1 2013-09-16 Vladimir Makarov <vmakarov@redhat.com>
4 * lra-constraints.c (undo_optional_reloads): Consider all optional
5 reload even if it did not get a hard reg.
7 2013-09-16 Teresa Johnson <tejohnson@google.com>
9 * dumpfile.c (dump_loc): Remove newline emission.
10 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
11 emission to dump_printf_loc calls where missing.
12 (vect_mark_for_runtime_alias_test): Ditto.
13 (vect_analyze_data_ref_dependence): Ditto.
14 (vect_analyze_data_ref_dependences): Ditto.
15 (vect_slp_analyze_data_ref_dependence): Ditto.
16 (vect_slp_analyze_data_ref_dependences): Ditto.
17 (vect_compute_data_ref_alignment): Ditto.
18 (vect_update_misalignment_for_peel): Ditto.
19 (vect_verify_datarefs_alignment): Ditto.
20 (vector_alignment_reachable_p): Ditto.
21 (vect_get_data_access_cost): Ditto.
22 (vect_enhance_data_refs_alignment): Ditto.
23 (vect_find_same_alignment_drs): Ditto.
24 (vect_analyze_data_refs_alignment): Ditto.
25 (vect_analyze_group_access): Ditto.
26 (vect_analyze_data_ref_access): Ditto.
27 (vect_analyze_data_ref_accesses): Ditto.
28 (vect_prune_runtime_alias_test_list): Ditto.
29 (vect_analyze_data_refs): Ditto.
30 (vect_create_addr_base_for_vector_ref): Ditto.
31 (vect_create_data_ref_ptr): Ditto.
32 (vect_grouped_store_supported): Ditto.
33 (vect_grouped_load_supported): Ditto.
34 * value-prof.c (check_counter): Ditto.
35 (check_ic_target): Ditto.
36 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
37 (vect_recog_widen_mult_pattern): Ditto.
38 (vect_recog_widen_sum_pattern): Ditto.
39 (vect_recog_over_widening_pattern): Ditto.
40 (vect_recog_widen_shift_pattern): Ditto.
41 (vect_recog_rotate_pattern): Ditto.
42 (vect_recog_vector_vector_shift_pattern): Ditto.
43 (vect_recog_divmod_pattern): Ditto.
44 (vect_recog_mixed_size_cond_pattern): Ditto.
45 (vect_recog_bool_pattern): Ditto.
46 (vect_pattern_recog_1): Ditto.
47 (vect_pattern_recog): Ditto.
48 * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
49 (vect_is_simple_iv_evolution): Ditto.
50 (vect_analyze_scalar_cycles_1): Ditto.
51 (vect_get_loop_niters): Ditto.
52 (vect_analyze_loