_FN): Add FN##IMAX case.
5659 * tree-vrp.c (extract_range_basic): Handle
5660 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
5661 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5662 fall thru to code calling set_value*.
5663 * builtins.c (expand_builtin): Remove *IMAX cases.
5664 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5665 if width is bigger than 2*HWI.
5667 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
5669 PR rtl-optimization/55342
5670 * lra-int.h (lra_subreg_reload_pseudos): New.
5671 * lra.c: Add undoing optional reloads to the block diagram.
5672 (lra_subreg_reload_pseudos): New.
5673 (lra_optional_reload_pseudos): Change comments.
5674 (lra): Init and clear lra_subreg_reload_pseudos. Clear
5675 lra_optional_reload_pseudos after undo transformations.
5676 * lra-assigns.c (pseudo_prefix_title): New.
5677 (lra_setup_reg_renumber): Use it.
5678 (spill_for): Ditto. Check subreg reload pseudos too.
5679 (assign_by_spills): Consider subreg reload pseudos too.
5680 * lra-constraints.c (simplify_operand_subreg): Use
5681 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5682 (curr_insn_transform): Recognize and do optional reloads.
5683 (undo_optional_reloads): New.
5684 (lra_undo_inheritance): Call undo_optional_reloads.
5686 2013-07-05 Thomas Quinot <quinot@adacore.com>
5688 * tree-complex.c (expand_complex_operations_1): Fix typo.
5690 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
5692 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5693 (tune_params): New member 'const vec_costs'.
5694 * config/aarch64/aarch64.c (generic_vector_cost): New.
5695 (generic_tunings): New member 'generic_vector_cost'.
5696 (aarch64_builtin_vectorization_cost): New.
5697 (aarch64_add_stmt_cost): New.
5698 (TARGET_VECTORIZE_ADD_STMT_COST): New.
5699 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5701 2013-07-03 Jakub Jelinek <jakub@redhat.com>
5704 * config/i386/predicates.md (vsib_address_operand): Disallow
5705 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5707 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
5710 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5711 expand_builtin_setjmp_receiver.
5712 (expand_label): Adjust, call expand_builtin_setjmp_receiver
5713 with NULL for the label parameter.
5714 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5715 the frame-pointer. Adjust comments.
5716 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5717 only if LABEL is non-NULL.
5719 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
5721 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5722 (ARM_ABI_AAPCS64): Ditto.
5724 (ARM_DEFAULT_ABI): Ditto.
5726 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
5728 * config/aarch64/aarch64-builtins.c
5729 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5730 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5732 * config/aarch64/aarch64-simd.md
5733 (aarch64_ld1<VALL:mode>): New.
5734 (aarch64_st1<VALL:mode>): Likewise.
5735 * config/aarch64/arm_neon.h
5736 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5738 2013-07-02 Sriraman Tallam <tmsriram@google.com>
5740 * config/i386/i386.c (gate_insert_vzeroupper): Check if
5742 (ix86_option_override_internal):Turn on all -mavx target flags by
5743 default as they are dependent on AVX anyway.
5745 2013-07-02 Cary Coutant <ccoutant@google.com>
5747 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5749 (loc_checksum_ordered): Likewise.
5750 (hash_loc_operands): Remove inline keyword.
5752 2013-07-02 Jakub Jelinek <jakub@redhat.com>
5754 PR tree-optimization/57741
5755 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5756 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5757 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5758 Allow REAL_CST step_exprs if flag_associative_math.
5759 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5761 2013-07-02 Ian Bolton <ian.bolton@arm.com>
5763 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5765 2013-07-02 Ian Bolton <ian.bolton@arm.com>
5767 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5769 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5771 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5773 (iorsi3_insn): Likewise.
5774 (arm_xorsi3): Likewise.
5776 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
5778 * arm.md (attribute "wtype"): Delete. Move attribute values from here
5780 (attribute "type"): ... here, and prefix with "wmmx_".
5781 (attribute "core_cycles"): Update for attribute changes.
5782 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5783 (tbcstv4hi): Likewise.
5784 (tbcstv2si): Likewise.
5785 (iwmmxt_iordi3): Likewise.
5786 (iwmmxt_xordi3): Likewise.
5787 (iwmmxt_anddi3): Likewise.
5788 (iwmmxt_nanddi3): Likewise.
5789 (iwmmxt_arm_movdi): Likewise.
5790 (iwmmxt_movsi_insn): Likewise.
5791 (mov<mode>_internal): Likewise.
5792 (and<mode>3_iwmmxt): Likewise.
5793 (ior<mode>3_iwmmxt): Likewise.
5794 (xor<mode>3_iwmmxt): Likewise.
5795 (add<mode>3_iwmmxt): Likewise.
5796 (ssaddv8qi3): Likewise.
5797 (ssaddv4hi3): Likewise.
5798 (ssaddv2si3): Likewise.
5799 (usaddv8qi3): Likewise.
5800 (usaddv4hi3): Likewise.
5801 (usaddv2si3): Likewise.
5802 (sub<mode>3_iwmmxt): Likewise.
5803 (sssubv8qi3): Likewise.
5804 (sssubv4hi3): Likewise.
5805 (sssubv2si3): Likewise.
5806 (ussubv8qi3): Likewise.
5807 (ussubv4hi3): Likewise.
5808 (ussubv2si3): Likewise.
5809 (mulv4hi3_iwmmxt): Likewise.
5810 (smulv4hi3_highpart): Likewise.
5811 (umulv4hi3_highpart): Likewise.
5812 (iwmmxt_wmacs): Likewise.
5813 (iwmmxt_wmacsz): Likewise.
5814 (iwmmxt_wmacu): Likewise.
5815 (iwmmxt_wmacuz): Likewise.
5816 (iwmmxt_clrdi): Likewise.
5817 (iwmmxt_clrv8qi): Likewise.
5818 (iwmmxt_clr4hi): Likewise.
5819 (iwmmxt_clr2si): Likewise.
5820 (iwmmxt_uavgrndv8qi3): Likewise.
5821 (iwmmxt_uavgrndv4hi3): Likewise.
5822 (iwmmxt_uavgv8qi3): Likewise.
5823 (iwmmxt_uavgv4hi3): Likewise.
5824 (iwmmxt_tinsrb): Likewise.
5825 (iwmmxt_tinsrh): Likewise.
5826 (iwmmxt_tinsrw): Likewise.
5827 (iwmmxt_textrmub): Likewise.
5828 (iwmmxt_textrmsb): Likewise.
5829 (iwmmxt_textrmuh): Likewise.
5830 (iwmmxt_textrmsh): Likewise.
5831 (iwmmxt_textrmw): Likewise.
5832 (iwmxxt_wshufh): Likewise.
5833 (eqv8qi3): Likewise.
5834 (eqv4hi3): Likewise.
5835 (eqv2si3): Likewise.
5836 (gtuv8qi3): Likewise.
5837 (gtuv4hi3): Likewise.
5838 (gtuv2si3): Likewise.
5839 (gtv8qi3): Likewise.
5840 (gtv4hi3): Likewise.
5841 (gtv2si3): Likewise.
5842 (smax<mode>3_iwmmxt): Likewise.
5843 (umax<mode>3_iwmmxt): Likewise.
5844 (smin<mode>3_iwmmxt): Likewise.
5845 (umin<mode>3_iwmmxt): Likewise.
5846 (iwmmxt_wpackhss): Likewise.
5847 (iwmmxt_wpackwss): Likewise.
5848 (iwmmxt_wpackdss): Likewise.
5849 (iwmmxt_wpackhus): Likewise.
5850 (iwmmxt_wpackwus): Likewise.
5851 (iwmmxt_wpackdus): Likewise.
5852 (iwmmxt_wunpckihb): Likewise.
5853 (iwmmxt_wunpckihh): Likewise.
5854 (iwmmxt_wunpckihw): Likewise.
5855 (iwmmxt_wunpckilb): Likewise.
5856 (iwmmxt_wunpckilh): Likewise.
5857 (iwmmxt_wunpckilw): Likewise.
5858 (iwmmxt_wunpckehub): Likewise.
5859 (iwmmxt_wunpckehuh): Likewise.
5860 (iwmmxt_wunpckehuw): Likewise.
5861 (iwmmxt_wunpckehsb): Likewise.
5862 (iwmmxt_wunpckehsh): Likewise.
5863 (iwmmxt_wunpckehsw): Likewise.
5864 (iwmmxt_wunpckelub): Likewise.
5865 (iwmmxt_wunpckeluh): Likewise.
5866 (iwmmxt_wunpckeluw): Likewise.
5867 (iwmmxt_wunpckelsb): Likewise.
5868 (iwmmxt_wunpckelsh): Likewise.
5869 (iwmmxt_wunpckelsw): Likewise.
5870 (ror<mode>3): Likewise.
5871 (ashr<mode>3_iwmmxt): Likewise.
5872 (lshr<mode>3_iwmmxt): Likewise.
5873 (ashl<mode>3_iwmmxt): Likewise.
5874 (ror<mode>3_di): Likewise.
5875 (ashr<mode>3_di): Likewise.
5876 (lshr<mode>3_di): Likewise.
5877 (ashl<mode>3_di): Likewise.
5878 (iwmmxt_wmadds): Likewise.
5879 (iwmmxt_wmaddu): Likewise.
5880 (iwmmxt_tmia): Likewise.
5881 (iwmmxt_tmiaph): Likewise.
5882 (iwmmxt_tmiabb): Likewise.
5883 (iwmmxt_tmiatb): Likewise.
5884 (iwmmxt_tmiabt): Likewise.
5885 (iwmmxt_tmiatt): Likewise.
5886 (iwmmxt_tmovmskb): Likewise.
5887 (iwmmxt_tmovmskh): Likewise.
5888 (iwmmxt_tmovmskw): Likewise.
5889 (iwmmxt_waccb): Likewise.
5890 (iwmmxt_wacch): Likewise.
5891 (iwmmxt_waccw): Likewise.
5892 (iwmmxt_waligni): Likewise.
5893 (iwmmxt_walignr): Likewise.
5894 (iwmmxt_walignr0): Likewise.
5895 (iwmmxt_walignr1): Likewise.
5896 (iwmmxt_walignr2): Likewise.
5897 (iwmmxt_walignr3): Likewise.
5898 (iwmmxt_wsadb): Likewise.
5899 (iwmmxt_wsadh): Likewise.
5900 (iwmmxt_wsadbz): Likewise.
5901 (iwmmxt_wsadhz): Likewise.
5902 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5903 (iwmmxt_wabsdiffb): Likewise.
5904 (iwmmxt_wabsdiffh): Likewise.
5905 (iwmmxt_wabsdiffw): Likewise.
5906 (iwmmxt_waddsubhx): Likewise
5907 (iwmmxt_wsubaddhx): Likewise.
5908 (addc<mode>3): Likewise.
5909 (iwmmxt_avg4): Likewise.
5910 (iwmmxt_avg4r): Likewise.
5911 (iwmmxt_wmaddsx): Likewise.
5912 (iwmmxt_wmaddux): Likewise.
5913 (iwmmxt_wmaddsn): Likewise.
5914 (iwmmxt_wmaddun): Likewise.
5915 (iwmmxt_wmulwsm): Likewise.
5916 (iwmmxt_wmulwum): Likewise.
5917 (iwmmxt_wmulsmr): Likewise.
5918 (iwmmxt_wmulumr): Likewise.
5919 (iwmmxt_wmulwsmr): Likewise.
5920 (iwmmxt_wmulwumr): Likewise.
5921 (iwmmxt_wmulwl): Likewise.
5922 (iwmmxt_wqmulm): Likewise.
5923 (iwmmxt_wqmulwm): Likewise.
5924 (iwmmxt_wqmulmr): Likewise.
5925 (iwmmxt_wqmulwmr): Likewise.
5926 (iwmmxt_waddbhusm): Likewise.
5927 (iwmmxt_waddbhusl): Likewise.
5928 (iwmmxt_wqmiabb): Likewise.
5929 (iwmmxt_wqmiabt): Likewise.
5930 (iwmmxt_wqmiatb): Likewise.
5931 (iwmmxt_wqmiatt): Likewise.
5932 (iwmmxt_wqmiabbn): Likewise.
5933 (iwmmxt_wqmiabtn): Likewise.
5934 (iwmmxt_wqmiatbn): Likewise.
5935 (iwmmxt_wqmiattn): Likewise.
5936 (iwmmxt_wmiabb): Likewise.
5937 (iwmmxt_wmiabt): Likewise.
5938 (iwmmxt_wmiatb): Likewise.
5939 (iwmmxt_wmiatt): Likewise.
5940 (iwmmxt_wmiabbn): Likewise.
5941 (iwmmxt_wmiabtn): Likewise.
5942 (iwmmxt_wmiatbn): Likewise.
5943 (iwmmxt_wmiattn): Likewise.
5944 (iwmmxt_wmiawbb): Likewise.
5945 (iwmmxt_wmiawbt): Likewise.
5946 (iwmmxt_wmiawtb): Likewise.
5947 (iwmmxt_wmiawtt): Likewise.
5948 (iwmmxt_wmiawbbn): Likewise.
5949 (iwmmxt_wmiawbtn): Likewise.
5950 (iwmmxt_wmiawtbn): Likewise.
5951 (iwmmxt_wmiawttn): Likewise.
5952 (iwmmxt_wmerge): Likewise.
5953 (iwmmxt_tandc<mode>3): Likewise.
5954 (iwmmxt_torc<mode>3): Likewise.
5955 (iwmmxt_torvsc<mode>3): Likewise.
5956 (iwmmxt_textrc<mode>3): Likewise.
5957 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5958 (wmmxt_pack): Likewise.
5959 (wmmxt_mult_c1): Likewise.
5960 (wmmxt_mult_c2): Likewise.
5961 (wmmxt_alu_c1): Likewise.
5962 (wmmxt_alu_c2): Likewise.
5963 (wmmxt_alu_c3): Likewise.
5964 (wmmxt_transfer_c1): Likewise.
5965 (wmmxt_transfer_c2): Likewise.
5966 (wmmxt_transfer_c3): Likewise.
5967 (marvell_f_iwmmxt_wstr): Likewise.
5968 (marvell_f_iwmmxt_wldr): Likewise.
5970 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
5972 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5974 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5977 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5978 * lra-constraints.c (need_for_split_p): Check call used hard regs
5979 living through calls.
5981 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5982 call used regs for call insn.
5984 2013-06-28 Jakub Jelinek <jakub@redhat.com>
5987 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5988 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5990 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
5992 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5994 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
5996 * lra-constraints.c (need_for_split_p): Check call used hard regs
5997 living through calls.
5999 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
6002 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
6003 to tie with any other modes. Eliminate Altivec vector mode tests,
6004 since these are a subset of ALTIVEC or VSX vector modes. Simplify
6005 code, to return 0 if testing MODE2 for a condition, if we've
6006 already tested MODE1 for the same condition.
6008 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
6010 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
6013 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
6015 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
6016 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
6018 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
6020 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
6022 (aarch64_symbolic_constant_p): Remove.
6023 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
6024 static. Fix line length and white space.
6025 (aarch64_symbolic_constant_p): Remove.
6026 * config/aarch64/predicates.md (aarch64_valid_symref):
6027 Use aarch64_classify_symbol_expression.
6029 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6031 * config/arm/constraints.md (Ts): New constraint.
6032 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
6034 (compare_scc): Use "Ts" constraint for operand 0.
6035 (ior_scc_scc): Likewise.
6036 (and_scc_scc): Likewise.
6037 (and_scc_scc_nodom): Likewise.
6038 (ior_scc_scc_cmp): Likewise for operand 7.
6039 (and_scc_scc_cmp): Likewise.
6040 * config/arm/thumb2.md (thumb2_movsi_insn):
6041 Add alternatives for 16-bit encodings.
6042 (thumb2_movhi_insn): Likewise.
6043 (thumb2_movsicc_insn): Likewise.
6044 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
6045 (thumb2_negscc): Use "Ts" constraint.
6046 Move mvn instruction outside cond_exec block.
6047 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
6048 for 16-bit encodings.
6050 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6052 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
6054 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
6055 (mulsi3subsi): Likewise.
6056 (mulsidi3adddi): Likewise.
6057 (mulsidi3_v6): Likewise.
6058 (umulsidi3_v6): Likewise.
6059 (umulsidi3adddi_v6): Likewise.
6060 (smulsi3_highpart_v6): Likewise.
6061 (umulsi3_highpart_v6): Likewise.
6062 (mulhisi3tb): Likewise.
6063 (mulhisi3bt): Likewise.
6064 (mulhisi3tt): Likewise.
6065 (maddhisi4): Likewise.
6066 (maddhisi4tb): Likewise.
6067 (maddhisi4tt): Likewise.
6068 (maddhidi4): Likewise.
6069 (maddhidi4tb): Likewise.
6070 (maddhidi4tt): Likewise.
6071 (zeroextractsi_compare0_scratch): Likewise.
6072 (insv_zero): Likewise.
6073 (insv_t2): Likewise.
6074 (anddi_notzesidi_di): Likewise.
6075 (anddi_notsesidi_di): Likewise.
6076 (andsi_notsi_si): Likewise.
6077 (iordi_zesidi_di): Likewise.
6078 (xordi_zesidi_di): Likewise.
6079 (andsi_iorsi3_notsi): Likewise.
6081 (smax_m1): Likewise.
6083 (not_shiftsi): Likewise.
6084 (unaligned_loadsi): Likewise.
6085 (unaligned_loadhis): Likewise.
6086 (unaligned_loadhiu): Likewise.
6087 (unaligned_storesi): Likewise.
6088 (unaligned_storehi): Likewise.
6089 (extv_reg): Likewise.
6090 (extzv_t2): Likewise.
6092 (udivsi3): Likewise.
6093 (arm_zero_extendhisi2addsi): Likewise.
6094 (arm_zero_extendqisi2addsi): Likewise.
6095 (compareqi_eq0): Likewise.
6096 (arm_extendhisi2_v6): Likewise.
6097 (arm_extendqisi2addsi): Likewise.
6098 (arm_movt): Likewise.
6099 (thumb2_ldrd): Likewise.
6100 (thumb2_ldrd_base): Likewise.
6101 (thumb2_ldrd_base_neg): Likewise.
6102 (thumb2_strd): Likewise.
6103 (thumb2_strd_base): Likewise.
6104 (thumb2_strd_base_neg): Likewise.
6105 (arm_negsi2): Add alternative for 16-bit encoding.
6106 (arm_one_cmplsi2): Likewise.
6108 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6110 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
6111 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
6112 (movdfcc): Likewise.
6113 * config/arm/vfp.md (*thumb2_movsf_vfp):
6114 Disable predication for arm_restrict_it.
6115 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
6116 (*thumb2_movdfcc_vfp): Likewise.
6117 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
6118 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
6119 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
6120 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
6121 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
6122 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
6123 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
6124 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
6125 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
6126 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
6127 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
6128 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
6129 Disable predication for arm_restrict_it.
6131 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
6133 * config/i386/bmiintrin.h (_bextr_u32): New.
6134 (_bextr_u64): Ditto.
6136 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
6138 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
6139 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
6140 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
6141 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
6142 * config/mips/n32-elf.h: ...this new file.
6144 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
6147 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
6148 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
6150 2013-06-27 Catherine Moore <clm@codesourcery.com>
6152 * config/mips/mips-tables.opt: Regenerate.
6153 * config/mips/mips-cpus.def: Add m14ke and m14kec.
6154 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
6155 * doc/invoke.texi: Add -m14kc.
6157 2013-06-27 Jakub Jelinek <jakub@redhat.com>
6160 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
6161 constraints of operand 1 and 2.
6164 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
6165 to match RTL canonicalization. Swap predicates and
6166 constraints of operand 1 and 2.
6168 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
6170 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
6171 Process OP_INOUT regs for splitting too.
6173 2013-06-27 Jakub Jelinek <jakub@redhat.com>
6175 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
6176 decl before the loop, initialize to NULL.
6177 (vectorizable_load): Initialize ptr_incr to NULL.
6179 2013-06-27 Martin Jambor <mjambor@suse.cz>
6182 * ipa-ref.h (ipa_maybe_record_reference): Declare.
6183 * ipa-ref.c (ipa_maybe_record_reference): New function.
6184 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
6185 * ipa-cp.c (create_specialized_node): Record potential references from
6187 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
6189 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
6191 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
6192 parameter 'mode' of type 'enum machine_mode mode'; change to pass
6193 'mode' to force_reg.
6194 (aarch64_add_offset): Update calls to aarch64_force_temporary.
6195 (aarch64_expand_mov_immediate): Likewise.
6197 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
6199 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
6200 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
6202 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6204 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
6205 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6206 (struct machine_function): Add tbegin_p.
6207 (s390_canonicalize_comparison): Fold CC mode compares to
6208 conditional jump if possible.
6209 (s390_emit_jump): Return the emitted jump.
6210 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
6211 Handle CCRAWmode compares.
6212 (s390_option_override): Default to -mhtm if available.
6213 (s390_reg_clobbered_rtx): Handle floating point regs as well.
6214 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
6215 FPRs instead of df_regs_ever_live_p.
6216 (s390_optimize_nonescaping_tx): New function.
6217 (s390_init_frame_layout): Extend clobbered_regs array to cover
6219 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
6220 (s390_expand_tbegin): New function.
6221 (enum s390_builtin): New enum definition.
6222 (code_for_builtin): New array definition.
6223 (s390_init_builtins): New function.
6224 (s390_expand_builtin): New function.
6225 (TARGET_INIT_BUILTINS): Define.
6226 (TARGET_EXPAND_BUILTIN): Define.
6227 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
6228 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
6229 (s390_alc_comparison): Likewise.
6230 * config/s390/s390-modes.def: Add CCRAWmode.
6231 * config/s390/s390.h (processor_flags): Add PF_TX.
6232 (TARGET_CPU_HTM): Define macro.
6233 (TARGET_HTM): Define macro.
6234 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
6235 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
6236 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
6237 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
6238 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
6239 (TBEGIN_MASK, TBEGINC_MASK): New constants.
6240 ("*cc_to_int"): Move up.
6241 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
6242 constants other than 0.
6243 ("*ccraw_to_int"): New insn and splitter definition.
6244 ("tbegin", "tbegin_nofloat", "tbegin_retry")
6245 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
6246 ("tx_assist"): New expander.
6247 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
6248 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
6249 * config/s390/s390.opt: Add -mhtm option.
6250 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
6251 * config/s390/htmxlintrin.h: New file.
6252 * config/s390/htmintrin.h: New file.
6253 * config/s390/s390intrin.h: New file.
6254 * doc/extend.texi: Document htm builtins.
6255 * config.gcc: Add the new header files to extra_headers.
6257 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
6259 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
6260 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
6262 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6263 Pat Haugen <pthaugen@us.ibm.com>
6264 Peter Bergner <bergner@vnet.ibm.com>
6266 * config/rs6000/power8.md: New.
6267 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
6268 setting for power8 entry.
6269 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
6270 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
6271 test for Power4/Power5 only.
6272 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
6274 (force_new_group): Adjust comment.
6275 * config/rs6000/rs6000.md: Include power8.md.
6277 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
6279 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
6280 * config/arm/arm-protos.h (arm_max_conditional_execute): New
6282 (tune_params): Update comment.
6283 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
6284 (arm_max_conditional_execute): New function.
6285 (thumb2_final_prescan_insn): Use max_insn_skipped and
6286 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
6288 2013-06-25 Jakub Jelinek <jakub@redhat.com>
6290 PR tree-optimization/57705
6291 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
6292 SSA_NAME step, provided that it is not defined inside the loop.
6293 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
6294 (get_initial_def_for_induction): Handle SSA_NAME IV step.
6296 2013-06-25 Martin Jambor <mjambor@suse.cz>
6299 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
6300 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
6302 (ipa_note_param_call): Initialize member_ptr flag.
6303 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
6304 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
6305 (ipa_write_indirect_edge_info): Stream member_ptr flag.
6306 (ipa_read_indirect_edge_info): Likewise.
6308 2013-06-25 Richard Biener <rguenther@suse.de>
6311 * passes.c (init_optimization_passes): Move pass_fold_builtins
6312 and pass_dce earlier with -Og.
6314 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
6316 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
6317 <BIT_FIELD_REF>: Remove trailing TAB.
6318 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
6321 2013-06-24 Martin Jambor <mjambor@suse.cz>
6323 PR tree-optimization/57358
6324 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
6325 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
6326 (ipa_analyze_params_uses): Generate pessimistic info when true.
6328 2013-06-24 Martin Jambor <mjambor@suse.cz>
6330 PR tree-optimization/57539
6331 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
6332 global.inlined_to of the new node to it. All callers changed.
6333 * ipa-inline-transform.c (clone_inlined_nodes): New variable
6334 inlining_into, pass it to cgraph_clone_node.
6335 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
6336 ipa_free_edge_args_substructures.
6337 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
6338 rdesc linked list. Do not assert rdesc edges have inlined caller.
6339 Assert we have found an rdesc in the rdesc list.
6341 2013-06-24 Richard Biener <rguenther@suse.de>
6343 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
6344 (pointer_set_lookup): Declare.
6345 (class pointer_map): New template class implementing a
6346 generic pointer to T map.
6347 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
6348 pointer_map<T>::contains, pointer_map<T>::insert,
6349 pointer_map<T>::traverse): New functions.
6350 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
6351 (pointer_set_lookup): New function.
6352 (pointer_set_contains): Use pointer_set_lookup.
6353 (pointer_set_insert): Likewise.
6354 (insert_aux): Remove.
6355 (struct pointer_map_t): Embed a pointer_set_t.
6356 (pointer_map_create): Adjust.
6357 (pointer_map_destroy): Likewise.
6358 (pointer_map_contains): Likewise.
6359 (pointer_map_insert): Likewise.
6360 (pointer_map_traverse): Likewise.
6361 * tree-streamer.h (struct streamer_tree_cache_d): Use a
6362 pointer_map<unsigned> instead of a pointer_map_t.
6363 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
6364 (streamer_tree_cache_lookup): Likewise.
6365 (streamer_tree_cache_create): Likewise.
6366 (streamer_tree_cache_delete): Likewise.
6367 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
6368 pointer_map<unsigned> instead of a pointer_map_t.
6369 (lto_init_tree_ref_encoder): Adjust.
6370 (lto_destroy_tree_ref_encoder): Likewise.
6371 * lto-section-out.c (lto_output_decl_index): Likewise.
6372 (lto_record_function_out_decl_state): Likewise.
6373 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
6375 2013-06-24 Richard Biener <rguenther@suse.de>
6377 PR tree-optimization/57488
6378 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
6380 2013-06-24 Alan Modra <amodra@gmail.com>
6382 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
6383 (gen_easy_altivec_constant): Likewise.
6384 * config/rs6000/predicates.md (easy_vector_constant_add_self,
6385 easy_vector_constant_msb): Likewise.
6387 2013-06-23 Jakub Jelinek <jakub@redhat.com>
6390 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
6391 add missing return true.
6393 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
6396 * config/sh/predicates.md (general_extend_operand): Invoke
6397 general_movsrc_operand for memory operands.
6398 (general_movsrc_operand): Allow reg+reg addressing, do not use
6399 general_operand for memory operands.
6401 2013-06-23 Sriraman Tallam <tmsriram@google.com>
6403 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
6404 when current target options does not apply.
6405 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
6406 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
6407 * config/i386/bmiintrin.h: Pass appropriate target
6408 attributes to header.
6409 * config/i386/mmintrin.h: Ditto.
6410 * config/i386/nmmintrin.h: Ditto.
6411 * config/i386/avx2intrin.h: Ditto.
6412 * config/i386/fxsrintrin.h: Ditto.
6413 * config/i386/tbmintrin.h: Ditto.
6414 * config/i386/xsaveintrin.h: Ditto.
6415 * config/i386/f16cintrin.h: Ditto.
6416 * config/i386/xtestintrin.h: Ditto.
6417 * config/i386/xsaveoptintrin.h: Ditto.
6418 * config/i386/bmi2intrin.h: Ditto.
6419 * config/i386/lzcntintrin.h: Ditto.
6420 * config/i386/smmintrin.h: Ditto.
6421 * config/i386/wmmintrin.h: Ditto.
6422 * config/i386/x86intrin.h: Remove all header include guards.
6423 * config/i386/prfchwintrin.h: Ditto.
6424 * config/i386/pmmintrin.h: Ditto.
6425 * config/i386/tmmintrin.h: Ditto.
6426 * config/i386/xmmintrin.h: Ditto.
6427 * config/i386/popcntintrin.h: Ditto.
6428 * config/i386/rdseedintrin.h: Ditto.
6429 * config/i386/ammintrin.h: Ditto.
6430 * config/i386/emmintrin.h: Ditto.
6431 * config/i386/immintrin.h: Remove all header include guards.
6432 * config/i386/fma4intrin.h: Ditto.
6433 * config/i386/lwpintrin.h: Ditto.
6434 * config/i386/xopintrin.h: Ditto.
6435 * config/i386/ia32intrin.h: Ditto.
6436 * config/i386/avxintrin.h: Ditto.
6437 * config/i386/rtmintrin.h: Ditto.
6438 * config/i386/fmaintrin.h: Ditto.
6439 * config/i386/mm3dnow.h: Ditto.
6441 2013-06-22 Sriraman Tallam <tmsriram@google.com>
6443 * common/config/i386/i386-common.c: Handle LZCNT.
6445 2013-06-22 Andi Kleen <ak@linux.intel.com>
6447 * doc/extend.texi: Use __atomic_store_n instead of
6448 __atomic_store in HLE example.
6450 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
6452 * config/sh/sh.c: Remove <cstdlib> workaround.
6454 2013-06-21 Andi Kleen <ak@linux.intel.com>
6456 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
6458 2013-06-21 Andi Kleen <ak@linux.intel.com>
6460 * doc/extend.texi: Document that __atomic_clear and
6461 __atomic_test_and_set should only be used with bool.
6463 2013-06-20 Jan Hubicka <jh@suse.cz>
6465 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
6467 * tree.c (decls_same_for_odr): New function.
6468 (same_for_edr): New function.
6469 (types_same_for_odr): New function.
6470 (get_binfo_at_offset): Use it.
6471 * tree.h (types_same_for_odr): Declare.
6473 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
6474 Jason Merrill <jason@redhat.com>
6476 * system.h: Include <cstdlib> as well as <stdlib.h>.
6478 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
6481 * config/i386/i386.c (construct_container): Report error if
6482 long double is used with disabled x87 float returns.
6484 2013-06-20 Jan Hubicka <jh@suse.cz>
6486 * lto-cgraph.c (input_symtab): Do not set cgraph state.
6488 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
6490 PR rtl-optimization/57425
6491 PR rtl-optimization/57569
6492 * alias.c (write_dependence_p): Remove parameters mem_mode and
6493 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
6494 Changed all callers.
6495 (canon_anti_dependence): Get comments and semantics in sync.
6496 Add parameter mem_canonicalized. Changed all callers.
6497 * rtl.h (canon_anti_dependence): Update prototype.
6499 2013-06-20 Richard Biener <rguenther@suse.de>
6501 * data-streamer-in.c (streamer_read_uhwi): Optimize single
6502 byte case, inline streamer_read_uchar and defer section
6505 2013-06-20 Richard Biener <rguenther@suse.de>
6507 PR tree-optimization/57584
6508 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
6509 SSA names into the expanded expression that take part in
6510 abnormal coalescing.
6512 2013-06-19 Sharad Singhai <singhai@google.com>
6514 * gcov.c (print_usage): Handle new option.
6515 (process_args): Ditto.
6516 (get_gcov_intermediate_filename): New function.
6517 (output_intermediate_file): New function.
6518 (output_gcov_file): New function
6519 (generate_results): Handle new option.
6520 (release_function): Relase demangled name.
6521 (read_graph_file): Handle demangled name.
6522 (output_lines): Ditto.
6523 * doc/gcov.texi: Document gcov intermediate format.
6525 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
6528 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6529 (lra_emit_add): Use the functions. Add comment about Y as an
6532 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
6535 * collect2.c (collect_atexit): New.
6536 (collect_exit): Delete.
6537 (main): Register collect_atexit with atexit.
6538 (collect_wait): Change collect_exit to exit.
6540 * collect2.h (collect_exit): Delete.
6541 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
6543 2013-06-19 Wei Mi <wmi@google.com>
6545 PR rtl-optimization/57518
6546 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
6547 if regno is used in paradoxical subreg.
6548 (update_equiv_regs): Check pdx_subregs[regno] before
6549 set a reg to be equivalent with a mem.
6551 2013-06-19 Matthias Klose <doko@ubuntu.com>
6554 * file-find.h (find_a_file): Add a mode parameter.
6555 * file-find.c (find_a_file): Likewise.
6556 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
6557 with X_OK for the executables.
6558 * collect2.c (main): Call find_a_file with X_OK.
6560 2013-06-19 Steve Ellcey <sellcey@mips.com>
6563 * config/mips/mips.md (casesi_internal_mips16_<mode>):
6564 Use NEXT_INSN instead of next_real_insn.
6566 2013-06-19 Jan Hubicka <jh@suse.cz>
6568 * cgraph.h (const_value_known_p): Replace by ...
6569 (ctor_for_folding): .. this one.
6570 * cgraphunit.c (process_function_and_variable_attributes): Use it.
6571 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
6572 * expr.c (expand_expr_real_1): Likewise.
6573 (string_constant): Likewise.
6574 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
6575 * ipa.c (process_references): Likewise.
6576 (symtab_remove_unreachable_nodes): Likewise.
6577 * ipa-inline-analysis.c (param_change_prob): Likewise.
6578 * gimple-fold.c (canonicalize_constructor_val): Likewise.
6579 (get_base_constructor): Likwise.
6580 * varpool.c (varpool_remove_node): Likewise.
6581 (varpool_remove_initializer): LIkewise.
6582 (dump_varpool_node): LIkwise.
6583 (const_value_known_p): Rewrite to ...
6584 (ctor_for_folding): ... this one.
6586 2013-06-19 Jakub Jelinek <jakub@redhat.com>
6589 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
6590 PERSONALITY in $PATH derived prefixes.
6592 2013-06-19 Jeff Law <law@redhat.com>
6594 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
6597 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6598 (simplify_bitwise_binary): Use it to simpify certain binary ops on
6601 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
6603 * config/arm/vfp.md: Move VFP instruction classification documentation
6605 * config/arm/arm.md: ... here. Update instruction classification
6608 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
6610 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6611 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6612 pattern. Use more efficient sequences on ARMv5 and Thumb2.
6614 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
6617 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6618 with NEXT_INSN. Use tablejump_p to check for jump table data
6621 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
6624 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6625 that now in C++ the value is correct per the C++ standards.
6627 2013-06-19 Richard Biener <rguenther@suse.de>
6629 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6632 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6635 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6638 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6639 with next_active_insn.
6641 2013-06-18 Sriraman Tallam <tmsriram@google.com>
6643 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6644 functions are inlined during failures to flag an error.
6645 * tree-inline.c (expand_call_inline): Allow the error to be flagged
6646 in early inline pass.
6648 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
6650 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6653 2013-06-18 Julian Brown <julian@codesourcery.com>
6655 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6656 Permit virtual register pre-reload if !strict.
6657 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6659 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6661 * config/arm/neon.md (movmisalign<mode>): Use
6662 neon_perm_struct_or_reg_operand instead of
6663 neon_struct_or_register_operand.
6664 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6665 neon_permissive_struct_operand instead of neon_struct_operand.
6666 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6667 neon_vector_mem_operand.
6668 * config/arm/predicates.md (neon_struct_operand): Adjust call to
6669 neon_vector_mem_operand.
6670 (neon_permissive_struct_operand): New.
6671 (neon_struct_or_register_operand): Rename to...
6672 (neon_perm_struct_or_reg_operand): This. Adjust call to
6673 neon_vector_mem_operand.
6675 2013-06-18 Richard Biener <rguenther@suse.de>
6677 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6678 * lto-streamer.h: Include pointer-set.h.
6679 (struct lto_decl_slot): Remove.
6680 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6681 Remove next_index entry.
6682 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6683 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6684 (lto_init_tree_ref_encoder): Adjust.
6685 (lto_destroy_tree_ref_encoder): Likewise.
6686 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6687 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6688 (lto_output_decl_index): Adjust.
6689 (lto_new_out_decl_state): Likewise.
6690 (lto_record_function_out_decl_state): Likewise.
6691 * lto-streamer-out.c (copy_function): Likewise.
6693 2013-06-18 Richard Biener <rguenther@suse.de>
6695 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6696 * cgraphunit.c: Include cfgloop.h.
6697 (init_lowered_empty_function): Initialize the loop tree.
6698 (assemble_thunk): Insert new BBs into loops.
6700 2013-06-18 Richard Biener <rguenther@suse.de>
6702 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6703 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6704 the map from cache entry to cache index optional.
6705 (streamer_tree_cache_replace_tree): Adjust accordingly.
6706 (streamer_tree_cache_append): Likewise.
6707 (streamer_tree_cache_delete): Likewise.
6708 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6709 streamer cache map from cache entry to cache index.
6710 * lto-streamer-out.c (create_output_block): Adjust.
6712 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
6714 * config/arm/arm.md (attribute "insn"): Move multiplication and
6715 division attributes to...
6716 (attribute "type"): ... here. Remove mult.
6717 (attribute "mul32"): New attribute.
6718 (attribute "mul64"): Add umaal.
6719 (*arm_mulsi3): Update attributes.
6720 (*arm_mulsi3_v6): Likewise.
6721 (*thumb_mulsi3): Likewise.
6722 (*thumb_mulsi3_v6): Likewise.
6723 (*mulsi3_compare0): Likewise.
6724 (*mulsi3_compare0_v6): Likewise.
6725 (*mulsi_compare0_scratch): Likewise.
6726 (*mulsi_compare0_scratch_v6): Likewise.
6727 (*mulsi3addsi): Likewise.
6728 (*mulsi3addsi_v6): Likewise.
6729 (*mulsi3addsi_compare0): Likewise.
6730 (*mulsi3addsi_compare0_v6): Likewise.
6731 (*mulsi3addsi_compare0_scratch): Likewise.
6732 (*mulsi3addsi_compare0_scratch_v6): Likewise.
6733 (*mulsi3subsi): Likewise.
6734 (*mulsidi3adddi): Likewise.
6735 (*mulsi3addsi_v6): Likewise.
6736 (*mulsidi3adddi_v6): Likewise.
6737 (*mulsidi3_nov6): Likewise.
6738 (*mulsidi3_v6): Likewise.
6739 (*umulsidi3_nov6): Likewise.
6740 (*umulsidi3_v6): Likewise.
6741 (*umulsidi3adddi): Likewise.
6742 (*umulsidi3adddi_v6): Likewise.
6743 (*smulsi3_highpart_nov6): Likewise.
6744 (*smulsi3_highpart_v6): Likewise.
6745 (*umulsi3_highpart_nov6): Likewise.
6746 (*umulsi3_highpart_v6): Likewise.
6747 (mulhisi3): Likewise.
6748 (*mulhisi3tb): Likewise.
6749 (*mulhisi3bt): Likewise.
6750 (*mulhisi3tt): Likewise.
6751 (maddhisi4): Likewise.
6752 (*maddhisi4tb): Likewise.
6753 (*maddhisi4tt): Likewise.
6754 (maddhidi4): Likewise.
6755 (*maddhidi4tb): Likewise.
6756 (*maddhidi4tt): Likewise.
6758 (udivsi3): Likewise.
6759 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6760 (thumb2_mulsi_short_compare0): Likewise.
6761 (thumb2_mulsi_short_compare0_scratch): Likewise.
6762 * config/arm/arm1020e.md (1020mult1): Update attribute change.
6763 (1020mult2): Likewise.
6764 (1020mult3): Likewise.
6765 (1020mult4): Likewise.
6766 (1020mult5): Likewise.
6767 (1020mult6): Likewise.
6768 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6770 (cortex_a15_mult64): Likewise.
6771 (cortex_a15_sdiv): Likewise.
6772 (cortex_a15_udiv): Likewise.
6773 * config/arm/arm1026ejs.md (mult1): Update attribute change.
6779 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6780 (pj4_ir_div): Likewise.
6781 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6782 (11_mult2): Likewise.
6783 (11_mult3): Likewise.
6784 (11_mult4): Likewise.
6785 (11_mult5): Likewise.
6786 (11_mult6): Likewise.
6787 (11_mult7): Likewise.
6788 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6789 (cortex_a8_mla): Likewise.
6790 (cortex_a8_mull): Likewise.
6791 (cortex_a8_smulwy): Likewise.
6792 (cortex_a8_smlald): Likewise.
6793 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6794 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6795 (cortex_r4_mul_3): Likewise.
6796 (cortex_r4_mla_4): Likewise.
6797 (cortex_r4_mla_3): Likewise.
6798 (cortex_r4_smlald): Likewise.
6799 (cortex_r4_mull): Likewise.
6800 (cortex_r4_sdiv): Likewise.
6801 (cortex_r4_udiv): Likewise.
6802 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6803 (cortex_a7_idiv): Likewise.
6804 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6805 (9_mult2): Likewise.
6806 (9_mult3): Likewise.
6807 (9_mult4): Likewise.
6808 (9_mult5): Likewise.
6809 (9_mult6): Likewise.
6810 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6811 (cortex_a53_sdiv): Likewise.
6812 (cortex_a53_udiv): Likewise.
6813 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6814 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6815 (mp626_mult2): Likewise.
6816 (mp626_mult3): Likewise.
6817 (mp626_mult4): Likewise.
6818 * config/arm/fa526.md (526_mult1): Update attribute change.
6819 (526_mult2): Likewise.
6820 * config/arm/arm-generic.md (mult): Update attribute change.
6821 (mult_ldsched_strongarm): Likewise.
6822 (mult_ldsched): Likewise.
6823 (multi_cycle): Likewise.
6824 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6825 * config/arm/fa606te.md (606te_mult1): Update attribute change.
6826 (606te_mult2): Likewise.
6827 (606te_mult3): Likewise.
6828 (606te_mult4): Likewise.
6829 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6830 (cortex_a9_mac16): Likewise.
6831 (cortex_a9_multiply): Likewise.
6832 (cortex_a9_mac): Likewise.
6833 (cortex_a9_multiply_long): Likewise.
6834 * config/arm/fa626te.md (626te_mult1): Update attribute change.
6835 (626te_mult2): Likewise.
6836 (626te_mult3): Likewise.
6837 (626te_mult4): Likewise.
6839 2013-06-18 Richard Biener <rguenther@suse.de>
6842 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6844 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6847 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6848 with next_active_insn.
6850 2013-06-18 Alan Modra <amodra@gmail.com>
6852 * config/rs6000/rs6000.h (enum data_align): New.
6853 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6854 (DATA_ABI_ALIGNMENT): Define.
6855 (CONSTANT_ALIGNMENT): Correct comment.
6856 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6857 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6859 2013-06-17 David Malcolm <dmalcolm@redhat.com>
6861 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6862 ATTRIBUTE_UNUSED marking.
6864 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
6866 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6867 alternative and update.
6868 (aarch64_dup_lanedi): Delete.
6869 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6870 * config/aarch64/aarch64-simd-builtins.def: Update.
6872 2013-06-17 Richard Biener <rguenther@suse.de>
6874 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6875 (lto_input_scc): Declare.
6876 (lto_input_tree_1): Likewise.
6877 (struct lto_stats_d): Add num_tree_bodies_output and
6878 num_pickle_refs_output.
6879 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6880 (lto_read_tree_1): Split out from ...
6881 (lto_read_tree): ... this.
6882 (lto_input_scc): New function.
6883 (lto_input_tree_1): Split out from ...
6884 (lto_input_tree): ... this. Handle LTO_tree_scc.
6885 (lto_data_in_create): Create the streamer cache without hashes.
6886 * lto-streamer-out.c (create_output_block): Create the streamer
6887 cache with hashes when not doing WPA.
6888 (lto_write_tree_1): Split out from ...
6889 (lto_write_tree): ... this.
6890 (get_symbol_initial_value): New function.
6891 (lto_output_tree_1): Split out from ...
6892 (lto_output_tree): ... this. Write trees as series of SCCs
6893 using a DFS walk via DFS_write_tree.
6894 (struct sccs, struct scc_entry): New types.
6895 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6896 (DFS_write_tree_body): New function.
6897 (DFS_write_tree): Likewise.
6898 (hash_tree): Likewise.
6899 (scc_entry_compare): Likewise.
6900 (hash_scc): Likewise.
6901 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6902 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6903 TREE_CHAIN as regular reference.
6904 (streamer_read_integer_cst): Remove.
6905 (streamer_get_pickled_tree): Adjust.
6906 * tree-streamer-out.c (streamer_write_chain): Disable streaming
6907 of DECL_EXTERNALs in BLOCK_VARS for now.
6908 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6910 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6911 Add hash value argument and record that if hashes are recorded
6913 (streamer_tree_cache_insert_1): Adjust.
6914 (streamer_tree_cache_insert): Likewise.
6915 (streamer_tree_cache_insert_at): Rename to ...
6916 (streamer_tree_cache_replace_tree): ... this and adjust.
6917 (streamer_tree_cache_append): Adjust.
6918 (record_common_node): Likewise.
6919 (streamer_tree_cache_create): Add argument whether to
6920 record hash values together with trees.
6921 (streamer_tree_cache_delete): Adjust.
6922 * tree-streamer.h (struct streamer_tree_cache_d): Add
6924 (streamer_read_integer_cst): Remove.
6925 (streamer_tree_cache_insert): Adjust.
6926 (streamer_tree_cache_append): Likewise.
6927 (streamer_tree_cache_insert_at): Rename to ...
6928 (streamer_tree_cache_replace_tree): ... this and adjust.
6929 (streamer_tree_cache_create): Add argument whether to record hashes.
6930 (streamer_tree_cache_get): Rename to ...
6931 (streamer_tree_cache_get_tree): ... this.
6932 (streamer_tree_cache_get_hash): New function.
6933 * tree.c (cache_integer_cst): New function.
6934 * tree.h (cache_integer_cst): Declare.
6935 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6936 * lto-symtab.c (lto_varpool_replace_node): Only release
6937 DECL_INITIAL of non-prevailing decls.
6938 * varpool.c (varpool_remove_initializer): Do not release
6939 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6941 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
6943 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6944 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6945 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6946 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6947 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6948 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6949 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6950 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6951 instead of TARGET_64BIT.
6952 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6953 Require ISA_HAS_<D>DIV.
6955 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
6957 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6958 (mips*-*-linux*): Move default with_llsc setting to where other
6960 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6962 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6963 Likewise. Remove default with_tune setting. Move default float
6964 setting to its own block. Handle with_llsc in the same block as above.
6966 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
6968 PR rtl-optimization/57425
6969 PR rtl-optimization/57569
6970 * alias.c (write_dependence_p): Add new parameters mem_mode,
6971 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
6972 Changed all callers.
6973 (canon_anti_dependence): New function.
6974 * cse.c (check_dependence): Use canon_anti_dependence.
6975 * cselib.c (cselib_invalidate_mem): Likewise.
6976 * rtl.h (canon_anti_dependence): Declare.
6978 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
6980 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6981 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6982 ".set mips3" for 64-bit targets.
6984 2013-06-15 Dehao Chen <dehao@google.com>
6986 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6987 * gimple-low.c (gimple_check_call_matching_types): Likewise.
6988 (gimple_check_call_args): Likewise.
6989 * value-prof.c (check_ic_target): Likewise.
6990 * ipa-inline.c (early_inliner): Likewise.
6991 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6992 * cgraph.c (cgraph_create_edge_1): Likewise.
6993 (cgraph_make_edge_direct): Likewise.
6995 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6998 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6999 rs6000_output_move_128bit to handle emitting quad memory
7000 operations. Set attribute length to 8 bytes.
7002 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
7004 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
7006 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
7007 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
7008 (aarch64_<su>mlsl<mode>): Likewise.
7010 2013-06-14 Mike Stump <mikestump@comcast.net>
7012 * Makefile.in (TARGET_H): Add insn-codes.h.
7014 2013-06-14 Alan Modra <amodra@gmail.com>
7018 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
7019 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
7020 bitfield expansion when EXPAND_MEMORY.
7021 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
7023 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7025 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
7026 test for clearing quad memory on 32-bit later.
7028 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
7030 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
7031 (fold_negate_expr): Likewise.
7032 (fold_real_zero_addition_p): Handle vectors.
7033 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
7035 2013-06-14 Alan Modra <amodra@gmail.com>
7037 * varasm.c (force_const_mem): Revert 2013-06-07 change.
7039 2013-06-13 Jan Hubicka <jh@suse.cz>
7041 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
7042 Local comdats are not externally visible.
7043 * symtab.c (dump_symtab_base): Dump externally visible.
7044 (verify_symtab_base): Verify back links in the symtab hash.
7046 2013-06-13 Bin Cheng <bin.cheng@arm.com>
7048 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
7049 CONVERT_EXPR as equal nodes.
7051 2013-06-13 Bin Cheng <bin.cheng@arm.com>
7053 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
7055 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
7057 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
7058 Generalize to complex and vector.
7059 * tree.c (build_all_ones_cst): New function.
7060 * tree.h (build_all_ones_cst): Declare it.
7062 2013-06-13 Alan Modra <amodra@gmail.com>
7064 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
7065 * config/rs6000/rs6000.md (signbittf2): New insn.
7066 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
7067 (abstf2_internal, cmptf_internal2): Likewise.
7068 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
7070 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7071 Pat Haugen <pthaugen@us.ibm.com>
7072 Peter Bergner <bergner@vnet.ibm.com>
7074 * config/rs6000/rs6000.c (emit_load_locked): Add support for
7075 power8 byte, half-word, and quad-word atomic instructions.
7076 (emit_store_conditional): Likewise.
7077 (rs6000_expand_atomic_compare_and_swap): Likewise.
7078 (rs6000_expand_atomic_op): Likewise.
7080 * config/rs6000/sync.md (larx): Add new modes for power8.
7082 (AINT): New mode iterator to include TImode as well as normal
7083 integer modes on power8.
7084 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
7085 that VSX registers are not considered. Use AINT mode iterator
7086 instead of INT1 to allow inclusion of quad word atomic operations
7088 (load_locked<mode>): Likewise.
7089 (store_conditional<mode>): Likewise.
7090 (atomic_compare_and_swap<mode>): Likewise.
7091 (atomic_exchange<mode>): Likewise.
7092 (atomic_nand<mode>): Likewise.
7093 (atomic_fetch_<fetchop_name><mode>): Likewise.
7094 (atomic_nand_fetch<mode>): Likewise.
7095 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
7097 (ATOMIC): On power8, add QImode, HImode modes.
7098 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
7099 modes that promote to SImode.
7100 (load_lockedti): Convert TImode arguments to PTImode, so that we
7101 get a guaranteed even/odd register pair.
7102 (load_lockedpti): Likewise.
7103 (store_conditionalti): Likewise.
7104 (store_conditionalpti): Likewise.
7106 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
7107 atomic load/store instructions.
7110 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
7112 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
7114 (insn_count): New attribute, with most cases extracted from...
7115 (length): ...here. Redefine most cases in terms of insn_count.
7116 (single_insn): Delete.
7117 (can_delay): Use insn_count to check for single instructions.
7118 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
7119 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
7120 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
7121 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
7122 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
7123 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
7124 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
7125 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
7126 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
7127 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
7128 rather than "length".
7129 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
7130 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
7131 Use "insn_count" rather than "length".
7132 * config/mips/mips-dsp.md
7133 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
7134 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
7137 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
7139 PR tree-optimization/57361
7140 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
7142 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
7144 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
7146 (aarch64_simd_combine<mode>): New instruction expansion.
7147 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
7149 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
7150 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
7152 2013-06-12 Jan Hubicka <jh@suse.cz>
7154 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
7155 decl has when in streaming stage.
7156 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
7157 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
7159 2013-06-12 Roland Stigge <stigge@antcom.de>
7162 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
7164 2013-06-12 Jakub Jelinek <jakub@redhat.com>
7166 PR tree-optimization/57537
7167 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
7168 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
7170 2013-06-12 Richard Biener <rguenther@suse.de>
7172 * data-streamer.h (streamer_write_char_stream): CSE
7173 obs->current_pointer.
7174 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
7175 streamer_write_char_stream manually and optimize the resulting loop.
7176 (streamer_write_hwi_stream): Likewise.
7178 2013-06-12 Jan Hubicka <jh@suse.cz>
7180 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
7181 * cgraph.h (varpool_create_empty_node): Declare.
7182 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
7184 * symtab.c (symtab_unregister_node): Be lax about missin entries
7186 (symtab_get_node): Update comment.
7187 * varpool.c (varpool_create_empty_node): Break out from ...
7188 (varpool_node_for_decl): ... here.
7189 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
7191 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
7193 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
7194 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
7195 part. Use straight-line flow at the end.
7196 <COMPONENT_REF>: Remove superfluous else.
7197 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
7199 2013-06-12 Jakub Jelinek <jakub@redhat.com>
7202 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
7203 target hook even for !TREE_PUBLIC decls. If no resolution info
7204 is available, return false for common and external decls.
7206 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
7208 * config/rl78/constraints.md (U): New constraint.
7209 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
7212 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
7215 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
7216 to allow returning address to AT_PLATFORM name.
7218 2013-06-11 Jan Hubicka <jh@suse.cz>
7220 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
7221 * cgraph.h (symtab_node_base): Add weakref flag.
7222 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
7223 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
7224 (output_weakrefs): Use weakref flag.
7225 * fold-const.c (simple_operand_p): Handle WEAK.
7226 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
7227 * ipa.c (varpool_externally_visible_p): Drop weakref.
7228 (function_and_variable_visibility): Update comment; fix weakref
7229 sanity checks; do not clear DECL_WEAK on them.
7230 * lto-cgraph.c (lto_output_node): update.
7231 (lto_output_varpool_node): Update.
7232 (input_overwrite_node): Update.
7233 (input_node): Update.
7234 (input_varpool_node): Update.
7235 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
7236 (lto_symtab_merge_symbols): Add sanity check.
7237 (lto_symtab_prevailing_decl): Do not special case weakrefs.
7238 * passes.c (rest_of_decl_compilation): Set static flag, too.
7239 * symtab.c (dump_symtab_base): Dump weakref.
7240 (verify_symtab_base): Sanity check weakrefs.
7241 (symtab_make_decl_local): Remove duplicated code.
7242 (symtab_alias_ultimate_target): Simplify.
7243 * varpool.c (varpool_create_variable_alias): Set weakref flag.
7245 2013-06-11 DJ Delorie <dj@redhat.com>
7247 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
7248 (rl78_unwind_word_mode): New.
7250 2013-06-11 David Malcolm <dmalcolm@redhat.com>
7252 * final.c (debug_prefix_maps): Make static.
7254 2013-06-11 David Malcolm <dmalcolm@redhat.com>
7256 * function.c (initial_trampoline): Remove stray copy.
7258 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
7260 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
7262 2013-06-11 Martin Jambor <mjambor@suse.cz>
7264 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
7265 within bounds at the beginning of the function.
7267 2013-06-11 Alan Modra <amodra@gmail.com>
7269 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
7271 (get_named_section): Don't NULL !DECL_P decl.
7273 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
7275 * doc/invoke.texi (core-avx2): Document.
7277 (atom): Updated with MOVBE.
7279 2013-06-11 Richard Biener <rguenther@suse.de>
7281 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
7283 2013-06-11 Anton Blanchard <anton@samba.org>
7285 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
7286 correct shift value in little-endian mode.
7288 2013-06-11 Jakub Jelinek <jakub@redhat.com>
7291 * varasm.c (get_variable_align): Move #endif to the right place.
7293 2013-06-10 Cary Coutant <ccoutant@google.com>
7295 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
7296 for hash so that hash table traversal order is deterministic.
7298 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7299 Pat Haugen <pthaugen@us.ibm.com>
7300 Peter Bergner <bergner@vnet.ibm.com>
7302 * config/rs6000/vector.md (GPR move splitter): Do not split moves
7303 of vectors in GPRS if they are direct moves or quad word load or
7306 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
7308 (direct_move_p): Likewise.
7309 (quad_load_store_p): Likewise.
7311 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
7312 classes into bins based on the physical register type.
7313 (reg_class_to_reg_type): Likewise.
7314 (IS_STD_REG_TYPE): Likewise.
7315 (IS_FP_VECT_REG_TYPE): Likewise.
7316 (reload_fpr_gpr): Arrays to determine what insn to use if we can
7317 use direct move instructions.
7318 (reload_gpr_vsx): Likewise.
7319 (reload_vsx_gpr): Likewise.
7320 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
7321 information that is a simplification of register classes. Also
7322 precalculate direct move reload helpers.
7323 (direct_move_p): New function to return true if the operation can
7324 be done as a direct move instruciton.
7325 (quad_load_store_p): New function to return true if the operation
7326 is a quad memory operation.
7327 (rs6000_legitimize_address): If quad memory, only allow register
7328 indirect for TImode addresses.
7329 (rs6000_legitimate_address_p): Likewise.
7330 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
7331 (rs6000_reload_register_type): Likewise.
7332 (register_to_reg_type): Return register type.
7333 (rs6000_secondary_reload_simple_move): New helper function for
7334 secondary reload and secondary memory needed to identify anything
7335 that is a simple move, and does not need reloading.
7336 (rs6000_secondary_reload_direct_move): New helper function for
7337 secondary reload to identify cases that can be done with several
7338 instructions via the direct move instructions.
7339 (rs6000_secondary_reload_move): New helper function for secondary
7340 reload to identify moves between register types that can be done.
7341 (rs6000_secondary_reload): Add support for quad memory operations
7342 and for direct move.
7343 (rs6000_secondary_memory_needed): Likewise.
7344 (rs6000_debug_secondary_memory_needed): Change argument names.
7345 (rs6000_output_move_128bit): New function to return the move to
7346 use for 128-bit moves, including knowing about the various
7347 limitations of quad memory operations.
7349 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
7350 memory operations. call rs6000_output_move_128bit for the actual
7351 instruciton(s) to generate.
7352 (vsx_movti_64bit): Likewise.
7354 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
7355 (UNSPEC_P8V_MTVSRWZ): Likewise.
7356 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
7357 (UNSPEC_P8V_MTVSRD): Likewise.
7358 (UNSPEC_P8V_XXPERMDI): Likewise.
7359 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
7360 (UNSPEC_FUSION_GPR): Likewise.
7361 (FMOVE128_GPR): New iterator for direct move.
7362 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
7365 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
7366 loads and direct move instructions.
7367 (zero_extendsidi2_lfiwzx): Likewise.
7368 (extendsidi2_lfiwax): Likewise.
7369 (extendsidi2_nocell): Likewise.
7370 (floatsi<mode>2_lfiwax): Likewise.
7372 (floatunssi<mode>2_lfiwzx): Likewise.
7374 (fix_trunc<mode>_stfiwx): Likewise.
7375 (fixuns_trunc<mode>_stfiwx): Likewise.
7376 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
7377 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
7378 (parity<mode>2_cmpb): Set length/type attr.
7379 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
7380 for 'mr.' to fast_compare.
7381 (bpermd_<mode>): Change type attr to popcnt.
7382 (p8_fmrgow_<mode>): New insns for power8 direct move support.
7383 (p8_mtvsrwz_1): Likewise.
7384 (p8_mtvsrwz_2): Likewise.
7385 (reload_fpr_from_gpr<mode>): Likewise.
7386 (p8_mtvsrd_1): Likewise.
7387 (p8_mtvsrd_2): Likewise.
7388 (p8_xxpermdi_<mode>): Likewise.
7389 (reload_vsx_from_gpr<mode>): Likewise.
7390 (reload_vsx_from_gprsf): Likewise.
7391 (p8_mfvsrd_3_<mode>): LIkewise.
7392 (reload_gpr_from_vsx<mode>): Likewise.
7393 (reload_gpr_from_vsxsf): Likewise.
7394 (p8_mfvsrd_4_disf): Likewise.
7395 (multi-word GPR splits): Do not split direct moves or quad memory
7398 2013-06-10 David Malcolm <dmalcolm@redhat.com>
7400 * tree-into-ssa.c (interesting_blocks): Make static.
7402 2013-06-10 Jakub Jelinek <jakub@redhat.com>
7405 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
7406 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
7407 Use DATA_ABI_ALIGNMENT for that case instead if defined.
7408 (get_variable_align): New function.
7409 (get_variable_section, emit_bss, emit_common,
7410 assemble_variable_contents, place_block_symbol): Use
7411 get_variable_align instead of DECL_ALIGN.
7412 (assemble_noswitch_variable): Add align argument, use it
7413 instead of DECL_ALIGN.
7414 (assemble_variable): Adjust caller. Use get_variable_align
7415 instead of DECL_ALIGN.
7416 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
7418 (DATA_ABI_ALIGNMENT): Define.
7419 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
7420 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
7421 opt is false, only return the psABI mandated alignment increase.
7422 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
7423 (DATA_ABI_ALIGNMENT): ... this.
7424 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
7425 (DATA_ABI_ALIGNMENT): ... this.
7426 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
7427 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
7428 (DATA_ABI_ALIGNMENT): ... this.
7429 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
7430 * doc/tm.texi: Regenerated.
7432 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
7434 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
7435 cmp_code to construct REG_EQUAL note.
7437 2013-06-09 Jakub Jelinek <jakub@redhat.com>
7440 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
7441 that operands[2] doesn't overlap with operands[0].
7443 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
7444 Jan Hubicka <jh@suse.cz>
7446 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
7447 hack to mark symbols as used.
7449 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
7451 PR rtl-optimization/57559
7452 * lra-constraints.c (process_alt_operands): Don't discourage
7453 memory with known offset for offsetable memory constraint.
7454 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
7456 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
7458 * varasm.c (struct oc_local_state): Reorder fields.
7459 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
7460 and adjust accordingly.
7461 (output_constructor): Reorder initialization code and adjust call to
7462 output_constructor_bitfield.
7464 2013-06-07 Jan Hubicka <jh@suse.cz>
7466 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
7468 2013-06-07 David Malcolm <dmalcolm@redhat.com>
7470 * tree-object-size.c (unknown): Make const.
7472 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
7474 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
7475 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
7476 for last alternative in the cpu_facility attribute.
7478 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7481 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
7482 (xordi3): Change operand 2 constraint to arm_xordi_operand.
7483 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
7484 * config/arm/constraints.md (Dg): New constraint.
7485 * config/arm/neon.md (xordi3_neon): Remove.
7486 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
7487 * config/arm/predicates.md (arm_xordi_operand): New predicate.
7489 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7491 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
7492 Clean up alternatives.
7494 2013-06-07 Alan Modra <amodra@gmail.com>
7496 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
7499 2013-06-07 Alan Modra <amodra@gmail.com>
7501 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
7503 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7505 * config/arm/constraints.md (Df): New constraint.
7506 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
7507 Correct length attribute for last two alternatives.
7509 2013-06-07 Alan Modra <amodra@gmail.com>
7511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7512 override user -mfp-in-toc.
7513 (offsettable_ok_by_alignment): Consider just the current access
7514 rather than the whole object, unless BLKmode. Handle
7515 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
7516 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
7517 for -mcmodel=medium.
7518 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
7519 override user -mfp-in-toc or -msum-in-toc. Default to
7520 -mno-fp-in-toc for -mcmodel=medium.
7522 2013-06-06 DJ Delorie <dj@redhat.com>
7524 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
7525 TARGET_VALID_POINTER_MODE.
7527 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
7528 Pat Haugen <pthaugen@us.ibm.com>
7529 Peter Bergner <bergner@vnet.ibm.com>
7531 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7532 Document new power8 builtins.
7534 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
7535 condition code register, to allow 128-bit logical operations to be
7536 done in the VSX or GPR registers.
7537 (nor<mode>3): Use the canonical form for nor.
7538 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
7539 vclz*, and vpopcnt* vector instructions.
7540 (nand<mode>3): Likewise.
7541 (orc<mode>3): Likewise.
7542 (clz<mode>2): LIkewise.
7543 (popcount<mode>2): Likewise.
7545 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
7546 that only the GPRs are recognized.
7548 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7549 support for new power8 builtins.
7551 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
7553 (xscvdpspn): Likewise.
7559 (vpopcnt): Likewise.
7560 (vpopcntb): Likewise.
7561 (vpopcnth): Likewise.
7562 (vpopcntw): Likewise.
7563 (vpopcntd): Likewise.
7568 (eqv_v16qi3): Likewise.
7569 (eqv_v8hi3): Likewise.
7570 (eqv_v4si3): Likewise.
7571 (eqv_v2di3): Likewise.
7572 (eqv_v4sf3): Likewise.
7573 (eqv_v2df3): Likewise.
7575 (nand_v16qi3): Likewise.
7576 (nand_v8hi3): Likewise.
7577 (nand_v4si3): Likewise.
7578 (nand_v2di3): Likewise.
7579 (nand_v4sf3): Likewise.
7580 (nand_v2df3): Likewise.
7582 (orc_v16qi3): Likewise.
7583 (orc_v8hi3): Likewise.
7584 (orc_v4si3): Likewise.
7585 (orc_v2di3): Likewise.
7586 (orc_v4sf3): Likewise.
7587 (orc_v2df3): Likewise.
7589 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
7590 allow power8 quad mode in 64-bit.
7591 (rs6000_builtin_vectorized_function): Add support to vectorize
7592 ISA 2.07 count leading zeros, population count builtins.
7593 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
7594 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
7595 (builtin_function_type): Add vgbbd builtin function which takes an
7597 (altivec_expand_vec_perm_const): Add support for new power8 merge
7600 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7601 that does not include TImdoe for use with 32-bit.
7602 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7604 (UNSPEC_VSX_CVDPSPN): Likewise.
7605 (vsx_xscvdpspn): Likewise.
7606 (vsx_xscvspdpn): Likewise.
7607 (vsx_xscvdpspn_scalar): Likewise.
7608 (vsx_xscvspdpn_directmove): Likewise.
7609 (vsx_and<mode>3): Split logical operations into 32-bit and
7610 64-bit. Add support to do logical operations on TImode as well as
7611 VSX vector types. Allow logical operations to be done in either
7612 VSX registers or in general purpose registers in 64-bit mode. Add
7613 splitters if GPRs were used. For AND, add clobber of CCmode to
7614 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
7616 (vsx_and<mode>3_32bit): Likewise.
7617 (vsx_and<mode>3_64bit): Likewise.
7618 (vsx_ior<mode>3): Likewise.
7619 (vsx_ior<mode>3_32bit): Likewise.
7620 (vsx_ior<mode>3_64bit): Likewise.
7621 (vsx_xor<mode>3): Likewise.
7622 (vsx_xor<mode>3_32bit): Likewise.
7623 (vsx_xor<mode>3_64bit): Likewise.
7624 (vsx_one_cmpl<mode>2): Likewise.
7625 (vsx_one_cmpl<mode>2_32bit): Likewise.
7626 (vsx_one_cmpl<mode>2_64bit): Likewise.
7627 (vsx_nor<mode>3): Likewise.
7628 (vsx_nor<mode>3_32bit): Likewise.
7629 (vsx_nor<mode>3_64bit): Likewise.
7630 (vsx_andc<mode>3): Likewise.
7631 (vsx_andc<mode>3_32bit): Likewise.
7632 (vsx_andc<mode>3_64bit): Likewise.
7633 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7634 and xxlorc instructions.
7635 (vsx_eqv<mode>3_64bit): Likewise.
7636 (vsx_nand<mode>3_32bit): Likewise.
7637 (vsx_nand<mode>3_64bit): Likewise.
7638 (vsx_orc<mode>3_32bit): Likewise.
7639 (vsx_orc<mode>3_64bit): Likewise.
7641 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7643 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7645 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7646 (p8_vmrgow): Likewise.
7647 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7648 GPRs to be split under VSX.
7649 (p8v_clz<mode>2): Add power8 count leading zero support.
7650 (p8v_popcount<mode>2): Add power8 population count support.
7651 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7654 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7657 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7659 (vec_nand): Likewise.
7660 (vec_vclz): Likewise.
7661 (vec_vclzb): Likewise.
7662 (vec_vclzd): Likewise.
7663 (vec_vclzh): Likewise.
7664 (vec_vclzw): Likewise.
7665 (vec_vgbbd): Likewise.
7666 (vec_vmrgew): Likewise.
7667 (vec_vmrgow): Likewise.
7668 (vec_vpopcnt): Likewise.
7669 (vec_vpopcntb): Likewise.
7670 (vec_vpopcntd): Likewise.
7671 (vec_vpopcnth): Likewise.
7672 (vec_vpopcntw): Likewise.
7674 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7676 PR rtl-optimization/57468
7677 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7680 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7682 PR rtl-optimization/57459
7683 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7684 type when setting live regs.
7686 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
7688 * config/s390/s390.opt (mlra): New option.
7689 * config/s390/s390.c (s390_decompose_address): Check displacement
7690 for all registers for LRA.
7691 (s390_secondary_reload): Don't used secondary reloads for LRA.
7692 (s390_lra_p): New function.
7693 (TARGET_LRA_P): Define.
7694 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7695 of attribute cpu_facility to zarch for the last alternative.
7696 (*cmpmem_short): Ditto.
7698 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
7700 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7701 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
7702 (arm_expand_prologue): Likewise.
7704 2013-06-06 Teresa Johnson <tejohnson@google.com>
7707 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7708 as this is now done by redirect_edge_and_branch_force.
7709 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7710 barriers, and fix interaction with splitting.
7711 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7712 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7713 reflect changes made in the routine.
7714 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7715 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7716 since this is called in cfglayout mode, and replace partition fixup
7717 with assert as that is now done by force_nonfallthru_and_redirect.
7718 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7719 already be marked with region crossing note.
7720 (insert_section_boundary_note): Make non-static, gate on flag
7721 has_bb_partition, rewrite to also check for multiple partitions.
7722 (rest_of_handle_reorder_blocks): Remove call to
7723 insert_section_boundary_note, now done later during free_cfg.
7724 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7725 * bb-reorder.h (insert_section_boundary_note): Declare.
7726 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7727 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7728 invoke insert_section_boundary_note.
7729 (try_redirect_by_replacing_jump): Remove unnecessary
7730 check for region crossing note.
7731 (fixup_partition_crossing): New function.
7732 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7733 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7734 in non-cfglayout mode.
7735 (force_nonfallthru_and_redirect): Fixup partition boundaries,
7736 remove old code that tried to do this. Emit barrier correctly
7737 when we are in cfglayout mode.
7738 (last_bb_in_partition): New function.
7739 (rtl_split_edge): Correctly fixup partition boundaries.
7740 (commit_one_edge_insertion): Remove old code that tried to
7741 fixup region crossing edge since this is now handled in
7742 split_block, and set up insertion point correctly since
7743 block may now end in a jump.
7744 (verify_hot_cold_block_grouping): Guard against checking when not in
7745 linearized RTL mode.
7746 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7748 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7749 rtl_verify_flow_info, so not called in cfglayout mode.
7750 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7751 (fixup_reorder_chain): Remove old code that attempted to fixup region
7752 crossing note as this is now handled in force_nonfallthru_and_redirect.
7753 (duplicate_insn_chain): Don't duplicate switch section notes.
7754 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7756 * basic-block.h (emit_barrier_after_bb): Declare.
7758 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7760 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7761 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7762 arm_usatsihi): Adjust alternatives for arm_restrict_it.
7764 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7766 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7768 * config/arm/ldmstm.md: Regenerate.
7770 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7772 * config/arm/sync.md (atomic_loaddi_1):
7773 Disable predication for arm_restrict_it.
7774 (arm_load_exclusive<mode>): Likewise.
7775 (arm_load_exclusivesi): Likewise.
7776 (arm_load_exclusivedi): Likewise.
7777 (arm_load_acquire_exclusive<mode>): Likewise.
7778 (arm_load_acquire_exclusivesi): Likewise.
7779 (arm_load_acquire_exclusivedi): Likewise.
7780 (arm_store_exclusive<mode>): Likewise.
7781 (arm_store_exclusive<mode>): Likewise.
7782 (arm_store_release_exclusivedi): Likewise.
7783 (arm_store_release_exclusive<mode>): Likewise.
7785 2013-06-06 Richard Biener <rguenther@suse.de>
7787 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7789 (lto_tag_is_tree_code_p): Adjust.
7790 (lto_tag_is_gimple_code_p): Likewise.
7791 (lto_gimple_code_to_tag): Likewise.
7792 (lto_tag_to_gimple_code): Likewise.
7793 (lto_tree_code_to_tag): Likewise.
7794 (lto_tag_to_tree_code): Likewise.
7795 * data-streamer.h (streamer_write_hwi_in_range): Use
7796 uhwi streaming to stream the normalized range.
7797 (streamer_read_hwi_in_range): Likewise.
7799 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7801 * config/arm/arm.md (enabled_for_depr_it): New attribute.
7802 (predicable_short_it): Likewise.
7803 (predicated): Likewise.
7804 (enabled): Handle above.
7805 (define_cond_exec): Set predicated attribute to yes.
7807 2013-06-05 Mike Stump <mikestump@comcast.net>
7809 * gdbinit.in (__FUNCTION__): Add.
7811 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
7813 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7814 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7816 2013-06-05 Jan Hubicka <jh@suse.cz>
7818 * varasm.c (mark_decl_referenced): Revert the removal until targets
7821 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
7823 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7824 instead of mark_decl_referenced.
7826 2013-06-05 Jan Hubicka <jh@suse.cz>
7828 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7829 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7830 and symtab_used_from_object_file_p.
7831 (cgraph_make_node_local_1): Clear forced_by_abi.
7832 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7833 * cgraph.h (symtab_node_base): Add forced_by_abi.
7834 (decide_is_variable_needed): Remove.
7835 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7836 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7837 (decide_is_symbol_needed): ... this one; handle symbols in general;
7838 always analyze virtuals; honnor forced_by_abi.
7839 (cgraph_finalize_function): Update.
7840 (varpool_finalize_decl): Update.
7841 (symbol_defined_and_needed): Remove.
7842 (analyze_functions): Update.
7843 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7844 output_refs, input_overwrite_node): Handle forced_by_abi.
7845 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7846 (address_taken_from_non_vtable_p): ... this one.
7847 (comdat_can_be_unshared_p_1): New function.
7848 (cgraph_comdat_can_be_unshared_p): Rename to ...
7849 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7850 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7851 (function_and_variable_visibility): Clear forced_by_abi as needed.
7852 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7853 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7854 * symtab.c (dump_symtab_base): Dump forced_by_abi.
7855 * varpool.c (decide_is_variable_needed): Remove.
7857 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7859 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7860 (arm_option_override): Override arm_restrict_it where appropriate.
7861 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7862 * config/arm/arm.opt (mrestrict-it): New command-line option.
7863 * doc/invoke.texi: Document -mrestrict-it.
7865 2013-06-05 David Malcolm <dmalcolm@redhat.com>
7867 * tsan.c (tsan_atomic_table): Make const.
7869 2013-06-05 Richard Biener <rguenther@suse.de>
7871 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7872 index associated with the tree we are supposed to replace.
7873 * tree-streamer-out.c (pack_ts_base_value_fields): Output
7874 TREE_ASM_WRITTEN as zero for everything but SSA names.
7876 2013-06-05 David Malcolm <dmalcolm@redhat.com>
7878 * tree-ssa-structalias.c (call_stmt_vars): Make static.
7880 2013-06-04 Jan Hubicka <jh@suse.cz>
7882 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7883 (input_node, input_varpool_node): Handle correctly external same
7885 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7886 nodes at ltrans stage.
7888 2013-06-04 Jan Hubicka <jh@suse.cz>
7890 * ipa-inline.c (update_caller_keys): Fix availability test.
7891 (update_callee_keys): Likewise.
7892 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7893 to follow ELF standard.
7895 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
7897 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7898 (mips64r5900el-*-elf*): New configurations.
7899 * config/mips/mips-cpus.def (r5900): New processor.
7900 * config/mips/mips-tables.opt: Regenerate.
7901 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7902 (mips_issue_rate): Handle PROCESSOR_R5900.
7903 (mips_reorg_process_insns): Force reorder mode for the R5900.
7904 * config/mips/mips.h (TARGET_MIPS5900): Define.
7905 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7907 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7909 * config/mips/mips.md (processor): Add r5900.
7910 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7912 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7914 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7915 into function to generate MOVI instruction.
7916 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7917 (aarch64_preferred_simd_mode): Turn into wrapper.
7918 (aarch64_output_scalar_simd_mov_immediate): New function.
7919 * config/aarch64/aarch64-protos.h: Add prototype for above.
7921 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7923 * config/aarch64/aarch64.c (simd_immediate_info): Remove
7924 element_char member.
7925 (sizetochar): Return signed char.
7926 (aarch64_simd_valid_immediate): Remove elchar and other
7927 unnecessary variables.
7928 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7929 Calculate element_char as required.
7930 * config/aarch64/aarch64-protos.h: Update and move prototype
7931 for aarch64_output_simd_mov_immediate.
7932 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7935 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7937 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7938 information completed by aarch64_simd_valid_immediate.
7939 (aarch64_legitimate_constant_p): Update arguments.
7940 (aarch64_simd_valid_immediate): Work with struct rather than many
7942 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7943 (aarch64_simd_make_constant): Update arguments.
7944 (aarch64_output_simd_mov_immediate): Work with struct rather than
7945 many pointers. Output immediate directly rather than as operand.
7946 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7948 * config/aarch64/constraints.md (Dn): Update arguments.
7950 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7952 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7954 (aarch64_simd_immediate_valid_for_move): Remove.
7955 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7956 (aarch64_simd_make_constant): Update call.
7957 (aarch64_output_simd_mov_immediate): Update call.
7958 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7960 * config/aarch64/constraints.md (Dn): Update call.
7962 2013-06-04 Ian Bolton <ian.bolton@arm.com>
7964 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7965 return type to bool for prototype.
7966 (aarch64_legitimate_constant_p): Check for true instead of not -1.
7967 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7968 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7970 2013-06-04 Catherine Moore <clm@codesourcery.com>
7972 * config/mips/mips.opt (meva): New.
7973 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7974 (ASM_SPEC): Handle -meva.
7975 * doc/invoke.texi (meva): Document.
7977 2013-06-04 Alan Modra <amodra@gmail.com>
7979 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7982 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7984 * rtl.def: Add extra fourth optional field to define_cond_exec.
7985 * gensupport.c (process_one_cond_exec): Process attributes from
7987 * doc/md.texi: Document fourth field in define_cond_exec.
7989 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
7991 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7992 out the processing order as in store_bit_field_1.
7994 2013-06-04 Jan Hubicka <jh@suse.cz>
7997 * cgraphunit.c (cgraph_process_same_body_aliases): Create
7998 non-VAR_DECL node if it does not exist yet.
8000 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
8002 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
8003 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
8004 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
8005 target_cpu_default setting.
8007 2013-06-03 Teresa Johnson <tejohnson@google.com>
8009 * dumpfile.c (opt_info_switch_p): Change -fopt-info
8010 default to -fopt-info=optimized instead of all.
8011 * doc/invoke.texi: Ditto.
8012 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
8013 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
8014 (execute_vect_slp): Emit BB vectorization success under
8015 MSG_OPTIMIZED_LOCATIONS.
8016 * tree-vect-slp.c (vect_slp_transform_bb): Change
8017 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
8018 * tree-vect-loop.c (vect_transform_loop): Ditto.
8020 2013-06-03 Jason Merrill <jason@redhat.com>
8023 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8024 Use TARGET_EXPR for C++.
8026 2013-06-03 Jakub Jelinek <jakub@redhat.com>
8028 PR rtl-optimization/57268
8029 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
8030 if DEBUG_INSN_P (insn).
8033 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
8035 PR rtl-optimization/57268
8036 * sched-deps.c (sched_analyze_2): Flush dependence lists if
8037 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8039 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
8041 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
8042 (ix86_avoid_lea_for_addr): Likewise.
8043 (exact_dependency_1): Likewise.
8044 (ix86_adjust_cost): Likewise.
8045 (swap_top_of_ready_list): Fix formatting and !reload_completed check
8047 (do_reorder_for_imul): Fix typo, formatting and
8048 !reload_completed check removed.
8049 (ix86_sched_reorder): Fix typo and formatting.
8050 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
8053 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
8055 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
8057 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
8059 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
8060 <STRING_CST>: Likewise.
8061 <VECTOR_CST>: Likewise.
8063 2013-06-01 Janus Weil <janus@gcc.gnu.org>
8064 Mikael Morin <mikael@gcc.gnu.org>
8066 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
8067 * config.in: Regenerated.
8068 * configure: Regenerated.
8070 2013-06-01 Jan Hubicka <jh@suse.cz>
8073 * cgraphunit.c (compile): When weakref is not supported,
8074 set up transparent aliases before final output pass.
8075 * varasm.c (assemble_alias): Do not try to do it here.
8077 2013-06-01 Jan Hubicka <jh@suse.cz>
8080 * passes.c (for_per_function): Skip unanalyzed functions.
8082 2013-06-01 Jan Hubicka <jh@suse.cz>
8084 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
8085 (lto_symtab_merge_symbols_1): ... this one.
8086 (lto_symtab_merge_cgraph_nodes): Rename to ...
8087 (lto_symtab_merge_symbols): ... this one; simplify.
8088 * cgraph.c (same_body_aliases_done): Rename to ...
8089 (cpp_implicit_aliases_done): ... this one.
8090 (cgraph_create_function_alias): Update.
8091 (cgraph_same_body_alias): Update.
8092 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
8093 (verify_edge_corresponds_to_fndecl): Simplify.
8094 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
8095 (cgraph_node): Remove same_body_alias.
8096 (varpool_node): Remove alias_of and extra_name_alias.
8097 (same_body_aliases_done): Rename to ..
8098 (cpp_implicit_aliases_done): ... this one.
8099 (symtab_alias_ultimate_target): Add default parameter.
8100 (symtab_resolve_alias): New function.
8101 (fixup_same_cpp_alias_visibility): Declare.
8102 (cgraph_function_node): Add default parameter.
8103 (cgraph_node_asm_name): Likewise.
8104 (cgraph_function_or_thunk_node): Add default parameter; do
8105 not ICE when it is NULL.
8106 (varpool_variable_node): Likewise.
8107 * tree-emutls.c (create_emultls_var): Update.
8108 (ipa_lower_emutls): Update.
8109 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
8110 (cgraph_reset_node): Reset alias info.
8111 (cgraph_finalize_function): Update.
8112 (fixup_same_cpp_alias_visibility): Move to symtab.c.
8113 (analyze_function): Simplify.
8114 (cgraph_process_same_body_aliases): Simplify.
8115 (analyze_functions): Fixup same body aliases.
8116 (handle_alias_pairs): Simplify.
8117 (assemble_thunk): Update.
8118 (assemble_thunks_and_aliases): Update.
8119 (output_weakrefs): Rewrite.
8120 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
8121 (lto_output_varpool_node): Likewise.
8122 (compute_ltrans_boundary): Remve assert.
8123 (get_alias_symbol): New functoin.
8124 (input_node): Rewrite alias handling.
8125 (input_varpool_node): Likewise.
8126 * ipa-pure-const.c (propagate_pure_const): Fix formating.
8127 * ipa.c (process_references): Handle weakrefs correctly.
8128 (symtab_remove_unreachable_nodes): Likewise.
8129 * trans-mem.c (get_cg_data): Update.
8130 (ipa_tm_create_version_alias): Update.
8131 (ipa_tm_execute): Update.
8132 * symtab.c (dump_symtab_base): Dump aliases.
8133 (verify_symtab_base): Verify aliases.
8134 (symtab_node_availability): New function.
8135 (symtab_alias_ultimate_target): Simplify.
8136 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
8137 handle all the fixup cases.
8138 (symtab_resolve_alias): New function.
8139 * passes.c (ipa_write_summaries): Handle weakrefs.
8140 * varpool.c (varpool_analyze_node): Simplify.
8141 (assemble_aliases): Update.
8142 (varpool_create_variable_alias): Simplify.
8143 (varpool_extra_name_alias): Simplify.
8144 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
8145 (lto_symtab_merge_symbols): ... this one.
8147 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
8150 PR rtl-optimization/57268
8151 * sched-deps.c (sched_analyze_2): Flush dependence lists if
8152 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8154 2013-06-01 Tobias Burnus <burnus@net-b.de>
8157 2013-05-31 Tobias Burnus <burnus@net-b.de>
8160 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
8163 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
8165 PR rtl-optimization/57268
8166 * sched-deps.c (sched_analyze_2): Flush dependence lists if
8167 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
8169 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
8171 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
8172 unordered comparison operators when -fno-trapping-math is in effect
8174 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
8175 and implement unordered comparison operators properly on the e500.
8177 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
8179 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
8180 for constant scalar integers.
8181 (simplify_relational_operation_1): Likewise.
8183 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
8185 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
8186 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
8189 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
8190 Igor Zamyatin <igor.zamyatin@intel.com>
8192 Silvermont (SLM) architecture performance tuning.
8193 * config/i386/i386.h (enum ix86_tune_indices): Add
8194 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
8195 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
8197 * config/i386/i386.c (initial_ix86_tune_features)
8198 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
8199 (ix86_lea_outperforms): Handle Silvermont tuning.
8200 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
8202 (ix86_use_lea_for_mov): Likewise.
8203 (ix86_avoid_lea_for_addr): Likewise.
8204 (ix86_lea_for_add_ok): Likewise.
8205 (exact_dependency_1): New function.
8206 (exact_store_load_dependency): Likewise.
8207 (ix86_adjust_cost): Handle Silvermont tuning.
8208 (do_reoder_for_imul): Likewise.
8209 (swap_top_of_ready_list): New function.
8210 (ix86_sched_reorder): Changed to handle Silvermont tuning.
8212 * config/i386/i386.md (peepholes that split memory operand in fp
8215 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
8217 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8218 Remove un-necessary braces.
8220 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
8222 * config/aarch64/aarch64.c (aarch64_classify_symbol):
8223 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
8225 2013-05-31 Tobias Burnus <burnus@net-b.de>
8228 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
8230 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8233 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
8234 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
8235 * config/arm/neon.md (iordi3_neon): Remove.
8236 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
8237 * config/arm/predicates.md (imm_for_neon_logic_operand):
8238 Move to earlier in the file.
8239 (neon_logic_op2): Likewise.
8240 (arm_iordi_operand_neon): New predicate.
8242 2013-05-31 Richard Biener <rguenther@suse.de>
8244 PR tree-optimization/57478
8245 PR tree-optimization/57453
8246 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
8249 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
8251 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
8252 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
8254 2013-05-30 Tobias Burnus <burnus@net-b.de>
8255 Thomas Koenig <tkoenig@gcc.gnu.org>
8258 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
8259 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
8261 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
8263 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
8265 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
8267 * target.def (register_usage_leveling_p): New hook.
8268 * targhooks.c (default_register_usage_leveling_p): New.
8269 * targhooks.h (default_register_usage_leveling_p): New prototype.
8270 * lra-assigns.c (register_usage_leveling_p): Use the hook.
8271 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
8272 * doc/tm.texi: Update.
8273 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
8275 2013-05-30 Ian Bolton <ian.bolton@arm.com>
8277 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
8278 (*insv_reg<mode>): New define_insn.
8280 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
8282 PR rtl-optimization/57439
8283 * postreload.c (move2add_valid_value_p): Check that we have
8284 a zero subreg_regno_offset when accessing the register in
8287 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
8288 Igor Zamyatin <igor.zamyatin@intel.com>
8290 Silvermont (SLM) architecture pipeline model, tuning and
8292 * config.gcc: Add slm config options and target.
8294 * config/i386/slm.md: New.
8296 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
8298 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): New case
8300 (ix86_target_macros_internal): Likewise.
8302 * gcc/config/i386/i386.c (slm_cost): New cost.
8303 (m_SLM): New macro flag.
8304 (initial_ix86_tune_features): Set m_SLM.
8305 (x86_accumulate_outgoing_args): Likewise.
8306 (x86_arch_always_fancy_math_387): Likewise.
8307 (processor_target_table): Add slm cost.
8308 (cpu_names): Add slm cpu name.
8309 (x86_option_override_internal): Set SLM ISA.
8310 (ix86_issue_rate): New case PROCESSOR_SLM.
8311 (ia32_multipass_dfa_lookahead): Likewise.
8312 (fold_builtin_cpu): Add slm.
8314 * config/i386/i386.h (TARGET_SLM): New target macro.
8315 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
8316 (processor_type): Add PROCESSOR_SLM.
8318 * config/i386/i386.md (cpu): Add new value "slm".
8319 (slm.md): Include slm.md.
8321 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
8322 Zhenqiang Chen <zhenqiang.chen@linaro.org>
8324 * config/arm/arm-protos.h: Add and update function protos.
8325 * config/arm/arm.c (use_simple_return_p): New added.
8326 (thumb2_expand_return): Check simple_return flag.
8327 * config/arm/arm.md: Add simple_return and conditional simple_return.
8328 * config/arm/iterators.md: Add iterator for return and simple_return.
8330 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
8332 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
8333 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
8334 (arm_emit_vfp_multi_reg_pop): Likewise.
8335 (thumb2_emit_ldrd_pop): Likewise.
8336 (arm_expand_epilogue): Add misc REG_CFA notes.
8337 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
8339 2013-05-29 Lawrence Crowl <crowl@google.com>
8341 * config/arm/t-arm: Update for below.
8343 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
8344 Change type to hash_table. Update dependent calls and types.
8346 * config/i386/t-cygming: Update for below.
8348 * config/i386/t-interix: Update for below.
8350 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
8351 Change type to hash_table. Update dependent calls and types.
8352 (i386_find_on_wrapper_list::wrappers): Likewise.
8354 * config/ia64/t-ia64: Update for below.
8356 * config/ia64/ia64.c (bundle_state_table):
8357 Change type to hash_table. Update dependent calls and types.
8359 * config/mips/mips.c (mips_reorg_process_insns::htab):
8360 Change type to hash_table. Update dependent calls and types.
8362 * config/sol2.c (solaris_comdat_htab):
8363 Change type to hash_table. Update dependent calls and types.
8365 * config/t-sol2: Update for above.
8367 2013-05-29 Teresa Johnson <tejohnson@google.com>
8369 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
8370 functions are not yet marked as defined.
8372 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8373 Pat Haugen <pthaugen@us.ibm.com>
8374 Peter Bergner <bergner@vnet.ibm.com>
8376 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
8380 (vrotl<mode>3): Likewise.
8381 (vashl<mode>3): Likewise.
8382 (vlshr<mode>3): Likewise.
8383 (vashr<mode>3): Likewise.
8385 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8386 support for power8 V2DI builtins.
8388 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
8389 power8 V2DI builtins.
8390 (vupkhsw): Likewise.
8391 (vupklsw): Likewise.
8392 (vaddudm): Likewise.
8397 (vpkudum): Likewise.
8398 (vpksdss): Likewise.
8399 (vpkudus): Likewise.
8400 (vpksdus): Likewise.
8405 (vsubudm): Likewise.
8406 (vcmpequd): Likewise.
8407 (vcmpgtsd): Likewise.
8408 (vcmpgtud): Likewise.
8409 (vcmpequd_p): Likewise.
8410 (vcmpgtsd_p): Likewise.
8411 (vcmpgtud_p): Likewise.
8412 (vupkhsw): Likewise.
8413 (vupklsw): Likewise.
8414 (vaddudm): Likewise.
8419 (vpksdss): Likewise.
8420 (vpksdus): Likewise.
8421 (vpkudum): Likewise.
8422 (vpkudus): Likewise.
8427 (vsubudm): Likewise.
8429 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
8430 support for power8 V2DI instructions.
8432 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
8433 power8 V2DI instructions. Combine pack and unpack insns to use an
8434 iterator for each mode. Check whether a particular mode supports
8435 Altivec instructions instead of just checking TARGET_ALTIVEC.
8436 (UNSPEC_VPKUWUM): Likewise.
8437 (UNSPEC_VPKSHSS): Likewise.
8438 (UNSPEC_VPKSWSS): Likewise.
8439 (UNSPEC_VPKUHUS): Likewise.
8440 (UNSPEC_VPKSHUS): Likewise.
8441 (UNSPEC_VPKUWUS): Likewise.
8442 (UNSPEC_VPKSWUS): Likewise.
8443 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
8444 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
8445 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
8446 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
8447 (UNSPEC_VUPKHSB): Likewise.
8448 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
8449 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
8450 (UNSPEC_VUPKHSH): Likewise.
8451 (UNSPEC_VUPKLSB): Likewise.
8452 (UNSPEC_VUPKLSH): Likewise.
8454 (VI_char): Likewise.
8455 (VI_scalar): Likewise.
8456 (VI_unit): Likewise.
8458 (VP_small): Likewise.
8459 (VP_small_lc): Likewise.
8460 (VU_char): Likewise.
8461 (add<mode>3): Likewise.
8462 (altivec_vaddcuw): Likewise.
8463 (altivec_vaddu<VI_char>s): Likewise.
8464 (altivec_vadds<VI_char>s): Likewise.
8465 (sub<mode>3): Likewise.
8466 (altivec_vsubcuw): Likewise.
8467 (altivec_vsubu<VI_char>s): Likewise.
8468 (altivec_vsubs<VI_char>s): Likewise.
8469 (altivec_vavgs<VI_char>): Likewise.
8470 (altivec_vcmpbfp): Likewise.
8471 (altivec_eq<mode>): Likewise.
8472 (altivec_gt<mode>): Likewise.
8473 (altivec_gtu<mode>): Likewise.
8474 (umax<mode>3): Likewise.
8475 (smax<mode>3): Likewise.
8476 (umin<mode>3): Likewise.
8477 (smin<mode>3): Likewise.
8478 (altivec_vpkuhum): Likewise.
8479 (altivec_vpkuwum): Likewise.
8480 (altivec_vpkshss): Likewise.
8481 (altivec_vpkswss): Likewise.
8482 (altivec_vpkuhus): Likewise.
8483 (altivec_vpkshus): Likewise.
8484 (altivec_vpkuwus): Likewise.
8485 (altivec_vpkswus): Likewise.
8486 (altivec_vpks<VI_char>ss): Likewise.
8487 (altivec_vpks<VI_char>us): Likewise.
8488 (altivec_vpku<VI_char>us): Likewise.
8489 (altivec_vpku<VI_char>um): Likewise.
8490 (altivec_vrl<VI_char>): Likewise.
8491 (altivec_vsl<VI_char>): Likewise.
8492 (altivec_vsr<VI_char>): Likewise.
8493 (altivec_vsra<VI_char>): Likewise.
8494 (altivec_vsldoi_<mode>): Likewise.
8495 (altivec_vupkhsb): Likewise.
8496 (altivec_vupkhs<VU_char>): Likewise.
8497 (altivec_vupkls<VU_char>): Likewise.
8498 (altivec_vupkhsh): Likewise.
8499 (altivec_vupklsb): Likewise.
8500 (altivec_vupklsh): Likewise.
8501 (altivec_vcmpequ<VI_char>_p): Likewise.
8502 (altivec_vcmpgts<VI_char>_p): Likewise.
8503 (altivec_vcmpgtu<VI_char>_p): Likewise.
8504 (abs<mode>2): Likewise.
8505 (vec_unpacks_hi_v16qi): Likewise.
8506 (vec_unpacks_hi_v8hi): Likewise.
8507 (vec_unpacks_lo_v16qi): Likewise.
8508 (vec_unpacks_hi_<VP_small_lc>): Likewise.
8509 (vec_unpacks_lo_v8hi): Likewise.
8510 (vec_unpacks_lo_<VP_small_lc>): Likewise.
8511 (vec_pack_trunc_v8h): Likewise.
8512 (vec_pack_trunc_v4si): Likewise.
8513 (vec_pack_trunc_<mode>): Likewise.
8515 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
8517 (vec_vmaxsd): Likewise.
8518 (vec_vmaxud): Likewise.
8519 (vec_vminsd): Likewise.
8520 (vec_vminud): Likewise.
8521 (vec_vpksdss): Likewise.
8522 (vec_vpksdus): Likewise.
8523 (vec_vpkudum): Likewise.
8524 (vec_vpkudus): Likewise.
8525 (vec_vrld): Likewise.
8526 (vec_vsld): Likewise.
8527 (vec_vsrad): Likewise.
8528 (vec_vsrd): Likewise.
8529 (vec_vsubudm): Likewise.
8530 (vec_vupkhsw): Likewise.
8531 (vec_vupklsw): Likewise.
8533 2013-05-29 Jan Hubicka <jh@suse.cz>
8535 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
8536 flags; reorder rest of fields in more consistent way.
8537 (varpool_node): Remove analyzed, finalized and alias.
8538 (cgraph_ndoe): Likewise.
8539 (symtab_alias_ultimate_target): New function.
8540 (cgraph_function_node): Move offline.
8541 (cgraph_reset_node): Declare.
8542 (cgraph_comdat_can_be_unshared_p): Remove.
8543 (varpool_remove_initializer): Declare.
8544 (varpool_first_defined_variable, varpool_next_defined_variable
8545 cgraph_first_defined_function, cgraph_next_defined_function): Update.
8546 (cgraph_function_with_gimple_body_p): Update.
8547 (varpool_all_refs_explicit_p): Update.
8548 (symtab_alias_target): New function.
8549 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
8550 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
8551 (cgraph_function_or_thunk_node): Simplify using
8552 symtab_alias_ultimate_target.
8553 (varpool_variable_node): Likewise.
8554 * cgraph.c (cgraph_create_function_alias): Update.
8555 (cgraph_add_thunk): Update.
8556 (cgraph_remove_node): Update.
8557 (dump_cgraph_node): Do not dump removed flags.
8558 (cgraph_function_body_availability): Update.
8559 (cgraph_propagate_frequency): Update.
8560 (verify_cgraph_node): Check sanity of local flag.
8561 (cgraph_function_node): Move here from cgraph.h; revamp for
8562 cgraph_function_or_thunk_node.
8563 * lto-symtab.c (lto_varpool_replace_node): Update.
8564 (lto_symtab_resolve_can_prevail_p): Update.
8565 (lto_symtab_merge_cgraph_nodes): Update.
8566 * ipa-cp.c (determine_versionability, initialize_node_lattices,
8567 propagate_constants_accross_call, devirtualization_time_bonus,
8568 ipcp_propagate_stage): Update.
8569 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
8570 * ipa-inline-transform.c (clone_inlined_nodes,
8571 preserve_function_body_p): Update.
8572 * ipa-reference.c (propagate): Update.
8573 (write_node_summary_p): Update.
8574 * toplev.c (wrapup_global_declaration_2): Update.
8575 * cgraphunit.c (cgraph_analyze_function): Rename to ...
8576 (analyze_function) ... this one.
8577 (cgraph_process_new_functions): Update.
8578 (cgraph_reset_node): Export.
8579 (cgraph_finalize_function): Update.
8580 (cgraph_add_new_function): Update.
8581 (process_function_and_variable_attributes): Update.
8582 (varpool_finalize_decl): Update.
8583 (symbol_finalized): Remove.
8584 (symbol_finalized_and_needed): Rename to ...
8585 (symbol_defined_and_needed): ... update.
8586 (cgraph_analyze_functions): Update.
8587 (handle_alias_pairs): Update.
8588 (mark_functions_to_output): Update.
8589 (assemble_thunk): Update.
8590 (output_in_order): Update.
8591 (output_weakrefs): Update.
8592 (finalize_compilation_unit): Update.
8593 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
8594 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
8595 input_node, input_varpool_node): Update.
8596 * dbxout.c (dbxout_expand_expr): Update.
8597 * cgraphclones.c (cgraph_clone_node): Update.
8598 (cgraph_copy_node_for_versioning): Update.
8599 (cgraph_materialize_clone): Update.
8600 (cgraph_materialize_all_clones): Update.
8601 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8602 propagate_pure_const, propagate_nothrow): Update.
8603 * lto-streamer-out.c (lto_output, write_symbol): Update.
8604 * ipa-utils.c (ipa_reverse_postorder): Update.
8605 * ipa-inline.c (can_inline_edge_p): Update.
8606 (update_caller_keys, ipa_inline): Update.
8607 * dwarf2out.c (reference_to_unused,
8608 premark_types_used_by_global_vars_helper): Update.
8609 * tree-eh.c (tree_could_trap_p): Update.
8610 * ipa-split.c (consider_split, execute_split_functions): Update.
8611 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8612 has_addr_references_p): Update; move ahead in file for better
8614 (process_references): Simplify.
8615 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8617 (cgraph_comdat_can_be_unshared_p): Make static.
8618 (cgraph_externally_visible_p): Update.
8619 (varpool_externally_visible_p): Update.
8620 (function_and_variable_visibility): Update.
8621 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8622 ipa_tm_mark_force_output_node): Update.
8623 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8624 estimate_edge_devirt_benefit, inline_generate_summary,
8625 inline_write_summary): Update.
8626 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8627 * ipa-prop.c (ipa_compute_jump_functions): Update.
8628 (ipa_print_node_params, ipa_prop_read_section,
8629 ipa_update_after_lto_read, read_replacements_section): Update.
8630 * varasm.c (mark_decl_referenced): Update.
8631 (assemble_alias, dump_tm_clone_pairs): Update.
8632 * tree-inline.c (copy_bb): Update.
8633 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8635 * symtab.c (dump_symtab_base): Print new flags.
8636 (verify_symtab_base): Verify new flags.
8637 (symtab_alias_ultimate_target): New function.
8638 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8639 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8641 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8643 * i386.c (ix86_get_function_versions_dispatcher,
8644 ix86_generate_version_dispatcher_body): Update.
8645 (fold_builtin_cpu): Use varpool_add_new_variable.
8646 * varpool.c (varpool_remove_initializer): Break out from ...
8647 (varpool_remove_node): ... this one.
8648 (dump_varpool_node, varpool_node_for_asm,
8649 cgraph_variable_initializer_availability, varpool_analyze_node,
8650 varpool_assemble_decl, varpool_remove_unreferenced_decls,
8651 varpool_finalize_named_section_flags, varpool_create_variable_alias):
8654 2013-05-29 Jan Hubicka <jh@suse.cz>
8656 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8658 2013-05-29 Easwaran Raman <eraman@google.com>
8660 PR tree-optimization/57442
8661 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8662 when control exits the main loop.
8664 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
8666 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8668 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8669 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8670 * rx/t-rx: Add rx100 under multi library matches option for nofpu
8673 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8675 PR tree-optimization/57441
8676 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8677 Don't limit size of incr_vec to number of candidates.
8679 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
8681 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8682 and mips16 directories.
8683 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8684 (MULTILIB_DIRNAMES): Ditto.
8685 (MULTILIB_EXCEPTIONS): Add new exceptions.
8686 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8687 (MULTILIB_DIRNAMES): Ditto.
8688 (MULTILIB_EXCEPTIONS): Add new exceptions.
8690 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8691 Marcus Shawcroft <marcus.shawcroft@arm.com>
8693 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8694 SYMBOL_TINY_ABSOLUTE.
8695 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8696 SYMBOL_TINY_ABSOLUTE.
8697 (aarch64_expand_mov_immediate): Likewise.
8698 (aarch64_classify_symbol): Likewise.
8699 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8700 Permit SYMBOL_TINY_ABSOLUTE.
8701 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8703 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
8704 Marcus Shawcroft <marcus.shawcroft@arm.com>
8706 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8707 Refactor if/switch. Replace gcc_assert with if.
8709 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
8711 * config/i386/i386.c (initial_ix86_tune_features): Enable
8712 FP Reassociation for AMD bdver1 and bdver2.
8714 2013-05-29 Martin Jambor <mjambor@suse.cz>
8716 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8717 and IMAGPART_EXPR do not occur within other handled_components.
8719 2013-05-29 Richard Biener <rguenther@suse.de>
8721 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8722 access on whether the use is in the BB we currently try to
8724 (vect_bb_vectorization_profitable_p): Pass the BB we currently
8725 vectorize to vect_bb_slp_scalar_cost.
8727 2013-05-29 Richard Biener <rguenther@suse.de>
8729 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8730 computing scalar cost offsetted by stmts that are kept live
8732 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8733 for computation of scalar cost.
8735 2013-05-28 Steve Ellcey <sellcey@mips.com>
8737 * config/mips/mips-cpus.def (mips32r2): Change processor type.
8739 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
8741 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8742 array notation built-in reduction functions.
8743 * doc/passes.texi (Passes): Added documentation about changes done
8745 * doc/invoke.texi (C Dialect Options): Added documentation about
8746 the -fcilkplus flag.
8747 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8748 (BUILTINS_DEF): Depend on cilkplus.def.
8749 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
8750 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8751 * cilkplus.def: New file.
8753 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
8755 PR rtl-optimization/57439
8756 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8758 2013-05-28 Easwaran Raman <eraman@google.com>
8760 PR tree-optimization/57337
8761 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8762 (find_insert_point): Correctly identify the insertion point
8763 when two statements with the same UID is compared.
8765 2013-05-28 Richard Biener <rguenther@suse.de>
8767 PR tree-optimization/56787
8768 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8769 from the list of data references.
8770 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8772 (vect_analyze_loop_operations): Likewise.
8773 (vect_transform_loop): Remove clobbers.
8775 2013-05-28 Martin Jambor <mjambor@suse.cz>
8777 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8778 and REALPART_EXPRs have scalar type.
8780 2013-05-28 Richard Biener <rguenther@suse.de>
8782 PR tree-optimization/57411
8783 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8785 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8786 virtual operand propagation.
8788 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
8790 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8791 destination register for bmasksi_vis.
8792 (vector_init_bshuffle): Likewise.
8793 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8795 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
8797 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8798 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8799 mode if the instruction isn't available in the original mode.
8800 * config/sparc/sparc.opt (mfix-ut699): New option.
8801 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8802 (divdf3): Turn into expander.
8803 (divdf3_nofix): New insn.
8804 (divdf3_fix): Likewise.
8805 (divsf3): Disable if -mfix-ut699.
8806 (sqrtdf2): Turn into expander.
8807 (sqrtdf2_nofix): New insn.
8808 (sqrtdf2_fix): Likewise.
8809 (sqrtsf2): Disable if -mfix-ut699.
8811 2013-05-27 Richard Biener <rguenther@suse.de>
8814 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8815 block for the new loop.
8817 2013-05-27 Richard Biener <rguenther@suse.de>
8819 PR tree-optimization/57343
8820 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8821 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8822 (number_of_iterations_cond): Do not build the folded tree.
8824 2013-05-27 Richard Biener <rguenther@suse.de>
8828 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8829 OEP_CONSTANT_ADDRESS_OF retained.
8831 PR tree-optimization/57417
8832 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8834 (set_ssa_val_to): Compare addresses using
8835 get_addr_base_and_unit_offset.
8837 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
8839 PR rtl-optimization/56833
8840 * postreload.c (move2add_record_mode): New function.
8841 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8842 (move2add_use_add2_insn): Use move2add_record_sym_value.
8843 (move2add_use_add3_insn): Likewise.
8844 (reload_cse_move2add): Use move2add_valid_value_p and
8845 move2add_record_mode. Invalidate call-clobbered and REG_INC
8846 affected regs by setting reg_mode to VOIDmode.
8847 (move2add_note_store): Don't pretend the inside of a SUBREG is
8848 the actual destination. Invalidate single/leading registers by
8849 setting reg_mode to VOIDmode.
8850 Use move2add_record_sym_value, move2add_valid_value_p and
8851 move2add_record_mode.
8853 2013-05-27 Richard Biener <rguenther@suse.de>
8855 PR tree-optimization/57396
8856 * tree-affine.c (double_int_constant_multiple_p): Properly
8857 return false for val == 0 and div != 0.
8859 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8861 * config/mips/mips.h: Use #elif in preprocessor conditions.
8863 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8866 * config/mips/constraints.md (kl): New constraint.
8867 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8868 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
8869 constraint for operand 0. Split after CSE for MIPS16. Emit a move
8871 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8873 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
8876 * config/mips/mips.c (mips_can_inline_p): New function.
8877 (TARGET_CAN_INLINE_P): Define.
8879 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
8881 * sched-int.h (ds_t, dw_t): Make unsigned int.
8882 Fix documentation that describes how all the ds_t bits are used.
8883 Reserve the last bit for delayed-branch scheduling.
8884 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8885 (BITS_PER_DEP_WEAK): Fix definition and documentation.
8886 (gen_dep_weak_1): Remove prototype.
8887 * sched-deps.c (get_dep_weak_1): Make static.
8888 * target.def (speculate_insn, needs_block_p, gen_spec_check,
8889 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8890 * doc/tm.texi: Regenerate.
8891 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8893 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
8896 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8898 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
8899 Sandra Loosemore <sandra@codesourcery.com>
8901 * config.gcc (powerpc-*): Allow native for with-cpu.
8903 2013-05-24 Jeff Law <law@redhat.com>
8905 PR tree-optimization/57124
8906 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8907 conversion feeding a condition if the range has an overflow
8908 if -fstrict-overflow. Add warnings for when we do make the
8911 2013-05-24 Dehao Chen <dehao@google.com>
8913 * tree-cfg.c (locus_discrim_map): Fix the typo.
8914 (locus_discrim_hasher): Likewise.
8915 (locus_discrim_hasher::hash): Likewise.
8916 (locus_discrim_hasher::equal): Likewise.
8918 2013-05-24 Martin Jambor <mjambor@suse.cz>
8920 PR tree-optimization/57294
8921 * cgraph.h (ipa_record_stmt_references): Declare.
8922 * cgraphbuild.c (ipa_record_stmt_references): New function.
8923 (build_cgraph_edges): Use ipa_record_stmt_references.
8924 (rebuild_cgraph_edges): Likewise.
8925 (cgraph_rebuild_references): Likewise.
8926 * ipa-prop.c (ipa_modify_call_arguments): Discard references
8927 associated with the old statement and build references from the
8928 newly built statements.
8929 * ipa-ref.c (ipa_remove_stmt_references): New function.
8930 * ipa-ref.h (ipa_remove_stmt_references): Declare.
8932 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
8934 * lra-constraints.c (emit_spill_move): Use smaller mode for
8936 (check_and_process_move): Consider mem-reg moves for secondary
8938 (curr_insn_transform): Don't lose insns emitted before for
8939 secondary memory moves.
8940 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
8941 reg set up in the current insn.
8943 2013-05-24 Dehao Chen <dehao@google.com>
8945 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8947 (locus_descrim_hasher::equal): Likewise.
8948 (build_gimple_cfg): New discriminator assignment algorithm.
8949 (make_edges): Likewise.
8950 (next_discriminator_for_locus): Likewise.
8951 (same_line_p): Likewise.
8952 (assign_discriminators): Likewise.
8953 (make_cond_expr_edges): Likewise.
8954 (make_gimple_switch_edges): Likewise.
8955 (make_goto_expr_edges): Likewise.
8956 (make_gimple_asm_edges): Likewise.
8958 2013-05-24 Ian Bolton <ian.bolton@arm.com>
8960 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8961 X format specifier to only display bottom 16 bits.
8962 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8963 immediate to match for operand 2, since it will be masked.
8965 2013-05-24 Richard Biener <rguenther@suse.de>
8967 PR tree-optimization/57287
8968 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8969 all SSA names that occur in abnormal PHIs.
8971 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
8974 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8975 that index is not negative.
8977 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
8979 PR rtl-optimization/55177
8980 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8981 (simplify_byte_swapping_operation): New.
8982 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8983 (simplify_relational_operation_1): Deal with BSWAP.
8985 2013-05-23 Richard Henderson <rth@redhat.com>
8988 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8989 (ix86_reorg): Call it.
8991 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
8994 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8995 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8996 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8998 2013-05-23 Christian Bruel <christian.bruel@st.com>
9001 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
9003 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
9004 Marcus Shawcroft <marcus.shawcroft@arm.com>
9006 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
9007 * config/aarch64/constraints.md (Usa): Remove.
9008 * doc/md.texi (AArch64 Usa): Remove.
9010 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
9011 Marcus Shawcroft <marcus.shawcroft@arm.com>
9013 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
9014 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
9015 * config/aarch64/predicates.md (aarch64_const_address): Remove.
9016 (aarch64_mov_operand): Use aarch64_mov_operand_p.
9018 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
9020 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
9021 instruction (AdvSIMD).
9022 * config/aarch64/aarch64-builtins.c
9023 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
9024 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
9026 2013-05-23 Martin Jambor <mjambor@suse.cz>
9029 * tree.h (contains_bitfld_component_ref_p): Declare.
9030 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
9031 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
9033 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
9034 not access a bit-field. Assert all final offsets are byte-aligned.
9036 2013-05-23 Richard Biener <rguenther@suse.de>
9038 PR tree-optimization/57380
9039 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
9040 least one invariant or re-used load.
9041 * passes.c (init_optimization_passes): Move pass_phiprop before
9044 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
9046 * config/aarch64/aarch64-simd.md
9047 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
9049 2013-05-23 Richard Biener <rguenther@suse.de>
9052 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
9053 OEP_CONSTANT_ADDRESS_OF retained.
9055 2013-05-23 Jakub Jelinek <jakub@redhat.com>
9058 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
9059 don't lower unit. Handle unit not being always BITS_PER_WORD.
9061 2013-05-23 Richard Biener <rguenther@suse.de>
9063 PR rtl-optimization/57341
9064 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
9065 instead of true_dependence.
9067 2013-05-22 David Malcolm <dmalcolm@redhat.com>
9069 * bb-reorder.c (branch_threshold): Make const.
9070 (exec_threshold): Ditto.
9072 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
9073 Pat Haugen <pthaugen@us.ibm.com>
9074 Peter Bergner <bergner@vnet.ibm.com>
9076 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
9077 documentation for the power8 crypto builtins.
9079 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
9081 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
9082 macros for defining power8 builtin functions.
9083 (BU_P8V_AV_2): Likewise.
9084 (BU_P8V_AV_P): Likewise.
9085 (BU_P8V_VSX_1): Likewise.
9086 (BU_P8V_OVERLOAD_1): Likewise.
9087 (BU_P8V_OVERLOAD_2): Likewise.
9088 (BU_CRYPTO_1): Likewise.
9089 (BU_CRYPTO_2): Likewise.
9090 (BU_CRYPTO_3): Likewise.
9091 (BU_CRYPTO_OVERLOAD_1): Likewise.
9092 (BU_CRYPTO_OVERLOAD_2): Likewise.
9093 (XSCVSPDP): Fix typo, point to the correct instruction.
9094 (VCIPHER): Add power8 crypto builtins.
9095 (VCIPHERLAST): Likewise.
9096 (VNCIPHER): Likewise.
9097 (VNCIPHERLAST): Likewise.
9098 (VPMSUMB): Likewise.
9099 (VPMSUMH): Likewise.
9100 (VPMSUMW): Likewise.
9101 (VPERMXOR_V2DI): Likewise.
9102 (VPERMXOR_V4SI: Likewise.
9103 (VPERMXOR_V8HI: Likewise.
9104 (VPERMXOR_V16QI: Likewise.
9105 (VSHASIGMAW): Likewise.
9106 (VSHASIGMAD): Likewise.
9108 (VPERMXOR): Likewise.
9109 (VSHASIGMA): Likewise.
9111 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
9112 __CRYPTO__ if the crypto instructions are available.
9113 (altivec_overloaded_builtins): Add support for overloaded power8
9116 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
9117 support for power8 crypto builtins.
9118 (builtin_function_type): Likewise.
9119 (altivec_init_builtins): Add support for builtins that take vector
9120 long long (V2DI) arguments.
9122 * config/rs6000/crypto.md: New file, define power8 crypto
9125 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
9126 Pat Haugen <pthaugen@us.ibm.com>
9127 Peter Bergner <bergner@vnet.ibm.com>
9129 * doc/invoke.texi (Option Summary): Add power8 options.
9130 (RS/6000 and PowerPC Options): Likewise.
9132 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
9133 constraints.md instead of rs6000.h. Reorder w* constraints. Add
9134 wm, wn, wr documentation.
9136 * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
9137 registers if direct move instructions are enabled.
9138 (wn): New constraint for no registers.
9139 (wq): New constraint for quad word even GPR registers.
9140 (wr): New constraint if 64-bit instructions are enabled.
9141 (wv): New constraint if power8 vector instructions are enabled.
9142 (wQ): New constraint for quad word memory locations.
9144 * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
9145 constraint for 0..15 for crypto instructions.
9146 (gpc_reg_operand): If VSX allow registers in VSX registers as well
9147 as GPR and floating point registers.
9148 (int_reg_operand): New predicate to match only GPR registers.
9149 (base_reg_operand): New predicate to match base registers.
9150 (quad_int_reg_operand): New predicate to match even GPR registers
9151 for quad memory operations.
9152 (vsx_reg_or_cint_operand): New predicate to allow vector logical
9153 operations in both GPR and VSX registers.
9154 (quad_memory_operand): New predicate for quad memory operations.
9155 (reg_or_indexed_operand): New predicate for direct move support.
9157 * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
9158 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
9159 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
9160 (POWERPC_MASKS): Add power8 options.
9161 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
9164 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
9165 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
9167 * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
9168 (-mpower8-fusion): New power8 options.
9169 (-mpower8-fusion-sign): Likewise.
9170 (-mpower8-vector): Likewise.
9171 (-mcrypto): Likewise.
9172 (-mdirect-move): Likewise.
9173 (-mquad-memory): Likewise.
9175 * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
9177 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
9179 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
9180 (rs6000_debug_vector_unit): Add p8_vector.
9181 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
9182 definitions. Also print fusion state.
9183 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
9184 (rs6000_builtin_mask_calculate): Add power8 builtin support.
9185 (rs6000_option_override_internal): Add support for power8.
9186 (rs6000_common_init_builtins): Add debugging for skipped builtins
9188 (rs6000_adjust_cost): Add power8 support.
9189 (rs6000_issue_rate): Likewise.
9190 (insn_must_be_first_in_group): Likewise.
9191 (insn_must_be_last_in_group): Likewise.
9192 (force_new_group): Likewise.
9193 (rs6000_register_move_cost): Likewise.
9194 (rs6000_opt_masks): Likewise.
9196 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
9197 power8 capable assembler, default to power7 options.
9198 (TARGET_DIRECT_MOVE): Likewise.
9199 (TARGET_CRYPTO): Likewise.
9200 (TARGET_P8_VECTOR): Likewise.
9201 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
9202 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
9203 (VECTOR_MEM_P8_VECTOR_P): Likewise.
9204 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
9205 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
9206 (TARGET_XSCVDPSPN): Likewise.
9207 (TARGET_XSCVSPDPN): Likewsie.
9208 (TARGET_SYNC_HI_QI): Likewise.
9209 (TARGET_SYNC_TI): Likewise.
9210 (MASK_CRYPTO): Likewise.
9211 (MASK_DIRECT_MOVE): Likewise.
9212 (MASK_P8_FUSION): Likewise.
9213 (MASK_P8_VECTOR): Likewise.
9214 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
9215 temporary used by some of the direct move instructions to get two FP
9216 temporary registers does not force creation of a stack frame.
9217 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
9218 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
9219 that any VSX registers are tieable, even if they are also an
9220 Altivec vector mode.
9221 (r6000_reg_class_enum): Add wm, wr, wv constraints.
9222 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
9223 (RS6000_BTM_CRYPTO): Likewise.
9224 (RS6000_BTM_COMMON): Likewise.
9226 * config/rs6000/rs6000.md (cpu attribute): Add power8.
9227 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
9228 (enum rs6000_vector): Add power8 vector support.
9230 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9234 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
9235 (any_sibcall_could_use_r3): this and handle indirect calls.
9236 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
9238 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9240 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
9242 2013-05-22 Richard Biener <rguenther@suse.de>
9245 * profile.c (branch_prob): Do not split blocks that are
9246 abnormally receiving from ECF_RETURNS_TWICE functions.
9248 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
9250 * recog.c (offsettable_address_addr_space_p): Fix calculation of
9251 address mode. Move pointer mode initialization to the same place.
9253 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
9255 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
9256 while it has any effect.
9258 2013-05-21 Easwaran Raman <eraman@google.com>
9260 PR tree-optimization/57322
9261 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
9262 UID of the statement added to the BB to be 1.
9264 2013-05-21 Jakub Jelinek <jakub@redhat.com>
9266 PR tree-optimization/57331
9267 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
9268 of conversion from pointer type to integral type with integer.
9270 2013-05-21 Martin Jambor <mjambor@suse.cz>
9273 * ipa-prop.c (ipa_read_node_info): Process param_used and
9274 controlled_uses in the same order as when writing.
9276 2013-05-21 Magnus Granberg <baldrick@free.fr>
9279 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
9281 2013-05-21 Richard Biener <rguenther@suse.de>
9283 PR tree-optimization/57318
9284 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
9285 estimate stmts with side-effects as likely eliminated.
9287 2013-05-21 Richard Biener <rguenther@suse.de>
9289 PR tree-optimization/57330
9290 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
9291 preserve the call stmts fntype.
9293 2013-05-21 Richard Biener <rguenther@suse.de>
9295 PR tree-optimization/57303
9296 * tree-ssa-sink.c (statement_sink_location): Improve killing
9297 stmt detection and properly handle self-assignments.
9299 2013-05-21 Christian Bruel <christian.bruel@st.com>
9301 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
9302 spanning registers. LEAF_REG_REMAP is supported only for contiguous
9303 registers. Set register size out of the PARALLEL loop.
9305 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
9308 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
9309 (*fmasf4, *fmasf4_media): New insns.
9311 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9313 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
9314 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
9315 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
9316 (mips_idiv_insns): Update the comments to say that the returned
9317 instruction counts are in units of BASE_INSN_LENGTH.
9318 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
9319 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
9320 using 2 rather than 4 as the length of indirect MIPS16 and
9321 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
9322 length of a NOP. Don't divide MIPS16 lengths by 2.
9323 (mips16_split_long_branches): Assume a branch is long if the
9324 length is greater than 4 rather than 8.
9325 * config/mips/mips.md (length): Give MIPS16 lengths directly,
9326 rather than multiplying them by 2. Multiply instruction counts
9327 by BASE_INSN_LENGTH rather than 4.
9328 (*jump_mips16, tls_get_tp_mips16_<mode>)
9329 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
9331 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9333 * config/mips/mips.md (extended_mips16): Remove branch case.
9334 (length): Remove duplicated extended_mips16 test.
9336 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
9338 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
9340 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
9342 * recog.h (Recog_data): Rename to...
9343 (recog_data_d): ...this.
9344 (recog_data): Update accordingly.
9345 * recog.c (recog_data): Likewise.
9346 * reload.c (save_recog_data): Likewise.
9347 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
9348 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
9350 2013-05-17 Julian Brown <julian@codesourcery.com>
9352 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
9353 found in a REG_EQUAL note, invalidate it.
9355 2013-05-17 Easwaran Raman <eraman@google.com>
9357 * tree-ssa-reassoc.c (find_insert_point): New function.
9358 (insert_stmt_after): Likewise.
9359 (get_def_stmt): Likewise.
9360 (ensure_ops_are_available): Likewise.
9361 (not_dominated_by): Likewise.
9362 (rewrite_expr_tree): Do not move statements beyond what is
9363 necessary. Remove call to swap_ops_for_binary_stmt...
9364 (reassociate_bb): ... and move it here.
9365 (build_and_add_sum): Assign UIDs for new statements.
9366 (linearize_expr): Likewise.
9367 (do_reassoc): Renumber gimple statement UIDs.
9369 2013-05-17 Jan Hubicka <jh@suse.cz>
9371 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
9373 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
9374 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
9376 (output_weakrefs): Update.
9378 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
9379 Martin Jambor <mjambor@suse.cz>
9382 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
9383 value that corresponds to the given aggval is found in values vector.
9385 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
9387 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
9388 sse, sse2, sse3, ssse3 and sse4a flags to options.
9390 2013-05-17 David Malcolm <dmalcolm@redhat.com>
9392 * gengtype-state.c: (s_expr_writer): New class, to handle
9393 prettifying of output layout of s-expressions.
9394 (state_writer): New class, to write out gtype.state.
9395 (state_written_type_count): Move this variable into member data of
9397 (s_expr_writer::s_expr_writer): New code: constructor for new class
9398 (state_writer::state_writer(): ditto
9399 (s_expr_writer::write_new_line): New function
9400 (s_expr_writer::write_any_indent): ditto
9401 (s_expr_writer::begin_s_expr): ditto
9402 (s_expr_writer::end_s_expr): ditto
9403 (write_state_fileloc): convert to method of state_writer...
9404 (state_writer:: write_state_fileloc): ...and use methods of
9405 s_expr_writer to write indentation into the gtype.state output file
9406 to visually represent the hierarchical structure of the list
9408 (write_state_fields): ditto, renaming to...
9409 (state_writer::write_state_fields)
9410 (write_state_a_string): ditto, renaming to...
9411 (state_writer::write_state_a_string)
9412 (write_state_string_option): ditto, renaming to...
9413 (state_writer::write_state_string_option)
9414 (write_state_type_option): ditto, renaming to...
9415 (state_writer::write_state_type_option)
9416 (write_state_nested_option): ditto, renaming to...
9417 (state_writer::write_state_nested_option)
9418 (write_state_option): ditto, renaming to...
9419 (state_writer::write_state_option)
9420 (write_state_options): ditto, renaming to...
9421 (state_writer::write_state_options)
9422 (write_state_lang_bitmap): ditto, renaming to...
9423 (state_writer::write_state_lang_bitmap)
9424 (write_state_version): ditto, renaming to...
9425 (state_writer::write_state_version)
9426 (write_state_scalar_type): ditto, renaming to...
9427 (state_writer::write_state_scalar_type)
9428 (write_state_string_type): ditto, renaming to...
9429 (state_writer::write_state_string_type)
9430 (write_state_undefined_type): ditto, renaming to...
9431 (state_writer::write_state_undefined_type)
9432 (write_state_struct_union_type): ditto, renaming to...
9433 (state_writer::write_state_struct_union_type)
9434 (write_state_struct_type): ditto, renaming to...
9435 (state_writer::write_state_struct_type)
9436 (write_state_user_struct_type): ditto, renaming to...
9437 (state_writer::write_state_user_struct_type)
9438 (write_state_lang_struct_type): ditto, renaming to...
9439 (state_writer::write_state_lang_struct_type)
9440 (write_state_param_struct_type): ditto, renaming to...
9441 (state_writer::write_state_param_struct_type)
9442 (write_state_pointer_type): ditto, renaming to...
9443 (state_writer::write_state_pointer_type)
9444 (write_state_array_type): ditto, renaming to...
9445 (state_writer::write_state_array_type)
9446 (write_state_gc_used): ditto, renaming to...
9447 (state_writer::write_state_gc_used)
9448 (write_state_common_type_content): ditto, renaming to...
9449 (state_writer::write_state_common_type_content)
9450 (write_state_type): ditto, renaming to...
9451 (state_writer::write_state_type)
9452 (write_state_pair_list): ditto, renaming to...
9453 (state_writer::write_state_pair_list)
9454 (write_state_pair): ditto, renaming to...
9455 (state_writer::write_state_pair)
9456 (write_state_typedefs): ditto, renaming to...
9457 (state_writer::write_state_typedefs)
9458 (write_state_structures): ditto, renaming to...
9459 (state_writer::write_state_structures)
9460 (write_state_param_structs): ditto, renaming to...
9461 (state_writer::write_state_param_structs)
9462 (write_state_variables): ditto, renaming to...
9463 (state_writer::write_state_variables)
9464 (write_state_srcdir): ditto, renaming to...
9465 (state_writer::write_state_srcdir)
9466 (write_state_files_list): ditto, renaming to...
9467 (state_writer::write_state_files_list)
9468 (write_state_languages): ditto, renaming to...
9469 (state_writer::write_state_languages)
9470 (write_state): create a state_writer instance and use it when
9471 writing out the state file
9473 2013-05-17 Mike Stump <mikestump@comcast.net>
9475 PR rtl-optimization/57304
9476 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
9477 accessing DF_REF_REAL_LOC.
9479 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9481 PR rtl-optimization/57281
9482 PR rtl-optimization/57300
9483 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
9484 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
9485 what the other splitter did if the registers are dead.
9487 2013-05-17 Richard Biener <rguenther@suse.de>
9489 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
9492 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9494 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
9497 2013-05-17 Marek Polacek <polacek@redhat.com>
9499 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
9500 length when doing non-zero store of storing '\0' to '\0'.
9502 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9504 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
9505 vect_external_def oprnd1 with loop_vinfo, try to emit
9506 optional cast, negation and and stmts on the loop preheader
9507 edge instead of into the pattern def seq.
9509 PR tree-optimization/57051
9510 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
9511 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
9513 2013-05-16 Nick Clifton <nickc@redhat.com>
9515 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
9516 (rl78_is_naked_func): New function.
9517 (rl78_expand_prologue): Skip prologue generation for naked functions.
9518 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
9519 * doc/extend.texi (naked): Add RL78 to the list of processors
9520 that supports this attribute.
9522 2013-05-16 Jeff Law <law@redhat.com>
9524 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
9526 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
9528 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
9529 cache parameters using detect_caches_amd also for CYRIX,
9530 NSC and TM2 signatures.
9532 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
9533 Dzianis Kahanovich <mahatma@eu.by>
9537 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9538 VIA/Centaur processors and determine their cache parameters
9539 using detect_caches_amd.
9541 2013-05-16 Teresa Johnson <tejohnson@google.com>
9543 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
9544 (rtl_verify_edges): New function.
9545 (rtl_verify_bb_insns): Ditto.
9546 (rtl_verify_bb_pointers): Ditto.
9547 (rtl_verify_bb_insn_chain): Ditto.
9548 (rtl_verify_fallthru): Ditto.
9549 (rtl_verify_bb_layout): Ditto.
9550 (rtl_verify_flow_info_1): Outline checks into new functions.
9551 (rtl_verify_flow_info): Ditto.
9553 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
9555 * cfghooks.c (copy_bbs): Add update_dominance argument.
9556 * cfghooks.h (copy_bbs): Update prototype.
9557 * tree-cfg.c (gimple_duplicate_sese_region):
9558 Add update_dominance argument.
9559 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
9560 * tree-ssa-loop-ch.c (copy_loop_headers): Update
9561 gimple_duplicate_sese_region call.
9562 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
9563 Update copy_bbs call.
9564 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
9565 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
9567 2013-05-16 Jakub Jelinek <jakub@redhat.com>
9569 * tree-vectorizer.h (NUM_PATTERNS): Increment.
9570 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
9571 vect_recog_rotate_pattern.
9572 (vect_recog_rotate_pattern): New function.
9574 2013-05-16 Jason Merrill <jason@redhat.com>
9576 * Makefile.in (LLINKER): New variable.
9577 (mostlyclean): Remove link mutex.
9578 * configure.ac: Handle --enable-link-mutex.
9579 * lock-and-run.sh: New script.
9581 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9584 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
9587 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9589 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
9591 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
9593 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
9594 * config/arm/arm.c (next_consecutive_mem): New function.
9595 (gen_movmem_ldrd_strd): Likewise.
9596 * config/arm/arm.md (movmemqi): Update condition and code.
9597 (unaligned_loaddi, unaligned_storedi): New patterns.
9599 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9601 * config.gcc: Obsolete *-*-solaris2.9*.
9602 * doc/install.texi (Specific, *-*-solaris2*): Document it.
9604 2013-05-16 Richard Biener <rguenther@suse.de>
9606 * passes.c (init_optimization_passes): Move pass_parallelize_loops
9607 earlier, after GRAPHITE transforms and IV canonicalization.
9609 2013-05-16 Jakub Jelinek <jakub@redhat.com>
9611 * omp-low.c (extract_omp_for_data): For collapsed loops,
9612 if at least one of the loops is known at compile time to
9613 iterate zero times, set count to 0.
9614 (expand_omp_regimplify_p): New function.
9615 (expand_omp_for_generic): For collapsed loops, if at least
9616 one of the loops isn't known to iterate at least once,
9617 add runtime check with setting count to 0.
9618 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9619 For unsigned types if it isn't known at compile time that
9620 the loop will iterate at least once, add runtime check to bypass
9621 the whole loop if initial condition isn't true.
9623 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
9625 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9627 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
9630 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9631 transformations to avoid an infinite loop.
9633 2013-05-16 Marek Polacek <polacek@redhat.com>
9635 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9637 2013-05-15 Leif Ekblad <leif@rdos.net>
9639 * config/i386/i386.c (ix86_decompose_address): Use
9640 DEFAULT_TLS_SEG_REG to access TLS segment register.
9641 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9642 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9643 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9645 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
9648 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9649 sibling calls to functions that would normally be lazily bound,
9650 unless $gp is call-clobbered.
9652 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
9654 * config/i386/i386.c (ix86_option_override_internal): Update
9655 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
9656 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9657 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9658 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9660 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9662 2013-05-15 Andreas Schwab <schwab@suse.de>
9664 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9665 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
9668 2013-05-15 Teresa Johnson <tejohnson@google.com>
9670 * loop-unroll.c (report_unroll_peel): Check decision before
9671 emitting unroll/peel message.
9673 2013-05-15 Teresa Johnson <tejohnson@google.com>
9675 * function.h (has_bb_partition): New rtl_data flag.
9676 (bb_reorder_complete): Ditto.
9677 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9678 instead of flag_reorder_blocks_and_partition.
9679 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9680 with some enhancements.
9681 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9682 * bb-reorder.c (connect_traces): Check for has_bb_partition
9683 instead of flag_reorder_blocks_and_partition.
9684 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9685 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9686 verify_hot_cold_block_grouping.
9687 (partition_hot_cold_basic_blocks): Set has_bb_partition.
9689 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9692 * config/arm/predicates.md (call_insn_operand): New predicate.
9693 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
9694 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9695 if insn is not a tail call.
9696 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9698 * config/arm/arm.h (enum reg_class): New caller save register class.
9699 (REG_CLASS_NAMES): Likewise.
9700 (REG_CLASS_CONTENTS): Likewise.
9701 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9704 2013-05-15 Richard Biener <rguenther@suse.de>
9706 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9707 of MSG_OPTIMIZED_LOCATIONS.
9708 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9709 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9711 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9712 of MSG_OPTIMIZED_LOCATIONS.
9713 (execute_vect_slp): Likewise.
9714 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9715 (vect_create_cond_for_alias_checks): Likewise.
9716 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9717 (vect_recog_widen_mult_pattern): Likewise.
9718 (vect_recog_widen_sum_pattern): Likewise.
9719 (vect_recog_over_widening_pattern): Likewise.
9720 (vect_recog_widen_shift_pattern): Likewise.
9721 (vect_recog_vector_vector_shift_pattern): Likewise.
9722 (vect_recog_divmod_pattern): Likewise.
9723 (vect_recog_mixed_size_cond_pattern): Likewise.
9724 (vect_recog_bool_pattern): Likewise.
9725 (vect_pattern_recog_1): Likewise.
9727 2013-05-15 Martin Jambor <mjambor@suse.cz>
9729 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9730 non-functions to builtin_unreachable.
9731 * ipa-inline-transform.c (inline_call): Do not assert estimates were
9732 correct when new direct edges were discovered.
9734 2013-05-15 Martin Jambor <mjambor@suse.cz>
9736 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9737 header, print symbol order instead of node uid, print more information
9738 about indirect edge targets.
9739 (ipa_make_edge_direct_to_target): Print symbol order instead of node
9741 (ipa_make_edge_direct_to_target): Likewise.
9742 (remove_described_reference): Likewise.
9743 (propagate_controlled_uses): Likewise.
9744 (ipa_print_node_params): Also print symbol order.
9745 (ipcp_transform_function): Print symbol order instead of node uids.
9746 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9747 (cgraph_get_create_real_symbol_node): Likewise.
9748 * ipa-cp.c (print_lattice): Likewise.
9749 (print_all_lattices): Likewise.
9750 (determine_versionability): Likewise.
9751 (initialize_node_lattices): Likewise.
9752 (estimate_local_effects): Likewise.
9753 (update_profiling_info): Likewise.
9754 (create_specialized_node): Likewise.
9755 (perhaps_add_new_callers): Likewise.
9756 (decide_about_value): Likewise.
9757 (decide_whether_version_node): Likewise.
9758 (identify_dead_nodes): Likewise.
9759 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9760 (dump_inline_summary): Likewise.
9761 (estimate_node_size_and_time): Likewise.
9762 (inline_analyze_function): Likewise.
9763 * ipa-inline.c (report_inline_failed_reason): Likewise.
9764 (want_early_inline_function_p): Likewise.
9765 (edge_badness): Likewise.
9766 (update_edge_key): Likewise.
9767 (inline_small_functions): Likewise. Add dumping of order to two other
9769 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9770 instead of node uids.
9771 (propagate_pure_const): Likewise.
9772 (propagate_pure_const): Likewise.
9773 * ipa-utils.c (dump_cgraph_node_set): Likewise.
9774 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9775 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9777 * tree-pretty-print.c (dump_function_header): Likewise.
9778 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9779 Print symbol order instead of node uids.
9781 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9783 * config/s390/s390.c (s390_register_move_cost): Don't impose the
9784 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9786 2013-05-15 Richard Biener <rguenther@suse.de>
9788 PR tree-optimization/57275
9789 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9790 return value for fail to do runtime alias checks for gather loads.
9792 2013-05-15 Jan Hubicka <jh@suse.cz>
9796 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9797 weakrefs are not external.
9798 (lto_symtab_merge_decls): Fix thinko when dealing with
9799 non-lto_symtab decls.
9800 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9801 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9802 * varpool.c (dump_varpool_node): Dump more flags.
9804 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
9806 * config/i386/i386.c (processor_alias_table): Add instruction
9807 FSGSBASE for AMD bdver3 architecture.
9809 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9811 * tree.c (warn_deprecated_use): Print file:line using locus color.
9812 * diagnostic.c (diagnostic_report_current_module): Print file:line
9813 and file:line:column using locus color.
9815 2013-05-14 Mike Stump <mikestump@comcast.net>
9817 * gdbinit.in: Add __null.
9819 2013-05-14 Mike Stump <mikestump@comcast.net>
9821 * recog.h: Rename struct recog_data to Recog_data.
9822 * recog.c: Likewise.
9823 * reload.c (can_reload_into): Likewise.
9824 * config/picochip/picochip.c: Likewise.
9826 2013-05-14 Mike Stump <mikestump@comcast.net>
9828 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9830 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
9832 * resource.h (struct resources): Remove unch_memory member.
9833 (CLEAR_RESOURCE): Don't clear unch_memory.
9834 * resource.c (mark_referenced_resources): Don't set it.
9835 (mark_set_resources): Likewise.
9836 (mark_target_live_regs): Don't clear it.
9837 (init_resource_info): Likewise.
9838 * reorg.c (resource_conflicts_p): Don't compare it.
9839 (redundant_insn): Don't set it.
9841 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9843 * emit-rtl.c (next_label): Remove unused function.
9844 (skip_consecutive_labels, link_cc0_insns): Move to ...
9845 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9846 only place where these functions are used, and make them static.
9848 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
9850 * fold-const.c (fold_negate_expr): Handle vectors.
9851 (fold_truth_not_expr): Make it static.
9852 (fold_invert_truthvalue): New static function.
9853 (invert_truthvalue_loc): Handle vectors. Do not call
9854 fold_truth_not_expr directly.
9855 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9856 <TRUTH_NOT_EXPR>: Do not cast to boolean.
9857 (fold_comparison): Handle vector constants.
9858 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9859 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9860 * tree.h (fold_truth_not_expr): Remove declaration.
9862 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
9864 * config/aarch64/aarch64-simd.md
9865 (aarch64_vcond_internal<mode>): Rename to...
9866 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9867 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9868 float modes. Clarify all iterator modes.
9869 (vcond<mode><mode>): Use new name for vcond expanders.
9870 (vcond<v_cmp_result><mode>): Likewise.
9871 (vcondu<mode><mode>: Likewise.
9872 * config/aarch64/iterators.md (VDQF_COND): New.
9874 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
9877 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9878 variable for the shift amount. Check that we shift by non-negative
9881 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
9884 * config/arm/arm.h (EPILOGUE_USES): Only return true
9885 for LR_REGNUM after epilogue_completed.
9887 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
9889 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9892 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9895 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9896 and Solaris 11+/x86 with gld.
9897 * configure: Regenerate.
9899 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9901 * expmed.c (expand_shift_1): Canonicalize rotates by
9902 constant bitsize / 2 to bitsize - 1.
9903 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9904 case ROTATERT>: Likewise.
9907 2013-05-10 Jakub Jelinek <jakub@redhat.com>
9909 * config/i386/i386.md (rotateinv): New code attr.
9910 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9911 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9912 roll $31, %eax, etc.
9914 2013-05-14 Richard Biener <rguenther@suse.de>
9917 * tree-eh.c (sink_clobbers): Give up for successors with
9918 multiple predecessors and no virtual uses.
9920 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
9922 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9923 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9925 2013-05-14 Jakub Jelinek <jakub@redhat.com>
9928 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9929 the case when both op0 and op1 have VOIDmode.
9931 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
9933 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9934 in multiply-accumulate mode.
9936 2013-05-13 Guozhi Wei <carrot@google.com>
9938 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9940 2013-05-13 Kai Tietz <ktietz@redhat.com>
9943 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9944 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9945 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9946 * config/i386/i386.c (ix86_option_override_internal): Likewise.
9947 (ix86_expand_prologue): Likewise.
9948 (ix86_expand_split_stack_prologue): Likewise.
9949 (legitimate_pic_address_disp_p): Likewise.
9950 (legitimize_pic_address): Likewise.
9951 (legitimize_tls_address): Likewise.
9952 (legitimize_pe_coff_symbol): Likewise.
9953 (output_pic_addr_const): Likewise.
9954 (construct_plt_address): Likewise.
9955 (ix86_expand_call): Likewise.
9956 (x86_output_mi_thunk): Likewise.
9957 (x86_function_profiler): Likewise.
9959 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
9961 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9962 similar switch cases.
9963 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9964 (aarch64_simd_mov_to_<mode>low): Delete.
9965 (aarch64_simd_mov_to_<mode>high): Delete.
9966 (move_lo_quad_<mode>): Add w<-r alternative.
9967 (aarch64_simd_move_hi_quad_<mode>): Likewise.
9968 (aarch64_simd_mov_from_*): Update type attribute.
9969 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9972 2013-05-13 Jan Hubicka <jh@suse.cz>
9974 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9975 * config/i386/i386.c (ix86_compute_frame_layout,
9976 ix86_expand_epilogue, emit_i387_cw_initialization,
9977 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9978 ix86_local_alignment): Fix use of size/speed predicates.
9980 2013-05-13 Jakub Jelinek <jakub@redhat.com>
9982 PR tree-optimization/45216
9983 PR tree-optimization/57157
9984 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9985 the (-Y) & (B - 1) variant if OP is |.
9986 * expmed.c (expand_shift_1): For rotations by const0_rtx just
9987 return shifted. Use (-op1) & (prec - 1) as other_amount
9988 instead of prec - op1.
9990 2013-05-13 Martin Jambor <mjambor@suse.cz>
9993 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9994 (ipa_constant_data): New type.
9995 (ipa_jump_func): Use ipa_constant_data to hold information about
9996 constant jump functions.
9997 (ipa_get_jf_constant): Adjust to jump function type changes.
9998 (ipa_get_jf_constant_rdesc): New function.
9999 (ipa_param_descriptor): New field controlled_uses.
10000 (ipa_get_controlled_uses): New function.
10001 (ipa_set_controlled_uses): Likewise.
10002 * ipa-ref.h (ipa_find_reference): Declare.
10003 * ipa-prop.c (ipa_cst_ref_desc): New type.
10004 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
10006 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
10007 New parameter cs. Adjust all callers.
10008 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
10009 (remove_described_reference): New function.
10010 (jfunc_rdesc_usable): Likewise.
10011 (try_make_edge_direct_simple_call): Decrement controlled use count,
10012 attempt to remove reference if it hits zero.
10013 (combine_controlled_uses_counters): New function.
10014 (propagate_controlled_uses): Likewise.
10015 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
10016 (ipa_edge_duplication_hook): Duplicate reference descriptions.
10017 (ipa_print_node_params): Print described use counter.
10018 (ipa_write_jump_function): Adjust to jump function type changes.
10019 (ipa_read_jump_function): New parameter CS, pass it to
10020 ipa_set_jf_constant. Adjust caller.
10021 (ipa_write_node_info): Stream controlled use count
10022 (ipa_read_node_info): Likewise.
10023 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
10025 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
10026 count. Remove cloning-added reference if it reaches zero.
10027 * ipa-ref.c (ipa_find_reference): New function.
10029 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
10031 * config/i386/i386.c (processor_target_table): Modified default
10032 alignment values for AMD BD and BT architectures.
10034 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
10036 * tree-vect-generic.c (uniform_vector_p): Move ...
10037 * tree.c (uniform_vector_p): ... here.
10038 * tree.h (uniform_vector_p): Declare it.
10039 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
10042 2013-05-13 Jakub Jelinek <jakub@redhat.com>
10044 PR tree-optimization/57230
10045 * tree-ssa-strlen.c (handle_char_store): Record length for
10046 array store from STRING_CST.
10048 PR tree-optimization/57230
10049 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
10052 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
10054 * config/epiphany/epiphany.c (epiphany_init): Check size of
10055 NUM_MODES_FOR_MODE_SWITCHING.
10056 (epiphany_expand_prologue):
10057 Remove CONFIG_REGNUM initial value handling code.
10058 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
10059 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
10060 (emit_set_fp_mode, epiphany_mode_after): Likewise.
10061 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
10062 Don't return 1 for FP_MODE_NONE.
10063 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
10064 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
10065 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
10066 * config/epiphany/epiphany.md (save_config): New pattern.
10068 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
10070 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
10072 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
10074 * config/i386/i386.md (memory): Handle sseishft1.
10075 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
10076 (*vec_extractv2di_1): Ditto.
10078 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
10080 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
10083 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
10085 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
10086 Add mthumb/march=armv7-a multilib.
10087 Add mthumb/march=armv7-r multilib.
10088 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
10090 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
10092 * config/v850/t-rtems: Add more multilibs.
10094 2013-05-10 Richard Biener <rguenther@suse.de>
10096 PR tree-optimization/57214
10097 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
10098 not propagate from SSA names that occur in abnormal PHI nodes.
10100 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
10102 * stor-layout.c (element_precision): New function.
10103 * machmode.h (element_precision): Declare it.
10104 * tree.c (build_minus_one_cst): New function.
10105 (element_precision): Likewise.
10106 * tree.h (build_minus_one_cst): Declare new function.
10107 (element_precision): Likewise.
10108 * fold-const.c (operand_equal_p): Use element_precision.
10109 (fold_binary_loc): Handle vector types.
10110 * convert.c (convert_to_integer): Use element_precision.
10111 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
10114 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
10116 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
10117 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
10118 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
10119 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10120 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
10121 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
10122 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
10123 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
10124 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
10125 (Uuw8): New constraints.
10126 (Usb4): Move into alphabetical order.
10127 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
10128 (sd8_operand, ub8_operand, uw8_operand): New predicates.
10129 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
10130 previously unnamed patterns.
10131 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
10132 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
10133 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
10134 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
10135 of set_attr_alternative/if_then_else. Use extended_mips16 instead
10136 of specific lengths.
10138 2013-05-10 Jakub Jelinek <jakub@redhat.com>
10140 * config/i386/i386.md (rotateinv): New code attr.
10141 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
10142 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
10143 roll $31, %eax, etc.
10145 PR tree-optimization/45216
10146 PR tree-optimization/57157
10147 * tree-ssa-forwprop.c (simplify_rotate): New function.
10148 (ssa_forward_propagate_and_combine): Call it.
10150 2013-05-10 Richard Biener <rguenther@suse.de>
10152 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
10153 disable peeling when we version for aliasing.
10154 (vector_alignment_reachable_p): Honor explicit user alignment.
10155 (vect_supportable_dr_alignment): Likewise.
10156 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
10157 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
10158 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
10159 then peeling to arrange for the cost-model check to come first.
10161 2013-05-10 Alan Modra <amodra@gmail.com>
10163 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
10164 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
10165 * configure: Regenerate.
10167 2013-05-10 Alan Modra <amodra@gmail.com>
10170 * varasm.c (default_elf_select_section): Move !DECL_P check..
10171 (get_named_section): ..to here before calling get_section_name.
10173 (default_section_type_flags): Add DECL_P check.
10174 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
10175 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
10177 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
10179 * config/epiphany/epiphany.c (epiphany_expand_prologue):
10180 When using gen_stack_adjust_str with a register offset, add a
10181 REG_FRAME_RELATED_EXPR note.
10183 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
10185 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
10186 (*vec_extractv4si_zext_mem): Ditto.
10187 (*vec_extractv2di): Add 0->x and x->x alternatives.
10188 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
10189 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
10191 2013-05-09 Jason Merrill <jason@redhat.com>
10193 N3639 C++1y VLA support
10194 * gimplify.c (gimplify_vla_decl): Don't touch an existing
10197 * tree.c (build_constructor_va): New.
10198 * tree.h: Declare it.
10200 2013-05-09 Martin Jambor <mjambor@suse.cz>
10203 * gimple-fold.c (canonicalize_constructor_val): Call
10204 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
10206 2013-05-09 Jan Hubicka <jh@suse.cz>
10207 Richard Biener <rguenther@suse.de>
10210 * symtab.c (symtab_make_decl_local): Do not add private names.
10212 2013-05-09 Jan Hubicka <jh@suse.cz>
10215 * symtab.c (insert_to_assembler_name_hash): Handle clones.
10216 (unlink_from_assembler_name_hash): Likewise.
10217 (symtab_prevail_in_asm_name_hash, symtab_register_node,
10218 symtab_unregister_node, symtab_initialize_asm_name_hash,
10219 change_decl_assembler_name): Update.
10221 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
10223 * config/aarch64/aarch64.md: New movtf split.
10224 (*movtf_aarch64): Update.
10225 (aarch64_movdi_tilow): Handle TF modes and rename to
10226 aarch64_movdi_<mode>low.
10227 (aarch64_movdi_tihigh): Handle TF modes and rename to
10228 aarch64_movdi_<mode>high
10229 (aarch64_movtihigh_di): Handle TF modes and rename to
10230 aarch64_mov<mode>high_di
10231 (aarch64_movtilow_di): Handle TF modes and rename to
10232 aarch64_mov<mode>low_di
10233 (aarch64_movtilow_tilow): Remove spurious whitespace.
10234 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
10236 (aarch64_print_operand): Update.
10238 2013-05-09 Alan Modra <amodra@gmail.com>
10240 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
10242 * configure: Regenerate.
10244 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
10246 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
10247 splitter preparation statements.
10248 * config/i386/sse.md (*vec_extract* splitters): Ditto.
10249 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
10252 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10254 * gimple-ssa-strength-reduction.c (count_candidates): Change
10255 return value to int.
10256 (analyze_candidates_and_replace): Change type of length to int.
10258 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
10260 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
10261 (*vec_extract<mode>): Use VI12_128 mode iterator.
10262 (*vec_extract<mode>_mem): Ditto.
10263 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
10266 2013-05-08 Diego Novillo <dnovillo@google.com>
10271 2012-08-17 Diego Novillo <dnovillo@google.com>
10274 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
10275 * config.in: Regenerate.
10276 * configure: Regenerate.
10277 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
10279 2013-05-08 Jan Hubicka <jh@suse.cz>
10282 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
10283 * cgraph.h (symtab_node_base): Add unique_name.
10284 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10285 input_overwrite_node, input_varpool_node): Stream unique_name.
10286 * cgraphclones.c (cgraph_create_virtual_clone,
10287 cgraph_function_versioning): Set unique_name.
10288 * ipa.c (function_and_variable_visibility): Set unique_name.
10290 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10292 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
10293 (alloc_cand_and_find_basis): Restrict conditional candidate
10294 processing to CAND_MULTs.
10296 2013-05-08 Jan Hubicka <jh@suse.cz>
10299 lto-symtab.c (lto_symtab_symbol_p): New function.
10300 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
10301 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
10302 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
10303 Skip static symbols.
10305 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
10307 PR tree-optimization/57200
10308 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
10309 Only call inform if the preceding warning_at returns true.
10311 2013-05-07 Han Shen <shenhan@google.com>
10313 * cfgexpand.c (record_or_union_type_has_array_p): New function.
10314 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
10315 * common.opt (fstack-protector-strong): New option.
10316 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
10317 * doc/invoke.texi (Optimization Options): Document
10318 "-fstack-protector-strong".
10319 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
10321 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
10323 * config/mips/mips.c (mips_machine_reorg2): Return 0.
10325 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
10327 * ira.c (update_equiv_regs): Add insn having equiv memory even if
10328 it is not lhs of the insn.
10329 (setup_reg_equiv): Remove insn having equiv memory which it is not
10331 * lra-constraints.c (process_address): Try to improve generation
10332 code for address base + disp.
10333 (lra_constraints): Make correct the code for checking insn setting
10334 up backward equivalence. Remove insn only if it is in the init
10336 * lra-eliminations.c (update_reg_eliminate): Change return value.
10337 (lra_eliminate): Use the result.
10339 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
10341 * config/i386/sse.md (ssescalarnummask): New mode attribute.
10342 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
10343 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
10344 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
10345 register target operands.
10346 (*vec_extractv8hi_sse2): New pattern.
10347 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
10348 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
10349 (*vec_extract<mode>_mem): New insn and split pattern.
10351 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
10353 * config/arm/arm.c (arm_asan_shadow_offset): New function.
10354 (TARGET_ASAN_SHADOW_OFFSET): Define.
10355 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
10356 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
10358 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10360 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
10361 (incr_vec_index): Return -1 if increment not found.
10362 (create_add_on_incoming_edge): Assert if increment not found.
10363 (record_increment): Limit number of increments recorded.
10364 (all_phi_incrs_profitable): Return false if an increment not found.
10365 (replace_profitable_candidates): Don't process increments that were
10367 (analyze_candidates_and_replace): Limit size of incr_vec.
10369 2013-05-07 Richard Biener <rguenther@suse.de>
10371 * calls.c (special_function_p): setjmp-like functions are leaf.
10372 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
10373 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
10375 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
10377 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
10378 (aarch64_simd_mov<mode>): New expander.
10379 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
10380 (aarch64_simd_mov_to_<mode>high): Likewise.
10381 (aarch64_simd_mov_from_<mode>low): Likewise.
10382 (aarch64_simd_mov_from_<mode>high): Likewise.
10383 (aarch64_dup_lane<mode>): Update.
10384 (aarch64_dup_lanedi): New instruction pattern.
10385 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
10386 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
10388 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10390 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
10391 (replace_mult_candidate): Remove unnecessary argument; remove
10392 unnecessary parameter from call to introduce_cast_before_cand.
10393 (replace_unconditional_candidate): Remove unnecessary parameter
10394 from call to replace_mult_candidate.
10395 (replace_conditional_candidate): Likewise.
10396 (insert_initializers): Use make_temp_ssa_name.
10397 (introduce_cast_before_cand): Remove unnecessary argument; use
10398 make_temp_ssa_name.
10399 (replace_one_candidate): Remove unnecessary argument; remove
10400 unnecessary parameter from calls to introduce_cast_before_cand.
10401 (replace_profitable_candidates): Remove unnecessary parameters
10402 from calls to replace_one_candidate.
10404 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10406 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
10407 phi def as possibly hiding a basis for a CAND_ADD whose operands
10408 have been commuted in the analysis.
10409 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
10411 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
10413 * config/aarch64/aarch64.md
10414 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
10415 shift value between 0-4.
10417 2013-05-07 Richard Biener <rguenther@suse.de>
10419 * double-int.h (rshift): New overload.
10420 * double-int.c (rshift): New function.
10421 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
10422 (create_reference_ops_from_ref): Remove.
10423 (vn_reference_insert): Use shared ops for constructing the
10424 reference and copy it.
10426 2013-05-07 Richard Biener <rguenther@suse.de>
10428 PR middle-end/57190
10429 * tree-eh.c (sink_clobbers): Properly propagate
10430 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
10432 2013-05-07 Jakub Jelinek <jakub@redhat.com>
10434 PR tree-optimization/57149
10435 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
10436 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
10437 collect_phi_def_edges, execute_late_warn_uninitialized): Use
10438 uninit_undefined_value_p instead of ssa_undefined_value_p.
10441 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
10442 for modifier == EXPAND_INITIALIZER.
10444 2013-05-07 Anton Blanchard <anton@samba.org>
10446 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
10447 for powerpc64 little endian.
10448 * configure: Regenerate.
10450 2013-05-06 Graham Stott <grahams@btinternet.com>
10452 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
10453 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
10454 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
10455 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
10457 2013-05-06 Graham Stott <grahams@btinternet.com>
10459 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
10460 codes which allow non-lvalues.
10462 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
10464 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
10465 components are all 1s.
10466 (integer_minus_onep): New function.
10467 * tree.h (integer_minus_onep): Declare it.
10468 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
10469 integer_minus_onep instead of integer_all_onesp.
10471 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10474 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
10475 variations of these patterns.
10477 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
10479 * config/i386/i386.md (isa): Add x64_sse4 member.
10480 (enabled): Handle x64_sse4.
10481 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
10482 instruction for 64bit SSE4_1 targets. Update insn attributes.
10483 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
10484 instruction for SSE4_1 targets. Update insn attributes.
10485 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10486 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
10487 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
10489 (*vec_extractv4si): Rename from *sse4_1_pextrd.
10490 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
10491 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
10493 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10496 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
10498 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
10500 * final.c (do_assembler_dialects): Don't handle curly braces and
10501 vertical bar escaped by % as dialect delimiters.
10502 (output_asm_insn): Print curly braces and vertical bar if escaped
10503 by % and ASSEMBLER_DIALECT defined.
10504 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
10505 * doc/tm.texi: Regenerated.
10507 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
10509 * config/mips/mips.c: Include tree-pass.h.
10510 (mips_reorg): Split in pre- and post-dbr_schedule parts.
10511 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
10512 (pass_mips_machine_reorg2): New machine specific pass.
10513 (insert_pass_mips_machine_reorg2): New pass plugin definition.
10514 (mips_option_override): Register the new pass.
10515 * rtl.h (cleanup_barriers): Remove prototype.
10516 (dbr_schedule): Likewise.
10517 * jump.c (cleanup_barriers): Make static.
10518 * reorg.c (dbr_schedule): Likewise.
10520 2013-05-06 Richard Biener <rguenther@suse.de>
10522 PR tree-optimization/57185
10523 * tree-parloops.c (add_field_for_reduction): Handle anonymous
10524 SSA names properly.
10526 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
10529 * config/i386/i386.c (add_parameter_dependencies): Add dependence
10530 between "first_arg" and "insn", not "last" and "insn".
10532 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10534 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
10535 (find_candidates_in_block): Re-enable slsr_process_phi.
10536 (create_phi_basis): Fix double counting of candidate adjustment.
10538 2013-05-06 Richard Biener <rguenther@suse.de>
10540 PR middle-end/57147
10541 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
10542 the edge is also fallthru, preserve it and just clear the
10544 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
10545 also complex, preserve that and just clear the fallthru flag.
10546 * tree-inline.c (update_ssa_across_abnormal_edges): Also
10547 update virtual operands.
10549 2013-05-06 Alan Modra <amodra@gmail.com>
10551 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
10552 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
10553 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
10554 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
10555 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10556 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
10558 2013-05-06 Alan Modra <amodra@gmail.com>
10560 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
10561 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
10562 (DEFAULT_ASM_ENDIAN): Define.
10563 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10564 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
10565 Update -K PIC clause from sysv4.h.
10566 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
10567 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
10569 2013-05-06 Alan Modra <amodra@gmail.com>
10571 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
10572 twice for little-endian.
10573 (ashrdi3_no_power, ashrdi3): Support little-endian.
10575 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
10578 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
10579 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
10581 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
10582 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
10583 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
10586 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
10587 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
10589 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
10590 * config/pa/pa.opt: Make mbig-switch a no-op.
10591 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
10592 (CASE_VECTOR_MODE): Always return SImode.
10593 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
10594 for the !TARGET_BIG_SWITCH case.
10595 * config/pa/pa-linux.h: Likewise.
10596 * config/pa/pa-openbsd.h: Likewise.
10597 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
10598 * config/pa/pa.md (short_jump): Remove define_insn.
10599 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
10600 (casesi0): Remove define_insn.
10601 (type): Remove btable_branch.
10602 (pa_combine_type): Likewise.
10603 (in_nullified_branch_delay): Likewise.
10604 (in_call_delay): Likewise.
10605 (define_delay): Likewise.
10606 (define_insn_reservation "Z3"): Likewise.
10607 (define_insn_reservation "Z4"): Likewise.
10608 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10609 (pa_adjust_insn_length): Remove adjustment for btable branches.
10610 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10613 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
10615 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10616 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10617 Add m->r,x alternatives.
10618 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10619 splitters using SWI48x mode iterator.
10620 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
10621 TARGET_64BIT. Add m->x alternative.
10622 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10623 Add o->x alternative. Enable for TARGET_SSE.
10624 (sse_storeq): Remove expander.
10625 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10626 with memory input operand.
10627 (*vec_extractv2di_1 splitter): New.
10628 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10629 * config/i386/i386.md (ssevecmodelower): New mode attribute.
10631 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
10633 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
10634 (INT_LOWPART): Delete.
10635 (extract_MB): Adjust.
10636 (extract_ME): Adjust.
10637 (print_operand): Adjust.
10639 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
10641 * config/rs6000/predicates.md (reg_or_add_cint_operand,
10642 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10643 (reg_or_logical_cint_operand, easy_fp_constant,
10644 logical_const_operand): Delete "CONST_DOUBLE" case.
10645 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10646 "HOST_BITS_PER_WIDE_INT == 64" test.
10647 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
10648 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10649 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10650 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10652 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10653 CONST_DOUBLE DImode/VOIDmode case.
10654 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10655 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
10656 CONST_DOUBLE VOIDmode case.
10657 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10658 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10659 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10660 Delete CONST_DOUBLE case.
10661 (splitters for mov FMOVE64 const_double): Delete
10662 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
10663 "HOST_BITS_PER_WIDE_INT >= 64" test.
10664 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10666 (mov DI const_double): Delete.
10668 2013-05-04 Jakub Jelinek <jakub@redhat.com>
10670 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10671 on op shows all bits zero in mode of a lowpart subreg, return zero.
10673 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10676 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10677 to save TFmode registers and DImode to save TImode registers for
10678 caller save operations.
10679 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10680 mark being partially clobbered since they only use the first
10683 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10684 and TDmode only use the upper 64-bits of each VSX register.
10686 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10688 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10689 (find_candidates_in_block): Disable slsr_process_phi.
10691 2013-05-03 Guozhi Wei <carrot@google.com>
10693 * coverage.c (coverage_obj_init): Move the construction of gcov
10695 (build_init_ctor): ... here.
10697 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10699 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10700 (slsr_cand_d): Redefine def_phi.
10701 (stride_status, phi_adjust_status, count_phis_status): New enums.
10702 (find_phi_def): New.
10703 (find_basis_for_base_expr): New.
10704 (find_basis_for_candidate): Handle hidden bases.
10705 (alloc_cand_and_find_basis): Handle phi candidates.
10706 (slsr_process_phi): New.
10707 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10708 (create_mul_imm_cand): Likewise.
10709 (create_add_ssa_cand): Exclude phi base candidates.
10710 (create_add_imm_cand): Likewise.
10711 (slsr_process_cast): Likewise.
10712 (slsr_process_copy): Likewise.
10713 (find_candidates_in_block): Handle phi candidates.
10714 (dump_candidate): Likewise.
10715 (unconditional_cands): Delete.
10716 (unconditional_cands_with_known_stride_p): Delete.
10717 (phi_dependent_cand_p): New.
10718 (cand_increment): Handle phi-dependent candidates.
10719 (replace_dependent): Delete.
10720 (replace_mult_candidate): New.
10721 (replace_unconditional_candidate): New.
10722 (incr_vec_index): Move to avoid forward reference.
10723 (create_add_on_incoming_edge): New.
10724 (create_phi_basis): New.
10725 (replace_dependents): Delete.
10726 (replace_conditional_candidate): New.
10727 (phi_add_costs): New.
10728 (replace_uncond_cands_and_profitable_phis): New.
10729 (record_increment): Handle phi adjustments.
10730 (record_phi_increments): New.
10731 (record_increments): Handle phi adjustments.
10732 (phi_incr_cost): New.
10733 (lowest_cost_path): Handle phis.
10734 (total_savings): Likewise.
10735 (analyze_increments): Likewise.
10736 (ncd_with_phi): New.
10737 (ncd_of_cand_and_phis): New.
10738 (nearest_common_dominator_for_cands): Handle phi increments.
10739 (all_phi_incrs_profitable): New.
10740 (replace_profitable_candidates): Handle phi-dependent candidates.
10741 (analyze_candidates_and_replace): Likewise.
10743 2013-05-03 Teresa Johnson <tejohnson@google.com>
10746 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10747 do not exceed REG_BR_PROB_BASE.
10749 2013-05-03 Jeff Law <law@redhat.com>
10751 PR tree-optimization/57144
10752 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10753 operand of the condition will bit into the new type when eliminating
10754 a cast feeding a condition.
10756 2013-05-03 Jakub Jelinek <jakub@redhat.com>
10758 PR rtl-optimization/57130
10759 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10760 of COMPARE as in_code to the recursive call if needed.
10762 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
10764 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10765 (enabled): Handle new members.
10766 * config/i386/sse.md (*vec_concatv2si): Merge from
10767 *vec_concatv2si_sse2 and vec_concatv2si_sse.
10768 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10770 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
10772 PR tree-optimization/57027
10773 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10774 for fnms opportunity, check we got the prerequisite kind
10775 of tree / gimple before using accessor functions.
10777 2013-05-03 Richard Biener <rguenther@suse.de>
10779 * double-int.h (lshift): New overload without precision
10780 and arith argument.
10781 (operator *=, operator +=, operator -=): Move ...
10782 * double-int.c (operator *=, operator +=, operator -=): ... here
10783 and implement more efficiently.
10784 (mul_double_with_sign): Remove.
10785 (lshift_double): Adjust to take unsinged shift argument, push
10786 dispatching code to callers.
10787 (mul_double_wide_with_sign): Add early out for callers that
10788 are not interested in high parts or overflow.
10789 (lshift): New function.
10790 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10791 dispatch code here.
10792 (lrotate, rrotate): Use logical shifts.
10793 * expr.c (get_inner_reference): Use lshift.
10794 * fixed-value.c (do_fixed_divide): Likewise.
10795 * tree-dfa.c (get_ref_base_and_extent): Likewise.
10796 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10797 (indirect_refs_may_alias_p): Likewise.
10798 (stmt_kills_ref_p_1): Likewise.
10800 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
10802 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10804 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
10806 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10807 scalar form of FABD instruction.
10809 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
10811 * lra-constraints.c (process_alt_operands): Add checking alt
10812 number to choose the best alternative.
10814 2013-05-02 Richard Biener <rguenther@suse.de>
10816 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10817 bitmap and its handling.
10818 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10820 2013-05-02 Richard Biener <rguenther@suse.de>
10822 PR middle-end/57140
10823 * tree-inline.c (copy_loops): Properly handle removed loops.
10824 (copy_cfg_body): Mark destination loops for fixup if source
10825 loops needed fixup.
10827 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
10830 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10831 generating simple_return for naked functions.
10833 2013-05-02 Martin Jambor <mjambor@suse.cz>
10835 PR middle-end/56988
10836 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10837 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10839 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10840 ipa_agg_replacement_value structures.
10841 (known_aggs_to_agg_replacement_list): Likewise.
10842 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10843 (read_agg_replacement_chain): Likewise.
10844 (ipcp_transform_function): Also check that by_ref flags match.
10846 2013-05-02 Richard Biener <rguenther@suse.de>
10848 * graphds.h (struct graph): Add obstack member.
10849 * graphds.c (new_graph): Initialize obstack and allocate
10851 (add_edge): Allocate edge from the obstack.
10852 (free_graph): Free the obstack instead of all edges and vertices.
10854 2013-05-02 Teresa Johnson <tejohnson@google.com>
10856 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10858 * cfg.c (update_bb_profile_for_threading): Ditto.
10859 * tree-inline.c (copy_bb): Ditto.
10860 (copy_edges_for_bb): Ditto.
10861 (initialize_cfun): Ditto.
10862 (copy_cfg_body): Ditto.
10863 (expand_call_inline): Ditto.
10864 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10865 (estimate_node_size_and_time): Ditto.
10866 (inline_merge_summary): Ditto.
10867 * cgraphclones.c (cgraph_clone_edge): Ditto.
10868 (cgraph_clone_node): Ditto.
10869 * sched-rgn.c (compute_dom_prob_ps): Ditto.
10870 (compute_trg_info): Ditto.
10872 2013-05-02 Ian Bolton <ian.bolton@arm.com>
10874 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10875 S reg when fp attribute set.
10876 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10878 2013-05-02 Ian Bolton <ian.bolton@arm.com>
10880 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10882 (*and_one_cmplsi3_compare0_uxtw): Likewise.
10883 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10884 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10886 2013-05-02 Richard Biener <rguenther@suse.de>
10888 * tree-scalar-evolution.c (scev_info_hasher): Remove.
10889 (struct instantiate_cache_entry): New type.
10890 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10891 (struct instantiate_cache_type): New type.
10892 (set_instantiated_value, get_instantiated_value): Remove.
10893 (get_instantiated_value_entry): New function.
10894 (instantiate_scev_name): Use the new cache and adjust.
10895 (instantiate_scev_poly): Adjust.
10896 (instantiate_scev_binary): Likewise.
10897 (instantiate_array_ref): Likewise.
10898 (instantiate_scev_convert): Likewise.
10899 (instantiate_scev_not): Likewise.
10900 (instantiate_scev_3): Likewise.
10901 (instantiate_scev_2): Likewise.
10902 (instantiate_scev_r): Likewise.
10903 (instantiate_scev): Likewise.
10904 (resolve_mixers): Likewise.
10906 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
10909 * lra-constraints.c (best_small_class_operands_num): Remove.
10910 (process_alt_operands): Remove small_class_operands_num. Take
10911 small classes operands into losers and only if the operand is not
10912 matched. Modify debugging output.
10913 (curr_insn_transform): Remove best_small_class_operands_num.
10916 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10918 * config/aarch64/aarch64-builtins.c
10919 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10920 * config/aarch64/aarch64-simd-builtins.def
10921 (reduc_splus_): Add new modes.
10922 (reduc_uplus_): New.
10923 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10924 (reduc_uplus_v4sf): Likewise.
10925 (reduc_splus_v4sf): Likewise.
10926 (aarch64_addv<mode>): Likewise.
10927 (reduc_uplus_<mode>): Likewise.
10928 (reduc_splus_<mode>): Likewise.
10929 (aarch64_addvv2di): Likewise.
10930 (reduc_uplus_v2di): Likewise.
10931 (reduc_splus_v2di): Likewise.
10932 (aarch64_addvv2si): Likewise.
10933 (reduc_uplus_v2si): Likewise.
10934 (reduc_splus_v2si): Likewise.
10935 (reduc_<sur>plus_<mode>): New.
10936 (reduc_<sur>plus_v2di): Likewise.
10937 (reduc_<sur>plus_v2si): Likewise.
10938 (reduc_<sur>plus_v4sf): Likewise.
10939 (aarch64_addpv4sf): Likewise.
10940 * config/aarch64/arm_neon.h
10941 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10942 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10943 add UNSPEC_SADDV, UNSPEC_UADDV.
10945 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10947 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10949 * config/aarch64/arm_neon.h
10950 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10952 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10954 * config/aarch64/aarch64-builtins
10955 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10957 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10959 * config/aarch64/aarch64-simd-builtins.def
10960 (reduc_smax_): New.
10961 (reduc_smin_): Likewise.
10962 (reduc_umax_): Likewise.
10963 (reduc_umin_): Likewise.
10964 (reduc_smax_nan_): Likewise.
10965 (reduc_smin_nan_): Likewise.
10968 (smax): Update for V2SF, V4SF and V2DF modes.
10971 (smin_nan): Likewise.
10972 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10973 (<su><maxmin><mode>3): ...This, refactor.
10974 (s<maxmin><mode>3): New.
10975 (<maxmin_uns><mode>3): Likewise.
10976 (reduc_<maxmin_uns>_<mode>): Refactor.
10977 (reduc_<maxmin_uns>_v4sf): Likewise.
10978 (reduc_<maxmin_uns>_v2si): Likewise.
10979 (aarch64_<fmaxmin><mode>: Remove.
10980 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10982 (vmin<q>_f<32,64>): Likewise.
10983 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10985 (su): Add mappings for smax, smin, umax, umin.
10987 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10988 (FMAXMIN): Rename as...
10989 (FMAXMIN_UNS): ...This.
10991 (fmaxminv): Likewise.
10992 (fmaxmin): Likewise.
10994 (maxmin_uns_op): Likewise.
10996 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
10998 * config/aarch64/arm_neon.h
10999 (vac<ge, gt><sd>_f<32, 64>): Rename to...
11000 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
11001 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
11003 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
11005 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
11006 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
11008 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
11010 * config/aarch64/aarch64-simd.md
11011 (vcond<mode>_internal): Handle special cases for constant masks.
11012 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
11013 (vcondu<mode><mode>): Likewise.
11014 (vcond<v_cmp_result><mode>): New.
11016 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
11018 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
11019 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
11020 * config/aarch64/aarch64-simd-builtins.def
11021 (cmeq): Update to BUILTIN_VALLDI.
11026 * config/aarch64/arm_neon.h
11027 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
11028 to builtins or C as appropriate.
11030 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
11032 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
11034 (cmhi): Rename to...
11036 * config/aarch64/aarch64-simd.md
11037 (simd_mode): Add SF.
11038 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
11039 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
11040 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
11041 (cstore<mode>_neg): ...This.
11042 * config/aarch64/iterators.md
11044 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
11045 (COMPARISONS): New.
11046 (UCOMPARISONS): Likewise.
11047 (optab): Add missing comparisons.
11053 (VCMP_S): Likewise.
11054 (VCMP_U): Likewise.
11055 (V_cmp_result): Add DF, SF modes.
11056 (v_cmp_result): Likewise.
11058 (vmtype): Likewise.
11059 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
11061 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
11063 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
11064 define_insn to define_insn_and_split.
11065 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
11066 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
11067 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
11068 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
11069 (thumb2_negscc): Likewise.
11071 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
11073 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
11075 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
11077 * config/arm/thumb2.md: Remove trailing whitespaces.
11079 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
11081 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
11082 Use gen_int_mode rather than GEN_INT.
11084 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
11086 * value-prof.c (stream_in_histogram_value): Remove the strayed
11089 2013-04-30 Richard Biener <rguenther@suse.de>
11091 PR middle-end/57122
11092 * cfghooks.c (split_edge): Properly check for the loop latch edge.
11094 2013-04-30 Richard Biener <rguenther@suse.de>
11096 PR middle-end/57107
11097 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
11099 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
11101 PR rtl-optimization/56957
11102 PR rtl-optimization/57105
11103 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
11104 variable. Use just INSN_UID for determining whether an insn
11105 should be only disconnected from the insn stream.
11106 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
11108 2013-04-30 Jakub Jelinek <jakub@redhat.com>
11110 PR tree-optimization/57104
11111 * tsan.c (instrument_expr): Don't instrument accesses to
11112 DECL_HARD_REGISTER VAR_DECLs.
11114 2013-04-30 Richard Biener <rguenther@suse.de>
11116 * function.h (loops_for_fn): New inline function.
11117 (set_loops_for_fn): Likewise.
11118 * cfgloop.h (place_new_loop): Add struct function parameter.
11119 (get_loop): Likewise.
11120 (get_loops): Likewise.
11121 (number_of_loops): Likewise.
11122 (fel_next): Adjust.
11123 (fel_init): Likewise.
11124 * cfg.c (get_loop_copy): Adjust.
11125 * cfgloop.c (flow_loops_dump): Likewise.
11126 (record_loop_exits): Likewise.
11127 (verify_loop_structure): Likewise.
11128 * cfgloopanal.c (mark_irreducible_loops): Likewise.
11129 (estimate_reg_pressure_cost): Likewise.
11130 (mark_loop_exit_edges): Likewise.
11131 * cfgloopmanip.c (place_new_loop): Likewise.
11132 (add_loop): Likewise.
11133 (duplicate_loop): Likewise.
11134 * graph.c (draw_cfg_nodes): Likewise.
11135 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
11136 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
11137 (extract_affine_chrec): Likewise.
11138 (build_scop_iteration_domain): Likewise.
11139 * graphite.c (graphite_initialize): Likewise.
11140 * ira-build.c (create_loop_tree_nodes): Likewise.
11141 (more_one_region_p): Likewise.
11142 (rebuild_regno_allocno_maps): Likewise.
11143 (mark_loops_for_removal): Likewise.
11144 (mark_all_loops_for_removal): Likewise.
11145 (remove_unnecessary_regions): Likewise.
11146 (ira_build): Likewise.
11147 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
11148 * loop-init.c (fix_loop_structure): Likewise.
11149 (gate_rtl_move_loop_invariants): Likewise.
11150 (gate_rtl_unswitch): Likewise.
11151 (gate_rtl_unroll_and_peel_loops): Likewise.
11152 (rtl_doloop): Likewise.
11153 * lto-streamer-in.c (input_cfg): Likewise.
11154 * lto-streamer-out.c (output_cfg): Likewise.
11155 * modulo-sched.c (sms_schedule): Likewise.
11156 * predict.c (tree_estimate_probability): Likewise.
11157 (tree_estimate_probability_driver): Likewise.
11158 (estimate_loops): Likewise.
11159 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
11160 (move_sese_region_to_fn): Likewise.
11161 (debug_loop_num): Likewise.
11162 * tree-chrec.c (chrec_evaluate): Likewise.
11163 (hide_evolution_in_other_loops_than_loop): Likewise.
11164 (chrec_component_in_loop_num): Likewise.
11165 (reset_evolution_in_loop): Likewise.
11166 (evolution_function_is_invariant_rec_p): Likewise.
11167 * tree-if-conv.c (main_tree_if_conversion): Likewise.
11168 * tree-inline.c (copy_loops): Likewise.
11169 (copy_cfg_body): Likewise.
11170 (tree_function_versioning): Likewise.
11171 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
11172 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
11174 (add_to_evolution_1): Likewise.
11175 (scev_const_prop): Likewise.
11176 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
11177 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
11178 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
11179 (tree_ssa_lim_initialize): Likewise.
11180 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
11181 (verify_loop_closed_ssa): Likewise.
11182 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
11183 (tree_ssa_loop_im): Likewise.
11184 (tree_ssa_loop_unswitch): Likewise.
11185 (tree_vectorize): Likewise.
11186 (check_data_deps): Likewise.
11187 (tree_ssa_loop_ivcanon): Likewise.
11188 (tree_ssa_loop_bounds): Likewise.
11189 (tree_complete_unroll): Likewise.
11190 (tree_complete_unroll_inner): Likewise.
11191 (tree_parallelize_loops): Likewise.
11192 (tree_ssa_loop_prefetch): Likewise.
11193 (tree_ssa_loop_ivopts): Likewise.
11194 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
11195 * tree-vectorizer.c (vectorize_loops): Likewise.
11197 2013-04-29 Mike Frysinger <vapier@gentoo.org>
11199 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
11200 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
11201 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
11202 with EABI_LINK_SPEC.
11204 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11207 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
11210 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
11213 * lra-constraints.c (process_alt_operands): Discourage a bit more
11214 using memory for pseudos. Print cost dump for alternatives.
11215 Modify cost values for conflicts with early clobbers.
11216 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
11218 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11221 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
11223 2013-04-29 Ian Bolton <ian.bolton@arm.com>
11225 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
11226 from/to S register.
11227 (movdi_aarch64): Support LDR/STR from/to D register.
11229 2013-04-29 Ian Bolton <ian.bolton@arm.com>
11231 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
11232 or higher by default.
11234 2013-04-29 Richard Biener <rguenther@suse.de>
11236 PR middle-end/57075
11237 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
11238 even if not adding abnormal edges for calls that can make
11241 2013-04-29 Richard Biener <rguenther@suse.de>
11243 PR middle-end/57103
11244 * tree-cfg.c (move_stmt_op): Fix condition under which to update
11246 (move_stmt_r): Remove redundant checking.
11248 2013-04-29 Teresa Johnson <tejohnson@google.com>
11251 * basic-block.h (apply_scale): New function.
11252 (apply_probability): Use apply_scale.
11253 * gimple-streamer-in.c (input_bb): Ditto.
11254 * lto-streamer-in.c (input_cfg): Ditto.
11255 * lto-cgraph.c (merge_profile_summaries): Ditto.
11256 * tree-optimize.c (execute_fixup_cfg): Ditto.
11257 * tree-inline.c (copy_bb): Update comment to use apply_scale.
11258 (copy_edges_for_bb): Ditto.
11259 (copy_cfg_body): Ditto.
11261 2013-04-29 Tom de Vries <tom@codesourcery.com>
11263 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
11264 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
11265 (tail_merge_optimize): Handle current_loops == NULL.
11267 2013-04-26 Jeff Law <law@redhat.com>
11269 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
11270 (simplify_cond_using_ranges): Generalize code to simplify
11271 COND_EXPRs where one argument is a constant and the other
11272 is an SSA_NAME created by an integral type conversion.
11274 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11276 * config/arm/arm.md (store_minmaxsi): Use only when
11277 optimize_insn_for_size_p.
11279 2013-04-29 Christian Bruel <christian.bruel@st.com>
11282 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
11284 2013-04-29 Richard Biener <rguenther@suse.de>
11286 PR middle-end/57089
11287 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
11288 loop tree make sure to schedule a fixup for the child as well.
11289 (expand_omp_for_generic): Properly add loops.
11290 (expand_omp_for_static_nochunk): Likewise.
11291 (expand_omp_for_static_chunk): Likewise.
11292 (expand_omp_for): For the degenerate case fixup loops.
11293 (expand_omp_sections): Fix default bb placement in loops.
11294 (expand_omp_atomic_pipeline): Properly add loops.
11296 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11298 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
11300 2013-04-29 Tom de Vries <tom@codesourcery.com>
11302 * tree-ssa-tail-merge.c: Update header comment.
11304 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11306 * config/aarch64/arm_neon.h
11307 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
11308 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
11309 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
11310 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
11311 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
11312 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
11313 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
11314 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
11316 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11318 * config/aarch64/aarch64-simd.md
11319 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
11320 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
11321 fix_trunc, fixuns_trunc.
11322 (ftrunc<VDQF:mode>2): New.
11323 * config/aarch64/iterators.md (optab): Add fix, fixuns.
11324 (fix_trunc_optab): New.
11326 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11328 * config/aarch64/aarch64-builtins.c
11329 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
11330 iceilf, lround, iroundf.
11332 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
11335 * config/i386/i386.h (enum ix86_tune_indices)
11336 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
11337 New, split from X86_TUNE_INTER_UNIT_MOVES.
11338 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
11339 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
11340 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
11341 (TARGET_INTER_UNIT_MOVES): Remove.
11342 * config/i386/i386.c (initial_ix86_tune_features): Update.
11343 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
11344 (ix86_expand_convert_uns_didf_sse): Use
11345 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11346 (ix86_expand_vector_init_one_nonzero): Ditto.
11347 (ix86_expand_vector_init_interleave): Ditto.
11348 (inline_secondary_memory_needed): Return true for moves from SSE class
11349 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
11350 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11351 * config/i386/constraints.md (Yi, Ym): Depend on
11352 TARGET_INTER_UNIT_MOVES_TO_VEC.
11353 (Yj, Yn): New constraints.
11354 * config/i386/i386.md (*movdi_internal): Change constraints of
11355 operand 1 from Yi to Yj and from Ym to Yn.
11356 (*movsi_internal): Ditto.
11357 (*movdf_internal): Ditto.
11358 (*movsf_internal): Ditto.
11359 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
11360 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
11361 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
11362 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
11363 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
11364 * config/i386/sse.md (movdi_to_sse): Ditto.
11365 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
11366 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
11367 TARGET_INTER_UNIT_MOVES.
11368 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
11369 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
11370 instead of TARGET_INTER_UNIT_MOVES.
11371 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
11372 operand 1 from Yi to Yj and from Ym to Yn.
11374 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11376 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
11377 (float_truncate_hi_): Likewise.
11378 (float_extend_lo_): Likewise.
11379 (float_truncate_lo_): Likewise.
11380 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
11381 (aarch64_float_extend_lo_v2df): Likewise.
11382 (vec_unpacks_hi_v4sf): Likewise.
11383 (aarch64_float_truncate_lo_v2sf): Likewise.
11384 (aarch64_float_truncate_hi_v4sf): Likewise.
11385 (vec_pack_trunc_v2df): Likewise.
11386 (vec_pack_trunc_df): Likewise.
11388 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11390 * config/aarch64/aarch64-builtins.c
11391 (aarch64_fold_builtin): Fold float conversions.
11392 * config/aarch64/aarch64-simd-builtins.def
11393 (floatv2si, floatv4si, floatv2di): New.
11394 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
11395 * config/aarch64/aarch64-simd.md
11396 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
11397 * config/aarch64/iterators.md (FLOATUORS): New.
11398 (optab): Add float, floatuns.
11399 (su_optab): Likewise.
11401 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11403 * config/aarch64/aarch64-builtins.c
11404 (aarch64_builtin_vectorized_function): Use new names for
11406 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
11407 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
11408 (fcvtzu): Split as...
11409 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
11410 (fcvtas): Split as...
11411 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
11412 (fcvtau): Split as...
11413 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
11414 (fcvtps): Split as...
11415 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
11416 (fcvtpu): Split as...
11417 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
11418 (fcvtms): Split as...
11419 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
11420 (fcvtmu): Split as...
11421 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
11422 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
11423 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
11424 (lfrintnusf, lfrintnudf): Likewise.
11425 * config/aarch64/aarch64-simd.md
11426 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
11428 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
11429 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
11430 (fcvt_pattern): Likewise.
11432 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11434 * config/aarch64/aarch64-simd.md
11435 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
11436 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
11438 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11440 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
11441 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
11442 (vrnd<a,m,n,p>_f32): Implement using builtins.
11443 (vrnd<i,x><q>_f<32, 64>): New.
11445 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
11447 * config/aarch64/aarch64-builtins.c
11448 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
11449 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
11450 (frintz): Rename to...
11452 (frintp): Rename to...
11454 (frintm): Rename to...
11456 (frinti): Rename to...
11457 (nearbyint): ...this.
11458 (frintx): Rename to...
11460 (frinta): Rename to...
11462 * config/aarch64/aarch64-simd.md
11463 (aarch64_frint<frint_suffix><mode>): Delete.
11464 (<frint_pattern><mode>2): Convert to insn.
11465 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
11466 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
11467 (frint_pattern): Likewise.
11468 (frint_suffix): Likewise.
11470 2013-04-29 Richard Biener <rguenther@suse.de>
11472 PR tree-optimization/57081
11473 * loop-init.c: Include tree-flow.h.
11474 (loop_optimizer_finalize): Free number of iteration estimates.
11475 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
11477 2013-04-29 Jakub Jelinek <jakub@redhat.com>
11479 PR tree-optimization/57083
11480 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
11481 non-singleton shift count range, zero extend low_bound for uns case.
11483 * config/i386/predicates.md (general_vector_operand): New predicate.
11484 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
11485 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
11486 if they aren't nonimmediate operands. If their original values
11487 satisfy const_vector_equal_evenodd_p, don't shift them.
11488 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
11489 predicates. For the SSE4.1 case force operands[{1,2}] into registers
11490 if not nonimmediate_operand.
11491 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
11492 instead of register_operand.
11493 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
11495 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
11497 * stor-layout.c (finalize_size_functions): Allocate a structure and
11498 reset cfun before dumping the functions.
11500 2013-04-27 Jakub Jelinek <jakub@redhat.com>
11502 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
11505 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
11506 use xop_pmacsdqh if uns_p.
11507 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
11508 the immediate rotate count.
11510 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
11512 * rtl.h (struct rtx_def): Add comment for field jump.
11513 (LRA_SUBREG_P): New macro.
11514 * recog.c (register_operand): Check LRA_SUBREG_P.
11515 * lra.c (lra): Add note at the end of RTL code. Align non-empty
11517 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11518 (lra_final_code_change): Skip subreg change for operators.
11519 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11520 if there are no operand changes.
11521 * lra-constraints.c (curr_insn_set): New.
11522 (match_reload): Set LRA_SUBREG_P.
11523 (emit_spill_move): Ditto.
11524 (check_and_process_move): Use curr_insn_set. Process only single
11525 set insns. Don't initialize sec_mem_p and change_p.
11526 (simplify_operand_subreg): Use LRA_SUBREG_P.
11527 (reg_in_class_p): New function.
11528 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
11529 of #ifdef. Add code to remove cycling.
11530 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11531 non-null disp. Reload inner instead of disp when base and index
11532 are null. Try to put lo_sum into register.
11533 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11534 (check_and_process_move): Move code for move cost check to
11535 simple_move_p. Remove equiv_substitution.
11536 (simple_move_p): New function.
11537 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
11538 curr_insn_set. Call check_and_process_move only for single set
11539 insns. Use the new function. Move call of check_and_process_move
11540 after operand equiv substitution and address process.
11542 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11545 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
11546 with nonlocal goto receivers or returns twice calls, ignore
11547 unininitialized values from abnormal edges to nl goto receiver
11548 or returns twice call.
11550 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11552 PR tree-optimization/57051
11553 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
11554 and VEC_RSHIFT_EXPR if shift count is a multiple of element
11557 2013-04-26 Richard Biener <rguenther@suse.de>
11559 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
11560 (expand_omp_taskreg): Likewise. Mark loops for fixup.
11561 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
11562 (fixup_loop_arrays_after_move): New function.
11563 (move_sese_region_to_fn): Properly outline the loop tree parts
11564 of the SESE region.
11566 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
11568 * config/i386/i386.md (type, unit): Fix long lines.
11570 2013-04-26 Richard Biener <rguenther@suse.de>
11572 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
11573 (lto-streamer-out.o): Likewise.
11574 * cfgloop.c (init_loops_structure): Export, add struct function
11575 argument and adjust.
11576 (flow_loops_find): Adjust.
11577 * cfgloop.h (enum loop_estimation): Add EST_LAST.
11578 (init_loops_structure): Declare.
11579 * lto-streamer-in.c: Include cfgloop.h.
11580 (input_cfg): Input the loop tree.
11581 * lto-streamer-out.c: Include cfgloop.h.
11582 (output_cfg): Output the loop tree.
11583 (output_struct_function_base): Do not drop PROP_loops.
11585 2013-03-26 Richard Biener <rguenther@suse.de>
11587 * tree-cfg.c (execute_build_cfg): Build the loop tree.
11588 (pass_build_cfg): Provide PROP_loops.
11589 (move_sese_region_to_fn): Remove loops that are outlined into fn
11591 * tree-inline.c: Include cfgloop.h.
11592 (initialize_cfun): Do not drop PROP_loops.
11593 (copy_loops): New function.
11594 (copy_cfg_body): Copy loop structure.
11595 (tree_function_versioning): Initialize destination loop tree.
11596 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
11597 (pass_parallelize_loops): Do IL verification.
11598 * loop-init.c (loop_optimizer_init): Fixup loops if required.
11599 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
11600 the CFG make sure we fixup loops as well.
11601 * tree-ssa-tail-merge.c: Include cfgloop.h.
11602 (replace_block_by): When merging loop latches mark loops for fixup.
11603 * lto-streamer-out.c (output_struct_function_base): Drop
11604 PROP_loops for now.
11605 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11606 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11607 * ipa-split.c: Include cfgloop.h.
11608 (split_function): Add the new return block to the loop tree root.
11609 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11610 whether we have removed the forwarder block.
11611 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11612 * cfgloop.h (place_new_loop): Declare.
11613 * cfgloopmanip.c (place_new_loop): Export.
11614 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11615 (tree-switch-conversion.o): Likewise.
11616 (tree-complex.o): Likewise.
11617 (tree-inline.o): Likewise.
11618 (tree-ssa-tailmerge.o): Likewise.
11619 (ipa-split.o): Likewise.
11620 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11621 (tree-ssa-copy.o): Likewise.
11622 * tree-switch-conversion.c: Include cfgloop.h
11623 (process_switch): If we emit a bit-test cascade, schedule loops
11625 * tree-complex.c: Include cfgloop.h.
11626 (expand_complex_div_wide): Properly add new basic-blocks to loops.
11627 * asan.c: Include cfgloop.h.
11628 (create_cond_insert_point): Properly add new basic-blocks to
11629 loops, schedule loop fixup.
11630 * cfgloop.c (verify_loop_structure): Check that looks are not
11632 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11634 (expand_omp_for_generic): Likewise.
11635 (expand_omp_sections): Likewise.
11636 (expand_omp_atomic_pipeline): Schedule loops for fixup.
11637 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11638 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11639 is initialized, not when loops are present.
11640 * tree-parloops.c (parallelize_loops): Remove checking here.
11641 * passes.c (init_optimization_passes): Schedule a copy-propagation
11642 pass before complete unrolling of inner loops.
11644 2013-04-26 Jakub Jelinek <jakub@redhat.com>
11646 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11647 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11648 (colorize_init): Add argument to _WIN32 version.
11649 * toplev.c: Include diagnostic-color.h.
11650 (process_options): Default to -fdiagnostics-color=auto if
11651 GCC_COLORS env var is in the environment.
11652 * common.opt (fdiagnostics-color=): Add Var and Init.
11653 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11654 env var is in the environment, the default is auto rather than never.
11656 * diagnostic.h (file_name_as_prefix): Add context argument.
11657 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
11658 the string as locus.
11659 * langhooks.c (lhd_print_error_function): Adjust caller.
11661 2013-04-25 Lawrence Crowl <crowl@google.com>
11663 * var-tracking.c (shared_hash_def::htab):
11664 Change type to hash_table. Update dependent calls and types.
11666 2013-04-25 Lawrence Crowl <crowl@google.com>
11668 * Makefile.in: Update as needed below.
11670 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11671 Move declaration to after the type's method definitons.
11673 * attribs.c (htab_t scoped_attributes::attribute_hash):
11674 Change type to hash_table. Update dependent calls and types.
11676 * bitmap.c (htab_t bitmap_desc_hash):
11677 Change type to hash_table. Update dependent calls and types.
11679 * cselib.c (htab_t cselib_hash_table):
11680 Change type to hash_table. Update dependent calls and types.
11682 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11683 (hash_string_slot_node): Move implementation into lto-streamer.h
11684 struct string_slot_hasher.
11685 (eq_string_slot_node): Likewise.
11687 * data-streamer-out.c: Update output_block::string_hash_table
11688 dependent calls and types.
11690 * dwarf2cfi.c (htab_t trace_index):
11691 Change type to hash_table. Update dependent calls and types.
11693 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11694 Change type to hash_table. Update dependent calls and types.
11695 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11696 (htab_t optimize_external_refs::map): Likewise.
11697 (htab_t output_comp_unit::extern_map): Likewise.
11698 (htab_t output_comdat_type_unit::extern_map): Likewise.
11699 (htab_t output_macinfo::macinfo_htab): Likewise.
11700 (htab_t optimize_location_lists::htab): Likewise.
11701 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11703 * except.c (htab_t ehspec_hash_type):
11704 Change type to hash_table. Update dependent calls and types.
11705 (assign_filter_values::ttypes): Likewise.
11706 (assign_filter_values::ehspec): Likewise.
11707 (sjlj_assign_call_site_values::ar_hash): Likewise.
11708 (convert_to_eh_region_ranges::ar_hash): Likewise.
11710 * gcse.c (htab_t pre_ldst_table):
11711 Change type to hash_table. Update dependent calls and types.
11713 * ggc-common.c (htab_t saving_htab):
11714 Change type to hash_table. Update dependent calls and types.
11715 (htab_t loc_hash): Likewise.
11716 (htab_t ptr_hash): Likewise.
11717 (call_count): Rename ggc_call_count.
11718 (call_alloc): Rename ggc_call_alloc.
11719 (loc_descriptor): Rename make_loc_descriptor.
11720 (add_statistics): Rename ggc_add_statistics.
11722 * ggc-common.c (saving_htab):
11723 Change type to hash_table. Update dependent calls and types.
11725 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11726 (push_gimplify_context): Likewise.
11727 (pop_gimplify_context): Likewise.
11728 (struct gimple_temp_hash_elt): Added.
11729 (struct gimplify_hasher): Likewise.
11730 (struct gimplify_ctx.temp_htab):
11731 Change type to hash_table. Update dependent calls and types.
11733 * gimple-fold.c: Include gimplify-ctx.h.
11735 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11736 Change type to hash_table. Update dependent calls and types.
11737 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11738 avoid potential global name collision.
11740 * gimplify.c: Include gimplify-ctx.h.
11741 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11742 (htab_t gimplify_ctx::temp_htab):
11743 Update dependent calls and types for new type hash_table.
11744 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11745 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11747 * gimplify-ctx.h: New.
11748 (struct gimple_temp_hash_elt): Move from gimplify.c.
11749 (class gimplify_hasher): New.
11750 (struct gimplify_ctx): Move from gimple.h.
11751 (htab_t gimplify_ctx::temp_htab):
11752 Change type to hash_table. Update dependent calls and types.
11754 * graphite-clast-to-gimple.c: Include graphite-htab.h.
11755 (htab_t ivs_params::newivs_index):
11756 Change type to hash_table. Update dependent calls and types.
11757 (htab_t ivs_params::params_index): Likewise.
11758 (htab_t print_generated_program::params_index): Likewise.
11759 (htab_t gloog::newivs_index): Likewise.
11760 (htab_t gloog::params_index): Likewise.
11762 * graphite.c: Include graphite-htab.h.
11763 4htab_t graphite_transform_loops::bb_pbb_mapping):
11764 Change type to hash_table. Update dependent calls and types.
11766 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11767 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11768 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11770 * graphite-dependences.c: Include graphite-htab.h.
11771 (loop_is_parallel_p): Change hash table type of parameter.
11773 * graphite-htab.h: New.
11774 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11775 (extern find_pbb_via_hash): Move from graphite-poly.h.
11776 (extern loop_is_parallel_p): Move from graphite-poly.h.
11777 (extern get_loop_body_pbbs): Move from graphite-poly.h.
11779 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11780 (extern loop_is_parallel_p): Move to graphite-htab.h.
11781 (extern get_loop_body_pbbs): Move to graphite-htab.h.
11783 * haifa-sched.c (htab_t delay_htab):
11784 Change type to hash_table. Update dependent calls and types.
11785 (htab_t delay_htab_i2): Likewise.
11787 * ira-color.c (htab_t allocno_hard_regs_htab):
11788 Change type to hash_table. Update dependent calls and types.
11790 * ira-costs.c (htab_t cost_classes_htab):
11791 Change type to hash_table. Update dependent calls and types.
11793 * loop-invariant.c (htab_t merge_identical_invariants::eq):
11794 Change type to hash_table. Update dependent calls and types.
11796 * loop-iv.c (htab_t bivs):
11797 Change type to hash_table. Update dependent calls and types.
11799 * loop-unroll.c (htab_t opt_info::insns_to_split):
11800 Change type to hash_table. Update dependent calls and types.
11801 (htab_t opt_info::insns_with_var_to_expand): Likewise.
11803 * lto-streamer.h (struct string_slot): Move from data-streamer.h
11804 (struct string_slot_hasher): New.
11805 (htab_t output_block::string_hash_table):
11806 Change type to hash_table. Update dependent calls and types.
11808 * lto-streamer-in.c (freeing_string_slot_hasher): New.
11809 (htab_t file_name_hash_table):
11810 Change type to hash_table. Update dependent calls and types.
11812 * lto-streamer-out.c: Update output_block::string_hash_table dependent
11815 * lto-streamer.c (htab_t tree_htab):
11816 Change type to hash_table. Update dependent calls and types.
11818 * omp-low.c: Include gimplify-ctx.h.
11820 * passes.c (htab_t name_to_pass_map):
11821 Change type to hash_table. Update dependent calls and types.
11822 (pass_traverse): Rename to passes_pass_traverse.
11824 * plugin.c (htab_t event_tab):
11825 Change type to hash_table. Update dependent calls and types.
11827 * postreload-gcse.c (htab_t expr_table):
11828 Change type to hash_table. Update dependent calls and types.
11829 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11831 * sese.c (debug_rename_map_1): Make extern.
11832 (htab_t copy_bb_and_scalar_dependences::rename_map):
11833 Change type to hash_table. Update dependent calls and types.
11835 * sese.h (extern debug_rename_map): Move to .c file.
11837 * store-motion.c (htab_t store_motion_mems_table):
11838 Change type to hash_table. Update dependent calls and types.
11840 * trans-mem.c (htab_t tm_new_mem_hash):
11841 Change type to hash_table. Update dependent calls and types.
11843 * tree-browser.c (htab_t TB_up_ht):
11844 Change type to hash_table. Update dependent calls and types.
11846 * tree-cfg.c (htab_t discriminator_per_locus):
11847 Change type to hash_table. Update dependent calls and types.
11849 * tree-complex.c: Include tree-hasher.h
11850 (htab_t complex_variable_components):
11851 Change type to hash_table. Update dependent calls and types.
11853 * tree-eh.c (htab_t finally_tree):
11854 Change type to hash_table. Update dependent calls and types.
11856 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11857 struct int_tree_hasher.
11858 (extern int_tree_map_eq): Likewise.
11859 (uid_decl_map_hash): Removed.
11860 (extern decl_tree_map_eq): Likewise.
11862 * tree-hasher.h: New.
11863 (struct int_tree_hasher): New.
11864 (typedef int_tree_htab_type): New.
11866 * tree-inline.c: Include gimplify-ctx.h.
11868 * tree-mudflap.c: Include gimplify-ctx.h.
11870 * tree-parloops.c: Include tree-hasher.h.
11871 (htab_t eliminate_local_variables_stmt::decl_address):
11872 Change type to hash_table. Update dependent calls and types.
11873 (htab_t separate_decls_in_region::decl_copies): Likewise.
11875 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11876 Change type to hash_table. Update dependent calls and types.
11878 * tree-sra.c (candidates):
11879 Change type to hash_table. Update dependent calls and types.
11881 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11883 (int_tree_map_hash): Likewise.
11885 * tree-ssa-dom.c (htab_t avail_exprs):
11886 Change type to hash_table. Update dependent calls and types.
11888 * tree-ssa-live.c (var_map_base_init::tree_to_index):
11889 Change type to hash_table. Update dependent calls and types.
11891 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11892 Change type to hash_table. Update dependent calls and types.
11894 * tree-ssa-phiopt.c (seen_ssa_names):
11895 Change type to hash_table. Update dependent calls and types.
11897 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11898 Change type to hash_table. Update dependent calls and types.
11900 * tree-ssa-uncprop.c (equiv):
11901 Change type to hash_table. Update dependent calls and types.
11903 2013-04-25 Jakub Jelinek <jakub@redhat.com>
11905 PR rtl-optimization/57003
11906 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11907 call note_stores with kill_clobbered_value callback again after
11908 killing regs_invalidated_by_call.
11910 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11912 * config/aarch64/aarch64-simd.md
11913 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11914 (aarch64_simd_bsl<mode>): Likewise.
11915 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11917 2013-04-25 Marek Polacek <polacek@redhat.com>
11919 PR tree-optimization/57066
11920 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11922 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
11924 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11926 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11928 * config/aarch64/aarch64-builtins.c
11929 (aarch64_fold_builtin): New.
11930 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11931 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11932 * config/aarch64/aarch64-simd-builtins.def (abs): New.
11933 * config/aarch64/arm_neon.h
11934 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11936 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11937 Tejas Belagod <tejas.belagod@arm.com>
11939 * config/aarch64/aarch64-builtins.c
11940 (aarch64_gimple_fold_builtin): New.
11941 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11942 * config/aarch64/aarch64-simd-builtins.def (addv): New.
11943 * config/aarch64/aarch64-simd.md (addpv4sf): New.
11944 (addvv4sf): Update.
11945 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11947 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11949 * config/aarch64/aarch64.md
11950 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11952 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
11954 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11955 (*ngcsi_uxtw): New pattern.
11957 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11958 Julian Brown <julian@codesourcery.com>
11960 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11961 (TB_DREG): Add T_V4HF.
11962 (v4hf_UP): New macro.
11963 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11964 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11965 Handle initialisation of V4HF. Adjust initialisation of reinterpret
11967 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11968 (arm_vector_mode_supported_p): Handle V4HF.
11969 (arm_mangle_map): Handle V4HFmode.
11970 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11971 * config/arm/arm_neon_builtins.def: Add entries for
11972 vcvtv4hfv4sf, vcvtv4sfv4hf.
11973 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11974 (neon_vcvtv4hfv4sf): Likewise.
11975 * config/arm/neon-gen.ml: Handle half-precision floating point
11977 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11978 * config/arm/arm_neon.h: Regenerate.
11979 * config/arm/neon.ml (type elts): Add F16.
11980 (type vectype): Add T_float16x4, T_floatHF.
11981 (type vecmode): Add V4HF.
11982 (type features): Add Requires_FP_bit feature.
11983 (elt_width): Handle F16.
11984 (elt_class): Likewise.
11985 (elt_of_class_width): Likewise.
11986 (mode_of_elt): Refactor.
11987 (type_for_elt): Handle F16, fix error messages.
11988 (vectype_size): Handle T_float16x4.
11989 (vcvt_sh): New function.
11990 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11991 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11992 (string_of_mode): Handle V4HF.
11993 * doc/arm-neon-intrinsics.texi: Regenerate.
11995 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
11997 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11998 format specifier in 'X' case.
12000 2013-04-25 Alan Modra <amodra@gmail.com>
12003 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
12004 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
12005 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
12006 Repeat for many other rotate/shift and mask patterns using subregs.
12007 Name lshiftrt insns.
12008 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
12009 on WORDS_BIG_ENDIAN.
12011 2013-04-25 Alan Modra <amodra@gmail.com>
12013 * config.gcc: Support little-endian powerpc-linux targets.
12014 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
12015 (LINK_OS_LINUX_SPEC): Define.
12016 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
12017 Preserve MASK_LITTLE_ENDIAN.
12018 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
12019 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
12020 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
12021 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
12022 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
12023 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
12024 Correct fp word order for little-endian. Don't shift toc entries
12025 smaller than a word for little-endian.
12026 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
12027 (bswapdi2 splits): Correct low-part subreg for little-endian.
12028 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
12029 low/high where such is correct only for be.
12030 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
12031 little-endian for -mcall-aixdesc.
12033 2013-04-25 Alan Modra <amodra@gmail.com>
12035 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
12036 replace_equiv_address_nv.
12038 2013-04-25 Alan Modra <amodra@gmail.com>
12040 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
12042 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
12045 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
12046 * rtl.h (struct rtx_def): ...
12048 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
12050 PR rtl-optimizations/57046
12051 * lra-constraints (split_reg): Set up lra_risky_transformations_p
12052 for multi-reg splits.
12054 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
12056 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
12058 2013-04-24 Sterling Augustine <saugustine@google.com>
12060 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
12061 (comp_dir_string, debug_str_dwo_section): New.
12062 (DEBUG_STR_DWO_SECTION): Rename to ...
12063 (DEBUG_DWO_STR_SECTION): ... this.
12064 (DEBUG_NORM_STR_SECTION): Delete.
12065 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
12066 (DEBUG_STR_DWO_SECTION_FLAGS): New.
12067 (find_AT_string): Move most logic to ...
12068 (find_AT_string_in_table): ... here. New.
12069 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
12070 add_skeleton_AT_string. Delete logic.
12071 (output_skeleton_debug_sections): Remove call to
12072 add_top_level_skeleton_die_attrs.
12073 (add_comp_dir_attribute): Move logic to comp_dir_string.
12074 (dwarf2out_init): Initialize debug_str_dwo_section.
12075 (output_indirect_string): Call find_string_form.
12076 (output_indirect_strings): Rewrite.
12077 (prune_unused_types): Empty skeleton_debug_str_hash.
12078 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
12079 (dwarf2out_finish): Call output_indirect_strings.
12081 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
12083 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
12085 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
12087 * rtl.h (struct rtx_def): Add comment for field jump.
12088 (LRA_SUBREG_P): New macro.
12089 * recog.c (register_operand): Check LRA_SUBREG_P.
12090 * lra.c (lra): Add note at the end of RTL code. Align non-empty
12092 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
12093 (lra_final_code_change): Skip subreg change for operators.
12094 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
12095 if there are no operand changes.
12096 * lra-constraints.c (curr_insn_set): New.
12097 (match_reload): Set LRA_SUBREG_P.
12098 (emit_spill_move): Ditto.
12099 (check_and_process_move): Use curr_insn_set. Process only single
12100 set insns. Don't initialize sec_mem_p and change_p.
12101 (simplify_operand_subreg): Use LRA_SUBREG_P.
12102 (reg_in_class_p): New function.
12103 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
12104 of #ifdef. Add code to remove cycling.
12105 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
12106 non-null disp. Reload inner instead of disp when base and index
12107 are null. Try to put lo_sum into register.
12108 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
12109 (check_and_process_move): Move code for move cost check to
12110 simple_move_p. Remove equiv_substitution.
12111 (simple_move_p): New function.
12112 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
12113 curr_insn_set. Call check_and_process_move only for single set
12114 insns. Use the new function. Move call of check_and_process_move
12115 after operand equiv substitution and address process.
12117 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
12119 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
12120 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
12121 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
12123 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
12125 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
12127 2013-04-24 Marek Polacek <polacek@redhat.com>
12129 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
12130 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
12131 (select_loops_exit_conditions): Likewise.
12132 (number_of_iterations_for_all_loops): Likewise.
12133 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
12134 (scev_analysis): Likewise.
12136 2013-04-02 Catherine Moore <clm@codesourcery.com>
12137 Chao-ying Fu <fu@mips.com>
12139 * config/mips/micromips.md (jraddiusp): New pattern.
12140 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
12141 instruction if possible.
12143 2013-04-24 Alan Modra <amodra@gmail.com>
12145 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
12147 2013-04-24 Julian Brown <julian@codesourcery.com>
12148 Chung-Lin Tang <cltang@codesourcery.com>
12150 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
12151 dependency behavior in enumeration type DIE generation. Add TODO note
12152 to comments about future DW_FORM_sdata/udata re-work of related code.
12154 2013-04-23 Lawrence Crowl <crowl@google.com>
12156 * Makefile.in: Update as needed below.
12158 * hash-table.h (class hash_table):
12159 Correct many methods with parameter types compare_type to the correct
12160 value_type. (Correct code was unlikely to notice the change.)
12161 (hash_table::elements_with_deleted) New.
12162 (class hashtable::iterator): New.
12163 (hashtable::begin()): New.
12164 (hashtable::end()): New.
12165 (FOR_EACH_HASH_TABLE_ELEMENT): New.
12167 * statistics.c (statistics_hashes):
12168 Change type to hash_table. Update dependent calls and types.
12170 * tree-into-ssa.c (var_infos):
12171 Change type to hash_table. Update dependent calls and types.
12173 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
12174 Change type to hash_table. Update dependent calls and types.
12176 * tree-ssa-loop-im.c (struct mem_ref.refs):
12177 Change type to hash_table. Update dependent calls and types.
12179 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
12180 Change type to hash_table. Update dependent calls and types.
12182 * tree-ssa-sccvn.c (vn_tables_s::nary):
12183 Change type to hash_table. Update dependent calls and types.
12184 (vn_tables_s::phis): Likewise.
12185 (vn_tables_s::references): Likewise.
12187 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
12188 (vn_reference_eq): Update parameter and return types.
12190 * tree-ssa-structalias.c (pointer_equiv_class_table):
12191 Change type to hash_table. Update dependent calls and types.
12192 (location_equiv_class_table): Likewise.
12194 * tree-vect-data-refs.c: Consequential changes for making
12195 peeling a hash_table.
12197 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
12198 (destroy_loop_vec_info): Dependent hash_table update.
12200 * tree-vectorizer.h (peeling_htab):
12201 Change type to hash_table. Update dependent calls and types.
12203 2013-04-23 Shiva Chen <shiva0217@gmail.com>
12205 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
12206 to check the register content is equal or not.
12207 * lra-constraints.c (match_reload): Use lra_assign_reg_val
12208 to assign register content record.
12209 * lra-eliminations.c (update_reg_eliminate): Use
12210 lra_update_reg_val_offset to update register content offset.
12211 * lra-int.h (struct lra_reg): Add offset member.
12212 (lra_reg_val_equal_p): New static inline function.
12213 (lra_update_reg_val_offset): New static inline function.
12214 (lra_assign_reg_val): New static inline function.
12215 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
12216 to assign register content record.
12217 (initialize_lra_reg_info_element): Initial offset to zero.
12219 2013-04-23 Catherine Moore <clm@codesourcery.com>
12221 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
12222 operands. Record compression.
12224 2013-04-23 Xinliang David Li <davidxl@google.com>
12226 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
12228 2013-04-23 Richard Biener <rguenther@suse.de>
12230 PR middle-end/57036
12231 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
12232 parameter, only add abnormal goto edges from the copied body
12233 if the call could perform abnormal gotos.
12234 (copy_cfg_body): Adjust.
12236 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
12238 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
12240 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
12242 * coretypes.h (gimple_stmt_iterator): Add struct to make
12245 2013-04-23 Richard Biener <rguenther@suse.de>
12247 PR tree-optimization/57026
12248 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
12249 from SSA names occuring in abnormal PHI nodes.
12251 2013-04-22 Andi Kleen <ak@linux.intel.com>
12253 * lto/lto.c (print_lto_report_1): Fix LTO report names.
12255 2013-04-22 Andi Kleen <ak@linux.intel.com>
12257 * lto/lto.c (print_lto_report_1): Declare early.
12258 (read_cgraph_and_symbols): Call print_lto_report_1 early.
12260 2013-04-22 Andi Kleen <ak@linux.intel.com>
12262 * common.opt (-flto-report-wpa): Add.
12263 * doc/invoke.texi (-flto-report-wpa): Add.
12264 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
12267 2013-04-22 Xinliang David Li <davidxl@google.com>
12269 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
12270 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
12271 * Makefile.in: New dependency
12273 David Daney <ddaney.cavm@gmail.com>
12275 * configure.ac (gcc_cv_as_micromips_support): Use the
12276 --fatal-warnings option.
12277 * configure: Regenerate.
12279 2013-04-22 Marek Polacek <polacek@redhat.com>
12282 * tsan.c (instrument_expr): Don't instrument expression
12283 in case its size is zero.
12285 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
12289 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
12291 * config/alpha/alpha.c (TARGET_LRA_P): New define.
12293 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12295 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
12296 (gimple_stmt_iterator): New typedef.
12297 * gimple.h (gimple_stmt_iterator): Rename to...
12298 (gimple_stmt_iterator_d): ... This.
12299 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
12300 trees be valid for GIMPLE and GENERIC.
12301 (TARGET_GIMPLE_FOLD_BUILTIN): New.
12302 * gimple-fold.c (gimple_fold_call): Call target hook
12303 gimple_fold_builtin.
12304 * hooks.c (hook_bool_gsiptr_false): New.
12305 * hooks.h (hook_bool_gsiptr_false): New.
12306 * target.def (fold_stmt): New.
12307 * doc/tm.texi: Regenerate.
12309 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
12312 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
12313 a set sp if no stack realignment.
12315 2013-04-22 Nick Clifton <nickc@redhat.com>
12317 * config.gcc (tilegx-linux): Extend extra_objs rather than
12319 (tilepro-linux): Likewise.
12321 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12323 * config/aarch64/aarch64-builtins.c
12325 (CF0, CF1, CF2, CF3, CF4, CF10): New.
12326 (VAR<1-12>): Add MAP parameter.
12327 (BUILTIN_*): Likewise.
12328 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
12329 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
12330 (aarch64_ushl_n<mode>): Likewise.
12331 (aarch64_sshr_n<mode>): Likewise.
12332 (aarch64_ushr_n<mode>): Likewise.
12333 (aarch64_<maxmin><mode>): Likewise.
12334 (aarch64_sqrt<mode>): Likewise.
12335 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
12336 (vshr<q>_n_*): Likewise.
12338 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
12340 * config/aarch64/aarch64-builtins.c
12341 (aarch64_simd_builtin_type_mode): Handle SF types.
12343 (BUILTIN_GPF): Define.
12344 (aarch64_init_simd_builtins): Handle SF types.
12345 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
12346 (frecps): Likewise.
12347 (frecpx): Likewise.
12348 * config/aarch64/aarch64-simd.md
12349 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
12350 (aarch64_frecpe<mode>): New.
12351 (aarch64_frecps<mode>): Likewise.
12352 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
12353 (v8type): Add frecp<esx>.
12354 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
12355 (aarch64_frecps<mode>): Likewise.
12356 * config/aarch64/iterators.md (FRECP): New.
12357 (frecp_suffix): Likewise.
12358 * config/aarch64/arm_neon.h
12359 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
12361 2013-04-22 Christian Bruel <christian.bruel@st.com>
12364 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
12365 (REG_CLASS_NAMES): Idem.
12366 (REG_CLASS_CONTENTS): Idem.
12367 (REGCLASS_HAS_FP_REG): Idem.
12368 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
12369 (sh_conditional_register_usage): Idem.
12371 2013-04-21 Jeff Law <law@redhat.com>
12373 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
12374 (ssa_forward_propagate_and_combine): Use it.
12376 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
12378 * lra.c: Update the flow chart diagram.
12380 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
12382 PR rtl-optimization/56847
12383 * lra-constraints.c (process_alt_operands): Discourage alternative
12384 with non-matche doffsettable memory constraint fro memory with
12387 2013-04-19 Richard Biener <rguenther@suse.de>
12389 PR tree-optimization/56982
12390 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
12392 * gimplify.c (gimplify_call_expr): Notice special calls.
12393 (gimplify_modify_expr): Likewise.
12394 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
12395 abnormal control flow receivers.
12396 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
12397 in the same way as cfun->has_nonlocal_labels.
12398 (gimple_purge_dead_abnormal_call_edges): Likewise.
12399 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
12400 receivers start a basic-block.
12402 2013-04-19 Richard Biener <rguenther@suse.de>
12404 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
12406 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
12407 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
12408 (SLP_TREE_LOAD_PERMUTATION): Add.
12409 (vect_transform_slp_perm_load): Adjust prototype.
12410 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
12411 (vect_free_slp_instance): Likewise.
12412 (vect_create_new_slp_node): Likewise.
12413 (vect_supported_slp_permutation_p): Remove.
12414 (vect_slp_rearrange_stmts): Adjust.
12415 (vect_supported_load_permutation_p): Likewise. Inline
12416 vect_supported_slp_permutation_p here.
12417 (vect_analyze_slp_instance): Compute load permutations per
12418 slp node instead of per instance.
12419 (vect_get_slp_defs): Adjust.
12420 (vect_transform_slp_perm_load): Likewise.
12421 (vect_schedule_slp_instance): Remove redundant code.
12422 (vect_schedule_slp): Remove hack for PR56270, add it ...
12423 * tree-vect-stmts.c (vectorizable_load): ... here, do not
12424 CSE loads for SLP. Adjust.
12426 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
12428 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
12429 spelling in two comments.
12431 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
12434 * config/arm/arm.c (load_multiple_sequence): Require SP
12435 as base register for loads if SP is in the register list.
12437 2013-04-19 Martin Jambor <mjambor@suse.cz>
12439 PR tree-optimization/56718
12440 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
12441 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
12442 and made public. Adjusted all callers.
12443 (ipa_intraprocedural_devirtualization): New function.
12444 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
12445 (ipa_intraprocedural_devirtualization): Likewise.
12446 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
12448 2013-04-19 Richard Biener <rguenther@suse.de>
12450 PR tree-optimization/57000
12451 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
12453 2013-04-19 Terry Guo <terry.guo@arm.com>
12455 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
12457 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
12458 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
12459 (cortex_m4_fmacs): Use new reservations.
12460 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
12462 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
12464 PR rtl-optimization/56999
12465 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
12467 (lra_coalesce): Remove split_origin_bitmap and related code.
12468 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
12469 ranges if necessary.
12471 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
12473 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
12475 (ix86_expand_call): Remove clobbered_registers array and use
12476 x86_64_ms_sysv_extra_clobbered_registers instead.
12477 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
12479 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
12481 * config/i386/i386.md (*call_rex64_ms_sysv): Use
12482 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
12483 (*call_value_rex64_ms_sysv): Ditto.
12485 2013-04-18 Cary Coutant <ccoutant@google.com>
12487 * dwarf2out.c (output_pubnames): Check die_perennial_p of
12488 parent instead of die_mark.
12490 2013-04-18 Diego Novillo <dnovillo@google.com>
12492 * gimple.c (create_gimple_tmp): New.
12493 (get_expr_type): New.
12494 (build_assign): New.
12495 (build_type_cast): New.
12496 * gimple.h (enum ssa_mode): Define.
12497 (gimple_seq_set_location): New.
12498 * asan.c (build_check_stmt): Change some gimple_build_* calls
12499 to use build_assign and build_type_cast.
12501 2013-04-18 Richard Biener <rguenther@suse.de>
12503 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
12504 handle negative step. Remove redundant checks.
12505 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
12506 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
12507 for negative step and grouped loads fail to vectorize.
12509 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
12511 * emit-rtl.c (reset_insn_used_flags): New function.
12512 (reset_all_used_flags): Use it.
12513 (verify_insn_sharing): New function.
12514 (verify_rtl_sharing): Fix verification for SEQUENCEs.
12516 2013-04-18 Jakub Jelinek <jakub@redhat.com>
12518 PR tree-optimization/56984
12519 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
12520 and (x >> M) >= N don't register any assertion if N << M is the
12523 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
12525 * lower-subreg.c (resolve_simple_move): If called self-recursive,
12526 do not delete_insn insns that have not yet been emitted, only
12527 unlink them with remove_insn.
12528 * df-scan.c (df_insn_delete): Revert r197492.
12530 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
12532 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
12533 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
12535 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
12537 * config/arm/arm.md (movsicc_insn): Convert define_insn into
12538 define_insn_and_split.
12539 (and_scc,ior_scc,negscc): Likewise.
12540 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
12542 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
12544 * config/arm/arm.c (use_return_insn): Return 0 for targets that
12545 can benefit from using a sequence of LDRD instructions in epilogue
12546 instead of a single LDM instruction.
12548 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
12551 * doc/extend.texi: Fix typo.
12553 2013-04-17 Richard Biener <rguenther@suse.de>
12555 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
12556 (vect_build_slp_tree): ... here.
12557 (vect_build_slp_tree_1): Compute which stmts of the SLP group
12558 match. Remove special-casing of mismatched complex loads.
12559 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
12560 re-try the match with swapped commutative operands.
12561 (vect_supported_load_permutation_p): Remove special-casing of
12562 mismatched complex loads.
12563 (vect_analyze_slp_instance): Adjust.
12565 2013-04-17 Richard Biener <rguenther@suse.de>
12567 PR rtl-optimization/56921
12568 * cfgloop.h (struct loop): Add simple_loop_desc member.
12569 (struct niter_desc): Mark with GTY(()).
12570 (simple_loop_desc): Do not use aux field but simple_loop_desc.
12571 * loop-iv.c (get_simple_loop_desc): Likewise.
12572 (free_simple_loop_desc): Likewise.
12575 2013-04-16 Richard Biener <rguenther@suse.de>
12577 PR rtl-optimization/56921
12578 * loop-init.c (pass_rtl_move_loop_invariants): Add
12579 TODO_do_not_ggc_collect to todo_flags_finish.
12580 (pass_rtl_unswitch): Same.
12581 (pass_rtl_unroll_and_peel_loops): Same.
12582 (pass_rtl_doloop): Same.
12584 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
12586 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
12587 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
12588 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
12590 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
12591 * tree-streamer.c (record_common_node): Adjust reference in comment.
12593 2013-04-17 Terry Guo <terry.guo@arm.com>
12595 * config/arm/cortex-m4.md: Add a new bypass.
12597 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12599 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12601 (*subs_<optab><mode>_multp2): New pattern.
12602 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12603 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12605 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12607 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12608 (*subs_mul_imm_<mode>): New pattern.
12610 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
12613 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12614 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
12615 (vsx_movti_32bit): Change j->wa to O->wa.
12617 2013-04-16 Richard Biener <rguenther@suse.de>
12619 PR rtl-optimization/56921
12620 * loop-init.c (pass_rtl_move_loop_invariants): Add
12621 TODO_do_not_ggc_collect to todo_flags_finish.
12622 (pass_rtl_unswitch): Same.
12623 (pass_rtl_unroll_and_peel_loops): Same.
12624 (pass_rtl_doloop): Same.
12626 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
12628 * config/arm/arm.c (emit_multi_reg_push): New declaration
12629 for an existing function.
12630 (arm_emit_strd_push): New function.
12631 (arm_expand_prologue): Used here.
12632 (arm_emit_ldrd_pop): New function.
12633 (arm_expand_epilogue): Used here.
12634 (arm_get_frame_offsets): Update condition.
12635 (arm_emit_multi_reg_pop): Add a special case for load of a single
12636 register with writeback.
12638 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
12640 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12643 2013-04-16 Richard Biener <rguenther@suse.de>
12645 PR tree-optimization/56756
12646 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12647 (first_mem_ref_loc): New.
12648 (execute_sm): Place the load temporarily before a previous
12649 access instead of in the latch edge to ensure its SSA dependencies
12650 are defined at points dominating the load.
12652 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
12654 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12655 correct fix by moving header and footer insn to the footer of
12656 the merged basic block. Clear BB_END of the merged-away block.
12658 PR middle-end/43631
12659 * emit-rtl.c (make_note_raw): New function.
12660 (link_insn_into_chain): New static inline function.
12661 (add_insn): Use it.
12662 (add_insn_before, add_insn_after): Factor insn chain linking code...
12663 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12664 using link_insn_into_chain.
12665 (note_outside_basic_block_p): New helper function for emit_note_after
12666 and emit_note_before.
12667 (emit_note_after): Use nobb variant of add_insn_after if the note
12668 should not be contained in a basic block.
12669 (emit_note_before): Use nobb variant of add_insn_before if the note
12670 should not be contained in a basic block.
12671 (emit_note_copy): Use make_note_raw.
12672 (emit_note): Likewise.
12673 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12674 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12675 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12676 the moved barrier the tail of the basic block it follows.
12677 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12679 2013-04-15 Jakub Jelinek <jakub@redhat.com>
12681 PR tree-optimization/56962
12682 * gimple-ssa-strength-reduction.c (record_increment): Only set
12683 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12684 either rhs1 or rhs2 is equal to c->base_expr.
12686 2013-04-15 Richard Biener <rguenther@suse.de>
12688 PR tree-optimization/56933
12689 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12691 (GROUP_READ_WRITE_DEPENDENCE): Remove.
12692 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12693 * tree-vect-data-refs.c (vect_analyze_group_access): Move
12694 dependence check ...
12695 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12697 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12698 GROUP_READ_WRITE_DEPENDENCE.
12700 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12702 * emit-rtl.c (reset_all_used_flags): New function.
12703 (verify_rtl_sharing): Call reset_all_used_flags before and after
12704 performing the checks.
12706 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12708 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12709 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12710 * config/arm/constraints.md (De): New constraint.
12711 * config/arm/neon.md (anddi3_neon): Delete.
12712 (neon_vand<mode>): Expand to standard anddi3 pattern.
12713 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12714 Move earlier in the file.
12715 (neon_inv_logic_op2): Likewise.
12716 (arm_anddi_operand_neon): New predicate.
12718 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12720 * configure.ac (gcc_cv_ld_as_needed): Set
12721 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12722 Use -z ignore, -z record on *-*-solaris2*.
12723 (HAVE_LD_AS_NEEDED): Update comment.
12724 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12725 * configure: Regenerate.
12726 * config.in: Regenerate.
12727 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12728 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12729 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12730 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12731 equivalents. Fix markup.
12732 * doc/tm.texi: Regenerate.
12734 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
12736 * config/i386/i386.opt: New option mstack-protector-guard=.
12737 * config/i386/i386-opts.h: Add enum stack_protector_guard.
12738 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12739 TARGET_SSP_TLS_GUARD.
12740 * config/i386/i386.c (ix86_option_override_internal): Set
12741 ix86_stack_protector_guard.
12742 * config/i386/i386.md (stack_protect_set): Enable for
12743 TARGET_SSP_TLS_GUARD only.
12744 (stack_protect_set_<mode>): Ditto.
12745 (stack_protect_test): Ditto.
12746 (stack_protect_test_<mode>): Ditto.
12747 * doc/invoke.texi (i386 Option): Document.
12749 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
12752 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12753 (S_MODES): Set H_MODE bit.
12754 (SF_MODES): Set only S_MODE and SF_MODE bits.
12755 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12756 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12757 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12758 <MODE_FLOAT>: Likewise.
12760 2013-04-15 Joey Ye <joey.ye@arm.com>
12762 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12764 2013-04-15 Joey Ye <joey.ye@arm.com>
12766 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12768 (thumb_far_jump_used_p): Count instruction size and set
12771 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
12773 * reorg.c (fill_simple_delay_slots): Reindent block of code.
12774 * resource.c (mark_target_live_regs): Reformat conditional block.
12776 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
12778 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12779 notes, they are emitted only just before final.
12780 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12782 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
12784 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12785 * cfgrtl.c (delete_insn): Call it here instead.
12786 * lra-spills.c (lra_final_code_change): Use delete_insn.
12787 * haifa-sched.c (sched_remove_insn): Likewise.
12788 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12789 returning to the nop pool.
12790 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12791 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
12793 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
12795 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12796 * doc/tm.texi: Regenerated.
12798 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
12800 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12803 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
12805 * df-core.c (df_find_def): Compare register numbers.
12806 (df_find_use): Likewise.
12808 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
12811 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12812 lra_in_progress for return.
12814 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
12816 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12817 define_insn into define_insn_and_split and emit movsicc patterns.
12819 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
12821 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12823 2013-04-12 Richard Biener <rguenther@suse.de>
12825 * tree-pass.h (TODO_do_not_ggc_collect): New.
12826 * passes.c (execute_one_ipa_transform_pass): Honor
12827 TODO_do_not_ggc_collect.
12828 (execute_one_pass): Likewise.
12831 2013-04-10 Richard Biener <rguenther@suse.de>
12833 * passes.c (init_optimization_passes): Remove reload pass.
12834 * ira.c (do_reload): Merge into ...
12836 (rest_of_handle_reload): Remove.
12837 (pass_reload): Likewise.
12838 * config/i386/i386.c (ix86_option_override): Refer to ira instead
12839 of reload for vzeroupper pass placement.
12841 2013-04-12 Jakub Jelinek <jakub@redhat.com>
12843 PR tree-optimization/56918
12844 PR tree-optimization/56920
12845 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12846 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
12847 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
12848 use wide_mul_with_sign method.
12850 2013-04-12 Richard Biener <rguenther@suse.de>
12852 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12853 not be considered a gimple constant.
12855 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
12857 * fold-const.c (const_binop): Handle vector shifts by a scalar.
12858 (fold_binary_loc): Call const_binop also for mixed vector-scalar
12861 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
12862 Jakub Jelinek <jakub@redhat.com>
12864 * opts.c: Include diagnostic-color.h.
12865 (common_handle_option): Handle OPT_fdiagnostics_color_.
12866 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12867 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12868 (diagnostic-color.o): New.
12869 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12870 (diagnostic_color_rule): New enum.
12871 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12872 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12873 the location string.
12874 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12875 either NULL, or color kind.
12876 * diagnostic-color.c: New file.
12877 * diagnostic-color.h: New file.
12878 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12880 * doc/invoke.texi (-fdiagnostics-color): Document.
12881 * pretty-print.h (pp_show_color): Define.
12882 (struct pretty_print_info): Add show_color field.
12883 * diagnostic.c: Include diagnostic-color.h.
12884 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12885 macros. Colorize error:, warning: etc. strings and also the location
12887 (diagnostic_show_locus): Colorize the caret line.
12888 * pretty-print.c: Include diagnostic-color.h.
12889 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
12890 inside of %< %> quotes or quoted through q format modifier.
12892 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12894 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12896 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12898 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12899 code in CC_NZ mode.
12900 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12903 2013-04-11 Marek Polacek <polacek@redhat.com>
12905 PR tree-optimization/48184
12906 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12908 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
12910 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12911 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12912 (skip_simple_arithmetic): Tidy up.
12913 * tree.h (skip_simple_constant_arithmetic): Declare.
12915 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12917 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12919 2013-04-11 Richard Biener <rguenther@suse.de>
12921 * tree-vect-loop.c (get_initial_def_for_induction): Properly
12922 generate vector constants.
12924 2013-04-11 Richard Biener <rguenther@suse.de>
12926 PR tree-optimization/56878
12927 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12928 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12930 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12931 Prefer to align the DR with the most invariant base address.
12933 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
12935 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12938 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
12940 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12941 floating-point vector comparisons against 0.
12943 2013-04-11 Jakub Jelinek <jakub@redhat.com>
12945 PR tree-optimization/56899
12946 * fold-const.c (extract_muldiv_1): Apply distributive law
12947 only if TYPE_OVERFLOW_WRAPS (ctype).
12949 2013-04-11 Bin Cheng <bin.cheng@arm.com>
12952 * ira-costs.c (scan_one_insn): Check whether the source rtx of
12953 loading has side effect.
12955 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
12957 * config/sparc/sparc.c: Include tree-pass.h.
12958 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12959 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
12960 head of file. Change return type. Split off gate function.
12961 (sparc_gate_work_around_errata): New function.
12962 (pass_work_around_errata): New pass definition.
12963 (insert_pass_work_around_errata) New pass insert definition to
12964 insert pass_work_around_errata just after delayed-branch scheduling.
12965 (sparc_option_override): Insert the pass.
12966 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12968 2013-04-10 David S. Miller <davem@davemloft.net>
12970 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12971 or -mcpu=hypersparc.
12973 * target.def (cstore_mode): New hook.
12974 * target.h: Include insn-codes.h
12975 * targhooks.c: Likewise.
12976 (default_cstore_mode): New function.
12977 * targhooks.h: Declare it.
12978 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12979 * doc/tm.texi: Rebuild.
12980 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12981 target hook, rather than inspecting the insn_data.
12982 * config/sparc/sparc.c (sparc_cstore_mode): New function.
12983 (TARGET_CSTORE_MODE): Redefine.
12984 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12986 * config/sparc/predicates.md (cstore_result_operand): New special
12988 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12989 Use it for operand 0.
12990 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12991 (*snesi_special): Likewise.
12992 (*snesi_zero): Likewise.
12993 (*seqsi_zero): Likewise.
12994 (*sltu_insn): Likewise.
12995 (*sgeu_insn): Likewise.
12996 (*seqdi_special): Make operand 0 and comparison operation be of
12998 (*snedi_special): Likewise.
12999 (*snedi_special_vis3): Likewise.
13000 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
13001 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
13002 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
13003 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
13004 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
13005 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
13006 (*sltu_extend_sp64): Likewise.
13007 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
13008 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
13009 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
13010 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
13011 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
13013 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
13015 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
13016 (aarch64_start_file): Use the new function.
13018 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
13019 Jason Merrill <jason@redhat.com>
13021 * common.opt: Add -gdwarf.
13022 * opts.c (common_handle_option): Handle it.
13023 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
13025 2013-04-10 Richard Biener <rguenther@suse.de>
13027 * passes.c (execute_todo): Do not call ggc_collect conditional here.
13028 (execute_one_ipa_transform_pass): But unconditionally here.
13029 (execute_one_pass): And here.
13030 (init_optimization_passes): Remove reload pass.
13031 * tree-pass.h (TODO_ggc_collect): Remove.
13032 (pass_reload): Likewise.
13033 * ira.c (do_reload): Merge into ...
13035 (rest_of_handle_reload): Remove.
13036 (pass_reload): Likewise.
13037 * config/i386/i386.c (ix86_option_override): Refer to ira instead
13038 of reload for vzeroupper pass placement.
13039 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
13040 and todo_flags_finish of all passes.
13042 2013-04-10 Richard Biener <rguenther@suse.de>
13044 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
13045 first_const_oprnd field, rename first_def_type to first_op_type.
13046 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
13047 (vect_get_and_check_slp_defs): Always use the type of the
13048 operand. Allow mixed vect_external_def, vect_constant_def types.
13049 (vect_get_constant_vectors): Handle mixed vect_external_def,
13050 vect_constant_def types.
13052 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
13054 PR tree-optimization/55524
13055 * tree-ssa-math-opts.c
13056 (convert_mult_to_fma): Don't use an fms construct
13057 when we don't have an fms operation, but fnma, and it looks
13058 likely that we'll be able to use the latter.
13060 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
13062 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
13064 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
13065 inline fail caused by overwritable functions.
13067 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
13069 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
13070 unnecessary bits in the constant power of two case.
13072 2013-04-10 Richard Biener <rguenther@suse.de>
13074 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
13075 broken code swapping operands.
13076 (vect_build_slp_tree): Do not compute load permutations here.
13077 (vect_analyze_slp_instance): Compute load permutations here,
13078 after building the SLP tree.
13080 2013-04-09 Christian Bruel <christian.bruel@st.com>
13082 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
13083 of next/prev_real_insn.
13085 2013-04-09 Jan Hubicka <jh@suse.cz>
13087 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
13088 Drop aliased parameter.
13089 (function_and_variable_visibility): Do not handle alias pairs.
13090 * cgraph.c (varpool_externally_visible_p): Update prototype.
13091 * varpool.c (varpool_add_new_variable): Update.
13093 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13095 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
13097 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
13099 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
13101 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
13103 2013-04-09 Marek Polacek <polacek@redhat.com>
13105 PR tree-optimization/48762
13106 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
13108 2013-04-09 Richard Biener <rguenther@suse.de>
13110 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
13112 (vect_build_slp_tree): Likewise.
13113 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
13114 calculating the cost of a SLP instance.
13115 (vect_analyze_slp_instance): Use it from here, after building
13118 2013-04-09 Jakub Jelinek <jakub@redhat.com>
13120 PR middle-end/56883
13121 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
13122 expand_omp_for_static_chunk): Use simple_p = true in
13123 force_gimple_operand_gsi calls when assigning to addressable decls.
13125 2013-04-09 Jeff Law <law@redhat.com>
13127 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
13128 when the boolean was created by converting a wider object which
13129 had a boolean range.
13131 2013-04-09 Richard Biener <rguenther@suse.de>
13133 * tree-vectorizer.h (slp_void_p): Remove.
13134 (slp_tree): Typedef before _slp_tree declaration.
13135 (struct _slp_tree): Use a vector of slp_tree as children.
13136 (vect_get_place_in_interleaving_chain): Remove.
13137 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
13139 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
13141 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
13142 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
13143 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
13144 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
13145 Use slp_node instead of slp_void_p and adjust.
13147 2013-04-09 Richard Biener <rguenther@suse.de>
13149 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
13150 work that is not necessary.
13152 2013-04-09 Jakub Jelinek <jakub@redhat.com>
13154 PR tree-optimization/56854
13155 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
13156 forward into clobber stmts if it would change MEM_REF lhs into
13159 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
13161 * tree.c (type_hash_lookup, type_hash_add): Make static.
13162 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
13164 2013-04-09 Richard Biener <rguenther@suse.de>
13166 * tree.h (unsave_expr_now): Remove.
13167 * tree-inline.c (mark_local_for_remap_r): Remove.
13168 (unsave_expr_1): Likewise.
13169 (unsave_r): Likewise.
13170 (unsave_expr_now): Likewise.
13171 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
13172 (propagate_tree_value): Likewise.
13174 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
13176 * doc/rtl.texi (sequence): Rewrite documentation to match the
13177 current use of SEQUENCE rtl objects.
13178 * rtl.def (SEQUENCE): Likewise.
13180 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
13181 Update documentation.
13182 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
13183 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
13185 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
13187 2013-04-08 Teresa Johnson <tejohnson@google.com>
13189 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
13190 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
13192 (estimate_edge_size_and_time): Add comment to suggest using rounding
13194 (estimate_node_size_and_time): Ditto.
13195 (remap_edge_change_prob): Use helper rounding divide methods.
13196 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
13197 (gimple_mod_pow2_value_transform): Ditto.
13198 (gimple_mod_subtract_transform): Ditto.
13199 (gimple_ic_transform): Ditto.
13200 (gimple_stringops_transform): Ditto.
13201 * stmt.c (conditional_probability): Ditto.
13202 (emit_case_dispatch_table): Ditto.
13203 * lto-cgraph.c (merge_profile_summaries): Ditto.
13204 * tree-optimize.c (execute_fixup_cfg): Ditto.
13205 * cfgcleanup.c (try_forward_edges): Ditto.
13206 * cfgloopmanip.c (scale_loop_profile): Ditto.
13208 (duplicate_loop_to_header_edge): Ditto.
13209 (lv_adjust_loop_entry_edge): Ditto.
13210 * tree-vect-loop.c (vect_transform_loop): Ditto.
13211 * profile.c (compute_branch_probabilities): Ditto.
13212 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
13213 * lto-streamer-in.c (input_cfg): Ditto.
13214 * gimple-streamer-in.c (input_bb): Ditto.
13215 * ipa-cp.c (update_profiling_info): Ditto.
13216 (update_specialized_profile): Ditto.
13217 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
13218 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
13220 * sched-rgn.c (compute_dom_prob_ps): Ditto.
13221 (compute_trg_info): Ditto.
13222 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
13223 (purge_dead_edges): Ditto.
13224 * loop-unswitch.c (unswitch_loop): Ditto.
13225 * cgraphclones.c (cgraph_clone_edge): Ditto.
13226 (cgraph_clone_node): Ditto.
13227 * tree-inline.c (copy_bb): Ditto.
13228 (copy_edges_for_bb): Ditto.
13229 (initialize_cfun): Ditto.
13230 (copy_cfg_body): Ditto.
13231 (expand_call_inline): Ditto.
13233 2013-04-08 Kai Tietz <ktietz@redhat.com>
13235 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
13236 TARGET_CYGWIN64 by TARGET_64BIT.
13238 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
13240 * config/epiphany/epiphany.md (GPR_1): New constant.
13241 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
13242 * config/epiphany/epiphany.c (gen_compare_reg):
13243 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
13244 is already in place.
13245 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
13246 Don't require being called during rtl expansion; If y operlaps r0,
13248 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
13249 (epiphany_expand_epilogue): Likewise.
13251 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
13252 Don't use CC_FPmode for ORDERED / UNORDERED.
13253 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
13255 * config/epiphany/constraints.md (CnL): New constraint.
13256 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
13257 * config/epiphany/predicates.md (add_operand): Allow 1024.
13259 * config/epiphany/epiphany.md (logical_op): New code iterator.
13260 (op_mnc): New code attribute.
13261 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
13262 (mov_f+1, mov_f+2): New peephole2 patterns.
13264 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
13265 (cstoresi4): Also allow re-use of zero result when doing a NE
13266 comparison to a non-zero operand.
13267 Use (clobber (scratch)) for first insn if the gpr output is not needed.
13269 * config/epiphany/epiphany.md (<insn_opname>v2si3):
13270 Use gen_addsi3_i / gen_subsi3_i.
13272 2013-04-08 Jakub Jelinek <jakub@redhat.com>
13276 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
13277 contain anything but clobbers, at most one __builtin_stack_restore,
13278 optionally debug stmts and final resx, and if it has at least one
13279 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
13280 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
13281 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
13282 which isn't defaut definition, remove them.
13283 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
13284 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
13285 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
13286 with MEM_REF LHS with SSA_NAME address.
13288 2013-04-08 Jeff Law <law@redhat.com>
13290 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
13292 2013-04-08 Richard Biener <rguenther@suse.de>
13294 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
13296 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
13297 determined vector type.
13298 (vect_analyze_data_refs): Likewise.
13299 (vect_get_new_vect_var): Adjust.
13300 (vect_create_destination_var): Preserve SSA name versions.
13301 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
13302 not dump anything here.
13304 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
13306 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
13307 Add member lr_slot_known.
13308 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
13310 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
13311 Remove code that sets lr_slot_offset according to what a previous
13312 version of epiphany_emit_save_restore used to do.
13313 (epiphany_emit_save_restore): When doing an lr save or restore,
13314 set/verify lr_slot_known and lr_slot_offset.
13316 2013-04-08 Xinyu Qi <xyqi@marvell.com>
13319 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
13322 2013-04-08 Richard Biener <rguenther@suse.de>
13324 * alias.c (find_base_term): Fix thinko in previous change.
13326 2013-04-08 Jakub Jelinek <jakub@redhat.com>
13328 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
13329 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
13330 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
13331 if possible to compute val.
13332 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
13333 For QImode integers don't require anything about precision. Use
13334 const_with_all_bytes_same to find out if the constant doesn't have
13335 repeated bytes in it.
13337 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
13339 * config/s390/s390.c (s390_expand_insv): Only accept insertions
13342 2013-04-08 Marek Polacek <polacek@redhat.com>
13344 PR rtl-optimization/48182
13345 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
13348 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
13351 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
13352 nuses, make sure we have a label.
13354 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13357 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
13358 (rs6000_emit_swdiv_low_precision): Remove.
13359 (rs6000_emit_swdiv): Rewrite to handle between one and four
13360 iterations of Newton-Raphson generally; modify required number of
13361 iterations for some cases.
13362 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
13364 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
13366 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
13367 set-but-unused variable.
13369 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
13370 basic blocks of released function bodies garbage-collectable.
13372 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
13373 (struct rtl_opt_pass): Add TODO_df_finish.
13375 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
13377 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13379 * config/arm/constraints.md (q): New constraint.
13380 * config/arm/ldrdstrd.md: New file.
13381 * config/arm/arm.md (ldrdstrd.md) New include.
13382 (arm_movdi): Use "q" instead of "r" constraint
13383 for double-word memory access.
13384 (movdf_soft_insn): Likewise.
13385 * config/arm/vfp.md (movdi_vfp): Likewise.
13386 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
13387 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
13388 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
13389 (mem_ok_for_ldrd_strd): Likewise.
13390 (output_move_double): Update assertion.
13392 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13394 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
13396 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13398 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
13399 define_insn_and_split.
13400 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
13402 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13404 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
13405 define_insn_and_split.
13406 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
13407 (shiftsi3_compare): New pattern.
13408 (rrx): New pattern.
13409 * config/arm/unspecs.md (UNSPEC_RRX): New.
13411 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13413 * config/arm/arm.md (negdi_extendsidi): New pattern.
13414 (negdi_zero_extendsidi): Likewise.
13416 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13418 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
13419 define_insn_and_split.
13420 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
13421 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
13423 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13425 * config/arm/arm.md (arm_subdi3): Convert define_insn into
13426 define_insn_and_split.
13427 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
13428 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
13430 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13432 * config/arm/arm.md (subsi3_carryin): New pattern.
13433 (subsi3_carryin_const): Likewise.
13434 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
13435 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
13437 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13439 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
13441 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
13443 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
13444 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
13446 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13448 * config/arm/arm.c (arm_expand_builtin): Change fcode
13449 type to unsigned int.
13451 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13453 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
13455 2013-04-04 Ian Lance Taylor <iant@google.com>
13457 * doc/standards.texi (Standards): The Go frontend supports the Go 1
13460 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
13462 PR middle-end/56729
13463 * df-scan.c (df_insn_delete): Disable failing assert.
13465 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13467 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
13468 New function prototype.
13469 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
13470 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
13471 (arm_builtin_vectorized_function): New function.
13473 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13475 * config/arm/arm_neon_builtins.def: New file.
13476 * config/arm/arm.c (neon_builtin_data): Move contents to
13477 arm_neon_builtins.def.
13478 (enum arm_builtins): Include neon builtin definitions.
13479 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
13480 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
13482 2013-04-04 Marek Polacek <polacek@redhat.com>
13484 PR tree-optimization/48186
13485 * predict.c (maybe_hot_frequency_p): Return false if
13486 HOT_BB_FREQUENCY_FRACTION is 0.
13487 (cgraph_maybe_hot_edge_p): Likewise.
13489 2013-04-04 Richard Biener <rguenther@suse.de>
13491 PR tree-optimization/56826
13492 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
13495 2013-04-04 Richard Biener <rguenther@suse.de>
13497 PR tree-optimization/56213
13498 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
13499 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
13501 2013-04-04 Richard Biener <rguenther@suse.de>
13503 PR tree-optimization/56837
13504 * tree-loop-distribution.c (classify_partition): For non-zero
13505 values require that the value has the same precision as its
13506 mode to be useful as memset value.
13508 2013-04-03 Nick Clifton <nickc@redhat.com>
13510 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
13511 (fmssf4): Use fmsf.s on E3V5 architectures.
13512 (fnmasf4): Use fnmaf.s on E3V5 architectures.
13513 (fnmssf4): Use fnmsf.s on E3V5 architectures.
13515 2013-04-03 Jeff Law <law@redhat.com>
13517 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
13518 (lra-eliminations.o): Likewise.
13520 2013-04-03 Teresa Johnson <tejohnson@google.com>
13522 * gcov-io.c (compute_working_sets): Moved most of body of old
13523 compute_working_sets here from profile.c.
13524 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
13525 (gcov_working_set_t): Moved typedef here from basic-block.h
13526 (compute_working_set): Declare.
13527 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
13528 (get_working_sets): Renamed from compute_working_set,
13529 replace most of body with call to new compute_working_sets.
13530 (get_exec_counts): Replace call to compute_working_sets
13531 to get_working_sets.
13532 * profile.h (get_working_sets): Renamed from compute_working_set.
13533 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
13534 to get_working_sets.
13535 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
13536 * gcov-dump.c (dump_working_sets): New function.
13538 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
13540 * hwint.c (sext_hwi, zext_hwi): New functions.
13541 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
13542 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
13543 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
13544 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
13545 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
13546 (sext_hwi, zext_hwi): New functions.
13548 2013-04-03 Jeff Law <law@redhat.com>
13550 PR tree-optimization/56799
13551 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
13552 back test for widening conversion erroneously dropped in prior change.
13554 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13557 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
13558 instead of next_real_insn.
13560 2013-04-03 Marek Polacek <polacek@redhat.com>
13563 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13565 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13568 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
13570 (thumb1_output_casesi): Likewise.
13571 (thumb2_output_casesi): Likewise.
13573 2013-04-03 Richard Biener <rguenther@suse.de>
13575 PR tree-optimization/56817
13576 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
13578 (tree_unroll_loops_completely_1): ... new function to manually
13579 walk the loop tree, properly defering outer loops of unrolled
13580 loops to later iterations.
13582 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13584 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
13585 (vectorizable_load): Likewise.
13586 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
13587 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
13589 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13591 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
13594 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
13596 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
13598 2013-04-03 Bin Cheng <bin.cheng@arm.com>
13600 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13602 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13604 PR tree-optimization/56790
13605 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13608 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13610 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13612 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13613 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13616 2013-04-03 Jakub Jelinek <jakub@redhat.com>
13619 * tree.h (force_folding_builtin_constant_p): New decl.
13620 * builtins.c (force_folding_builtin_constant_p): New variable.
13621 (fold_builtin_constant_p): Fold immediately also if
13622 force_folding_builtin_constant_p.
13624 2013-04-03 Richard Biener <rguenther@suse.de>
13626 PR tree-optimization/56812
13627 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13628 DRs of the same interleaving chain are independent.
13630 2013-04-02 Jason Merrill <jason@redhat.com>
13632 * gdbinit.in (pbb): Use debug fn.
13634 2013-04-02 Lawrence Crowl <crowl@google.com>
13636 * sese.h (struct ivtype_map_elt_s): Remove unused.
13637 (extern debug_ivtype_map): Remove unused.
13638 (extern eq_ivtype_map_elts): Remove unused.
13639 * sese.c (debug_ivtype_map): Removed unused.
13640 (debug_ivtype_map_1): Removed unused.
13641 (debug_ivtype_elt): Remove unused.
13642 (eq_ivtype_map_elts): Remove unused.
13645 2013-04-02 Kai Tietz <ktietz@redhat.com>
13648 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13649 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13650 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13652 (legitimize_pe_coff_symbol): Likewise.
13653 (is_imported_p): New helper-function.
13654 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13655 for Windows x64 targets.
13656 (ix86_expand_prologue): Optimize for pe-coff targets.
13657 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13658 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13659 medium/large code-model.
13660 (legitimize_pic_address): Likewise.
13661 (legitimize_tls_address): Likewise.
13662 (ix86_expand_call): Likewise.
13663 (x86_output_mi_thunk): Likewise.
13664 (get_dllimport_decl): Add new beimport argument.
13665 (construct_plt_address): Don't assert for x64 pe-coff targets.
13666 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13668 (SYMBOL_FLAG_STUBVAR): New macro.
13669 (SYMBOL_REF_STUBVAR_P): Likewise.
13670 * config/i386/winnt.c (stub_list): New structure.
13671 (stub_head): New local variable.
13672 (i386_pe_record_stub): New function.
13673 (i386_pe_file_end): Emit refptr-stubs.
13675 2013-04-02 Jakub Jelinek <jakub@redhat.com>
13677 PR rtl-optimization/56745
13678 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13679 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13682 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13683 and both of them are MEM_REFs, just compare first argument for
13684 equality and attempt to deal even with differing offsets.
13687 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13688 of gimple_clobber_p to be MEM_REF.
13689 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13690 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
13691 after gimplification.
13692 * asan.c (get_mem_ref_of_assignment): Don't instrument
13693 gimple_clobber_p stmts.
13694 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13695 gimple_clobber_p stmt if they have MEM_REF lhs and
13696 are dead because of another gimple_clobber_p stmt.
13697 * tree-ssa-live.c (clear_unused_block_pointer): Treat
13698 gimple_clobber_p stmts like debug stmts.
13699 (remove_unused_locals): Remove clobbers with MEM_REF lhs
13700 that refer to unused VAR_DECLs or uninitialized values.
13701 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13702 gimple_clobber_p stmts if they refer to removed parameters.
13703 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13706 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
13708 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13709 using SWI48 mode attribute.
13711 2013-04-02 Wei Mi <wmi@google.com>
13713 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13714 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13715 *<rotate_insn><mode>3_mask in i386.md.
13717 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
13719 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13721 2013-04-02 Richard Biener <rguenther@suse.de>
13723 PR tree-optimization/56778
13724 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13725 Runtime alias tests are not supported for gather loads.
13726 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13727 stmts referenced from SSA operands before updating SSA form.
13729 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
13730 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13732 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13733 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13734 * config/arm/cortex-a53.md: New file.
13735 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13736 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13737 * config/arm/arm.c (arm_issue_rate): Likewise.
13738 * config/arm/arm-tune.md: Regenerate
13739 * config/arm/arm-tables.opt: Regenerate.
13740 * config/arm/arm-cores.def: Add cortex-a53.
13742 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
13744 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13747 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
13749 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13750 scalar load/store operations using B/H registers.
13751 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13753 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
13755 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13757 * config/aarch64/aarch64.c
13758 (aarch64_simd_scalar_immediate_valid_for_move): New.
13759 * config/aarch64/aarch64-protos.h
13760 (aarch64_simd_scalar_immediate_valid_for_move): New.
13761 * config/aarch64/constraints.md (Dh, Dq): New.
13762 * config/aarch64/iterators.md (hq): New.
13764 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
13766 * reorg.c (get_branch_condition): Deal with conditional returns.
13767 (fill_simple_delay_slots): Remove dead code dealing with jumps.
13769 2013-04-01 Wei Mi <wmi@google.com>
13771 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13772 Truncate operand 2 using %b asm operand modifier.
13773 (*<shift_insn><mode>3_mask): Ditto.
13774 (*<rotate_insn><mode>3_mask): Ditto.
13776 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
13778 PR middle-end/56798
13779 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13781 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
13783 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13785 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13787 2013-03-30 Lawrence Crowl <crowl@google.com>
13789 * dse.c (clear_alias_sets): Remove never set.
13790 (disqualified_clear_alias_sets): Remove never set.
13791 (clear_alias_mode_pool): Remove never set.
13792 (dse_step0): Remove condition that is never true.
13793 (canon_address): Remove condition that is never true.
13794 (dse_step7): Remove condition that is never true.
13795 (rest_of_handle_dse): Remove condition that is never true.
13796 (rest_of_handle_dse::did_global): Remove never read from above.
13797 (dse_step2_spill): Remove never called from above.
13798 (dse_step5_spill): Remove never called from above.
13800 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
13802 * doc/md.texi (Standard Names) <casesi>: Update documentation for
13803 JUMP_TABLE_DATA changes.
13804 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13805 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13806 (Insns) <jump_table_data>: New entry.
13807 * doc/tm.texi: Regenerate.
13809 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13811 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13812 for table jump at the end of a basic block using tablejump_p.
13813 * targhooks.c (default_invalid_within_doloop): Likewise.
13814 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13815 target hook implementation that is identical to the default hook.
13816 (rs6000_invalid_within_doloop): Remove.
13818 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13819 unused variable from tablejump_p call.
13821 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13822 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13823 (INSN_DELETED_P): Likewise.
13824 (emit_jump_table_data): New prototype.
13825 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13826 after 4th as unused.
13827 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13828 * sched-vis.c (print_insn): Likewise.
13829 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13830 insn for compatibility with back ends that use next_active_insn to
13831 identify jump table data.
13832 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13833 (remove_insn): Likewise.
13834 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13836 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13837 (emit_jump_table_data): New function.
13839 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13840 basic block, a JUMP_TABLE_DATA never is.
13841 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13842 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13843 off from code handling real insns.
13844 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13845 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13846 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13847 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13848 is not a NONDEBUG_INSN_P.
13849 * ira-costs.c (scan_one_insn): Likewise.
13850 * jump.c (mark_all_labels): Likewise.
13851 (mark_jump_label_1): Likewise.
13852 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13853 * lra.c (get_insn_freq): Expect all insns reaching here to be in
13855 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13856 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13857 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13858 JUMP_TABLE_DATA_P insns.
13859 (calculate_elim_costs_all_insns): Likewise.
13860 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13861 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13862 (delete_output_reload): Code style fixups.
13863 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13864 insn flags on this non-insn.
13865 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13866 as scheduling barriers, for pre-change compatibility.
13867 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13868 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13870 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13871 redundant JUMP_TABLE_DATA_P test.
13872 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13873 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13874 (frv_for_each_packet): Likewise.
13875 * config/i386/i386.c (min_insn_size): Likewise.
13876 (ix86_avoid_jump_mispredicts): Likewise.
13877 * config/m32r/m32r.c (m32r_is_insn): Likewise.
13878 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13879 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13880 (mips16_insn_length): Robustify.
13881 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13882 (mips16_split_long_branches): Likewise.
13883 * config/pa/pa.c (pa_combine_instructions): Likewise.
13884 * config/rs6000/rs6000.c (get_next_active_insn): Treat
13885 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13886 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13887 as contributing to pool range lengths.
13888 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13889 Remove redundant JUMP_TABLE_DATA_P test.
13890 (sh_loop_align): Likewise.
13891 (split_branches): Likewise.
13892 (sh_insn_length_adjustment): Likewise.
13893 * config/spu/spu.c (get_branch_target): Likewise.
13895 2013-03-29 Jan Hubicka <jh@suse.cz>
13897 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13898 gcov streaming; stream hot bb threshold to ltrans.
13899 * predict.c (get_hot_bb_threshold): Break out from ....
13900 (maybe_hot_count_p): ... here.
13901 (set_hot_bb_threshold): New function.
13902 * lto-section-in.c (lto_section_name): Add profile.
13903 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13904 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13905 and data-streamer.h
13906 (histogram_entry): New structure.
13907 (histogram, histogram_pool): New global vars.
13908 (histogram_hash): New structure.
13909 (histogram_hash::hash): New method.
13910 (histogram_hash::equal): Likewise.
13911 (account_time_size): New function.
13912 (cmp_counts): New function.
13913 (dump_histogram): New function.
13914 (ipa_profile_generate_summary): New function.
13915 (ipa_profile_write_summary): New function.
13916 (ipa_profile_read_summary): New function.
13917 (ipa_profile): Decide on threshold.
13918 (pass_ipa_profile): Add ipa_profile_write_summary and
13919 ipa_profile_read_summary.
13920 * Makefile.in (ipa.o): Update dependencies.
13921 * lto-streamer.h (LTO_section_ipa_profile): New section.
13923 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
13925 * tree.h (VAR_P): New.
13927 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
13930 * doc/invoke.texi ([-fwhole-program]): Fix typo.
13932 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
13934 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13935 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13936 (control_flow_insn_p): Likewise.
13937 * cfgrtl.c (duplicate_insn_chain): Likewise.
13938 * final.c (get_attr_length_1): Likewise.
13939 (shorten_branches): Likewise.
13940 (final_scan_insn): Likewise.
13941 * function.c (instantiate_virtual_regs): Likewise.
13942 * gcse.c (insert_insn_end_basic_block): Likewise.
13943 * ira-costs.c (scan_one_insn): Likewise.
13944 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13945 * lra.c (check_rtl): Likewise.
13946 * reload1.c (elimination_costs_in_insn): Likewise.
13947 * reorg.c (follow_jumps): Likewise.
13949 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13950 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13951 (thumb_far_jump_used_p): Likewise.
13952 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13953 (workaround_speculation): Likewise.
13954 (add_sched_insns_for_speculation): Likewise.
13955 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13956 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13957 (frv_for_each_packet): Likewise.
13958 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13959 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13960 (final_emit_insn_group_barriers): Likewise.
13961 * config/m32r/m32r.c (m32r_is_insn): Likewise.
13962 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13963 (mips16_insn_length): Likewise.
13964 * config/pa/pa.c (pa_reorg): Likewise.
13965 (pa_combine_instructions): Likewise.
13966 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13967 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13968 (sh_reorg): Likewise.
13969 (split_branches): Likewise.
13970 * config/spu/spu.c (get_branch_target): Likewise.
13972 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13975 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
13977 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13978 Fix declaration name.
13980 2013-03-28 Lawrence Crowl <crowl@google.com>
13982 * graphds.h (struct graph.indicies): Remove unused.
13983 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13984 (SCOP_ORIGINAL_PDDRS): Remove unused.
13985 * sese.h (extern insert_loop_close_phis): Removed unused.
13986 (extern insert_guard_phis): Removed unused.
13987 (extern ivtype_map_elt_info): Removed unused.
13988 (new_ivtype_map_elt): Removed unused.
13989 * sese.c (ivtype_map_elt_info): Removed unused.
13991 2013-03-28 Lawrence Crowl <crowl@google.com>
13993 * Makefile.in: Add several missing include dependences.
13995 (test-dump.o): New. This object is not added to any executable,
13996 but is present for ad-hoc testing.
13998 (debug (const bitmap_head_def &)): New.
13999 (debug (const bitmap_head_def *)): New.
14001 (extern debug (const bitmap_head_def &)): New.
14002 (extern debug (const bitmap_head_def *)): New.
14004 (debug (edge_def &)): New.
14005 (debug (edge_def *)): New.
14007 (debug (basic_block_def &)): New.
14008 (debug (basic_block_def *)): New.
14010 (dump_node (const_tree, int, FILE *)): Correct source file.
14012 (debug (die_struct &)): New.
14013 (debug (die_struct *)): New.
14015 (extern debug (die_struct &)): New.
14016 (extern debug (die_struct *)): New.
14017 * gimple-pretty-print.c
14018 (debug (gimple_statement_d &)): New.
14019 (debug (gimple_statement_d *)): New.
14020 * gimple-pretty-print.h
14021 (extern debug (gimple_statement_d &)): New.
14022 (extern debug (gimple_statement_d *)): New.
14024 (debug (ira_allocno_copy &)): New.
14025 (debug (ira_allocno_copy *)): New.
14026 (debug (ira_allocno &)): New.
14027 (debug (ira_allocno *)): New.
14029 (extern debug (ira_allocno_copy &)): New.
14030 (extern debug (ira_allocno_copy *)): New.
14031 (extern debug (ira_allocno &)): New.
14032 (extern debug (ira_allocno *)): New.
14034 (debug (live_range &)): New.
14035 (debug (live_range *)): New.
14037 (debug (lra_live_range &)): New.
14038 (debug (lra_live_range *)): New.
14040 (debug (lra_live_range &)): New.
14041 (debug (lra_live_range *)): New.
14043 (debug (omega_pb_d &)): New.
14044 (debug (omega_pb_d *)): New.
14046 (extern debug (omega_pb_d &)): New.
14047 (extern debug (omega_pb_d *)): New.
14049 (debug (const rtx_def &)): New.
14050 (debug (const rtx_def *)): New.
14052 (debug_tree (tree): Move within file.
14053 (debug_raw (const tree_node &)): New.
14054 (debug_raw (const tree_node *)): New.
14055 (dump_tree_via_hooks (const tree_node *, int)): New.
14056 (debug (const tree_node &)): New.
14057 (debug (const tree_node *)): New.
14058 (debug_verbose (const tree_node &)): New.
14059 (debug_verbose (const tree_node *)): New.
14060 (debug_head (const tree_node &)): New.
14061 (debug_head (const tree_node *)): New.
14062 (debug_body (const tree_node &)): New.
14063 (debug_body (const tree_node *)): New.
14064 (debug_vec_tree (tree): Move and reimplement in terms of dump.
14065 (debug (vec<tree, va_gc> &)): New.
14066 (debug (vec<tree, va_gc> *)): New.
14068 (extern debug (const rtx_def &)): New.
14069 (extern debug (const rtx_def *)): New.
14071 (debug_raw (simple_bitmap_def &)): New.
14072 (debug_raw (simple_bitmap_def *)): New.
14073 (debug (simple_bitmap_def &)): New.
14074 (debug (simple_bitmap_def *)): New.
14076 (extern debug (simple_bitmap_def &)): New.
14077 (extern debug (simple_bitmap_def *)): New.
14078 (extern debug_raw (simple_bitmap_def &)): New.
14079 (extern debug_raw (simple_bitmap_def *)): New.
14081 (debug (vinsn_def &)): New.
14082 (debug (vinsn_def *)): New.
14083 (debug_verbose (vinsn_def &)): New.
14084 (debug_verbose (vinsn_def *)): New.
14085 (debug (expr_def &)): New.
14086 (debug (expr_def *)): New.
14087 (debug_verbose (expr_def &)): New.
14088 (debug_verbose (expr_def *)): New.
14089 (debug (vec<rtx> &)): New.
14090 (debug (vec<rtx> *)): New.
14092 (extern debug (vinsn_def &)): New.
14093 (extern debug (vinsn_def *)): New.
14094 (extern debug_verbose (vinsn_def &)): New.
14095 (extern debug_verbose (vinsn_def *)): New.
14096 (extern debug (expr_def &)): New.
14097 (extern debug (expr_def *)): New.
14098 (extern debug_verbose (expr_def &)): New.
14099 (extern debug_verbose (expr_def *)): New.
14100 (extern debug (vec<rtx> &)): New.
14101 (extern debug (vec<rtx> *)): New.
14103 (_list_iter_cond_expr): Make inline instead of static.
14105 (debug (sreal &)): New.
14106 (debug (sreal *)): New.
14108 (extern debug (sreal &)): New.
14109 (extern debug (sreal *)): New.
14111 (extern debug_raw (const tree_node &)): New.
14112 (extern debug_raw (const tree_node *)): New.
14113 (extern debug (const tree_node &)): New.
14114 (extern debug (const tree_node *)): New.
14115 (extern debug_verbose (const tree_node &)): New.
14116 (extern debug_verbose (const tree_node *)): New.
14117 (extern debug_head (const tree_node &)): New.
14118 (extern debug_head (const tree_node *)): New.
14119 (extern debug_body (const tree_node &)): New.
14120 (extern debug_body (const tree_node *)): New.
14121 (extern debug (vec<tree, va_gc> &)): New.
14122 (extern debug (vec<tree, va_gc> *)): New.
14124 (debug (struct loop &)): New.
14125 (debug (struct loop *)): New.
14126 (debug_verbose (struct loop &)): New.
14127 (debug_verbose (struct loop *)): New.
14128 * tree-dump.c: Add header dependence.
14130 (extern debug (struct loop &)): New.
14131 (extern debug (struct loop *)): New.
14132 (extern debug_verbose (struct loop &)): New.
14133 (extern debug_verbose (struct loop *)): New.
14135 (debug (data_reference &)): New.
14136 (debug (data_reference *)): New.
14137 (debug (vec<data_reference_p> &)): New.
14138 (debug (vec<data_reference_p> *)): New.
14139 (debug (vec<ddr_p> &)): New.
14140 (debug (vec<ddr_p> *)): New.
14142 (extern debug (data_reference &)): New.
14143 (extern debug (data_reference *)): New.
14144 (extern debug (vec<data_reference_p> &)): New.
14145 (extern debug (vec<data_reference_p> *)): New.
14146 (extern debug (vec<ddr_p> &)): New.
14147 (extern debug (vec<ddr_p> *)): New.
14149 (debug (pt_solution &)): New.
14150 (debug (pt_solution *)): New.
14152 (extern debug (pt_solution &)): New.
14153 (extern debug (pt_solution *)): New.
14155 (debug (_var_map &)): New.
14156 (debug (_var_map *)): New.
14157 (debug (tree_live_info_d &)): New.
14158 (debug (tree_live_info_d *)): New.
14160 (extern debug (_var_map &)): New.
14161 (extern debug (_var_map *)): New.
14162 (extern debug (tree_live_info_d &)): New.
14163 (extern debug (tree_live_info_d *)): New.
14165 2013-03-28 Jan Hubicka <jh@suse.cz>
14167 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
14169 2013-03-28 Ian Bolton <ian.bolton@arm.com>
14171 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
14172 record only when desired or required.
14174 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
14176 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
14177 *vec_extractv2di_1_rex64. Use x64 isa attribute.
14179 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14181 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
14182 (*andsi3_compare0_uxtw): New pattern.
14183 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
14184 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
14186 2013-03-28 Jan Hubicka <jh@suse.cz>
14188 * data-streamer-in.c (streamer_read_gcov_count): New function.
14189 * gimple-streamer-out.c: Include value-prof.h.
14190 (output_gimple_stmt): Output histogram.
14191 (output_bb): Use streamer_write_gcov_count.
14192 * value-prof.c: Include data-streamer.h
14193 (dump_histogram_value): Add HIST_TYPE_MAX.
14194 (stream_out_histogram_value): New function.
14195 (stream_in_histogram_value): New function.
14196 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
14197 (stream_out_histogram_value, stream_in_histogram_value): Declare.
14198 * data-streamer-out.c (streamer_write_gcov_count): New function.
14199 (streamer_write_gcov_count_stream): New function.
14200 * lto-cgraph.c (lto_output_edge): Update counter streaming.
14201 (lto_output_node): Likewise.
14202 (input_node, input_edge): Likewise.
14203 * lto-streamer-out.c (output_cfg): Update streaming.
14204 * lto-streamer-in.c (input_cfg): Likewise.
14205 * data-streamer.h (streamer_write_gcov_count,
14206 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
14207 * gimple-streamer-in.c: Include value-prof.h
14208 (input_gimple_stmt): Input histograms.
14209 (input_bb): Update profile streaming.
14211 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
14213 * genmodes.c (emit_max_int): New function.
14214 (emit_insn_modes_h): Added call to emit_max_function.
14215 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
14217 * machmode.def: Fixed comment.
14219 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
14221 * combine.c (try_combine): Removed useless assert.
14222 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
14224 2013-03-28 Marek Polacek <polacek@redhat.com>
14225 Richard Biener <rguenther@suse.de>
14227 PR tree-optimization/56695
14228 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
14229 build signed result of a vector comparison.
14230 * tree-cfg.c (verify_gimple_comparison): Check that a result
14231 of a vector comparison has signed type.
14233 2013-03-28 Richard Biener <rguenther@suse.de>
14235 PR tree-optimization/37021
14236 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
14237 do not restrict gaps between groups.
14238 * tree-vect-stmts.c (vectorizable_load): Properly account for
14239 a gap between groups.
14241 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
14243 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
14244 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
14247 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
14249 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
14250 * doc/extend.texi (Named Address Spaces): Ditto.
14251 (Variable Attributes): Ditto.
14253 2013-03-27 Kai Tietz <ktietz@redhat.com>
14255 * config.build: Add support for cygwin x64 target.
14256 * config.gcc: Likewise.
14257 * config.host: Likewise.
14258 * configure.ac: Likewise
14259 * configure: Regenerated.
14261 2013-03-27 Kai Tietz <ktietz@redhat.com>
14263 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
14264 * config/i386/t-cygwin-w64: New file.
14265 * config/i386/cygwin-w64.h: New file.
14266 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
14267 and add support for x64-cygwin target.
14268 (CPP_SPEC): Likewise.
14269 (CXX_WRAP_SPEC_LIST): Undefine before define.
14270 (LIBGCJ_SONAME): Use 15 as version.
14272 2013-03-27 Richard Biener <rguenther@suse.de>
14274 PR tree-optimization/56716
14275 * tree-ssa-structalias.c (perform_var_substitution): Adjust
14276 dumping for ref nodes.
14278 2013-03-27 Martin Jambor <mjambor@suse.cz>
14280 PR tree-optimization/55334
14281 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
14282 restricted pointers to arrays.
14284 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
14286 * Makefile.in (.SUFFIXES): Add .cc.
14287 (.c.o): Apply same recipe for implicit rule .cc.o.
14289 2013-03-27 Richard Biener <rguenther@suse.de>
14291 PR tree-optimization/37021
14292 * tree-vect-data-refs.c (vect_check_strided_load): Allow
14293 REALPART/IMAGPART_EXPRs around the supported refs.
14294 * tree-ssa-structalias.c (find_func_aliases): Assume that
14295 floating-point values are not used to transfer pointers.
14297 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
14299 * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
14300 * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
14301 * gcc/doc/tm.texi: Regenerate.
14302 * gcc/targhooks.h (default_has_ifunc_p): New.
14303 * gcc/targhooks.c (default_has_ifunc_p): Ditto.
14304 * gcc/config/linux-protos.h: New file.
14305 * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
14306 this hook for linux which disables support of indirect functions in
14308 * gcc/config/linux-android.c: New file.
14309 * gcc/config/t-linux-android.c: Ditto.
14310 * gcc/config.gcc: Added new object file linux-android.o.
14311 * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
14312 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
14313 * gcc/varasm.c (do_assemble_alias): Likewise.
14314 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
14315 doesn't support indirect functions.
14316 * configure: Regenerate.
14318 2013-03-27 Bin Cheng <bin.cheng@arm.com>
14321 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
14322 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
14325 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
14327 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
14329 2013-03-27 Terry Guo <terry.guo@arm.com>
14331 * config/arm/arm-cores.def: Added core cortex-r7.
14332 * config/arm/arm-tune.md: Regenerated.
14333 * config/arm/arm-tables.opt: Regenerated.
14334 * doc/invoke.texi: Added entry for core cortex-r7.
14336 2013-03-27 Walter Lee <walt@tilera.com>
14338 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
14339 double-decrement of next_scratch_regno.
14341 2013-03-27 Walter Lee <walt@tilera.com>
14343 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
14345 (insn_v1mulus): Ditto.
14346 (insn_v2muls): Ditto.
14348 2013-03-27 Walter Lee <walt@tilera.com>
14350 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
14351 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
14353 2013-03-27 Walter Lee <walt@tilera.com>
14355 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
14356 (*sibcall_value): Ditto.
14358 2013-03-27 Walter Lee <walt@tilera.com>
14360 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
14361 (insn_mnz_v8qi): ... this ...
14362 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
14364 (insn_v<n>mnz): Replaced by ...
14365 (insn_v1mnz): ... this ...
14366 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
14368 (insn_mz_<mode>): Replaced by ...
14369 (insn_mz_v8qi): ... this ...
14370 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
14372 (insn_v<n>mz): Replaced by ...
14373 (insn_v1mz): ... this ...
14374 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
14377 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
14379 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
14381 2013-03-26 Roland McGrath <mcgrathr@google.com>
14383 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
14384 than fprintf with a non-constant, non-format string.
14386 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
14388 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
14389 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
14390 operand 0 predicate.
14391 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
14392 attribute. Use general_x64nomem_operand as operand 1 predicate.
14393 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
14394 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
14395 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
14396 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
14397 (mov<mode>_insv_1): Remove expander. Merge insn with
14398 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
14399 Use general_x64nomem_operand as operand 1 predicate.
14400 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
14401 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
14402 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
14403 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
14404 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
14405 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
14406 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
14407 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
14408 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
14409 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
14410 (general_x64nomem_operand): Ditto.
14412 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14414 * config/rtems.opt: Add -pthread option.
14416 2013-03-26 Richard Biener <rguenther@suse.de>
14418 * alias.c (find_base_term): Avoid redundant and not used recursion.
14419 (base_alias_check): Get the initial base term from the caller.
14420 (true_dependence_1): Compute and pass base terms to base_alias_check.
14421 (write_dependence_p): Likewise.
14422 (may_alias_p): Likewise.
14424 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
14426 * config/aarch64/aarch64.c (aarch64_classify_address): Support
14427 PC-relative load in SI modes and above only.
14429 2013-03-26 Xinyu Qi <xyqi@marvell.com>
14431 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
14432 * config/arm/iwmmxt.md (WCGR0): Update.
14433 (WCGR1, WCGR2, WCGR3): Likewise.
14435 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
14437 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
14438 Use x64 and nox64 isa attributes.
14440 2013-03-26 Richard Biener <rguenther@suse.de>
14442 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
14443 alignment computations and rely on get_object_alignment_1
14444 for the !TYPE_P case.
14445 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
14447 2013-03-26 Walter Lee <walt@tilera.com>
14449 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
14450 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
14452 2013-03-25 Jeff Law <law@redhat.com>
14454 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
14455 check for INTEGRAL_TYPE_P that was missing due to checking in
14456 wrong version of prior patch.
14458 2013-03-25 Walter Lee <walt@tilera.com>
14460 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
14461 TILEGX_INSN_SHUFFLEBYTES1.
14462 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
14464 (tilegx_builtins): Ditto.
14465 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
14467 2013-03-25 Walter Lee <walt@tilera.com>
14469 * config/tilegx/tilegx.md (floatsisf2): New pattern.
14470 (floatunssisf2): New pattern.
14471 (floatsidf2): New pattern.
14472 (floatunssidf2): New pattern.
14474 2013-03-25 Walter Lee <walt@tilera.com>
14476 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
14477 tests for constraint J, K, N, P.
14479 2013-03-25 Walter Lee <walt@tilera.com>
14481 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
14482 Use indirect/pcrel encoding.
14483 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
14486 2013-03-25 Steve Ellcey <sellcey@mips.com>
14488 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
14489 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
14490 * config/mips/mips.c (mips_option_override): Set IMADD default.
14491 * config/mips/mips.h (PTF_AVOID_IMADD): New.
14492 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
14493 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
14494 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
14495 * doc/invoke.texi (-mimadd/-mno-imadd): New.
14497 2013-03-25 Jeff Law <law@redhat.com>
14499 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
14500 slightly to avoid creating and folding useless trees. Simplify
14501 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
14503 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
14505 * config/i386/i386.md (*zero_extendsidi2): Merge with
14506 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
14507 * config/i386/predicates.md (x86_64_zext_operand): Rename from
14508 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
14509 targets. Clarify comment.
14511 2013-03-25 Martin Jambor <mjambor@suse.cz>
14513 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
14514 pass-through jump functions differently.
14515 (ipa_read_jump_function): Likewise. Also use setter functions to set
14518 2013-03-25 Martin Jambor <mjambor@suse.cz>
14520 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
14521 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
14523 (ipa_get_indirect_edge_target): New function.
14524 (devirtualization_time_bonus): New parameter known_aggs, pass it to
14525 ipa_get_indirect_edge_target. Update all callers.
14526 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
14527 ipa_get_indirect_edge_target_1 instead of calling
14528 ipa_get_indirect_edge_target.
14529 (create_specialized_node): Pass aggvlas to
14530 ipcp_discover_new_direct_edges.
14532 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14534 * config/arm/arm.md (f_sels, f_seld): New types.
14535 (*cmov<mode>): New pattern.
14536 * config/arm/predicates.md (arm_vsel_comparison_operator): New
14539 2013-03-25 Kai Tietz <ktietz@redhat.com>
14541 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
14542 POSIX-printf for mingw-hosted builds.
14544 2013-03-25 Richard Biener <rguenther@suse.de>
14546 PR middle-end/56694
14547 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
14548 must-not-throw stmt location.
14550 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14552 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
14553 Emit load-acquire versions when acq is true.
14554 (arm_emit_store_exclusive): Add rel parameter.
14555 Emit store-release versions when rel is true.
14556 (arm_split_compare_and_swap): Use acquire-release instructions
14558 of barriers when appropriate.
14559 (arm_split_atomic_op): Likewise.
14560 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
14561 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
14562 (VUNSPEC_SLX): Likewise.
14563 (VUNSPEC_LDA): Likewise.
14564 (VUNSPEC_STL): Likewise.
14565 * config/arm/sync.md (atomic_load<mode>): New pattern.
14566 (atomic_store<mode>): Likewise.
14567 (arm_load_acquire_exclusive<mode>): Likewise.
14568 (arm_load_acquire_exclusivesi): Likewise.
14569 (arm_load_acquire_exclusivedi): Likewise.
14570 (arm_store_release_exclusive<mode>): Likewise.
14572 2013-03-25 Catherine Moore <clm@codesourcery.com>
14574 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
14575 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
14576 * config/mip/predicates.md (lwsp_swsp_operand,
14577 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
14578 sb16_operand, db4_operand, db7_operand, ib3_operand,
14579 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
14580 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
14581 andi16_operand): New predicates.
14582 * config/mips/mips.md (compression): New attribute.
14583 (enabled): New attribute.
14584 (length): Consider compression in computing length.
14585 (shift_compression): New code attribute.
14586 (*add<mode>3): New operands. Record compression.
14587 (sub<mode>3): Likewise.
14588 (one_cmpl<mode>2): Likewise.
14589 (*and<mode>3): Likewise.
14590 (*ior<mode>3): Likewise.
14591 (unnamed pattern for xor): Likewise.
14592 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
14593 (*<optab><mode>3): Likewise.
14594 (*mov<mode>_internal: Likewise.
14595 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
14596 (mips_unsigned_immediate_p): New.
14597 (umips_lwsp_swsp_address_p): New.
14598 (m16_based_address_p): New.
14599 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
14600 (mips_unsigned_immediate_p): New prototype.
14601 (lwsp_swsp_address_p): New prototype.
14602 (m16_based_address_p): New prototype.
14603 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14604 (mips_signed_immediate_p): New function.
14605 (m16_based_address_p): New function.
14606 (lwsp_swsp_address_p): New function.
14607 (mips_print_operand_punctuation): Recognize short delay slot insns
14608 for microMIPS.add<mode>3"
14610 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14613 * config/arm/iterators.md (v_cmp_result): New mode attribute.
14614 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14616 2013-03-25 Richard Biener <rguenther@suse.de>
14618 PR tree-optimization/56689
14619 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14622 2013-03-25 Richard Biener <rguenther@suse.de>
14624 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14626 (memory_references): Likewise.
14627 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14628 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14629 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14630 (gather_mem_refs_in_loops): Fold into ...
14631 (analyze_memory_references): ... this. Move initialization
14632 to tree_ssa_lim_initialize.
14633 (fill_always_executed_in): Rename to ...
14634 (fill_always_executed_in_1): ... this.
14635 (fill_always_executed_in): Move contains_call computation to
14636 this new function from ...
14637 (tree_ssa_lim_initialize): ... here.
14638 (tree_ssa_lim): Call fill_always_executed_in.
14640 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
14642 * postreload.c (reload_combine): Fix code detecting returns.
14644 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
14646 * function.c (emit_use_return_register_into_block): On cc0 targets,
14647 do not emit the sequence between cc0 setter and user.
14649 2013-03-25 Kai Tietz <ktietz@redhat.com>
14651 * config/i386/predicates.md (local_symbolic_operand): Interpret
14652 dll-imported symbols as none-local.
14654 2013-03-25 Richard Biener <rguenther@suse.de>
14656 * tree-ssa-loop-im.c (struct depend): Remove.
14657 (struct lim_aux_data): Make depends a vec of gimples.
14658 (free_lim_aux_data): Adjust.
14659 (add_dependency): Likewise.
14660 (set_level): Likewise.
14662 2013-03-25 Richard Biener <rguenther@suse.de>
14664 PR middle-end/56434
14665 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14666 the pointer returned by calls with ECF_MALLOC set.
14668 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
14670 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14672 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
14674 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14675 using MMXMODE mode iterator.
14676 (*move<mode>_internal): Merge with *movv2sf_internal and
14677 *movv2sf_internal_rex64 using MMXMODE mode iterator.
14679 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
14681 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14682 (record_last_mem_set_info): Likewise.
14684 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14685 of XNEWVEC followed by memset.
14686 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14688 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
14690 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14691 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14692 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14693 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14694 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14695 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14696 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14697 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14698 BARRIER_P instead of GET_CODE.
14700 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14702 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14703 inaccuracy in the probing code.
14705 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14706 (ctrapdi4): Likewise.
14708 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14710 * calls.c (expand_call): Add missing guard to code handling return
14711 of non-BLKmode structures in MSB.
14712 * function.c (expand_function_end): Likewise.
14714 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
14716 * combine.c (try_combine): Adjust comment. Do not add the set of
14717 insn #0 if the destination indirectly is set or dies in insn #2.
14718 Tidy up code to distribute a new note.
14720 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
14722 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14723 also for alternatives 16 and 17.
14725 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
14727 * config/i386/sse.md (*mov<mode>_internal): Merge with
14728 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
14729 Emit insn template depending on type attribute. Use
14730 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14731 movd instead of movq mnemonic for interunit moves. Rewrite mode
14732 attribute calculation. Remove unit attribute calculation.
14733 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14734 Set prefix_data16 attribute for DImode ssemov types.
14735 Use Ym instead of y for SSE-MMX conversion alternatives.
14736 Reorder operand constraints.
14738 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
14740 * df.h (df_insn_delete): Adjust prototype.
14741 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14742 and let it decide whether mark the basic block dirty.
14743 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14744 * df-scan.c (df_insn_info_delete): New helper function, split
14745 off from df_insn_delete.
14746 (df_scan_free_bb_info): Use it.
14747 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14749 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
14750 that the insn is actually an insn and it has a non-NULL basic block.
14751 Do not mark basic block dirty if only deleting a DEBUG_INSN.
14753 2013-03-22 Richard Biener <rguenther@suse.de>
14755 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14757 (mem_ref_alloc): Do not allocate them.
14758 (refs_independent_p): Do not query or maintain a cache.
14760 2013-03-22 Richard Biener <rguenther@suse.de>
14762 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14763 (gather_mem_refs_in_loops): Do not compute it.
14764 (analyze_memory_references): Do not allocate it.
14765 (tree_ssa_lim_finalize): Do not free it.
14766 (for_all_locs_in_loop): Do not query all_refs_in_loop.
14768 2013-03-22 Richard Biener <rguenther@suse.de>
14770 * is-a.h (as_a): Use gcc_checking_assert.
14772 2013-03-22 Ian Bolton <ian.bolton@arm.com>
14774 * config/aarch64/aarch64.c (aarch64_print_operand): New
14775 format specifier for printing a constant in hex.
14776 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14777 format specifier for printing second operand.
14779 2013-03-22 Richard Biener <rguenther@suse.de>
14781 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14783 (gather_mem_refs_in_loops): Perform store accumulation here.
14784 (create_vop_ref_mapping_loop): Remove.
14785 (create_vop_ref_mapping): Likewise.
14786 (analyze_memory_references): Initialize refs_stored_in_loop.
14787 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14788 (record_indep_loop): Remove.
14789 (record_dep_loop): New function.
14790 (ref_indep_loop_p_1): Adjust to only walk over references
14791 in the loop, not its subloops.
14792 (ref_indep_loop_p): Rename to ...
14793 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14794 maintaining a more fine-grained cache.
14795 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14796 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14798 2013-03-22 Richard Biener <rguenther@suse.de>
14800 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14801 (struct mem_ref): Make accesses_in_loop a vec of a vec of
14802 aggregate mem_ref_loc.
14803 (free_mem_ref_locs): Inline into ...
14804 (memref_free): ... this and adjust.
14805 (mem_ref_alloc): Adjust.
14806 (mem_ref_locs_alloc): Remove.
14807 (record_mem_ref_loc): Adjust.
14808 (get_all_locs_in_loop): Rewrite into ...
14809 (for_all_locs_in_loop): ... this iterator.
14810 (rewrite_mem_ref_loc): New functor.
14811 (rewrite_mem_refs): Use for_all_locs_in_loop.
14812 (sm_set_flag_if_changed): New functor.
14813 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14814 (ref_always_accessed): New functor.
14815 (ref_always_accessed_p): Use for_all_locs_in_loop.
14817 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14819 * tree-pass.h (PROP_gimple_lvec): New.
14820 * passes.c (dump_properties): Handle PROP_gimple_lvec.
14821 (init_optimization_passes): Move pass_lower_vector.
14822 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14824 (pass_lower_vector): Provide PROP_gimple_lvec.
14825 (pass_lower_vector_ssa): Likewise.
14826 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14828 2013-03-21 Mark Wielaard <mjw@redhat.com>
14830 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14832 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
14834 * config/i386/i386.md (*movdi_internal): Disparage slightly
14835 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
14836 conversion alternatives.
14838 2013-03-21 Jakub Jelinek <jakub@redhat.com>
14840 PR middle-end/48087
14841 * diagnostic.def (DK_WERROR): New kind.
14842 * diagnostic.h (werrorcount): Define.
14843 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14844 promoted to DK_ERROR, increment DK_WERROR counter instead of
14846 * toplev.c (toplev_main): Call print_ignored_options even if
14847 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
14848 even if just werrorcount is non-zero.
14851 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14853 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14854 (string_cst_pool_decl): New function.
14855 (optimize_one_addr_into_implicit_ptr): New function.
14856 (resolve_addr_in_expr): Optimize DWARF location expression
14857 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14858 which doesn't live in memory, but has DW_AT_location or
14859 DW_AT_const_value, or refers to a string literal, into
14860 DW_OP_GNU_implicit_pointer.
14861 (optimize_location_into_implicit_ptr): New function.
14862 (resolve_addr): If removing DW_AT_location of a variable because
14863 it was DW_OP_addr of address of the variable, but the variable doesn't
14864 live in memory, try to emit const value attribute for the initializer.
14866 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14868 * tree.h (VECTOR_TYPE_P): New macro.
14869 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14870 TYPE_MODE): Use it.
14871 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14872 VEC_COND_EXPR cannot be lvalues.
14873 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14875 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
14877 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14878 Restrict the transformation to equal modes.
14880 2013-03-21 Richard Biener <rguenther@suse.de>
14882 PR tree-optimization/39326
14883 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14884 (MEM_ANALYZABLE): Adjust.
14885 (record_mem_ref_loc): Move bitmap ops ...
14886 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
14887 unanalyzable refs, do not record locations for it.
14888 (analyze_memory_references): Allocate ref zero as shared
14890 (refs_independent_p): Do not test for unanalyzed mems here.
14891 (ref_indep_loop_p_1): Special-case disambiguation against
14892 the unanalyzed ref.
14893 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14895 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
14897 * config/arm/arm-protos.h (tune_params): Add
14898 prefer_neon_for_64bits field.
14899 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14900 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14901 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14902 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14903 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14904 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14905 (arm_option_override): Handle -mneon-for-64bits new option.
14906 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14907 (prefer_neon_for_64bits): Declare new variable.
14908 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14909 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14910 (arch_enabled): Handle new arch types. Remove support for onlya8
14912 (one_cmpldi2): Use new arch names.
14913 (zero_extend<mode>di2, extend<mode>di2): Ditto.
14914 * config/arm/arm.opt (mneon-for-64bits): Add option.
14915 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14916 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14917 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14919 * doc/invoke.texi (-mneon-for-64bits): Document.
14921 2013-03-21 Richard Biener <rguenther@suse.de>
14923 PR tree-optimization/39326
14924 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14925 (sort_bbs_in_loop_postorder_cmp): New function.
14926 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14928 2013-03-21 Richard Biener <rguenther@suse.de>
14930 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14931 (vect_insert_into_interleaving_chain): Likewise.
14932 (vect_drs_dependent_in_basic_block): Inline ...
14933 (vect_slp_analyze_data_ref_dependence): ... here. New function,
14935 (vect_analyze_data_ref_dependence): ... here. Simplify.
14936 (vect_check_interleaving): Simplify.
14937 (vect_analyze_data_ref_dependences): Likewise. Split out ...
14938 (vect_slp_analyze_data_ref_dependences): ... this new function.
14939 (dr_group_sort_cmp): New function.
14940 (vect_analyze_data_ref_accesses): Compute data-reference groups
14941 here instead of in vect_analyze_data_ref_dependence. Use
14942 a more efficient algorithm.
14943 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14944 vect_slp_analyze_data_ref_dependences. Call
14945 vect_analyze_data_ref_accesses earlier.
14946 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14947 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14948 (vect_slp_analyze_data_ref_dependences): New prototype.
14950 2013-03-21 Richard Biener <rguenther@suse.de>
14952 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14953 ref is stored in the loop.
14954 (find_refs_for_sm): Walk only over all stores.
14955 (store_motion_loop): Allocate from lim_bitmap_obstack.
14956 (store_motion): Likewise.
14958 2013-03-21 Richard Biener <rguenther@suse.de>
14960 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14961 Update virtual SSA form.
14963 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14965 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14966 * configure: Regenerate.
14967 * config.in: Regenerate.
14968 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14969 if !HAVE_LD_EH_FRAME_CIEV3.
14971 2013-03-21 Richard Biener <rguenther@suse.de>
14973 * tree-cfg.c (verify_expr_no_block): New function.
14974 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14975 nor DECL_VALUE_EXPR have locations with associated blocks.
14976 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14977 (clear_unused_block_pointer): Remove code dealing with
14978 blocks in DECL_DEBUG_EXPR locations.
14980 2013-03-21 Richard Biener <rguenther@suse.de>
14982 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14983 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
14984 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14985 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14986 instead of DECL_DEBUG_EXPR_IS_FROM.
14987 * gimplify.c (gimplify_modify_expr): Likewise.
14988 * tree-cfg.c (verify_expr_location_1): Likewise.
14989 * tree-complex.c (create_one_component_var): Likewise.
14990 * tree-sra.c (create_access_replacement): Likewise.
14991 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14992 (clear_unused_block_pointer): Likewise.
14993 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14994 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14995 * var-tracking.c (var_debug_decl): Likewise.
14996 (track_expr_p): Likewise.
14997 * tree-inline.c (add_local_variables): Likewise. Set
14998 DECL_HAS_DEBUG_EXPR_P after copying it.
14999 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
15000 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
15002 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
15005 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
15006 * configure: Regenerate.
15007 * config.in: Regenerate.
15008 * config/i386/i386.md (*movdf_internal): Use
15009 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
15010 movd instead of movq mnemonic for interunit moves.
15011 (*movdi_internal): Ditto.
15013 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
15015 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
15016 (abd<mode>_3): New pattern.
15017 (aba<mode>_3): New pattern.
15018 (fabd<mode>_3): New pattern.
15020 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
15022 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
15023 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
15024 occurrence of REGISTER_PREFIX as its empty string.
15026 2013-03-20 Jeff Law <law@redhat.com>
15028 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
15029 addititional equivalences for equality comparisons between an SSA_NAME
15030 and a constant where the SSA_NAME was set from a widening conversion.
15032 2013-03-20 Walter Lee <walt@tilera.com>
15034 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
15036 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
15038 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
15039 depending on type attribute.
15040 (*movti_internal): Ditto.
15041 (*movtf_internal): Ditto.
15042 (*movxf_internal): Ditto.
15043 (*movdf_internal): Ditto.
15044 (*movsf_internal): Ditto.
15046 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
15048 * config/i386/i386.md (*movti_internal): Set prefix attribute to
15049 maybe_vex for sselog1 and ssemov types.
15050 (*movdi_internal): Reorder operand constraints.
15051 (*movsi_internal): Ditto. Set prefix attribute to
15052 maybe_vex for sselog1 and ssemov types.
15053 (*movtf_internal): Set prefix attribute to maybe_vex
15054 for sselog1 and ssemov types.
15055 (*movdf_internal): Ditto. Set prefix_data16 attribute for
15056 DImode ssemov types. Reorder operand constraints.
15057 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
15058 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
15059 attribute for SImode ssemov types. Reorder operand constraints.
15061 2013-03-20 Martin Jambor <mjambor@suse.cz>
15063 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
15064 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
15066 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
15068 * config/rs6000/predicates.md (indexed_address, update_address_mem
15069 update_indexed_address_mem): New predicates.
15070 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
15071 attribute for load/store instructions.
15072 * config/rs6000/dfp.md (movsd_store): Likewise.
15073 (movsd_load): Likewise.
15074 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
15075 (unnamed HI->DI extend define_insn): Likewise.
15076 (unnamed SI->DI extend define_insn): Likewise.
15077 (unnamed QI->SI extend define_insn): Likewise.
15078 (unnamed QI->HI extend define_insn): Likewise.
15079 (unnamed HI->SI extend define_insn): Likewise.
15080 (unnamed HI->SI extend define_insn): Likewise.
15081 (extendsfdf2_fpr): Likewise.
15082 (movsi_internal1): Likewise.
15083 (movsi_internal1_single): Likewise.
15084 (movhi_internal): Likewise.
15085 (movqi_internal): Likewise.
15086 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
15087 attribute for load/store instructions.
15088 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
15090 (mov<mode>_softfloat): Likewise.
15091 (mov<mode>_hardfloat32): Likewise.
15092 (mov<mode>_hardfloat64): Likewise.
15093 (mov<mode>_softfloat64): Likewise.
15094 (movdi_internal32): Likewise.
15095 (movdi_internal64): Likewise.
15096 (probe_stack_<mode>): Likewise.
15098 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15100 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
15101 floating point, and decimal floating point to reload iterator.
15103 * config/rs6000/constraints.md (wl constraint): New constraints to
15104 return FLOAT_REGS if certain options are used to reduce the number
15105 of separate patterns that exist in the file.
15106 (wx constraint): Likewise.
15107 (wz constraint): Likewise.
15109 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15110 -mdebug=reg, print wg, wl, wx, and wz constraints.
15111 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15112 Initialize the reload functions for 64-bit binary/decimal floating
15114 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15115 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15116 create the buffer on the stack to overcome not having a 32-bit
15118 (rs6000_emit_move): Likewise.
15119 (rs6000_secondary_memory_needed_rtx): Likewise.
15120 (rs6000_alloc_sdmode_stack_slot): Likewise.
15121 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15122 via xxlxor, just like DFmode 0.0.
15124 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
15125 define as 1 if we are running on a power7 or newer.
15126 (enum r6000_reg_class_enum): Add new constraints.
15128 * config/rs6000/dfp.md (movsd): Delete, combine with binary
15129 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
15130 with other moves by using conditional constraits (wg). Use LFIWZX
15131 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
15132 (movsd splitter): Likewise.
15133 (movsd_hardfloat): Likewise.
15134 (movsd_softfloat): Likewise.
15136 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
15137 binary and decimal floating point moves.
15138 (fmove_ok): New attributes to combine binary and decimal floating
15139 point moves, and to combine power6x (mfpgpr) moves along normal
15141 (real_value_to_target): Likewise.
15142 (f32_lr): Likewise.
15143 (f32_lm): Likewise.
15144 (f32_li): Likewise.
15145 (f32_sr): Likewise.
15146 (f32_sm): Likewise.
15147 (f32_si): Likewise.
15148 (movsf): Combine binary and decimal floating point moves. Combine
15149 power6x (mfpgpr) moves with other moves by using conditional
15150 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
15151 (mov<mode> for SFmode/SDmode); Likewise.
15152 (SFmode/SDmode splitters): Likewise.
15153 (movsf_hardfloat): Likewise.
15154 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
15155 (movsf_softfloat): Likewise.
15156 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
15158 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
15159 wx and wz constraints.
15161 * config/rs6000/constraints.md (wg constraint): New constraint to
15162 return FLOAT_REGS if -mmfpgpr (power6x) was used.
15164 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
15167 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
15168 -mdebug=reg, print wg, wl, wx, and wz constraints.
15169 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
15170 Initialize the reload functions for 64-bit binary/decimal floating
15172 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
15173 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
15174 create the buffer on the stack to overcome not having a 32-bit
15176 (rs6000_emit_move): Likewise.
15177 (rs6000_secondary_memory_needed_rtx): Likewise.
15178 (rs6000_alloc_sdmode_stack_slot): Likewise.
15179 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
15180 via xxlxor, just like DFmode 0.0.
15182 * config/rs6000/dfp.md (movdd): Delete, combine with binary
15183 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
15184 with other moves by using conditional constraits (wg). Use LFIWZX
15185 and STFIWX for loading SDmode on power7.
15186 (movdd splitters): Likewise.
15187 (movdd_hardfloat32): Likewise.
15188 (movdd_softfloat32): Likewise.
15189 (movdd_hardfloat64_mfpgpr): Likewise.
15190 (movdd_hardfloat64): Likewise.
15191 (movdd_softfloat64): Likewise.
15193 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
15194 64-bit binary and decimal floating point moves.
15195 (FMOVE64X): Likewise.
15196 (movdf): Combine 64-bit binary and decimal floating point moves.
15197 Combine power6x (mfpgpr) moves with other moves by using
15198 conditional constraits (wg).
15199 (mov<mode> for DFmode/DDmode): Likewise.
15200 (DFmode/DDmode splitters): Likewise.
15201 (movdf_hardfloat32): Likewise.
15202 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
15203 (movdf_softfloat32): Likewise.
15204 (movdf_hardfloat64_mfpgpr): Likewise.
15205 (movdf_hardfloat64): Likewise.
15206 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
15207 (movdf_softfloat64): Likewise.
15208 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
15209 (reload_<mode>_load): Move to later in the file so they aren't in
15210 the middle of the floating point move insns.
15211 (reload_<mode>_store): Likewise.
15213 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
15216 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
15217 constraint if -mdebug=reg.
15218 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
15219 Enable using dd reload support if needed.
15221 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
15222 binary and decimal floating point moves in rs6000.md.
15223 (movtd_internal): Likewise.
15225 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
15226 decimal floating point moves.
15228 (movtf_internal): Likewise.
15229 (mov<mode>_internal, TDmode/TFmode): Likewise.
15230 (movtf_softfloat): Likewise.
15231 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
15233 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
15234 movdi_internal64, using wg constraint for move direct operations.
15235 (movdi_internal64): Likewise.
15237 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
15238 MODES_TIEABLE_P for selected modes. Print the numerical value of
15239 the various virtual registers. Use GPR/FPR first/last values,
15240 instead of hard coding the register numbers. Print which modes
15241 have reload functions registered.
15242 (rs6000_option_override_internal): If -mdebug=reg, trace the options
15243 settings before/after setting cpu, target and subtarget settings.
15244 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
15245 and for secondary reload failures in rs6000_secondary_reload_inner.
15246 (rs6000_secondary_reload_fail): Likewise.
15247 (rs6000_secondary_reload_inner): Likewise.
15249 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
15250 macros for first/last GPR and FPR registers.
15251 (LAST_GPR_REGNO): Likewise.
15252 (FIRST_FPR_REGNO): Likewise.
15253 (LAST_FPR_REGNO): Likewise.
15255 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
15256 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
15257 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
15258 (vcond<mode><mode>): Likewise.
15259 (vcondu<mode><mode>): Likewise.
15260 (vector_gtu<mode>): Likewise.
15261 (vector_gte<mode>): Likewise.
15262 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
15263 to prevent the compiler from converting DImode operations to TImode.
15264 (ior<mode>3): Likewise.
15265 (and<mode>3): Likewise.
15266 (one_cmpl<mode>2): Likewise.
15267 (nor<mode>3): Likewise.
15268 (andc<mode>3): Likewise.
15270 * config/rs6000/constraints.md (wt constraint): New constraint
15271 that returns VSX_REGS if TImode is allowed in VSX registers.
15273 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
15274 constant under VSX.
15276 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
15277 similar to TImode, but it is restricted to being in the GPRs.
15279 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
15280 TImode to occupy a single VSX register.
15282 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
15283 -mvsx-timode for power7/power8.
15284 (power7 cpu): Likewise.
15285 (power8 cpu): Likewise.
15287 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
15288 sure that TFmode/TDmode take up two registers if they are ever
15289 allowed in the upper VSX registers.
15290 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
15292 (rs6000_init_hard_regno_mode_ok): Likewise.
15293 (rs6000_debug_reg_global): Add debugging for PTImode and wt
15294 constraint. Print if LRA is turned on.
15295 (rs6000_option_override_internal): Give an error if -mvsx-timode
15296 and VSX is not enabled.
15297 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
15298 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
15299 to reg+offset addressing. Use PTImode when checking offset
15300 addresses for validity.
15301 (reg_offset_addressing_ok_p): Likewise.
15302 (rs6000_legitimate_offset_address_p): Likewise.
15303 (rs6000_legitimize_address): Likewise.
15304 (rs6000_legitimize_reload_address): Likewise.
15305 (rs6000_legitimate_address_p): Likewise.
15306 (rs6000_eliminate_indexed_memrefs): Likewise.
15307 (rs6000_emit_move): Likewise.
15308 (rs6000_secondary_reload): Likewise.
15309 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
15310 reloads to fpr registers to continue to use reg+offset addressing,
15311 but 64-bit reloads to altivec registers need reg+reg addressing.
15312 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
15313 it. Treat LO_SUM like a PLUS operation.
15314 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
15315 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
15316 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
15317 registers to share a register with a smaller sized type, since VSX
15318 puts scalars in the upper 64-bits.
15319 (print_operand): Add support for PTImode.
15320 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
15321 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
15322 registers, but don't have arithmetic support.
15323 (rs6000_memory_move_cost): Add test for VSX.
15324 (rs6000_opt_masks): Add -mvsx-timode.
15326 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
15329 (VSr): Use wt constraint for TImode.
15330 (VSv): Drop TImode support.
15331 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
15332 (vsx_movti_64bit): Likewise.
15333 (vsx_movti_32bit): Likewise.
15334 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
15335 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
15336 one '?' on the appropriate output constraint. Do not allow TImode
15337 logical operations on 32-bit systems.
15338 (vsx_ior<mode>3): Likewise.
15339 (vsx_xor<mode>3): Likewise.
15340 (vsx_one_cmpl<mode>2): Likewise.
15341 (vsx_nor<mode>3): Likewise.
15342 (vsx_andc<mode>3): Likewise.
15343 (vsx_concat_<mode>): Likewise.
15344 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
15346 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
15347 OPTION_MASK_VSX_TIMODE.
15348 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
15349 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
15351 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
15352 (TI2 iterator): New iterator for TImode, PTImode.
15353 (wd mode attribute): Add values for vector types.
15354 (movti_string): Replace TI move operations with operations for TImode
15355 and PTImode. Add support for TImode being allowed in VSX registers.
15356 (mov<mode>_string, TImode/PTImode): Likewise.
15357 (movti_ppc64): Likewise.
15358 (mov<mode>_ppc64, TImode/PTImode): Likewise.
15359 (TI mode splitters): Likewise.
15361 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
15364 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
15366 PR tree-optimization/56355
15367 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
15368 Also handle integers with undefined overflow.
15370 2013-03-20 Catherine Moore <clm@codesourcery.com>
15371 Maciej W. Rozycki <macro@codesourcery.com>
15372 Tom de Vries <tom@codesourcery.com>
15373 Nathan Sidwell <nathan@codesourcery.com>
15374 Iain Sandoe <iain@codesourcery.com>
15375 Nathan Froyd <froydnj@codesourcery.com>
15376 Chao-ying Fu <fu@mips.com>
15378 * doc/extend.texi: (micromips, nomicromips, nocompression):
15379 Document new function attributes.
15380 * doc/invoke.texi (minterlink-compressed, mmicromips,
15381 m14k, m14ke, m14kec): Document new options.
15382 (minterlink-mips16): Update documentation.
15383 * doc/md.texi (ZC, ZD): Document new constraints.
15384 * configure.ac (gcc_cv_as_micromips): Check if linker
15385 supports the .set micromips directive.
15386 * configure: Regenerate.
15387 * config.in: Regenerate.
15388 * config/mips/mips-tables.opt: Regenerate.
15389 * config/mips/micromips.md: New file.
15390 * constraints.md (ZC, ZD): New constraints.
15391 * config/mips/predicates.md (movep_src_register): New predicate.
15392 (movep_src_operand): New predicate.
15393 (non_volatile_mem_operand): New predicate.
15394 * config/mips/mips.md (multimem): New type.
15395 (length): Differentiate between 17-bit and 18-bit branch offsets.
15396 (MOVEP1, MOVEP2): New mode iterator.
15397 (mov_<load>l): Use ZC constraint.
15398 (mov_<load>r): Likewise.
15399 (mov_<store>l): Likewise.
15400 (mov_<store>r): Likewise.
15401 (*branch_equality<mode>_inverted): Add microMIPS support.
15402 (*branch_equality<mode>): Likewise.
15403 (*jump_absolute): Likewise.
15404 (indirect_jump_<mode>): Likewise.
15405 (tablejump_<mode>): Likewise.
15406 (<optab>_internal): Likewise.
15407 (sibcall_internal): Likewise.
15408 (sibcall_value_internal): Likewise.
15409 (prefetch): Use constraint ZD.
15410 * config/mips/mips.opt (minterlink-compressed): New option.
15411 (minterlink-mips16): Now an alias for minterlink-compressed.
15412 (mmicromips): New option.
15413 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
15414 (compare_and_swap_12): Likewise.
15415 (sync_add<mode>): Likewise.
15416 (sync_<optab>_12): Likewise.
15417 (sync_old_<optab>_12): Likewise.
15418 (sync_new_<optab>_12): Likewise.
15419 (sync_nand_12): Likewise.
15420 (sync_old_nand_12): Likewise.
15421 (sync_new_nand_12): Likewise.
15422 (sync_sub<mode>): Likewise.
15423 (sync_old_add<mode>): Likewise.
15424 (sync_old_sub<mode>): Likewise.
15425 (sync_new_add<mode>): Likewise.
15426 (sync_new_sub<mode>): Likewise.
15427 (sync_<optab><mode>): Likewise.
15428 (sync_old_<optab><mode>): Likewise.
15429 (sync_new_<optab><mode>): Likewise.
15430 (sync_nand<mode>): Likewise.
15431 (sync_old_nand<mode>): Likewise.
15432 (sync_new_nand<mode>): Likewise.
15433 (sync_lock_test_and_set<mode>): Likewise.
15434 (test_and_set_12): Likewise.
15435 (atomic_compare_and_swap<mode>): Likewise.
15436 (atomic_exchange<mode>_llsc): Likewise.
15437 (atomic_fetch_add<mode>_llsc): Likewise.
15438 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
15439 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
15440 (umips_save_restore_pattern_p): Likewise.
15441 (umips_load_store_pair_p): Likewise.
15442 (umips_output_load_store_pair): Likewise.
15443 (umips_movep_target_p): Likewise.
15444 (umips_12bit_offset_address_p): Likewise.
15445 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
15446 (mips_base_mips16): Rename this...
15447 (mips_base_compression_flags): ...to this. Update all uses.
15448 (mips_attribute_table): Add micromips, nomicromips and nocompression.
15449 (mips_mips16_decl_p): Delete.
15450 (mips_nomips16_decl_p): Delete.
15451 (mips_get_compress_on_flags): New function.
15452 (mips_get_compress_off_flags): New function.
15453 (mips_get_compress_mode): New function.
15454 (mips_get_compress_on_name): New function.
15455 (mips_get_compress_off_name): New function.
15456 (mips_insert_attributes): Support multiple compression types.
15457 (mips_merge_decl_attributes): Likewise.
15458 (umips_12bit_offset_address_p): New function.
15459 (mips_start_function_definition): Emit .set micromips directive.
15460 (mips_call_may_need_jalx_p): New function.
15461 (mips_function_ok_for_sibcall): Add microMIPS support.
15462 (mips_print_operand_punctuation): Support short delay slots and
15464 (umips_swm_mask, umips_swm_encoding): New.
15465 (umips_build_save_restore): New function.
15466 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15467 (was_mips16_p): Remove.
15468 (old_compression_mode): New.
15469 (mips_set_compression_mode): New function.
15470 (mips_set_current_function): Add microMIPS support.
15471 (mips_option_override): Likewise.
15472 (umips_save_restore_pattern_p): New function.
15473 (umips_output_save_restore): New function.
15474 (umips_load_store_pair_p_1): New function.
15475 (umips_load_store_pair_p): New function.
15476 (umips_output_load_store_pair_1): New function.
15477 (umips_output_load_store_pair): New function.
15478 (umips_movep_target_p) New function.
15479 (mips_prepare_pch_save): Add microMIPS support.
15480 * config/mips/mips.h (TARGET_COMPRESSION): New.
15481 (TARGET_CPU_CPP_BUILTINS): Update macro
15482 to use new compression flags and to support microMIPS.
15483 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15484 (MIPS_ARCH_FLOAT_SPEC): Likewise.
15485 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15486 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15487 (ASM_SPEC): Support mmicromips and mno-micromips.
15488 (M16STORE_REG_P): New macro.
15489 (MIPS_CALL): Support TARGET_MICROMIPS.
15490 (MICROMIPS_J): New macro.
15491 (mips_base_mips16): Rename this...
15492 (mips_base_compression_flags): ...to this.
15493 (UMIPS_12BIT_OFFSET_P): New macro.
15494 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15495 (MULTILIB_DIRNAMES): Likewise.
15496 2013-03-20 Richard Biener <rguenther@suse.de>
15498 PR tree-optimization/56661
15499 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
15500 the result does not have to be distinct.
15502 2013-03-20 Richard Biener <rguenther@suse.de>
15504 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
15507 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15508 Steven Bosscher <steven@gcc.gnu.org>
15510 PR rtl-optimization/56605
15511 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
15513 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
15516 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
15517 that require movd instead of movq.
15519 2013-03-20 Richard Biener <rguenther@suse.de>
15521 * tree-ssa-structalias.c (struct variable_info): Add pointer
15522 to the first field of an aggregate with sub-vars. Make
15523 this and the pointer to the next subfield its ID.
15524 (vi_next): New function.
15525 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
15526 storedanything_id, integer_id): Increment by one.
15527 (new_var_info, get_call_vi, lookup_call_clobber_vi,
15528 get_call_clobber_vi): Adjust.
15529 (solution_set_expand): Simplify and speedup.
15530 (solution_set_add): Inline into ...
15531 (set_union_with_increment): ... this. Adjust accordingly.
15532 (do_sd_constraint): Likewise.
15533 (do_ds_constraint): Likewise.
15534 (do_complex_constraint): Simplify.
15535 (build_pred_graph): Adjust.
15536 (solve_graph): Likewise. Simplify and speedup.
15537 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
15538 get_constraint_for_component_ref, get_constraint_for_1,
15539 first_vi_for_offset, first_or_preceding_vi_for_offset,
15540 create_function_info_for, create_variable_info_for_1,
15541 create_variable_info_for, intra_create_variable_infos): Adjust.
15542 (init_base_vars): Push NULL for ID zero.
15543 (compute_points_to_sets): Adjust.
15545 2013-03-20 Richard Biener <rguenther@suse.de>
15547 * cfgloop.c (verify_loop_structure): Streamline and avoid
15548 ICEing on corrupt loop tree.
15549 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
15552 2013-03-20 Richard Biener <rguenther@suse.de>
15554 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
15555 check whether an SSA update is needed.
15557 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
15559 * config/mips/constraints.md (T): Rename to...
15563 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
15564 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
15566 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15568 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
15569 (*subsi3_carryin_uxtw): Likewise.
15571 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15573 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
15574 (*rorsi3_insn_uxtw): Likewise.
15576 2013-03-19 Ian Bolton <ian.bolton@arm.com>
15578 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
15579 (*extrsi5_insn_uxtw): Likewise.
15581 2013-03-19 Richard Biener <rguenther@suse.de>
15583 PR tree-optimization/56273
15584 * passes.c (init_optimization_passes): Move second VRP after DOM.
15586 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
15588 * config/i386/i386.md (*movti_internal): Merge from
15589 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
15590 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
15591 nox64 isa attributes.
15593 2013-03-18 Richard Biener <rguenther@suse.de>
15595 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
15597 (merge_node_constraints): Likewise.
15598 (build_succ_graph): Likewise.
15599 (valid_graph_edge): Inline into single caller.
15600 (unify_nodes): Likewise. Use bitmap_set_bit return value
15602 (scc_visit): Fix formatting and variable use.
15603 (do_sd_constraint): Use gcc_checking_assert.
15604 (do_ds_constraint): Likewise.
15605 (do_complex_constraint): Likewise.
15606 (condense_visit): Likewise. Cleanup.
15607 (dump_pred_graph): New function.
15608 (perform_var_substitution): Dump the pred-graph before
15609 variable substitution.
15610 (find_equivalent_node): Use gcc_checking_assert.
15611 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15613 2013-03-18 Richard Biener <rguenther@suse.de>
15615 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15616 Remove cond_expr_stmt_list argument and do not gimplify the
15618 (vect_loop_versioning): Adjust.
15619 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15620 Cleanup to use less temporaries.
15621 (vect_create_data_ref_ptr): Cleanup.
15623 2013-03-18 Jakub Jelinek <jakub@redhat.com>
15625 PR tree-optimization/56635
15626 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15627 require types_compatible_p types.
15629 2013-03-18 Nick Clifton <nickc@redhat.com>
15631 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15632 spurious backslash.
15634 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15635 Add missing line to comment describing function.
15637 2013-03-18 Richard Biener <rguenther@suse.de>
15639 PR tree-optimization/56210
15640 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15641 Handle string / character search functions.
15642 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15644 2013-03-18 Richard Biener <rguenther@suse.de>
15646 PR middle-end/56483
15647 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15648 and implement properly.
15649 * gimple.h (gimple_cond_single_var_p): Remove.
15651 2013-03-18 Richard Biener <rguenther@suse.de>
15653 * tree-data-ref.h (find_data_references_in_loop): Declare.
15654 * tree-data-ref.c (get_references_in_stmt): Use a stack
15655 vector pre-allocated in the callers.
15656 (find_data_references_in_stmt): Adjust.
15657 (graphite_find_data_references_in_stmt): Likewise.
15658 (create_rdg_vertices): Likewise.
15659 (find_data_references_in_loop): Export.
15660 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15661 Compute dependences here...
15662 (vect_analyze_data_refs): ...not here. When we encounter
15663 a non-vectorizable data reference in basic-block vectorization
15664 truncate the data reference vector. Do not bother to
15665 fixup data-dependence information for gather loads.
15666 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15667 of data references, as reported.
15669 2013-03-18 Richard Biener <rguenther@suse.de>
15671 PR tree-optimization/3713
15672 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
15673 has_constants and expr.
15674 (stmt_has_constants): Properly valueize SSA names when deciding
15675 whether the stmt has constants.
15677 2013-03-18 Richard Biener <rguenther@suse.de>
15679 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15680 whole function when there is nothing to do.
15681 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15682 * tree-vectorizer.c (vectorize_loops): Update virtual and
15683 loop-closed SSA once.
15684 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15686 2013-03-18 Richard Biener <rguenther@suse.de>
15688 PR middle-end/56113
15689 * domwalk.c (bb_postorder): New global static.
15690 (cmp_bb_postorder): New function.
15691 (walk_dominator_tree): Replace scheme imposing an order for
15692 visiting dominator sons by one sorting them at the time they
15693 are pushed on the stack.
15695 2013-03-18 Richard Biener <rguenther@suse.de>
15697 PR tree-optimization/39326
15698 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15699 (struct mem_ref): Replace mem member with ao_ref typed member.
15700 (MEM_ANALYZABLE): Adjust.
15701 (memref_eq): Likewise.
15702 (mem_ref_alloc): Likewise.
15703 (gather_mem_refs_stmt): Likewise.
15704 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15705 (execute_sm_if_changed_flag_set): Adjust.
15706 (execute_sm): Likewise.
15707 (ref_always_accessed_p): Likewise.
15708 (refs_independent_p): Likewise.
15709 (can_sm_ref_p): Likewise.
15711 2013-03-18 Jakub Jelinek <jakub@redhat.com>
15714 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15715 return 1 even for !unsignedp.
15717 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15719 * config/i386/i386.md (isa): Add x64 and nox64.
15720 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15721 (*pushtf): Enable *roF alternative for x64 isa only.
15722 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
15723 mode attribute of integer alternatives to DImode for TARGET_64BIT.
15724 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
15725 (*movtf_internal): Merge from *movtf_internal_rex64 and
15726 *movtf_internal_sse. Use x64 and nox64 isa attributes.
15727 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
15728 nox64 isa attributes.
15729 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
15730 nox64 isa attributes.
15731 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15733 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
15735 * config/alpha/alpha.c (TARGET_LRA_P): New define.
15737 2013-03-17 Jakub Jelinek <jakub@redhat.com>
15740 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15741 class names. Remove trailing comma after "ALL_REGS".
15743 2013-03-16 Jan Hubicka <jh@suse.cz>
15745 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15746 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15747 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15748 of cgraph_get_create_node.
15749 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15751 2013-03-16 Jason Merrill <jason@redhat.com>
15754 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15755 with DW_AT_default_value.
15757 2013-03-16 Jakub Jelinek <jakub@redhat.com>
15759 * BASE-VER: Set to 4.9.0.
15761 2013-03-14 Andi Kleen <ak@linux.intel.com>
15764 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15765 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15766 Document _x* TSX intrinsics.
15768 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
15769 David Holsgrove <david.holsgrove@xilinx.com>
15771 * configure.ac: Add MicroBlaze TLS support detection.
15772 * configure: Regenerate.
15773 * config/microblaze/microblaze-protos.h
15774 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15775 symbol_mentioned_p, label_mentioned_p): Add prototypes.
15776 * config/microblaze/microblaze.c (microblaze_address_type): Add
15777 ADDRESS_TLS and tls_reloc address types.
15778 (microblaze_address_info): Add tls_reloc.
15779 (TARGET_HAVE_TLS): Define.
15780 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15781 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15782 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15783 load_tls_operand, microblaze_call_tls_get_addr,
15784 microblaze_legitimize_tls_address): New functions.
15785 (microblaze_classify_unspec): Handle UNSPEC_TLS.
15786 (get_base_reg): Use microblaze_tls_symbol_p.
15787 (microblaze_classify_address): Handle TLS.
15788 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15789 label_mentioned_p and microblaze_tls_referenced_p.
15790 (microblaze_legitimize_address): Handle TLS.
15791 (microblaze_address_insns): Handle ADDRESS_TLS.
15792 (pic_address_needs_scratch): Handle TLS.
15793 (print_operand_address): Handle TLS.
15794 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15795 (microblaze_expand_move): Handle TLS.
15796 (microblaze_legitimate_constant_p): Check
15797 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15798 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15799 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15800 (PIC_OFFSET_TABLE_REGNUM): Set.
15801 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15802 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15803 (addsi3, movsi_internal2, movdf_internal): Update constraints
15804 * config/microblaze/predicates.md (arith_plus_operand): Define
15805 (move_operand): Redefine as move_src_operand,
15806 check microblaze_tls_referenced_p.
15808 2013-03-14 Ian Bolton <ian.bolton@arm.com>
15810 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15811 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15813 2013-03-14 Ian Bolton <ian.bolton@arm.com>
15815 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15818 2013-03-14 Jakub Jelinek <jakub@redhat.com>
15820 PR tree-optimization/53265
15821 * common.opt (Waggressive-loop-optimizations): New option.
15822 * tree-ssa-loop-niter.c: Include tree-pass.h.
15823 (do_warn_aggressive_loop_optimizations): New function.
15824 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
15825 if number_of_latch_executions returned constant.
15826 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15827 early. If number_of_latch_executions returned constant, set
15828 nb_iterations_upper_bound back to it.
15829 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15831 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15832 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15834 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15835 (MULTILIB_OSDIRNAMES): Set.
15836 * genmultilib: If defaultosdirname doesn't start with :: , set
15837 defaultosdirname2 instead, clear it and emit two . multilib_raw
15838 entries instead of just one.
15840 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
15842 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15843 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15844 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15845 (SUBTARGET_OVERRIDE_OPTIONS): New.
15847 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
15850 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15851 (musermode): Convert to Var(TARGET_USERMODE).
15852 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15853 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15854 * config/sh/sh.c (sh_option_override): Use
15855 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15856 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15858 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15860 (udivsi3_i4_single, divsi3_i4_single): Use
15861 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15863 2013-03-13 Dave Korn <dave.korn.cygwin@....>
15865 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15868 2013-03-13 Richard Biener <rguenther@suse.de>
15870 PR tree-optimization/56608
15871 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15872 calls when vectorizing basic-blocks.
15874 2013-03-13 Jakub Jelinek <jakub@redhat.com>
15877 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15880 2013-03-12 Jakub Jelinek <jakub@redhat.com>
15882 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15884 2013-03-11 Jan Hubicka <jh@suse.cz>
15887 * lto-streamer-out.c (output_symbol_p): Skip references from
15888 constructors of external variables.
15890 2013-03-11 Jan Hubicka <jh@suse.cz>
15892 PR middle-end/56571
15893 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15895 * emit-rtl.c (verify_rtx_sharing): Likewise.
15896 (copy_insn_1): Likewise.
15897 * rtl.c (copy_rtx): Likewise.
15899 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
15902 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15903 output_operand_lossage message.
15905 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
15908 * arm.c (shift_op): Validate RTL pattern on the fly.
15909 (arm_print_operand, case 'S'): Don't use shift_operator to validate
15912 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
15915 * config/pa/pa.md (call_value): Check for calls to powf and direct to
15916 new call patterns that clobber %fr12.
15917 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15918 split and postreload patterns.
15919 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15920 registers %fr12 and %fr12R as call used.
15922 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
15924 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15925 (canon_address, record_store, replace_read, check_mem_read_rtx,
15926 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15927 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15928 rest_of_handle_dse): Likewise.
15930 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
15932 PR middle-end/56524
15933 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15935 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15936 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15937 (save_optabs_if_changed): Replace with...
15938 (init_tree_optimization_optabs): ...this.
15939 * optabs.c (save_optabs_if_changed): Rename to...
15940 (init_tree_optimization_optabs): ...this. Take the optimization node
15941 as argument. Do nothing if the base optabs are already correct.
15942 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15943 to recompute optabs.
15944 * function.h (function): Remove optabs field.
15945 * function.c (invoke_set_current_function_hook): Call
15946 init_tree_optimization_optabs. Use the result to initialize
15949 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
15951 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15952 if GTMA_HAS_NO_INSTRUMENTATION.
15953 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15954 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15955 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15956 * gimple-pretty-print.c (dump_gimple_transaction): Handle
15957 GTMA_HAS_NO_INSTRUMENTATION.
15959 2013-03-08 Jakub Jelinek <jakub@redhat.com>
15961 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15964 2013-03-08 Marek Polacek <polacek@redhat.com>
15965 Jakub Jelinek <jakub@redhat.com>
15967 PR tree-optimization/56478
15968 * predict.c (is_comparison_with_loop_invariant_p): Change the
15969 type of loop_step to tree.
15970 (predict_loops): Adjust.
15971 (predict_iv_comparison): Perform the computations on double_ints.
15973 2013-03-08 Richard Biener <rguenther@suse.de>
15975 PR tree-optimization/56570
15976 * tree-cfg.c (verify_expr_location_1): Verify locations for
15978 * tree-sra.c (create_access_replacement): Strip locations
15979 from DECL_DEBUG_EXPRs.
15981 2013-03-08 Richard Biener <rguenther@suse.de>
15983 * tree-inline.c (expand_call_inline): Do not associate
15984 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15985 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15987 2013-03-08 Richard Biener <rguenther@suse.de>
15989 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15990 or block changes with -Og. Fix for location / block encoding
15991 changes and PHI arguments with locations.
15993 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
15995 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15997 (struct output_info): Likewise.
15998 (register_overhead): Remove bad gcc_assert.
15999 (bitmap_find_bit): If there is only a single bitmap element, do not
16000 count a miss as a search.
16001 (print_statistics): Update for counter type changes.
16002 (dump_bitmap_statistics): Likewise. Print headers such that they
16003 are properly lined up with the printed counters.
16005 2013-03-07 Jakub Jelinek <jakub@redhat.com>
16007 PR tree-optimization/56559
16008 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
16009 check that it has only a single use.
16011 2013-03-07 Richard Biener <rguenther@suse.de>
16013 * doc/invoke.texi (fwhole-program): Discourage use in combination
16016 2013-03-06 Jakub Jelinek <jakub@redhat.com>
16018 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
16020 PR tree-optimization/56539
16021 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
16022 instead of GSI_CONTINUE_LINKING as last argument to
16023 force_gimple_operand_gsi. Adjust function comment.
16025 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
16028 PR middle-end/56548
16029 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
16030 promoted mode, convert the result back to the original mode.
16032 2013-03-06 Richard Biener <rguenther@suse.de>
16034 PR middle-end/56294
16035 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
16036 (insert_updated_phi_nodes_compare_uids): New function.
16037 (update_ssa): Sort symbols_to_rename after UID before
16038 traversing it to insert PHI nodes.
16040 2013-03-06 Richard Biener <rguenther@suse.de>
16042 PR middle-end/50494
16043 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
16044 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
16047 2013-02-13 Richard Biener <rguenther@suse.de>
16050 * varasm.c (output_constant_def_1): Get the decl representing
16051 the constant as argument.
16052 (output_constant_def): Wrap output_constant_def_1.
16053 (make_decl_rtl): Use output_constant_def_1 with the decl
16054 representing the constant.
16055 (build_constant_desc): Optionally re-use a decl already
16056 representing the constant.
16057 (tree_output_constant_def): Adjust.
16059 2013-03-06 Joey Ye <joey.ye@arm.com>
16062 * gcc.c (convert_white_space): New function.
16063 (main): Handles white space in function name.
16065 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
16068 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
16069 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
16070 to SH_DIV_CALL_TABLE for TARGET_SH2.
16071 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
16073 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
16074 call-table options.
16076 2013-03-05 Sterling Augustine <saugustine@google.com>
16077 Cary Coutant <ccoutant@google.com>
16080 * dwarf2out.c (resolve_addr): Don't call
16081 remove_loc_list_addr_table_entries a second time for the same
16084 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16087 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
16088 (avoid_complex_debug_insns): New function.
16089 (expand_debug_locations): Call it.
16091 PR rtl-optimization/56484
16092 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
16093 lifetimes of hard registers on small register class machines.
16095 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
16097 * config/microblaze/microblaze-protos.h: Rename
16098 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
16099 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
16101 (microblaze_fast_interrupt_function_p): New function.
16102 (microblaze_is_interrupt_handler): Rename to
16103 microblaze_is_interrupt_variant and add fast_interrupt check.
16104 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
16105 (save_restore_insns): Likewise.
16106 (compute_frame_size): Likewise.
16107 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
16108 (microblaze_globalize_label): Likewise.
16109 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
16110 * config/microblaze/microblaze.md: Use wrapper
16111 microblaze_is_interrupt_variant.
16113 2013-03-05 Kai Tietz <ktietz@redhat.com>
16115 * sdbout.c (sdbout_one_type): Switch to current function's section
16116 supporting cold/hot.
16118 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
16120 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
16123 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16125 PR middle-end/56461
16126 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
16127 if VALGRIND_GET_VBITS is defined, temporarily make object
16128 memory all defined, and restore previous valgrind addressability
16129 and definability afterwards. Free this_object at the end.
16131 PR middle-end/56461
16132 * lra.c (lra): Call lra_clear_live_ranges if live_p,
16133 right before calling lra_create_live_ranges, also call it
16134 when clearing live_p. Only call lra_clear_live_ranges
16135 at the end if live_p.
16137 PR middle-end/56461
16138 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
16140 2013-03-05 Richard Biener <rguenther@suse.de>
16142 PR tree-optimization/56521
16143 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
16146 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
16149 * except.h (remove_unreachable_eh_regions): New prototype.
16150 * except.c (remove_eh_handler_splicer): New function, split out
16151 of remove_eh_handler.
16152 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
16153 warning about running it on many EH regions one at a time.
16154 (remove_unreachable_eh_regions_worker): New function, walk the
16155 EH tree in depth-first order and remove non-marked regions.
16156 (remove_unreachable_eh_regions): New function.
16157 * tree-eh.c (mark_reachable_handlers): New function, split out
16158 from remove_unreachable_handlers.
16159 (remove_unreachable_handlers): Use mark_reachable_handlers and
16160 remove_unreachable_eh_regions.
16161 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
16162 and remove_unreachable_eh_regions.
16164 2013-03-05 Richard Biener <rguenther@suse.de>
16166 PR middle-end/56525
16167 * loop-init.c (fix_loop_structure): Remove loops in two stages,
16168 not freeing them until the end.
16170 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
16172 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
16174 2013-03-05 Richard Biener <rguenther@suse.de>
16176 PR tree-optimization/56270
16177 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
16178 of loads after scheduling an SLP instance.
16180 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16182 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
16184 (check_gcc_parallelize): Run guality.exp as a separate job from
16185 vect.exp with unsorted.exp and $(dg_target_exps) separately from
16186 struct-layout-1.exp with stackalign.exp.
16188 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
16190 PR middle-end/56461
16191 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
16192 load_index sbitmap even if some bit in it isn't set.
16194 PR middle-end/56461
16195 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
16196 (discover_iteration_bound_by_body_walk): Change queues to
16197 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
16198 spelling in comment. Call safe_push on queues[bound_index] directly.
16199 Release queues[queue_index] in every iteration unconditionally.
16200 Release bounds vector.
16202 PR middle-end/56461
16203 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
16204 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
16205 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
16208 2013-03-05 Richard Biener <rguenther@suse.de>
16211 * tree-inline.c (remap_blocks_to_null): New function.
16212 (expand_call_inline): When expanding a call stmt without
16213 an associated block inline remap all callee blocks to NULL.
16215 2013-03-05 Jakub Jelinek <jakub@redhat.com>
16217 PR rtl-optimization/56494
16218 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
16219 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
16220 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
16222 PR middle-end/56461
16223 * sel-sched-ir.c (free_sched_pools): Release
16224 succs_info_pool.stack[succs_info_pool.max_top] vectors too
16225 if succs_info_pool.max_top isn't -1.
16228 * opts.c (opts_obstack, opts_concat): Moved to...
16229 * opts-common.c (opts_obstack, opts_concat): ... here.
16231 2013-03-04 Jakub Jelinek <jakub@redhat.com>
16233 PR middle-end/56461
16234 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
16236 2013-03-04 Martin Jambor <mjambor@suse.cz>
16238 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
16239 all appropriate places.
16241 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
16243 PR tree-optimization/56424
16244 * ipa-split.c (split_function): Do not set the RSO flag if result is
16245 not by reference and its type is a register type.
16247 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
16249 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
16250 (microblaze_legitimate_pic_operand): Likewise
16251 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
16252 new function microblaze_legitimate_pic_operand
16253 * config/microblaze/microblaze-protos.h
16254 (microblaze_legitimate_pic_operand): Declare.
16256 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16258 * config/microblaze/predicates.md (call_insn_simple_operand):
16259 New predicate for supported rtx code types.
16260 * config/microblaze/microblaze.md (call_internal1): Use
16261 call_insn_simple_operand predicate.
16263 2013-03-04 Jakub Jelinek <jakub@redhat.com>
16265 PR middle-end/56461
16266 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
16267 partitions.ordered_remove.
16269 PR middle-end/56461
16270 * tree-vect-stmts.c (vectorizable_conversion): Don't call
16271 vec_oprnds0.create (1) for modifier == NONE.
16273 PR middle-end/56461
16274 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
16275 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
16276 vec_oprnds1 right before pushing anything to it for
16279 PR middle-end/56461
16280 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
16281 set nbbs to 0 instead of having separate code path.
16282 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
16283 instead of false as last argument if returning NULL.
16285 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
16287 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
16288 the attribute is now called "target" instead of "option".
16289 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
16290 * doc/tm.texi.in (Target Attributes): Likewise document the correct
16291 attribute/pragma name for TARGET_OPTION_VALID_P and
16292 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
16293 * doc/tm.texi: Regenerated.
16295 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
16297 * config/microblaze/microblaze.c:
16298 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
16299 * config/microblaze/microblaze.h: Add -mxl-reorder to
16301 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
16302 instructions emitted if TARGET_REORDER.
16303 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
16304 or 0 for -m/-mno case, but initialises as 2 to detect default use case
16307 2013-03-01 Xinliang David Li <davidxl@google.com>
16309 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
16312 2013-03-01 Jakub Jelinek <jakub@redhat.com>
16314 PR middle-end/56461
16315 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
16316 vector even when returning true. Fix up function comment formatting.
16318 PR middle-end/56461
16319 * ira-build.c (ira_loop_nodes_count): New variable.
16320 (create_loop_tree_nodes): Initialize it.
16321 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
16323 PR middle-end/56461
16324 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
16325 method on dr_chain and result_chain.
16326 * tree-vect-stmts.c (vectorizable_store): Only call
16327 result_chain.create if j == 0.
16329 PR middle-end/56461
16330 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
16331 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
16332 before overwriting it.
16334 2013-03-01 Tobias Burnus <burnus@net-b.de>
16336 * doc/extended.texi (C Extensions): Change order in @menu
16338 (Other MIPS Built-in Functions): Move last MIPS entry before
16339 "picoChip Built-in Functions".
16340 (SH Built-in Functions): Move after RX Built-in Functions.
16341 * doc/gcc.texi (Introduction): Change order in @menu
16343 * doc/md.texi (Constraints): Ditto.
16344 * gty.texi (Type Information): Ditto.
16345 (User-provided marking routines for template types): Make
16347 * doc/invoke.texi (AArch64 Options): Move before
16348 "Adapteva Epiphany Options".
16350 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
16351 Jakub Jelinek <jakub@redhat.com>
16354 * asan.c (gate_asan): Lookup no_sanitize_address instead of
16355 no_address_safety_analysis attribute.
16356 * doc/extend.texi (no_address_safety_attribute): Rename to
16357 no_sanitize_address attribute, mention no_address_safety_analysis
16358 attribute as deprecated alias.
16360 2013-02-28 Jakub Jelinek <jakub@redhat.com>
16362 PR middle-end/56461
16363 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
16364 type to vec<vec<tree> > *.
16365 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
16366 to be vec<tree> instead of vec<tree> *, set vec_defs
16367 to vNULL and call vec_defs.create (number_of_vects), adjust other
16369 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
16370 vectorizable_condition): Adjust vect_get_slp_defs callers.
16372 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16374 * config/aarch64/aarch64.c
16375 (aarch64_float_const_representable): Remove unused variable.
16377 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16379 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
16381 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16383 * config/aarch64/aarch64-builtins.c
16384 (aarch64_init_simd_builtins): Make static.
16386 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
16388 * config/aarch64/aarch64.c
16389 (aarch64_simd_make_constant): Make static.
16391 2013-02-28 Martin Jambor <mjambor@suse.cz>
16393 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
16394 with no initialization to the RHS of debug statements.
16396 2013-02-28 Martin Jambor <mjambor@suse.cz>
16398 PR tree-optimization/56294
16399 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
16401 (get_access_replacement): Do not call create_access_replacement.
16402 Assert a replacement exists.
16403 (get_repl_default_def_ssa_name): Create the replacement declaration
16406 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16408 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
16409 final_end_function.
16411 2013-02-28 Marek Polacek <polacek@redhat.com>
16413 PR rtl-optimization/56466
16414 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
16415 if we're changing a loop.
16416 (peel_loops_completely): Likewise.
16418 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
16421 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
16423 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16426 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
16427 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
16428 INTX_FTYPE_FX, FX_FTYPE_INTX.
16429 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
16431 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16433 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
16434 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
16435 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
16436 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
16437 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
16438 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
16439 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
16440 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
16441 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
16442 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
16443 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
16444 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
16445 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
16446 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
16447 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
16448 (avrxmega6): Increase max flash segments from 5 to 6.
16449 * config/avr/t-multilib: Regenerate.
16450 * config/avr/avr-tables.opt: Regenerate.
16451 * doc/avr-mmcu.texi: Regenerate.
16453 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
16455 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
16456 (avr_device_to_arch): Rename to avr_device_to_ld.
16457 (avr_device_to_as): New prototype.
16458 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
16459 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
16460 * config/avr/driver-avr.c (avr_device_to_as): New.
16461 (avr_device_to_arch): Rename to avr_device_to_ld.
16463 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16465 PR middle-end/56461
16466 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
16467 method on dr_chain and result_chain.
16469 PR middle-end/56461
16470 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
16471 pointer_set_destroy on not_executed_last_iteration.
16473 PR middle-end/56461
16474 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
16476 PR middle-end/56461
16477 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
16478 FOR_EACH_DEFINED_FUNCTION when freeing state.
16480 PR middle-end/56461
16481 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
16483 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
16486 PR middle-end/56461
16487 * ipa-cp.c (decide_whether_version_node): Call vec_free on
16488 known_aggs[i].items and release known_aggs vector.
16490 PR middle-end/56461
16491 * ipa-reference.c (propagate): Free node_info even for alias nodes.
16493 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
16495 * config/microblaze/microblaze.c (microblaze_emit_compare):
16496 Use xor for EQ/NE comparisions.
16497 * config/microblaze/microblaze.md (cstoresf4): Add constraints
16498 (cbranchsf4): Adjust operator to comparison_operator.
16500 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16502 PR middle-end/56461
16503 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
16505 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
16506 vec_safe_push, always update *slot.
16507 (redirect_edge_var_map_clear): Use vec_free.
16508 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
16509 (free_var_map_entry): Use vec_free.
16510 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
16511 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
16513 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
16515 PR middle-end/45472
16516 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
16517 when the may_trap_p bit of the exprs being merged differs.
16518 Reorder tests for speculativeness in the logical and operator.
16520 2013-02-27 Jakub Jelinek <jakub@redhat.com>
16522 * incpath.c (add_standard_paths): Use reconcat instead of concat
16523 where appropriate and avoid leaking memory.
16525 * opts.h: Include obstack.h.
16526 (opts_concat): New prototype.
16527 (opts_obstack): New declaration.
16528 * opts.c (opts_concat): New function.
16529 (opts_obstack): New variable.
16530 (init_options_struct): Call gcc_init_obstack on opts_obstack.
16531 (finish_options): Use opts_concat instead of concat
16532 and XOBNEWVEC instead of XNEWVEC.
16533 * opts-common.c (generate_canonical_option, decode_cmdline_option,
16534 generate_option): Likewise.
16535 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
16536 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
16539 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
16540 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
16542 2013-02-26 Jakub Jelinek <jakub@redhat.com>
16544 PR middle-end/56461
16545 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
16547 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
16549 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
16550 (arm_block_move_unaligned_straight): Likewise.
16551 (arm_adjust_block_mem): Likewise.
16553 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
16556 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
16557 temp, cond and label.
16558 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
16561 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16562 * config/c6x/c6x.h (dbx_register_map): Update declaration.
16565 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16566 of prologue/epilogue functions.
16569 * config/tilegx/tilegx.c (tilegx_expand_prologue):
16570 Remove unused variable cfa_offset.
16571 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16574 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16575 type promotion to unsigned.
16578 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
16579 for HOST_WIDE_INT of 32 bit / same size as int.
16580 (arm_block_move_unaligned_straight): Likewise.
16581 (arm_adjust_block_mem): Likewise.
16584 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
16587 2013-02-26 Marek Polacek <polacek@redhat.com>
16589 PR tree-optimization/56426
16590 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
16592 2013-02-26 Richard Biener <rguenther@suse.de>
16595 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
16596 unused variable loops.
16598 2013-02-26 Jakub Jelinek <jakub@redhat.com>
16600 PR tree-optimization/56448
16601 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16602 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16603 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16604 later operands of the references, or even first operand for
16605 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16607 PR tree-optimization/56443
16608 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16609 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16610 to type_for_mode langhook.
16612 2013-02-25 Matt Turner <mattst88@gmail.com>
16614 * doc/invoke.texi: Document r4700.
16616 2013-02-25 Richard Biener <rguenther@suse.de>
16618 PR tree-optimization/56175
16619 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16621 (simplify_bitwise_binary): ... here. Also guard the conversion
16622 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16624 2013-02-25 Catherine Moore <clm@codesourcery.com>
16627 2013-02-24 Catherine Moore <clm@codesourcery.com>
16628 Maciej W. Rozycki <macro@codesourcery.com>
16629 Tom de Vries <tom@codesourcery.com>
16630 Nathan Sidwell <nathan@codesourcery.com>
16631 Iain Sandoe <iain@codesourcery.com>
16632 Nathan Froyd <froydnj@codesourcery.com>
16633 Chao-ying Fu <fu@mips.com>
16635 * doc/extend.texi: (micromips, nomicromips, nocompression):
16636 Document new function attributes.
16637 * doc/invoke.texi (minterlink-compressed, mmicromips,
16638 m14k, m14ke, m14kec): Document new options.
16639 (minterlink-mips16): Update documentation.
16640 * doc/md.texi (ZC, ZD): Document new constraints.
16641 * configure.ac (gcc_cv_as_micromips): Check if linker
16642 supports the .set micromips directive.
16643 * configure: Regenerate.
16644 * config.in: Regenerate.
16645 * config/mips/mips-tables.opt: Regenerate.
16646 * config/mips/micromips.md: New file.
16647 * constraints.md (ZC, AD): New constraints.
16648 * config/mips/predicates.md (movep_src_register): New predicate.
16649 (movep_src_operand): New predicate.
16650 (non_volatile_mem_operand): New predicate.
16651 * config/mips/mips.md (multimem): New type.
16652 (length): Differentiate between 17-bit and 18-bit branch offsets.
16653 (MOVEP1, MOVEP2): New mode iterator.
16654 (mov_<load>l): Use ZC constraint.
16655 (mov_<load>r): Likewise.
16656 (mov_<store>l): Likewise.
16657 (mov_<store>r): Likewise.
16658 (*branch_equality<mode>_inverted): Add microMIPS support.
16659 (*branch_equality<mode>): Likewise.
16660 (*jump_absolute): Likewise.
16661 (indirect_jump_<mode>): Likewise.
16662 (tablejump_<mode>): Likewise.
16663 (<optab>_internal): Likewise.
16664 (sibcall_internal): Likewise.
16665 (sibcall_value_internal): Likewise.
16666 (prefetch): Use constraint ZD.
16667 * config/mips/mips.opt (minterlink-compressed): New option.
16668 (minterlink-mips16): Now an alias for minterlink-compressed.
16669 (mmicromips): New option.
16670 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16671 (compare_and_swap_12): Likewise.
16672 (sync_add<mode>): Likewise.
16673 (sync_<optab>_12): Likewise.
16674 (sync_old_<optab>_12): Likewise.
16675 (sync_new_<optab>_12): Likewise.
16676 (sync_nand_12): Likewise.
16677 (sync_old_nand_12): Likewise.
16678 (sync_new_nand_12): Likewise.
16679 (sync_sub<mode>): Likewise.
16680 (sync_old_add<mode>): Likewise.
16681 (sync_old_sub<mode>): Likewise.
16682 (sync_new_add<mode>): Likewise.
16683 (sync_new_sub<mode>): Likewise.
16684 (sync_<optab><mode>): Likewise.
16685 (sync_old_<optab><mode>): Likewise.
16686 (sync_new_<optab><mode>): Likewise.
16687 (sync_nand<mode>): Likewise.
16688 (sync_old_nand<mode>): Likewise.
16689 (sync_new_nand<mode>): Likewise.
16690 (sync_lock_test_and_set<mode>): Likewise.
16691 (test_and_set_12): Likewise.
16692 (atomic_compare_and_swap<mode>): Likewise.
16693 (atomic_exchange<mode>_llsc): Likewise.
16694 (atomic_fetch_add<mode>_llsc): Likewise.
16695 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16696 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16697 (umips_save_restore_pattern_p): Likewise.
16698 (umips_load_store_pair_p): Likewise.
16699 (umips_output_load_store_pair): Likewise.
16700 (umips_movep_target_p): Likewise.
16701 (umips_12bit_offset_address_p): Likewise.
16702 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16703 (mips_base_mips16): Rename this...
16704 (mips_base_compression_flags): ...to this. Update all uses.
16705 (mips_attribute_table): Add micromips, nomicromips and nocompression.
16706 (mips_mips16_decl_p): Delete.
16707 (mips_nomips16_decl_p): Delete.
16708 (mips_get_compress_on_flags): New function.
16709 (mips_get_compress_off_flags): New function.
16710 (mips_get_compress_mode): New function.
16711 (mips_get_compress_on_name): New function.
16712 (mips_get_compress_off_name): New function.
16713 (mips_insert_attributes): Support multiple compression types.
16714 (mips_merge_decl_attributes): Likewise.
16715 (umips_12bit_offset_address_p): New function.
16716 (mips_start_function_definition): Emit .set micromips directive.
16717 (mips_call_may_need_jalx_p): New function.
16718 (mips_function_ok_for_sibcall): Add microMIPS support.
16719 (mips_print_operand_punctuation): Support short delay slots and
16721 (umips_swm_mask, umips_swm_encoding): New.
16722 (umips_build_save_restore): New function.
16723 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16724 (was_mips16_p): Remove.
16725 (old_compression_mode): New.
16726 (mips_set_compression_mode): New function.
16727 (mips_set_current_function): Add microMIPS support.
16728 (mips_option_override): Likewise.
16729 (umips_save_restore_pattern_p): New function.
16730 (umips_output_save_restore): New function.
16731 (umips_load_store_pair_p_1): New function.
16732 (umips_load_store_pair_p): New function.
16733 (umips_output_load_store_pair_1): New function.
16734 (umips_output_load_store_pair): New function.
16735 (umips_movep_target_p) New function.
16736 (mips_prepare_pch_save): Add microMIPS support.
16737 * config/mips/mips.h (TARGET_COMPRESSION): New.
16738 (TARGET_CPU_CPP_BUILTINS): Update macro
16739 to use new compression flags and to support microMIPS.
16740 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16741 (MIPS_ARCH_FLOAT_SPEC): Likewise.
16742 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16743 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16744 (ASM_SPEC): Support mmicromips and mno-micromips.
16745 (M16STORE_REG_P): New macro.
16746 (MIPS_CALL): Support TARGET_MICROMIPS.
16747 (MICROMIPS_J): New macro.
16748 (mips_base_mips16): Rename this...
16749 (mips_base_compression_flags): ...to this.
16750 (UMIPS_12BIT_OFFSET_P): New macro.
16751 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16752 (MULTILIB_DIRNAMES): Likewise.
16754 2013-02-25 Tom de Vries <tom@codesourcery.com>
16756 PR rtl-optimization/56131
16757 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16758 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16759 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16761 2013-02-25 Tobias Burnus <burnus@net-b.de>
16763 * doc/invoke.texi (-fsanitize=): Move from optimization
16764 to debugging options.
16766 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
16768 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16770 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
16771 Alexander Monakov <amonakov@ispras.ru>
16773 PR middle-end/56077
16774 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16775 flush pending lists also on non-jumps. Adjust comment.
16777 2013-02-24 Catherine Moore <clm@codesourcery.com>
16778 Maciej W. Rozycki <macro@codesourcery.com>
16779 Tom de Vries <tom@codesourcery.com>
16780 Nathan Sidwell <nathan@codesourcery.com>
16781 Iain Sandoe <iain@codesourcery.com>
16782 Nathan Froyd <froydnj@codesourcery.com>
16783 Chao-ying Fu <fu@mips.com>
16785 * doc/extend.texi: (micromips, nomicromips, nocompression):
16786 Document new function attributes.
16787 * doc/invoke.texi (minterlink-compressed, mmicromips,
16788 m14k, m14ke, m14kec): Document new options.
16789 (minterlink-mips16): Update documentation.
16790 * doc/md.texi (ZC, ZD): Document new constraints.
16791 * configure.ac (gcc_cv_as_micromips): Check if linker
16792 supports the .set micromips directive.
16793 * configure: Regenerate.
16794 * config.in: Regenerate.
16795 * config/mips/mips-tables.opt: Regenerate.
16796 * config/mips/micromips.md: New file.
16797 * constraints.md (ZC, AD): New constraints.
16798 * config/mips/predicates.md (movep_src_register): New predicate.
16799 (movep_src_operand): New predicate.
16800 (non_volatile_mem_operand): New predicate.
16801 * config/mips/mips.md (multimem): New type.
16802 (length): Differentiate between 17-bit and 18-bit branch offsets.
16803 (MOVEP1, MOVEP2): New mode iterator.
16804 (mov_<load>l): Use ZC constraint.
16805 (mov_<load>r): Likewise.
16806 (mov_<store>l): Likewise.
16807 (mov_<store>r): Likewise.
16808 (*branch_equality<mode>_inverted): Add microMIPS support.
16809 (*branch_equality<mode>): Likewise.
16810 (*jump_absolute): Likewise.
16811 (indirect_jump_<mode>): Likewise.
16812 (tablejump_<mode>): Likewise.
16813 (<optab>_internal): Likewise.
16814 (sibcall_internal): Likewise.
16815 (sibcall_value_internal): Likewise.
16816 (prefetch): Use constraint ZD.
16817 * config/mips/mips.opt (minterlink-compressed): New option.
16818 (minterlink-mips16): Now an alias for minterlink-compressed.
16819 (mmicromips): New option.
16820 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16821 (compare_and_swap_12): Likewise.
16822 (sync_add<mode>): Likewise.
16823 (sync_<optab>_12): Likewise.
16824 (sync_old_<optab>_12): Likewise.
16825 (sync_new_<optab>_12): Likewise.
16826 (sync_nand_12): Likewise.
16827 (sync_old_nand_12): Likewise.
16828 (sync_new_nand_12): Likewise.
16829 (sync_sub<mode>): Likewise.
16830 (sync_old_add<mode>): Likewise.
16831 (sync_old_sub<mode>): Likewise.
16832 (sync_new_add<mode>): Likewise.
16833 (sync_new_sub<mode>): Likewise.
16834 (sync_<optab><mode>): Likewise.
16835 (sync_old_<optab><mode>): Likewise.
16836 (sync_new_<optab><mode>): Likewise.
16837 (sync_nand<mode>): Likewise.
16838 (sync_old_nand<mode>): Likewise.
16839 (sync_new_nand<mode>): Likewise.
16840 (sync_lock_test_and_set<mode>): Likewise.
16841 (test_and_set_12): Likewise.
16842 (atomic_compare_and_swap<mode>): Likewise.
16843 (atomic_exchange<mode>_llsc): Likewise.
16844 (atomic_fetch_add<mode>_llsc): Likewise.
16845 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16846 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16847 (umips_save_restore_pattern_p): Likewise.
16848 (umips_load_store_pair_p): Likewise.
16849 (umips_output_load_store_pair): Likewise.
16850 (umips_movep_target_p): Likewise.
16851 (umips_12bit_offset_address_p): Likewise.
16852 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16853 (mips_base_mips16): Rename this...
16854 (mips_base_compression_flags): ...to this. Update all uses.
16855 (mips_attribute_table): Add micromips, nomicromips and nocompression.
16856 (mips_mips16_decl_p): Delete.
16857 (mips_nomips16_decl_p): Delete.
16858 (mips_get_compress_on_flags): New function.
16859 (mips_get_compress_off_flags): New function.
16860 (mips_get_compress_mode): New function.
16861 (mips_get_compress_on_name): New function.
16862 (mips_get_compress_off_name): New function.
16863 (mips_insert_attributes): Support multiple compression types.
16864 (mips_merge_decl_attributes): Likewise.
16865 (umips_12bit_offset_address_p): New function.
16866 (mips_start_function_definition): Emit .set micromips directive.
16867 (mips_call_may_need_jalx_p): New function.
16868 (mips_function_ok_for_sibcall): Add microMIPS support.
16869 (mips_print_operand_punctuation): Support short delay slots and
16871 (umips_swm_mask, umips_swm_encoding): New.
16872 (umips_build_save_restore): New function.
16873 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16874 (was_mips16_p): Remove.
16875 (old_compression_mode): New.
16876 (mips_set_compression_mode): New function.
16877 (mips_set_current_function): Add microMIPS support.
16878 (mips_option_override): Likewise.
16879 (umips_save_restore_pattern_p): New function.
16880 (umips_output_save_restore): New function.
16881 (umips_load_store_pair_p_1): New function.
16882 (umips_load_store_pair_p): New function.
16883 (umips_output_load_store_pair_1): New function.
16884 (umips_output_load_store_pair): New function.
16885 (umips_movep_target_p) New function.
16886 (mips_prepare_pch_save): Add microMIPS support.
16887 * config/mips/mips.h (TARGET_COMPRESSION): New.
16888 (TARGET_CPU_CPP_BUILTINS): Update macro
16889 to use new compression flags and to support microMIPS.
16890 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16891 (MIPS_ARCH_FLOAT_SPEC): Likewise.
16892 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16893 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16894 (ASM_SPEC): Support mmicromips and mno-micromips.
16895 (M16STORE_REG_P): New macro.
16896 (MIPS_CALL): Support TARGET_MICROMIPS.
16897 (MICROMIPS_J): New macro.
16898 (mips_base_mips16): Rename this...
16899 (mips_base_compression_flags): ...to this.
16900 (UMIPS_12BIT_OFFSET_P): New macro.
16901 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16902 (MULTILIB_DIRNAMES): Likewise.
16904 2013-02-24 Jakub Jelinek <jakub@redhat.com>
16907 * target-globals.c (save_target_globals): For init_reg_sets and
16908 target_reinit remporarily set this_fn_optabs to this_target_optabs.
16910 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
16912 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16913 * config/aarch64/t-aarch64
16914 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16916 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
16918 PR inline-asm/56148
16919 * lra-constraints.c (process_alt_operands): Reload operand
16920 conflicting with earlier clobber only if no more other conflicting
16923 2013-02-22 Jakub Jelinek <jakub@redhat.com>
16926 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16927 if not linking a shared library.
16929 2013-02-22 Seth LaForge <sethml@google.com>
16931 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16933 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
16935 * config/arm/arm.md (split for extendsidi): Update condition.
16936 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16937 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16938 (qhs_zextenddi_cstr): Likewise.
16940 2013-02-21 Jakub Jelinek <jakub@redhat.com>
16942 PR middle-end/56420
16943 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16944 avoid signed wrapping.
16945 (expand_mult): Handle properly multiplication by
16946 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
16947 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
16948 in the compiler if coeff is HOST_WIDE_INT_MIN.
16949 (expand_divmod): Don't make ext_op1 static, change it's type to
16950 uhwi. Avoid undefined behavior in -INTVAL (op1).
16952 PR rtl-optimization/50339
16953 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16955 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16956 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16957 into splitting_ashiftrt field.
16958 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16960 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16963 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
16965 PR middle-end/56108
16966 * trans-mem.c (execute_tm_mark): Do not expand transactions that
16967 are sure to go irrevocable.
16969 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
16971 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16972 scalars are valid operands.
16974 2013-02-21 Martin Jambor <mjambor@suse.cz>
16976 PR tree-optimization/56310
16977 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16978 only matching indices and non-negative final offsets.
16979 (intersect_aggregates_with_edge): Pass src_idx to
16980 agg_replacements_to_vector. Pass src_idx insstead of index to
16981 intersect_with_agg_replacements.
16983 2013-02-21 Martin Jambor <mjambor@suse.cz>
16985 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16986 instead of hard-wired defaults.
16988 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
16990 * doc/invoke.texi (MIPS Options): Update documentation of the
16991 floating-point multiply-accumulate instruction restrictions.
16993 2013-02-21 Kostya Serebryany <kcc@google.com>
16995 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16996 asan_shadow_offset on x86_64 linux.
16998 2013-02-21 Richard Biener <rguenther@suse.de>
17000 PR tree-optimization/56415
17002 2013-02-11 Richard Biener <rguenther@suse.de>
17004 PR tree-optimization/56273
17005 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17008 2013-02-21 Jakub Jelinek <jakub@redhat.com>
17011 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
17014 PR inline-asm/56405
17015 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
17016 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
17018 2013-02-20 Jan Hubicka <jh@suse.cz>
17020 PR tree-optimization/56265
17021 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
17022 when target is referenced for first time.
17024 2013-02-20 Richard Biener <rguenther@suse.de>
17026 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
17027 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
17028 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
17029 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
17030 not return anything.
17031 (rename_ssa_copies): Do not remove unused locals.
17032 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
17033 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
17034 * passes.c (execute_function_todo): Do not schedule unused locals
17035 removal if cleanup_tree_cfg did something.
17036 * tree-ssa-live.c (remove_unused_locals): Dump statistics
17037 about the number of removed locals.
17039 2013-02-20 Richard Biener <rguenther@suse.de>
17041 PR tree-optimization/56398
17042 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
17044 2013-02-20 Martin Jambor <mjambor@suse.cz>
17046 PR tree-optimization/55334
17047 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
17048 restricted pointers to arrays.
17050 2013-02-20 Richard Biener <rguenther@suse.de>
17051 Jakub Jelinek <jakub@redhat.com>
17053 PR tree-optimization/56396
17054 * tree-ssa-ccp.c (n_const_val): New static variable.
17055 (get_value): Return NULL for SSA names we don't have a lattice
17057 (ccp_initialize): Initialize n_const_val.
17058 * tree-ssa-copy.c (n_copy_of): New static variable.
17059 (init_copy_prop): Initialize n_copy_of.
17060 (get_value): Return NULL_TREE for SSA names we don't have a
17063 2013-02-20 Martin Jambor <mjambor@suse.cz>
17065 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
17067 2013-02-20 Richard Biener <rguenther@suse.de>
17069 * genpreds.c (write_lookup_constraint): Do not compare first
17070 letter of the constraint again.
17072 2013-02-20 Richard Biener <rguenther@suse.de>
17074 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
17076 (get_use_iv_cost): Terminate hashtable walk when coming across
17079 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
17081 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
17082 reassociation for avx2 targets.
17084 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17086 * config/microblaze/microblaze.c: microblaze_has_clz = 0
17087 Add version check for v8.10.a to enable microblaze_has_clz
17088 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
17089 version and TARGET_PATTERN_COMPARE check
17090 * config/microblaze/microblaze.md: New clzsi2 instruction
17092 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17094 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
17095 function before branching.
17097 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
17099 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
17101 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
17103 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
17105 PR middle-end/55889
17106 * sel-sched.c: Include ira.h.
17107 (implicit_clobber_conflict_p): New function.
17108 (moveup_expr): Use it.
17109 * Makefile.in (sel-sched.o): Depend on ira.h.
17111 2013-02-19 Richard Biener <rguenther@suse.de>
17113 PR tree-optimization/56384
17114 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
17115 (vn_hash_type): Split out from ...
17116 (vn_hash_constant_with_type): ... here.
17117 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
17118 (vn_phi_eq): Compare types from vn_phi_s structure.
17119 (vn_phi_lookup): Populate vn_phi_s type.
17120 (vn_phi_insert): Likewise.
17122 2013-02-19 Jakub Jelinek <jakub@redhat.com>
17124 PR tree-optimization/56350
17125 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
17126 if haven't found reduction or nested cycle operand, rather than
17127 asserting we must find it.
17129 PR tree-optimization/56381
17130 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
17133 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
17134 Jakub Jelinek <jakub@redhat.com>
17137 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
17138 (swap_optab_enable): Same.
17139 (init_all_optabs): Use argument instead of global.
17140 * tree.h (struct tree_optimization_option): New field target_optabs.
17141 * expr.h (init_all_optabs): Add argument to prototype.
17142 (TREE_OPTIMIZATION_OPTABS): New.
17143 (save_optabs_if_changed): Protoize.
17144 * optabs.h: Declare this_fn_optabs.
17145 * optabs.c (save_optabs_if_changed): New.
17146 Declare this_fn_optabs.
17147 (init_optabs): Add argument to init_all_optabs() call.
17148 * function.c (invoke_set_current_function_hook): Handle per
17150 * function.h (struct function): New field optabs.
17151 * config/mips/mips.c (mips_set_mips16_mode): Handle when
17152 optimization_current_node has changed.
17153 * target-globals.h (save_target_globals_default_opts): Protoize.
17154 * target-globals.c (save_target_globals_default_opts): New.
17156 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17159 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
17160 registers %fr12 and %fr12R as call used.
17163 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
17164 and HImode, require all displacements to be an integer multiple of
17166 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
17167 only allow QImode and HImode when reload is in progress and strict is
17168 true. Likewise for symbolic addresses. Use base14_operand to check
17169 displacements in REG+BASE addresses.
17171 2013-02-18 Richard Biener <rguenther@suse.de>
17173 PR tree-optimization/56366
17174 * tree-vect-loop.c (get_initial_def_for_induction): Properly
17175 handle sign-conversion of outer-loop initial induction value.
17177 2013-02-18 Richard Biener <rguenther@suse.de>
17179 PR middle-end/56349
17180 * cfghooks.c (merge_blocks): If we merge a latch into another
17181 block adjust references to it.
17182 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
17183 (verify_loop_structure): Verify that a recorded latch is in fact
17186 2013-02-18 Richard Biener <rguenther@suse.de>
17188 PR tree-optimization/56321
17189 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
17190 order SSA name release and virtual operand unlinking.
17192 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17194 * config/microblaze/microblaze.md (save_stack_block): Define.
17195 (restore_stack_block): Likewise.
17197 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17199 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
17200 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
17201 * config/microblaze/microblaze.c (microblaze_option_override):
17202 Bail out early for PIC modes when target does not support PIC.
17204 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17206 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
17207 Replace with a microblaze version.
17208 (microblaze_trampoline_init): Adapt for microblaze.
17209 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
17212 2013-02-16 Jakub Jelinek <jakub@redhat.com>
17213 Dodji Seketeli <dodji@redhat.com>
17216 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
17217 (instrument_mem_region_access): Do not forget to always put
17218 instrumentation of the of 'base' and 'base + len' in a "if (len !=
17219 0) statement, even for cases where either 'base' or 'base + len'
17220 are not instrumented -- because they have been previously
17221 instrumented. Simplify the logic by putting all the statements
17222 instrument 'base + len' inside a sequence, and then insert that
17223 sequence right before the current insertion point. Then, to
17224 instrument 'base + len', just get an iterator on that statement.
17225 And do not forget to update the pointer to iterator the function
17226 received as argument.
17228 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
17230 PR rtl-optimization/56348
17231 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
17233 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
17235 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
17236 (clean_graph_dump_file): Pass base to start_graph_dump.
17238 2013-02-14 Richard Henderson <rth@redhat.com>
17241 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
17243 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
17245 * collect2-aix.h: Define F_LOADONLY.
17247 2013-02-14 Richard Biener <rguenther@suse.de>
17250 * varasm.c (output_constant_def_1): Get the decl representing
17251 the constant as argument.
17252 (output_constant_def): Wrap output_constant_def_1.
17253 (make_decl_rtl): Use output_constant_def_1 with the decl
17254 representing the constant.
17255 (build_constant_desc): Optionally re-use a decl already
17256 representing the constant.
17257 (tree_output_constant_def): Adjust.
17259 2013-02-14 Dodji Seketeli <dodji@redhat.com>
17262 * asan.c (instrument_builtin_call): Really put the length of the
17263 second source argument into src1_len.
17265 2013-02-13 Jakub Jelinek <jakub@redhat.com>
17267 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
17268 argument. If it is false, don't create edge from then_bb to
17270 (insert_if_then_before_iter): Pass true to it.
17271 (build_check_stmt): Pass false to it.
17272 (transform_statements): Flush hash table only on extended basic
17273 block boundaries, rather than at the beginning of every bb.
17274 Don't flush hash table on nonfreeing_call_p calls.
17275 * tree-flow.h (nonfreeing_call_p): New prototype.
17276 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
17278 2013-02-13 David S. Miller <davem@davemloft.net>
17280 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
17282 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
17285 * ira.c (max_regno_before_ira): Move from ...
17287 (fix_reg_equiv_init): Use max_regno_before_ira instead of
17290 2013-02-13 Jakub Jelinek <jakub@redhat.com>
17292 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
17294 2013-02-13 Richard Biener <rguenther@suse.de>
17297 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
17298 globals in MEM_REFs.
17300 2013-02-13 Richard Biener <rguenther@suse.de>
17302 * loop-init.c (loop_optimizer_init): Clear loop state when
17303 re-initializing preserved loops.
17304 * loop-unswitch.c (unswitch_single_loop): Return whether
17305 we unswitched the loop. Do not verify loop state here.
17306 (unswitch_loops): When we unswitched a loop discover new loops.
17308 2013-02-13 Kostya Serebryany <kcc@google.com>
17310 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
17312 * sanitizer.def: Rename __asan_init to __asan_init_v1.
17314 2013-02-12 Dodji Seketeli <dodji@redhat.com>
17316 Avoid instrumenting duplicated memory access in the same basic block
17317 * Makefile.in (asan.o): Add new dependency on hash-table.h
17318 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
17319 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
17320 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
17321 (free_mem_ref_resources, has_mem_ref_been_instrumented)
17322 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
17323 (get_mem_ref_of_assignment): New functions.
17324 (get_mem_refs_of_builtin_call): Extract from
17325 instrument_builtin_call and tweak a little bit to make it fit with
17327 (instrument_builtin_call): Use the new
17328 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
17329 of is_gimple_builtin_call.
17330 (instrument_derefs, instrument_mem_region_access): Insert the
17331 instrumented memory reference into the hash table.
17332 (maybe_instrument_assignment): Renamed instrument_assignment into
17333 this, and change it to advance the iterator when instrumentation
17334 actually happened and return true in that case. This makes it
17335 homogeneous with maybe_instrument_assignment, and thus give a
17336 chance to callers to be more 'regular'.
17337 (transform_statements): Clear the memory reference hash table
17338 whenever we enter a new BB, when we cross a function call, or when
17339 we are done transforming statements. Use
17340 maybe_instrument_assignment instead of instrumentation. No more
17341 need to special case maybe_instrument_assignment and advance the
17342 iterator after calling it; it's now handled just like
17343 maybe_instrument_call. Update comment.
17345 2013-02-13 Richard Biener <rguenther@suse.de>
17347 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
17348 Fix loop discovery code.
17350 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
17352 PR inline-asm/56148
17353 * lra-constraints.c (process_alt_operands): Match early clobber
17354 operand with itself. Check conflicts with earlyclobber only if
17355 the operand is not reloaded. Prefer to reload conflicting operand
17356 if earlyclobber and matching operands are the same.
17358 2013-02-12 Richard Biener <rguenther@suse.de>
17361 * lto-streamer-out.c (write_symbol): Do not output symbols
17362 for hard register variables.
17364 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
17367 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
17368 (umulsidi3_insn, mulsidi3_insn): New insns.
17370 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
17372 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
17373 (struct tune_params): Add vec_costs field.
17374 * config/arm/arm.c (arm_builtin_vectorization_cost)
17375 (arm_add_stmt_cost): New functions.
17376 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
17377 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
17378 (arm_default_vec_cost): New struct of type cpu_vec_costs.
17379 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
17380 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
17381 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
17382 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
17384 2013-02-12 Richard Biener <rguenther@suse.de>
17387 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
17388 decls again if possible.
17390 2013-02-12 Richard Biener <rguenther@suse.de>
17392 PR middle-end/56288
17393 * tree-ssa.c (verify_ssa_name): Fix check, move
17394 SSA_NAME_IN_FREE_LIST check up.
17396 2013-02-12 Jakub Jelinek <jakub@redhat.com>
17397 Steven Bosscher <steven@gcc.gnu.org>
17399 PR rtl-optimization/56151
17400 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
17401 equal to op0 or op1, and last_insn pattern is CODE operation
17402 with MEM dest and one of the operands matches that MEM.
17404 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
17406 * doc/extend.texi: Document Function Multiversioning and "default"
17407 parameter string to target attribute.
17408 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
17409 target attribute parameter is "default".
17410 (ix86_compare_version_priority): Remove checks for target attribute.
17411 (ix86_mangle_function_version_assembler_name): Change error to sorry.
17412 Remove check for target attribute equal to NULL. Add assert.
17413 (ix86_generate_version_dispatcher_body): Change error to sorry.
17415 2013-02-11 Iain Sandoe <iain@codesourcery.com>
17416 Jack Howarth <howarth@bromo.med.uc.edu>
17417 Patrick Marlier <patrick.marlier@gmail.com>
17420 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
17421 define ENDFILE_SPEC as TM_DESTRUCTOR.
17422 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
17424 2013-02-11 Alexander Potapenko <glider@google.com>
17425 Jack Howarth <howarth@bromo.med.uc.edu>
17426 Jakub Jelinek <jakub@redhat.com>
17429 * config/darwin.c (cdtor_record): Rename ctor_record.
17430 (sort_cdtor_records): Rename sort_ctor_records.
17431 (finalize_dtors): New routine to sort destructors by
17432 priority before use in assemble_integer.
17433 (machopic_asm_out_destructor): Use finalize_dtors if needed.
17435 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
17437 PR rtl-optimization/56275
17438 * simplify-rtx.c (avoid_constant_pool_reference): Check that
17439 offset is non-negative and less than cmode size before
17440 calling simplify_subreg.
17442 2013-02-11 Richard Biener <rguenther@suse.de>
17444 PR tree-optimization/56264
17445 * cfgloop.h (fix_loop_structure): Adjust prototype.
17446 * loop-init.c (fix_loop_structure): Return the number of
17447 newly discovered loops.
17448 * tree-cfgcleanup.c (repair_loop_structures): When new loops
17449 are discovered, do a full loop-closed SSA rewrite.
17451 2013-02-11 Richard Biener <rguenther@suse.de>
17453 PR tree-optimization/56273
17454 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17456 (check_array_ref): Fix missing newline in dumps.
17457 (search_for_addr_array): Likewise.
17459 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
17461 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
17463 2013-02-09 Jakub Jelinek <jakub@redhat.com>
17466 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
17468 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
17470 PR rtl-optimization/56246
17471 * lra-constraints.c (simplify_operand_subreg): Try to reuse
17473 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
17474 constraints are satisfied.
17476 2013-02-08 Jeff Law <law@redhat.com>
17479 * emit-rtl.c (reg_is_parm_p): New function.
17480 * regs.h (reg_is_parm_p): New prototype.
17481 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
17482 callee-clobbered registers.
17484 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
17487 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
17488 If there is no implicit builtin declaration, just return NULL.
17490 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17492 * config/i386/sse.md (FMAMODEM): New mode iterator.
17493 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
17494 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
17496 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17498 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
17499 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
17500 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
17502 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17504 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
17505 (microblaze*-*-elf): Likewise.
17506 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
17508 * config/microblaze/microblaze-c.c: Add builtin defines for
17509 _LITTLE_ENDIAN and _BIG_ENDIAN.
17510 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
17511 add to TARGET_DEFAULT flags.
17512 Expand ASM_SPEC and LINK_SPEC.
17513 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
17514 * config/microblaze/microblaze.md: Update extendsidi2 and
17515 movdi_internal instructions to use low-order / high-order reg
17517 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
17518 options and inversemask / mask of LITTLE_ENDIAN.
17519 * config/microblaze/t-microblaze: Expand multilib options to
17520 include mlittle-endian (le) and update exceptions patterns.
17522 2013-02-08 Jakub Jelinek <jakub@redhat.com>
17524 PR rtl-optimization/56195
17525 * lra-constraints.c (get_reload_reg): Don't reuse regs
17526 if they have smaller mode than requested, if they have
17527 wider mode than requested, try to return a SUBREG.
17529 PR tree-optimization/56250
17530 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
17531 if type is unsigned and code isn't MULT_EXPR.
17533 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
17535 PR tree-optimization/56064
17536 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
17537 bits according to mode.
17538 * fixed-value.h (fixed_from_double_int)
17539 (const_fixed_from_double_int): Adjust comments.
17541 2013-02-08 Richard Biener <rguenther@suse.de>
17544 * lto-streamer.h (struct data_in): Remove current_file, current_line
17545 and current_col members.
17546 * lto-streamer-out.c (lto_output_location): Stream changed bits
17547 en-block for efficiency.
17548 * lto-streamer-in.c (clear_line_info): Remove.
17549 (lto_input_location): Cache current file, line and column
17550 globally via local statics. Read changed bits en-block.
17551 (input_function): Do not call clear_line_info.
17552 (lto_read_body): Likewise.
17553 (lto_input_toplevel_asms): Likewise.
17555 2013-02-08 Michael Matz <matz@suse.de>
17557 PR tree-optimization/52448
17558 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
17559 (nt_call_phase): New static.
17560 (add_or_mark_expr): Only mark accesses with newer phase than any
17562 (nonfreeing_call_p): New.
17563 (nt_init_block): Update nt_call_phase, mark blocks as visited.
17564 (nt_fini_block): Keep blocks marked as visited.
17565 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
17567 2013-02-08 Richard Biener <rguenther@suse.de>
17569 * ira.c (ira): Free broken dominator information.
17571 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
17573 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
17575 2013-02-08 Marek Polacek <polacek@redhat.com>
17577 * cfgloop.c (verify_loop_structure): Add more checking of headers.
17579 2013-02-08 Richard Biener <rguenther@suse.de>
17581 PR middle-end/56181
17582 * cfgloop.h (flow_loops_find): Adjust.
17583 (bb_loop_header_p): Declare.
17584 * cfgloop.c (bb_loop_header_p): New function split out from ...
17585 (flow_loops_find): ... here. Adjust function signature,
17586 support incremental loop structure update.
17587 (verify_loop_structure): Cleanup. Verify a loop is a loop.
17588 * cfgloopmanip.c (fix_loop_structure): Move ...
17589 * loop-init.c (fix_loop_structure): ... here.
17590 (apply_loop_flags): Split out from ...
17591 (loop_optimizer_init): ... here.
17592 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
17593 in incremental mode, only remove dead loops here.
17595 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
17598 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
17599 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
17600 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17601 (*round<mode>3.libgcc): New insns for fixed-modes.
17602 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17603 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17604 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17605 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17606 implementations. Define to __builtin_avr_absFX,
17607 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17608 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17609 __builtin_avr_countlsFX, respectively.
17610 * config/avr/avr-c.c (target.h): Include it.
17611 (enum avr_builtin_id): New enum.
17612 (avr_resolve_overloaded_builtin): New static function.
17613 (avr_register_target_pragmas): Use it to set
17614 targetm.resolve_overloaded_builtin.
17615 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17616 tree nodes used by DEF_BUILTIN.
17617 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17618 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17619 <AVR_BUILTIN_xxBITS>: Same.
17621 2013-02-08 Richard Biener <rguenther@suse.de>
17623 * cfgloop.c (verify_loop_structure): Properly handle
17624 a loop exiting to another loop header.
17625 * ira-int.h (ira_loops): Remove.
17626 * ira.c (ira_loops): Remove.
17627 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17628 (do_reload): Use loop_optimizer_finalize.
17629 * ira-build.c (create_loop_tree_nodes): Use get_loops and
17630 number_of_loops to access the loop tree.
17631 (more_one_region_p): Likewise.
17632 (finish_loop_tree_nodes): Likewise.
17633 (rebuild_regno_allocno_maps): Likewise.
17634 (mark_loops_for_removal): Likewise.
17635 (mark_all_loops_for_removal): Likewise.
17636 (remove_unnecessary_regions): Likewise.
17637 (ira_build): Likewise.
17638 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17640 2013-02-08 Richard Biener <rguenther@suse.de>
17642 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17643 * ipa-pure-const.c (analyze_function): Avoid calling
17644 mark_irreducible_loops twice.
17645 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17647 2013-02-07 David S. Miller <davem@davemloft.net>
17649 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17651 * var-tracking.c (vt_add_function_parameter): Test the presence of
17652 HAVE_window_save properly and do not remap argument registers when
17653 we have a leaf function.
17655 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
17658 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17660 * config/i386/i386.c (ix86_print_operand): Ditto.
17662 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
17664 * lra-constraints.c (process_alt_operands): Fix recently added comment.
17666 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
17668 PR rtl-optimization/56225
17669 * lra-constraints.c (process_alt_operands): Check that reload hard
17670 reg can hold value for strict_low_part.
17672 2013-02-07 Jakub Jelinek <jakub@redhat.com>
17675 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17676 dwarf2out_end_function.
17677 (in_first_function_p, maybe_at_text_label_p,
17678 first_loclabel_num_not_at_text_label): New variables.
17679 (dwarf2out_var_location): In the first function find out
17680 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17681 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17684 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
17686 PR rtl-optimization/56178
17687 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17688 SUBREG of a register. Tidy up related block of code.
17689 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17690 note if the source is a register or a SUBREG of a register.
17692 2013-02-07 Jakub Jelinek <jakub@redhat.com>
17695 * config/rs6000/rs6000.md (ptrm): New mode attr.
17696 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17697 call_value_indirect_aix<pttrsize>,
17698 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17701 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
17703 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17704 if -bnortl. Convert to strcmp and strncmp.
17706 2013-02-07 Alan Modra <amodra@gmail.com>
17709 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17710 addresses won't wrap when offsetting.
17711 (rs6000_secondary_reload): Provide secondary reloads needed for
17712 wrapping LO_SUM addresses.
17714 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
17716 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17717 MACH, just __MACH__.
17719 2013-02-06 Richard Biener <rguenther@suse.de>
17721 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17722 instead of calling fix_loop_structure.
17724 2013-02-06 Jakub Jelinek <jakub@redhat.com>
17726 PR middle-end/56217
17727 * omp-low.c (use_pointer_for_field): Return false if
17728 lower_send_shared_vars doesn't generate any copy-out code.
17730 2013-02-06 Tom de Vries <tom@codesourcery.com>
17732 PR rtl-optimization/56131
17733 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17734 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
17735 of the label is NULL. Add comment.
17737 2013-02-05 Jakub Jelinek <jakub@redhat.com>
17739 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17742 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17743 (STATIC_LIBTSAN_LIBS): Likewise.
17744 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17745 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17746 is defined, don't add anything else beyond that.
17747 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17748 (LINK_COMMAND_SPEC): Use them.
17750 PR tree-optimization/56205
17751 * tree-stdarg.c (check_all_va_list_escapes): Return true if
17752 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17753 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17755 2013-02-05 Richard Biener <rguenther@suse.de>
17757 PR tree-optimization/53342
17758 PR tree-optimization/53185
17759 * tree-vectorizer.h (vect_check_strided_load): Remove.
17760 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17761 not disallow peeling for vectorized strided loads.
17762 (vect_check_strided_load): Make static and simplify.
17763 (vect_analyze_data_refs): Adjust.
17764 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17765 correctly when vectorizing strided loads.
17767 2013-02-05 Richard Biener <rguenther@suse.de>
17769 * doc/install.texi: Refer to ISL, not PPL.
17771 2013-02-05 Jan Hubicka <jh@suse.cz>
17773 PR tree-optimization/55789
17774 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17776 2013-02-05 Jan Hubicka <jh@suse.cz>
17778 PR tree-optimization/55789
17779 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17780 the dead call anyway.
17782 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
17785 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17787 2013-02-04 Alexander Potapenko <glider@google.com>
17788 Jack Howarth <howarth@bromo.med.uc.edu>
17789 Jakub Jelinek <jakub@redhat.com>
17792 * config/darwin.c (sort_ctor_records): Stabilized qsort
17793 on constructor priority by using original position.
17794 (finalize_ctors): New routine to sort constructors by
17795 priority before use in assemble_integer.
17796 (machopic_asm_out_constructor): Use finalize_ctors if needed.
17798 2013-02-04 Jakub Jelinek <jakub@redhat.com>
17801 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17802 about visibility on artificial decls.
17803 * config/sol2.c (solaris_assemble_visibility): Likewise.
17805 2013-02-04 Kai Tietz <ktietz@redhat.com>
17808 * config/i386/i386.c (function_value_ms_64): Add additional valtype
17809 argument and improve checking of return-argument types for 16-byte
17811 (ix86_function_value_1): Add additional valtype argument on call
17812 of function_value_64.
17813 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17814 handling infunction_value_64 function.
17816 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
17818 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17820 2013-02-04 Richard Biener <rguenther@suse.de>
17822 PR tree-optimization/56188
17823 * tree-ssa-structalias.c (label_visit): Consider case with
17824 initially non-empty points-to set.
17825 (perform_var_substitution): Dump node mapping and clean up.
17827 2013-02-04 Richard Guenther <rguenther@suse.de>
17830 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17831 node prevail as last resort.
17832 (lto_symtab_merge_decls): Remove guard on LTRANS here.
17833 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17835 2013-02-04 Richard Biener <rguenther@suse.de>
17837 PR tree-optimization/56113
17838 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17840 (equiv_class_lookup_or_add): ... this.
17841 (label_visit): Adjust and fix error in previous patch.
17842 (perform_var_substitution): Adjust.
17844 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
17846 * config/sh/divtab.c: Fix formatting and comments throughout the file.
17847 * config/sh/sh4-300.md: Likewise.
17848 * config/sh/sh4a.md: Likewise.
17849 * config/sh/constraints.md: Likewise.
17850 * config/sh/sh.md: Likewise.
17851 * config/sh/netbsd-elf.h: Likewise.
17852 * config/sh/predicates.md: Likewise.
17853 * config/sh/sh-protos.h: Likewise.
17854 * config/sh/ushmedia.h: Likewise.
17855 * config/sh/linux.h: Likewise.
17856 * config/sh/sh.c: Likewise.
17857 * config/sh/superh.h: Likewise.
17858 * config/sh/elf.h: Likewise.
17859 * config/sh/sh4.md: Likewise.
17860 * config/sh/sh.h: Likewise.
17862 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
17864 * config/pa/constraints.md: Adjust unused letters. Change "T"
17865 constraint to match_test floating_point_store_memory_operand().
17866 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17867 (base14_operand): New.
17868 (floating_point_store_memory_operand): New.
17869 (integer_store_memory_operand): Revise to use base14_operand and
17870 reg_plus_base_memory_operand.
17871 (move_dest_operand): Allow symbolic_memory_operands.
17872 (symbolic_memory_operand): Check for LO_SOM.
17873 (symbolic_operand): Change default case to break.
17874 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17875 CONST_DOUBLE values to be reloaded by putting them into memory when
17876 the destination is a floating point register.
17877 (movdf): Remove code to handle CONST_DOUBLE.
17879 (reload_indf_r1): New.
17880 (reload_insf_r1): New.
17881 Consistently use "Q" and "T" constraints with integer and floating
17882 point move instructions, respectively.
17883 (movdi): Remove FAIL.
17884 Change predicate for source operand unamed DImode move from
17885 general_operand to move_src_operand.
17886 (umulsidi3): Change predicate for destination operand to
17888 Likewise for similar unamed patterns.
17889 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17890 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17891 (hppa_legitimize_address): Simplify mask calculation.
17892 (pa_emit_move_sequence): Revised handling of secondary reloads from
17893 REG+D addresses for floating point loads and stores. Directly handle
17894 loading CONST0_RTX (mode) to a floating point register.
17895 (pa_secondary_reload): Handle reloading DF and SFmode constant values
17896 to floating point registers. Don't restrict secondary reloads to
17897 floating point registers to integer modes. Revise some comments and
17899 (TARGET_LEGITIMATE_ADDRESS_P): Define.
17900 (pa_legitimate_address_p): New.
17901 (pa_legitimize_reload_address): New.
17902 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17903 (STRICT_REG_OK_FOR_BASE_P): New.
17904 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
17905 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17907 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
17908 Andrew Dixie <andrewd@gentrack.com>
17910 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17913 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
17915 * expmed.c (extract_bit_field_1): Pass the full width of the
17916 structure to get_best_reg_extraction_insn.
17918 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
17921 * configure.ac (use_cxa_atexit): Add AIX.
17922 * configure: Regenerate.
17924 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17926 2013-02-01 Jakub Jelinek <jakub@redhat.com>
17929 * final.c (need_profile_function): New variable.
17930 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17931 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17932 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17933 notes, targetm.asm_out.function_prologue doesn't emit anything,
17934 HAVE_prologue and profiler should be emitted before prologue,
17935 set need_profile_function instead of emitting it.
17936 (final_scan_insn): If need_profile_function, emit
17937 profile_function on the first NOTE_INSN_BASIC_BLOCK or
17938 NOTE_INSN_FUNCTION_BEG note.
17940 2013-02-01 Richard Henderson <rth@redhat.com>
17942 * config/rs6000/rs6000.md (smulditi3): New.
17945 * config/alpha/alpha.md (umulditi3): New.
17947 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
17949 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17950 (ASM_OUTPUT_ALIGNED_LOCAL): New.
17952 2013-02-01 Richard Biener <rguenther@suse.de>
17954 PR tree-optimization/56113
17955 * tree-ssa-structalias.c (label_visit): Reduce work for
17956 single-predecessor nodes.
17958 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
17960 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17961 range isn't testing for zero.
17963 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
17965 PR middle-end/56113
17966 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17968 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
17969 Nick Clifton <nickc@redhat.com>
17971 * config/v850/constraints.md (Q): Define as a memory constraint.
17972 * config/v850/predicates.md (label_ref_operand): New predicate.
17973 (e3v5_shift_operand): New predicate.
17974 (ior_operator): New predicate.
17975 * config/v850/t-v850: Add e3v5 multilib.
17976 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17977 (v850_gen_movdi): Prototype.
17978 * config/v850/v850.c: Add support for e3v5 architecture.
17979 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17981 (construct_save_jarl): Add e3v5 long JARL support.
17982 (v850_adjust_insn_length): New function. Adjust length of call
17983 insns when using e3v5 instructions.
17984 (v850_gen_movdi): New function: Generate instructions to move a
17986 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17987 (CPP_SPEC): Define __v850e3v5__ as appropriate.
17988 (TARGET_USE_FPU): Enable for e3v5.
17989 (CONST_OK_FOR_W): New macro.
17990 (ADJUST_INSN_LENGTH): Define.
17991 * config/v850/v850.md (UNSPEC_LOOP): Define.
17992 (attr cpu): Add v850e3v5.
17993 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17994 (movdi): New pattern.
17995 (movdi_internal): New pattern.
17996 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17997 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17998 (cstoresf4): Likewise.
17999 (cstoredf4): Likewise.
18000 (insv): New pattern.
18001 (rotlso3_a): New pattern.
18002 (rotlsi3_b): New pattern
18003 (rotlsi3_v850e3v5): New pattern.
18004 (doloop_begin): New pattern.
18005 (fix_loop_counter): New pattern.
18006 (doloop_end): New pattern.
18007 (branch_normal): Add e3v5 long branch support.
18008 (branch_invert): Likewise.
18009 (branch_z_normal): Likewise.
18010 (branch_z_invert): Likewise.
18011 (branch_nz_normal): Likewise.
18012 (branch_nz_invert): Likewise.
18013 (call_internal_short): Add e3v5 register-indirect JARL support.
18014 (call_internal_long): Likewise.
18015 (call_value_internal_short): Likewise.
18016 (call_value_internal_long): Likewise.
18017 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
18018 (mloop): New option.
18019 * config.gcc: Add support for configuring v840e3v5 target.
18020 * doc/invoke.texi: Document new v850 specific command line options.
18022 2013-01-31 Paul Koning <ni1d@arrl.net>
18026 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
18027 children if parent is a class.
18028 (prune_unused_types_prune): Don't add DW_AT_declaration.
18030 2013-01-31 Richard Biener <rguenther@suse.de>
18032 PR tree-optimization/56157
18033 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
18034 match up operand with SLP child.
18036 2013-01-31 Jason Merrill <jason@redhat.com>
18039 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
18040 parameters the first time.
18041 (gen_scheduled_generic_parms_dies): Check completeness here.
18043 2013-01-31 Richard Biener <rguenther@suse.de>
18045 PR middle-end/53073
18046 * common.opt (faggressive-loop-optimizations): New flag,
18047 enabled by default.
18048 * doc/invoke.texi (faggressive-loop-optimizations): Document.
18049 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
18050 infer_loop_bounds_from_undefined by it.
18052 2013-01-31 Richard Biener <rguenther@suse.de>
18054 PR tree-optimization/56150
18055 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
18056 visit virtual operands.
18057 (find_uses_to_rename_bb): Likewise.
18059 2013-01-31 Richard Biener <rguenther@suse.de>
18061 PR tree-optimization/56150
18062 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
18063 mixed store non-store stmts.
18065 2013-01-30 Jakub Jelinek <jakub@redhat.com>
18068 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
18069 LIBASAN_EARLY_SPEC is defined.
18070 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
18071 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
18073 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
18076 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
18077 invalid args instead of ICEing on it.
18078 (ix86_valid_target_attribute_tree): Return error_mark_node if
18079 ix86_valid_target_attribute_inner_p failed.
18080 (ix86_valid_target_attribute_p): Return false only if
18081 ix86_valid_target_attribute_tree returned error_mark_node. Allow
18082 target("default") attribute.
18083 (sorted_attr_string): Change argument from const char * to tree,
18084 merge in all target attribute arguments rather than just one.
18085 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
18086 instead of free. Avoid using strcat.
18087 (ix86_mangle_function_version_assembler_name): Mangle
18088 target("default") as if no target attribute is present. Adjust
18089 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
18090 instead of xmalloc and XDELETEVEC instead of free.
18091 (ix86_function_versions): Don't return true if one of the decls
18092 doesn't have target attribute. If they don't and one of the decls
18093 is DECL_FUNCTION_VERSIONED, report an error. Adjust
18094 sorted_attr_string caller. Use XDELETEVEC instead of free.
18095 (ix86_supports_function_versions): Remove.
18096 (make_name): Fix up formatting.
18097 (make_dispatcher_decl): Remove resolver_name and its initialization.
18098 Avoid leaking memory.
18099 (is_function_default_version): Return true if there is
18100 target("default") attribute rather than no target attribute at all.
18101 (make_resolver_func): Avoid leaking memory.
18102 (ix86_generate_version_dispatcher_body): Likewise.
18103 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
18104 * target.def (supports_function_versions): Remove.
18105 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
18106 * doc/tm.texi: Regenerated.
18108 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
18110 PR rtl-optimization/56144
18111 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
18112 for values with side effects.
18114 2013-01-30 Richard Biener <rguenther@suse.de>
18116 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
18117 (sparseset_pop): Likewise.
18118 * cfganal.c (compute_idf): Likewise. Increase work-stack size
18119 to be able to use quick_push in the worker loop.
18121 2013-01-30 Marek Polacek <polacek@redhat.com>
18123 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
18125 2013-01-30 Richard Biener <rguenther@suse.de>
18128 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
18130 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
18132 PR tree-optimization/56064
18133 * fixed-value.c (fixed_from_double_int): New function.
18134 * fixed-value.h (fixed_from_double_int): New prototype.
18135 (const_fixed_from_double_int): New static inline function.
18136 * fold-const.c (native_interpret_fixed): New static function.
18137 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
18138 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
18139 (native_encode_fixed): New static function.
18140 (native_encode_expr) <FIXED_CST>: Use it.
18141 (native_interpret_int): Move double_int worker code to...
18142 * double-int.c (double_int::from_buffer): ...this new static method.
18143 * double-int.h (double_int::from_buffer): Prototype it.
18145 2013-01-30 Richard Biener <rguenther@suse.de>
18147 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
18148 New pointer-map and obstack.
18149 (init_alias_vars): Allocate pointer-map and obstack.
18150 (delete_points_to_sets): Free them.
18151 (find_what_var_points_to): Cache result.
18152 (find_what_p_points_to): Adjust for changed interface of
18153 find_what_var_points_to.
18154 (compute_points_to_sets): Likewise.
18155 (ipa_pta_execute): Likewise.
18157 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18159 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
18160 * configure: Regenerate.
18161 * config.in: Regenerate.
18162 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
18163 #nobits/#progbits if supported.
18165 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
18168 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
18169 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
18170 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
18172 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18174 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
18175 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
18177 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18179 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
18180 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
18182 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18184 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
18186 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
18187 * config/arm/cortex-a7.md: New bypasses using
18188 arm_mac_accumulator_is_result.
18190 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18192 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
18193 (cortex_a7_neon_mla): Likewise.
18194 (cortex_a7_fpfmad): New reservation.
18195 (cortex_a7_fpmacs): Use ffmas and update required units.
18196 (cortex_a7_fpmuld): Update required units and latency.
18197 (cortex_a7_fpmacd): Likewise.
18198 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
18199 (cortex_a7_neon). Likewise.
18200 (bypass) Update participating units.
18202 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
18204 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
18205 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
18207 * config/arm/vfp11.md (vfp_farith): Use ffmas.
18208 (vfp_fmul): Use ffmad.
18209 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
18210 (cortex_r4_fmacd): Use ffmad.
18211 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
18212 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
18213 (cortex_a9_fmacd): Use ffmad.
18214 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
18215 (cortex_a8_vfp_macd): Use ffmad.
18216 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
18217 (cortex_a5_fpmacd): Use ffmad.
18218 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
18219 (cortex_a15_vfp_macd): Use ffmad.
18220 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
18222 2013-01-29 Jason Merrill <jason@redhat.com>
18225 * varasm.c (default_assemble_visibility): Don't warn about
18226 visibility on artificial decls.
18228 2013-01-29 Richard Biener <rguenther@suse.de>
18230 PR tree-optimization/56113
18231 * tree-ssa-structalias.c (equiv_class_lookup): Also return
18233 (label_visit): Free duplicate bitmaps and record the leader instead.
18234 (perform_var_substitution): Adjust.
18236 2013-01-29 Richard Biener <rguenther@suse.de>
18238 PR tree-optimization/55270
18239 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
18240 the CFG, schedule loops for fixup.
18242 2013-01-29 Nick Clifton <nickc@redhat.com>
18244 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
18247 2013-01-28 Leif Ekblad <leif@rdos.net>
18249 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
18250 * config/i386/i386.h (TARGET_RDOS): New macro.
18251 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
18252 * config/i386/i386.c (ix86_option_override_internal): For 64bit
18253 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
18254 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
18255 DEFAULT_LARGE_SECTION_THRESHOLD.
18256 * config/i386/i386.md (R14_REG, R15_REG): New constants.
18257 * config/i386/rdos.h: New file.
18258 * config/i386/rdos64.h: New file.
18260 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
18263 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
18266 2013-01-28 Jakub Jelinek <jakub@redhat.com>
18268 PR rtl-optimization/56117
18269 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
18270 call cselib_lookup_from_insn on the MEM before calling
18271 add_insn_mem_dependence.
18273 2013-01-28 Richard Biener <rguenther@suse.de>
18275 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
18276 to a stmt that didn't have one.
18277 (copy_phis_for_bb): Likewise for PHI arguments.
18278 (copy_debug_stmt): Likewise for debug stmts.
18280 2013-01-28 Richard Biener <rguenther@suse.de>
18282 PR tree-optimization/56034
18283 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
18284 (partition_builtin_p): Adjust.
18285 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
18286 it is the last partition.
18287 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
18288 up the vertex for the definition.
18289 (classify_partition): Classify whether a partition is a
18290 PKIND_REDUCTION, thus has uses outside of the loop.
18291 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
18292 Merge all PKIND_REDUCTION partitions into the last partition.
18293 (tree_loop_distribution): Seed partitions from reductions as well.
18295 2013-01-28 Jakub Jelinek <jakub@redhat.com>
18297 PR tree-optimization/56125
18298 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
18299 pow(x,c) into sqrt(x) * powi(x, n/2) or
18300 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
18301 optimizing for size.
18302 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
18303 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
18306 PR tree-optimization/56094
18307 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
18308 to UNKNOWN_LOCATION while gimplifying expr.
18310 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
18313 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
18314 operand 0 in movabs insn template for -masm=intel asm alternative.
18315 (*movabs<mode>_2): Ditto for operand 1.
18317 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
18320 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
18323 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
18325 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
18328 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
18330 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
18331 Undef to avoid warning.
18333 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
18335 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
18336 * configure: Regenerate.
18338 2013-01-25 Jakub Jelinek <jakub@redhat.com>
18340 PR tree-optimization/56098
18341 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
18342 for stmts with volatile ops.
18343 (cond_store_replacement): Don't optimize if assign has volatile ops.
18344 (cond_if_else_store_replacement_1): Don't optimize if either
18345 then_assign or else_assign have volatile ops.
18346 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
18349 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18351 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
18353 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18355 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
18356 missing ':' in asm example.
18358 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
18360 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
18361 entries into lane and laneq entries.
18362 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
18363 Remove AdvSIMD scalar modes.
18364 (aarch64_sq<r>dmulh_laneq<mode>): New.
18365 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
18367 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
18368 builtin implementations to relfect changes in RTL in aarch64-simd.md.
18369 * config/aarch64/iterators.md (VCOND): New.
18372 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
18375 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
18376 Add NULL LIBNAME argument to existing definitions.
18377 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
18378 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
18379 * config/avr/avr.c (DEF_BUILTIN): Same.
18380 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
18381 (avr_expand_builtin): Expand to a vanilla call if a libgcc
18382 implementation is available (DECL_ASSEMBLER_NAME is set).
18383 (avr_fold_absfx): New static function.
18384 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
18385 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
18386 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
18387 AVR_BUILTIN_ABSLLK.
18388 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
18389 (abshk, absk, abslk, absllk): Provide as static inline functions.
18391 2013-01-25 Marek Polacek <polacek@redhat.com>
18393 PR tree-optimization/56035
18394 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
18396 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
18398 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
18399 (*movtf_internal_rex64): Add (!o,C) alternative
18400 (*movxf_internal_rex64): Ditto.
18401 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
18403 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
18405 * doc/invoke.texi: fix typo.
18406 * doc/objc.texi: fix typo.
18408 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
18410 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
18411 for the first two alternatives.
18413 2013-01-24 Diego Novillo <dnovillo@google.com>
18415 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
18416 (ggc-zone.o): Remove.
18417 * configure.ac: Remove option --with-gc.
18418 * configure: Re-generate.
18419 * doc/install.texi: Remove documentation for --with-gc.
18420 * gengtype.c (write_enum_defn): Remove. Update all users.
18421 (write_Types_process_field): Remove generation of gt_e_* argument.
18422 (output_type_enum): Remove. Update all users.
18423 (write_enum_defn): Remove. Update all users.
18424 (enum alloc_zone): Remove. Update all users.
18425 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
18426 * ggc-common.c (ggc_splay_alloc): Remove first argument.
18427 Update all callers.
18428 (struct ptr_data): Remove field TYPE. Update all users.
18429 (gt_pch_note_object): Remove argument TYPE. Update all users.
18430 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
18432 * ggc-none.c (ggc_alloc_typed_stat): Remove.
18433 (struct alloc_zone): Remove.
18434 (ggc_internal_alloc_zone_stat): Remove.
18435 (ggc_internal_cleared_alloc_zone_stat): Remove.
18436 * ggc-page.c (ggc_alloc_typed_stat): Remove.
18437 (ggc_pch_count_object): Remove last argument. Update all users.
18438 (ggc_pch_alloc_object): Remove last argument. Update all users.
18439 (struct alloc_zone): Remove.
18440 * ggc-zone.c: Remove.
18441 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
18442 (struct alloc_zone): Remove.
18443 (ggc_alloc_typed_stat): Remove.
18444 (ggc_alloc_typed): Remove.
18445 (ggc_splay_alloc): Remove first argument.
18446 (rtl_zone): Remove. Update all users.
18447 (tree_zone): Remove. Update all users.
18448 (tree_id_zone): Remove. Update all users.
18449 (ggc_internal_zone_alloc_stat): Remove. Update all users.
18450 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
18451 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
18452 * tree-ssanames.c: Remove references to zone allocator in comments.
18454 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
18456 * config/avr/avr.c (avr_out_fract): Make register numbers that
18457 might be outside of source operand signed.
18459 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
18461 * config/i386/constraints.md (Yf): New constraint.
18462 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
18463 of f constraint to conditionaly disable x87 register preferences.
18464 (*movdf_internal): Ditto.
18465 (*movsf_internal): Ditto.
18467 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
18469 PR inline-asm/55934
18470 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
18471 that have operands with impossible constraints.
18472 Add a FIXME for a speed-up opportunity.
18473 * lra-constraints.c (process_alt_operands): Verify that a class
18474 selected from constraints on asms is valid for the operand mode.
18475 (curr_insn_transform): Remove incorrect comment.
18477 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
18479 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
18480 TOC operand is a valid symbol ref in the constant pool.
18482 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
18484 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
18486 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
18489 * config/avr/stdfix.h: New file.
18490 * t-avr (stdfix-gcc.h): New rule to build it.
18491 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
18493 2013-01-23 Kostya Serebryany <kcc@google.com>
18495 * config/darwin.h: remove dependency on
18496 CoreFoundation (asan on Mac OS).
18498 2013-01-23 Jakub Jelinek <jakub@redhat.com>
18501 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
18502 instead of cmpdi_operand for first comparison operand.
18503 Don't assert that comparison operands aren't both constants.
18505 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
18507 * doc/install.texi (Downloading the Source): Update references to
18508 downloading separate components.
18510 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
18512 * doc/extend.texi (__int128): Improve grammar.
18514 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
18517 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
18518 alternative to (o,r).
18519 (*movdi_internal_rex64): Remove (!o,n) alternative.
18520 (DImode immediate->memory splitter): Remove.
18521 (DImode immediate->memory peephole2): Remove.
18522 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
18523 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
18524 alternative to (!o,*r).
18525 (*movtf_internal_sse): New pattern.
18526 (*movxf_internal_rex64): New pattern.
18527 (*movxf_internal): Disable for TARGET_64BIT.
18528 (*movdf_internal_rex64): Remove (!o,F) alternative.
18530 2013-01-22 Jakub Jelinek <jakub@redhat.com>
18532 PR middle-end/56074
18533 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
18534 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
18535 * tree-vect-loop-manip.c (find_loop_location): Also ignore
18536 stmt locations where LOCATION_LOCUS of the stmt location is
18537 UNKNOWN_LOCATION or BUILTINS_LOCATION.
18540 * config/i386/i386.md (UNSPEC_STOS): New.
18541 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
18542 *strsetqi_1): Add UNSPEC_STOS.
18544 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
18547 * doc/invoke.texi: Remove left over -Wsynth example.
18549 2013-01-21 Jakub Jelinek <jakub@redhat.com>
18551 PR tree-optimization/56051
18552 * fold-const.c (fold_binary_loc): Don't fold
18553 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
18554 a narrowing conversion, or widening conversion from signed
18557 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
18559 PR rtl-optimization/56023
18560 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
18561 dependent on debug instruction.
18563 2013-01-21 Martin Jambor <mjambor@suse.cz>
18565 PR middle-end/56022
18566 * function.c (allocate_struct_function): Call
18567 invoke_set_current_function_hook earlier.
18569 2013-01-21 Jakub Jelinek <jakub@redhat.com>
18571 * reload1.c (init_reload): Only initialize reload_obstack
18572 during the first call.
18574 2013-01-21 Marek Polacek <polacek@redhat.com>
18576 * cfgloop.c (verify_loop_structure): Fix up grammar.
18578 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
18580 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
18581 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
18583 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18586 * config/arm/marvell-pj4.md: Update copyright year.
18587 Fix up use of alu to alu_reg and simple_alu_imm.
18589 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
18591 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
18593 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
18596 * lra-constraints.c (curr_insn_transform): Don't reuse original
18597 insn for secondary memory move when memory mode should be different.
18599 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18601 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18602 atomic_storedi_1): New patterns.
18604 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
18606 btver2 pipeline descriptions.
18607 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18609 * config/i386/i386.md (btver2_decode): New type attributes.
18610 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18612 * config/i386/btver2.md: New file describing btver2 pipelines.
18614 2013-01-19 Andrew Pinski <apinski@cavium.com>
18616 PR tree-optimization/52631
18617 * tree-ssa-sccvn (visit_use): Before looking up the original
18618 statement, try looking up the simplified expression.
18620 2013-01-19 Anthony Green <green@moxielogic.com>
18622 * config/moxie/moxie.c (moxie_expand_prologue): Set
18623 current_function_static_stack_size.
18625 2013-01-18 Jakub Jelinek <jakub@redhat.com>
18627 PR tree-optimization/56029
18628 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18629 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18631 2013-01-18 Sharad Singhai <singhai@google.com>
18633 PR tree-optimization/55995
18634 * dumpfile.c (dump_loc): Print location only if available.
18635 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18637 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
18640 * lra-constraints.c (curr_insn_transform): Reuse original insn for
18641 secondary memory move.
18642 (inherit_reload_reg): Use rclass instead of cl for
18643 check_secondary_memory_needed_p.
18645 2013-01-18 Jakub Jelinek <jakub@redhat.com>
18647 PR middle-end/56015
18648 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18649 the case where writing real complex part of target modifies op1.
18651 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
18653 * config/aarch64/aarch64-simd.md
18654 (aarch64_vcond_internal<mode>): Handle unordered cases.
18655 * config/aarch64/iterators.md (v_cmp_result): New.
18657 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
18658 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18660 * config/arm/marvell-pj4.md: New file.
18661 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18662 * config/arm/arm.md (generic_sched): Add marvell_pj4.
18663 (generic_vfp): Likewise.
18664 * config/arm/arm-cores.def: Add marvell-pj4.
18665 * config/arm/arm-tune.md: Regenerate.
18666 * config/arm/arm-tables.opt: Regenerate.
18667 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18668 * doc/invoke.texi: Document marvell-pj4.
18670 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
18672 * config/aarch64/arm_neon.h: Map scalar types to standard types.
18674 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18679 * var-tracking.c (negative_power_of_two_p): New.
18680 (global_get_addr_cache, local_get_addr_cache): New.
18681 (get_addr_from_global_cache, get_addr_from_local_cache): New.
18682 (vt_canonicalize_addr): Rewrite using the above. Adjust the
18684 (vt_stack_offset_p): Remove.
18685 (vt_canon_true_dep): Always canonicalize loc's address.
18686 (clobber_overlapping_mems): Make sure we have a MEM.
18687 (local_get_addr_clear_given_value): New.
18688 (val_reset): Clear local cached entries.
18689 (compute_bb_dataflow): Create and release the local cache.
18690 Disable duplicate MEMs clobbering.
18691 (emit_notes_in_bb): Clobber MEMs likewise.
18692 (vt_emit_notes): Create and release the local cache.
18693 (vt_initialize, vt_finalize): Create and release the global
18694 cache, respectively.
18695 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18697 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18699 PR libmudflap/53359
18700 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18701 not found in the symtab.
18703 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
18706 PR rtl-optimization/55547
18707 PR rtl-optimization/53827
18710 * alias.c (offset_overlap_p): New, factored out of...
18711 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
18712 the conservative special case for symbolic constants. Don't
18713 adjust zero sizes on alignment.
18715 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
18717 PR rtl-optimization/52573
18718 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18719 REG_UNUSED for the same register.
18721 2013-01-17 Richard Biener <rguenther@suse.de>
18722 Marek Polacek <polacek@redhat.com>
18724 PR rtl-optimization/55833
18725 * loop-unswitch.c (unswitch_loops): Move loop verification...
18726 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
18727 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18728 Set it to true when we're removing a loop from hierarchy tree in
18729 an irreducible region.
18730 (fix_bb_placements): Adjust caller.
18731 (fix_loop_placements): Likewise.
18733 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
18735 * config/avr/builtins.def (DEF_BUILTIN): Factor out
18736 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18737 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18738 Remove ID. Adjust comments.
18739 * config/avr/avr-c.c (avr_builtin_name): Remove.
18740 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18741 * config/avr/avr.c (avr_tolower): New static function.
18742 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
18743 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18744 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18747 2013-01-17 Jan Hubicka <jh@suse.cz>
18749 PR tree-optimization/55273
18750 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18752 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
18755 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18756 store through atomic_store<mode>_1.
18757 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18759 2013-01-17 Martin Jambor <mjambor@suse.cz>
18761 PR tree-optimizations/55264
18762 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18763 for virtual methods.
18764 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18765 virtual methods before inlining is over.
18766 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18768 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18771 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
18773 PR rtl-optimization/56005
18774 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18775 pending reads for prefetch.
18777 2013-01-16 Ian Bolton <ian.bolton@arm.com>
18779 * config/aarch64/aarch64.md
18780 (*cstoresi_neg_uxtw): New pattern.
18781 (*cmovsi_insn_uxtw): New pattern.
18782 (*<optab>si3_uxtw): New pattern.
18783 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18784 (*<optab>si3_insn_uxtw): New pattern.
18785 (*bswapsi2_uxtw): New pattern.
18787 2013-01-16 Richard Biener <rguenther@suse.de>
18789 * tree-inline.c (tree_function_versioning): Remove set but
18790 never used variable.
18792 2013-01-16 Richard Biener <rguenther@suse.de>
18794 PR tree-optimization/55964
18795 * tree-flow.h (rename_variables_in_loop): Remove.
18796 (rename_variables_in_bb): Likewise.
18797 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18798 (copy_loop_before): Adjust and delete update-ssa status.
18799 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18800 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
18801 (rename_variables_in_loop): Remove.
18802 (slpeel_update_phis_for_duplicate_loop): Likewise.
18803 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18804 use available cfg machinery instead of duplicating it.
18805 Update PHI nodes and perform poor-mans SSA update here.
18806 (slpeel_tree_peel_loop_to_edge): Adjust.
18808 2013-01-16 Richard Biener <rguenther@suse.de>
18810 PR tree-optimization/54767
18811 PR tree-optimization/53465
18812 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18813 (vrp_visit_phi_node): For PHI arguments coming via backedges
18814 drop all symbolical range information.
18815 (execute_vrp): Compute backedges.
18817 2013-01-16 Richard Biener <rguenther@suse.de>
18819 * doc/install.texi: Update CLooG and ISL requirements to
18822 2013-01-16 Christian Bruel <christian.bruel@st.com>
18825 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18826 (broken_move): Handle UNSPECV_SP_SWITCH_B.
18827 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18829 2013-01-16 DJ Delorie <dj@redhat.com>
18831 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18832 (UNSPECV_SP_SWITCH_E): New.
18833 (sp_switch_1): Change to an unspec.
18834 (sp_switch_2): Change to an unspec. Don't use post-inc when we
18837 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
18839 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18840 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18841 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18842 (expand_mem_thread_fence): Ditto.
18843 (expand_mem_signal_fence): Ditto.
18844 (expand_atomic_load): Ditto.
18845 (expand_atomic_store): Ditto.
18847 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
18849 PR rtl-optimization/55547
18850 PR rtl-optimization/53827
18853 * alias.c (memrefs_conflict_p): Set sizes to negative after
18856 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18859 * function.c (thread_prologue_and_epilogue_insns): Always
18860 add crtl->drap_reg to set_up_by_prologue.set, even if
18861 stack_realign_drap is false.
18863 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
18865 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18866 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18867 *call): Fix indention.
18869 2013-01-15 Tom de Vries <tom@codesourcery.com>
18872 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18875 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
18877 PR rtl-optimization/55153
18878 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18880 2013-01-15 Martin Jambor <mjambor@suse.cz>
18882 PR tree-optimization/55920
18883 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18884 accesses as grp_to_be_debug_replaced.
18886 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18888 PR tree-optimization/55920
18889 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18890 there is non-useless type conversion needed from debug rhs to lhs,
18891 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18893 2013-01-15 Joseph Myers <joseph@codesourcery.com>
18894 Mikael Pettersson <mikpe@it.uu.se>
18897 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18899 (ASM_OUTPUT_CASE_LABEL): Remove.
18900 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18901 * final.c (shorten_branches): Update alignment of labels before
18902 jump tables if CASE_VECTOR_SHORTEN_MODE.
18904 2013-01-15 Richard Biener <rguenther@suse.de>
18907 * system.h: Do not include gmp.h for building host tools.
18909 2013-01-15 Richard Biener <rguenther@suse.de>
18911 PR middle-end/55882
18912 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18913 account for bitpos when computing alignment.
18915 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
18917 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18918 (ix86_target_macros_internal): Likewise.
18920 * config/i386/i386.c (m_CORE2I7): Removed.
18921 (m_CORE_HASWELL): New macro.
18922 (m_CORE_ALL): Likewise.
18923 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18924 (initial_ix86_arch_features): Likewise.
18925 (processor_target_table): Initializations for Core avx2.
18926 (cpu_names): New names "core-avx2".
18927 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18928 PROCESSOR_CORE_HASWELL.
18929 (ix86_issue_rate): New case.
18930 (ia32_multipass_dfa_lookahead): Likewise.
18931 (ix86_sched_init_global): Likewise.
18933 * config/i386/i386.h (TARGET_HASWELL): New macro.
18934 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18935 (processor_type): New PROCESSOR_HASWELL.
18937 2013-01-15 Jakub Jelinek <jakub@redhat.com>
18939 PR tree-optimization/55955
18940 * tree-vect-loop.c (vectorizable_reduction): Give up early on
18941 *SHIFT_EXPR and *ROTATE_EXPR codes.
18943 PR tree-optimization/48766
18944 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18945 -ftrapv disable -fwrapv.
18947 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18950 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18951 etc. to 1 and not to __flash.
18952 Use LL suffix for __INT24_MAX__ with -mint8.
18953 Use ULL suffix for __UINT24_MAX__ with -mint8.
18955 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18957 * config/avr/avr-arch.h
18958 (struct base_arch_s): Use typedef avr_arch_t instead.
18959 (struct arch_info_s): Use typedef avr_arch_info_t instead.
18960 (struct mcu_type_s): Use typedef avr_mcu_t instead.
18961 * config/avr/avr.c: Same.
18962 * config/avr/avr-devices.c: Same.
18963 * config/avr/driver-avr.c: Same.
18964 * config/avr/gen-avr-mmcu-texi.c: Same.
18965 * config/avr/avr-mcus.def: Adjust comment.
18967 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
18969 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18970 * config/aarch64/iterators.md (VALLDI): New.
18972 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
18973 Andi Kleen <ak@linux.intel.com>
18976 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18977 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18980 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
18982 * config/avr/avr-stdint.h: Remove trailing blanks.
18983 * config/avr/avr-log.h: Same.
18984 * config/avr/avr-arch.h: Same.
18985 * config/avr/avr-devices.c: Same.
18986 * config/avr/avr-dimode.md: Same.
18987 * config/avr/predicates.md: Same.
18988 * config/avr/avr-c.c: Same. And fix typo.
18990 * config/avr/avr-protos.h: Same. And:
18991 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18992 (init_cumulative_args): Rename to avr_init_cumulative_args.
18993 (expand_prologue): Rename to avr_expand_prologue.
18994 (expand_epilogue): Rename to avr_expand_epilogue.
18995 (adjust_insn_length): Rename to avr_adjust_insn_length.
18996 (notice_update_cc): Rename to avr_notice_update_cc.
18997 (final_prescan_insn): Rename to avr_final_prescan_insn.
18998 * config/avr/avr.c: Same.
18999 * config/avr/avr.h: Same.
19000 * config/avr/avr.md: Remove trailing blanks.
19001 (prologue): Use avr_expand_prologue.
19002 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
19004 2013-01-14 Richard Biener <rguenther@suse.de>
19006 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
19007 verify_location, collect_subblocks): New functions.
19008 (verify_gimple_in_cfg): Verify that locations only reference
19009 BLOCKs in the functions BLOCK tree.
19011 2013-01-14 Richard Biener <rguenther@suse.de>
19013 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
19015 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
19017 (insert_out_of_ssa_copy_on_edge): Likewise.
19018 (rewrite_close_phi_out_of_ssa): Likewise.
19019 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
19021 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
19022 propagated constants.
19023 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
19026 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
19028 * config/avr/avr-modes.def: Add GPL copyright notice.
19030 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
19032 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
19033 MEMMODEL_MASK to determine memory model.
19034 (atomic_store<mode>): Ditto from operands[2].
19035 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
19037 2013-01-13 Jakub Jelinek <jakub@redhat.com>
19040 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
19041 (fold_gimple_assign): Don't call unshare_expr here.
19042 (fold_ctor_reference): Call unshare_expr.
19044 2013-01-13 Terry Guo <terry.guo@arm.com>
19046 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
19047 * doc/fragments.texi: Document MULTILIB_REUSE.
19048 * gcc.c (multilib_reuse): New internal spec.
19049 (set_multilib_dir): Also search multilib from multilib_reuse.
19050 * genmultilib (tmpmultilib3): Refactor code.
19051 (tmpmultilib4): Ditto.
19052 (multilib_reuse): New multilib argument.
19054 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
19056 * Makefile.in: Update copyright.
19058 2013-01-12 Tom de Vries <tom@codesourcery.com>
19060 PR middle-end/55890
19061 * calls.c (expand_call): Check if arg_nr is valid.
19063 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19065 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
19066 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
19067 documentation. Add missing '__' in front of
19068 __builtin_ia32_packssdw256.
19070 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
19073 * config/s390/s390.c (s390_preferred_reload_class): Do not return
19074 NO_REGS for larl operands.
19075 (s390_reload_larl_operand): Use s390_load_address instead of
19078 2013-01-11 Richard Biener <rguenther@suse.de>
19080 * tree-cfg.c (verify_node_sharing_1): Split out from ...
19081 (verify_node_sharing): ... here.
19082 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
19084 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
19086 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
19087 Substitute TREECHECKING.
19088 * configure: Regenerate.
19089 * Makefile.in (TREECHECKING): New.
19091 2013-01-11 Richard Guenther <rguenther@suse.de>
19093 PR tree-optimization/44061
19094 * tree-vrp.c (extract_range_basic): Compute zero as
19095 value-range for __builtin_constant_p of function parameters.
19097 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
19099 Update copyright years.
19101 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
19103 PR rtl-optimization/55672
19104 * lra-eliminations.c (mark_not_eliminable): Permit addition with
19105 const to be eliminable.
19107 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
19109 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
19110 * configure: Regenerate.
19112 2013-01-10 Richard Biener <rguenther@suse.de>
19114 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
19116 2013-01-10 Richard Biener <rguenther@suse.de>
19119 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
19120 locations for virtual PHI arguments.
19121 (rewrite_update_phi_arguments): Likewise.
19123 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
19125 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
19128 2013-01-10 Jakub Jelinek <jakub@redhat.com>
19130 PR tree-optimization/55921
19131 * tree-complex.c (expand_complex_asm): New function.
19132 (expand_complex_operations_1): Call it for GIMPLE_ASM.
19134 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
19137 * config/s390/s390.c (s390_symref_operand_p)
19138 (s390_loadrelative_operand_p): Merge the two functions.
19139 (s390_check_qrst_address, print_operand_address): Add parameters
19140 to s390_loadrelative_operand_p invokation.
19141 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
19142 (s390_reload_larl_operand, s390_secondary_reload): Use
19143 s390_loadrelative_operand_p instead of s390_symref_operand_p.
19144 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
19146 2013-01-09 Mike Stump <mikestump@comcast.net>
19148 * dse.c (record_store): Remove unnecessary assert.
19150 2013-01-09 Jan Hubicka <jh@suse.cz>
19152 PR tree-optimization/55569
19153 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
19154 * cfgloop.h (scale_loop_profile): Likewise.
19156 2013-01-09 Jan Hubicka <jh@suse.cz>
19159 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
19161 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
19163 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
19165 PR middle-end/55114
19166 * expr.h (maybe_emit_group_store): Declare.
19167 * expr.c (maybe_emit_group_store): New function.
19168 * builtins.c (expand_builtin_int_roundingfn): Call it.
19169 (expand_builtin_int_roundingfn_2): Likewise.
19171 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
19173 PR rtl-optimization/55829
19174 * lra-constraints.c (match_reload): Add code for absent output.
19175 (curr_insn_transform): Add code for reloads of matched inputs
19178 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
19180 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
19181 attribute of movddup insn to DF.
19182 (*vec_interleave_lowv2df): Ditto.
19183 (vec_dupv2df): Ditto.
19185 2013-01-09 Jan Hubicka <jh@suse.cz>
19187 PR tree-optimiation/55875
19188 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
19189 EVERY_ITERATION parameter.
19190 (number_of_iterations_exit): Check if exit is executed every iteration.
19191 (idx_infer_loop_bounds): Similarly here.
19192 (n_of_executions_at_most): Simplify
19193 to only test for cases where statement is dominated by the
19194 particular bound; handle correctly the "postdominance" test.
19195 (scev_probably_wraps_p): Use max loop iterations info
19196 as a global bound first.
19198 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
19199 Nick Clifton <nickc@redhat.com>
19201 * config/v850/v850.md (cbranchsf4): New pattern.
19202 (cstoresf4): New pattern.
19203 (cbranchdf4): New pattern.
19204 (cstoredf4): New pattern.
19205 (movsicc): Disallow floating point comparisons.
19206 (cmpsf_le_insn): Fix order of operators.
19207 (cmpsf_lt_insn): Likewise.
19208 (cmpsf_eq_insn): Likewise.
19209 (cmpdf_le_insn): Likewise.
19210 (cmpdf_lt_insn): Likewise.
19211 (cmpdf_eq_insn): Likewise.
19212 (cmpsf_ge_insn): Use LE comparison.
19213 (cmpdf_ge_insn): Likewise.
19214 (cmpsf_gt_insn): Use LT comparison.
19215 (cmpdf_gt_insn): Likewise.
19216 (cmpsf_ne_insn): Delete pattern.
19217 (cmpdf_ne_insn): Delete pattern.
19218 * config/v850/v850.c (v850_gen_float_compare): Use
19219 gen_cmpdf_eq_insn for NE comparison.
19220 (v850_float_z_comparison_operator)
19221 (v850_float_nz_comparison_operator): Move from here ...
19222 * config/v850/predicates.md: ... to here. Move GT and GE
19223 comparisons into v850_float_z_comparison_operator.
19224 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
19226 (v850_float_nz_comparison_operator): Likewise.
19228 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19230 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
19231 with calls to gen_insvsi/gen_insvdi.
19233 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
19235 * config/i386/i386.c (initial_ix86_tune_features): Set up
19236 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
19238 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
19239 Jakub Jelinek <jakub@redhat.com>
19241 PR tree-optimization/48189
19242 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
19243 If nitercst is 0, don't predict the exit edge.
19245 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
19247 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
19248 in asm_fprintf with reg_names.
19249 (aarch64_print_operand_address): Likewise.
19250 (aarch64_return_addr): Likewise.
19251 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
19253 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19255 * config/pa/pa.h (VAL_U6_BITS_P): Define.
19256 (INT_U6_BITS): Likewise.
19257 * config/pa/predicates.md (uint6_operand): New predicate.
19258 (shift5_operand, shift6_operand): Likewise.
19259 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
19261 (lshrdi3): Use shift6_operand.
19262 (shrpsi4, shrpdi4): New insn patterns.
19263 (extzv): Delete expander.
19264 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
19265 predicates in unamed zero extract patterns. Tighten common constraint.
19266 (extv): Delete expander.
19267 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
19268 predicates in unamed sign extract patterns. Tighten common constraint.
19269 (insv): Delete expander.
19270 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
19271 predicates in unamed insert patterns. Tighten common constraint.
19272 Change uint32_operand predicate to uint6_operand predicate in unamed
19273 DImode pattern to insert constant values of type 1...1xxxx.
19275 2013-01-04 Jan Hubicka <jh@suse.cz>
19277 PR tree-optimization/55823
19278 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
19281 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19282 Uros Bizjak <ubizjak@gmail.com>
19284 PR rtl-optimization/55845
19285 * df-problems.c (can_move_insns_across): Stop scanning at
19286 volatile_insn_p source instruction or give up if
19287 across_from .. across_to range contains any volatile_insn_p
19290 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
19292 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
19293 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
19295 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
19296 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
19298 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19301 * asan.c (asan_clear_shadow): New function.
19302 (asan_emit_stack_protection): Use it.
19304 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
19306 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
19307 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
19308 with tab instead of space.
19310 2013-01-08 Nick Clifton <nickc@redhat.com>
19312 * config/rl78/rl78.c (rl78_expand_prologue): Always select
19313 register bank 0 at the start of an interrupt handler.
19314 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
19317 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
19319 * config/aarch64/aarch64-simd.md
19320 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
19321 (aarch64_simd_bsl): Likewise.
19322 (aarch64_vcond_internal<mode>): Likewise.
19323 (vcond<mode><mode>): Likewise.
19324 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
19325 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
19327 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
19329 * config/aarch64/aarch64-builtins.c
19330 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
19332 2013-01-08 Martin Jambor <mjambor@suse.cz>
19335 * tree-sra.c (analyze_access_subtree): Return true also after
19336 potentially creating a debug-only replacement.
19338 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19340 PR middle-end/55890
19341 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
19343 PR tree-optimization/54120
19344 * tree-vrp.c (range_fits_type_p): Don't allow
19345 src_precision < precision from signed vr to unsigned_p
19346 if vr->min or vr->max is negative.
19347 (simplify_float_conversion_using_ranges): Test can_float_p
19348 against CODE_FOR_nothing.
19350 2013-01-08 Jakub Jelinek <jakub@redhat.com>
19351 Richard Biener <rguenther@suse.de>
19353 PR middle-end/55851
19354 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
19355 types instead of just INTEGER_TYPE types.
19357 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
19359 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
19362 2013-01-07 Steve Ellcey <sellcey@mips.com>
19365 * config/mips/mips.opt: Change mad to mmad to match documentation.
19367 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19370 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
19371 .progmemx.data now.
19373 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19376 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
19377 (avr_addrspace_t): Add .section_name field.
19378 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
19380 (avr_addrspace): Same. Initialize .section_name. Remove last
19381 NULL entry. Put __memx into .progmemx.data.
19382 (progmem_section_prefix): Remove.
19383 (avr_asm_init_sections): No need to initialize progmem_section.
19384 (avr_asm_named_section): Use avr_addrspace[].section_name to get
19385 section name prefix.
19386 (avr_asm_select_section): Ditto. And use get_unnamed_section to
19387 retrieve the progmem section.
19388 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
19389 boundary to run over avr_addrspace[].
19390 (avr_register_target_pragmas): Ditto.
19392 2013-01-06 Jakub Jelinek <jakub@redhat.com>
19394 * varasm.c (output_constant_def_contents): For asan_protect_global
19395 protected strings, adjust DECL_ALIGN if needed, before testing for
19397 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
19398 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
19400 (output_object_block): For asan protected decls, emit asan padding
19401 after their contents.
19402 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
19403 (asan_finish_file): Test it here instead.
19405 2013-01-07 Nick Clifton <nickc@redhat.com>
19406 Matthias Klose <doko@debian.org>
19407 Doug Kwan <dougkwan@google.com>
19408 H.J. Lu <hongjiu.lu@intel.com>
19411 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
19413 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
19415 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
19417 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
19419 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
19421 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19424 * doc/install.texi (Cross-Compiler-Specific Options): Document
19427 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
19429 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
19430 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
19431 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
19432 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
19433 vqmovun_high_s64): Fix source operand number and update copyright.
19435 2013-01-07 Richard Biener <rguenther@suse.de>
19437 PR middle-end/55890
19438 * gimple.h (gimple_call_builtin_p): New overload.
19439 * gimple.c (validate_call): New function.
19440 (gimple_call_builtin_p): Likewise.
19441 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19442 Use gimple_call_builtin_p.
19443 (find_func_clobbers): Likewise.
19444 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
19445 (strlen_optimize_stmt): Likewise.
19447 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
19449 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
19450 (vld1q_dup_*): Likewise.
19451 (vld1_*): Likewise.
19452 (vld1q_*): Likewise.
19453 (vld1_lane_*): Likewise.
19454 (vld1q_lane_*): Likewise.
19456 2013-01-07 Richard Biener <rguenther@suse.de>
19458 * lto-streamer.h (LTO_minor_version): Bump to 2.
19460 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
19462 * config/aarch64/aarch64-protos.h
19463 (aarch64_const_double_zero_rtx_p): Rename to...
19464 (aarch64_float_const_zero_rtx_p): ...this.
19465 (aarch64_float_const_representable_p): New.
19466 (aarch64_output_simd_mov_immediate): Likewise.
19467 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
19468 move immediate case.
19469 * config/aarch64/aarch64.c
19470 (aarch64_const_double_zero_rtx_p): Rename to...
19471 (aarch64_float_const_zero_rtx_p): ...this.
19472 (aarch64_print_operand): Allow printing of new constants.
19473 (aarch64_valid_floating_const): New.
19474 (aarch64_legitimate_constant_p): Check for valid floating-point
19476 (aarch64_simd_valid_immediate): Likewise.
19477 (aarch64_vect_float_const_representable_p): New.
19478 (aarch64_float_const_representable_p): Likewise.
19479 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
19480 (aarch64_output_simd_mov_immediate): New.
19481 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
19482 (*movdf_aarch64): Likewise.
19483 * config/aarch64/constraints.md (Ufc): New.
19484 (Y): call aarch64_float_const_zero_rtx.
19485 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
19487 2013-01-07 Richard Biener <rguenther@suse.de>
19489 PR tree-optimization/55888
19490 PR tree-optimization/55862
19491 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
19492 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
19493 not if it is contained therein.
19495 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19497 * config/avr/t-avr: Typo.
19499 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
19502 * config/avr/t-avr: Don't automatically rebuild
19503 $(srcdir)/config/avr/t-multilib
19504 $(srcdir)/config/avr/avr-tables.opt
19505 $(srcdir)/doc/avr-mmcu.texi
19506 (avr-mcus): New phony target to build them on request.
19507 (s-avr-mlib, s-avr-mmcu-texi): Remove.
19508 * avr/avr-mcus.def: Adjust comments.
19510 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
19512 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
19514 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
19516 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
19518 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
19520 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
19522 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
19524 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
19525 to generate profiling.
19526 * config/rs6000/aix64.h (LIB_SPEC): Same.
19528 2013-01-04 Andrew Pinski <apinski@cavium.com>
19530 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
19532 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
19534 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
19536 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
19538 (ix86_expand_move): Ditto.
19539 (ix86_zero_extend_to_Pmode): Ditto.
19540 (ix86_expand_call): Ditto.
19541 (ix86_expand_special_args_builtin): Ditto.
19542 (ix86_expand_builtin): Ditto.
19544 2013-01-04 Richard Biener <rguenther@suse.de>
19546 PR tree-optimization/55862
19547 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
19548 translating them through PHI nodes.
19550 2013-01-04 Martin Jambor <mjambor@suse.cz>
19552 PR tree-optimization/55755
19553 * tree-sra.c (sra_modify_assign): Do not check that an access has no
19554 children when trying to avoid producing a VIEW_CONVERT_EXPR.
19556 2013-01-04 Marek Polacek <polacek@redhat.com>
19558 PR middle-end/55859
19559 * opts.c (default_options_optimization): Clarify error message.
19561 2013-01-04 Richard Biener <rguenther@suse.de>
19563 PR middle-end/55863
19564 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
19567 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19570 * config/pa/pa.md (movsi): Revert previous change.
19571 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
19574 2013-01-03 Richard Henderson <rth@redhat.com>
19576 * config/i386/i386.c (ix86_expand_move): Always assign to op1
19577 after eliminating TLS symbols.
19579 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
19582 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
19583 * graphite-poly.c (debug_gmp_value): Likewise.
19585 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
19588 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
19589 selected code model, define __code_mode_small__, __code_model_medium__,
19590 __code_model_large__, __code_model_32__ or __code_model_kernel__.
19591 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
19592 xchg temporary register with %k. Declare temporary register as
19594 [__x86_64__]: For medium and large code models, preserve %rbx register.
19596 2013-01-03 Richard Biener <rguenther@suse.de>
19598 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
19599 (dump_subscript): Adjust.
19600 (finalize_ddr_dependent): Do not dump redundant info.
19601 (analyze_siv_subscript): Adjust.
19602 (subscript_dependence_tester): Likewise.
19603 (compute_affine_dependence): Likewise.
19605 2013-01-03 Richard Biener <rguenther@suse.de>
19608 2013-01-03 Richard Biener <rguenther@suse.de>
19610 PR tree-optimization/55857
19611 * tree-vect-stmts.c (vectorizable_load): Do not setup
19612 re-alignment for invariant loads.
19614 2013-01-02 Richard Biener <rguenther@suse.de>
19616 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19617 invariant load do not generate a vector load from the scalar location.
19619 2013-01-03 Richard Biener <rguenther@suse.de>
19621 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19622 for not vectorizing.
19623 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19624 not build INDIRECT_REFs, call get_name once only.
19625 (vect_create_data_ref_ptr): Likewise. Dump base object kind
19626 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19628 2013-01-03 Richard Biener <rguenther@suse.de>
19630 PR tree-optimization/55857
19631 * tree-vect-stmts.c (vectorizable_load): Do not setup
19632 re-alignment for invariant loads.
19634 2013-01-03 Richard Biener <rguenther@suse.de>
19637 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19638 prefer a built-in decl.
19640 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19642 * gcc.c (process_command): Update copyright notice dates.
19643 * gcov.c (print_version): Likewise.
19644 * gcov-dump.c (print_version): Likewise.
19646 PR rtl-optimization/55838
19647 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19648 iv0.step, iv1.step and step.
19650 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19651 Marc Glisse <marc.glisse@inria.fr>
19653 PR tree-optimization/55832
19654 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19655 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19656 integer_{one,zero}_node.
19658 2013-01-03 Jakub Jelinek <jakub@redhat.com>
19661 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19662 * var-tracking.c (reverse_op): Don't add reverse ops to
19663 VALUEs that have already
19664 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19666 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19668 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19670 2013-01-02 Teresa Johnson <tejohnson@google.com>
19672 * dumpfile.c (dump_loc): Print filename with location.
19673 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19674 new location_t parameter to emit complete unroll message with
19675 new dump framework.
19676 (canonicalize_loop_induction_variables): Compute loops location
19677 and pass to try_unroll_loop_completely.
19678 * loop-unroll.c (report_unroll_peel): New function.
19679 (peel_loops_completely): Use new dump format with location
19680 for main dumpfile message, and invoke report_unroll_peel on success.
19681 (decide_unrolling_and_peeling): Ditto.
19682 (decide_peel_once_rolling): Remove old dumpfile message subsumed
19683 by report_unroll_peel.
19684 (decide_peel_completely): Ditto.
19685 (decide_unroll_constant_iterations): Ditto.
19686 (decide_unroll_runtime_iterations): Ditto.
19687 (decide_peel_simple): Ditto.
19688 (decide_unroll_stupid): Ditto.
19689 * cfgloop.c (get_loop_location): New function.
19690 * cfgloop.h (get_loop_location): Declare.
19692 2013-01-02 Sriraman Tallam <tmsriram@google.com>
19694 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19697 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19699 PR middle-end/55198
19700 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19701 BLKmode objects when EXPAND_MEMORY is specified.
19703 2013-01-02 Sriraman Tallam <tmsriram@google.com>
19705 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19707 (fold_builtin_cpu): Do not share cpu model decls across statements.
19709 2013-01-02 Jason Merrill <jason@redhat.com>
19712 * tree.c (build_array_type_1): Revert earlier change.
19714 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
19716 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19718 * config/aarch64/aarch64-tune.md: Re-generate.
19720 2013-01-02 Richard Biener <rguenther@suse.de>
19722 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19723 invariant load do not generate a vector load from the scalar location.
19725 2013-01-02 Richard Biener <rguenther@suse.de>
19728 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19729 * configure: Regenerate.
19731 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
19733 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19734 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19735 (expand_builtin_int_roundingfn_2): Keep the original target around
19736 for the fallback case.
19738 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
19740 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19741 to be clear for sign changes.
19743 2013-01-01 Jan Hubicka <jh@suse.cz>
19745 * ipa-inline-analysis.c: Fix formatting.
19747 2013-01-01 Jakub Jelinek <jakub@redhat.com>
19749 PR tree-optimization/55831
19750 * tree-vect-loop.c (get_initial_def_for_induction): Use
19751 gsi_after_labels instead of gsi_start_bb.
19753 Copyright (C) 2013 Free Software Foundation, Inc.
19755 Copying and distribution of this file, with or without modification,
19756 are permitted in any medium without royalty provided the copyright
19757 notice and this notice are preserved.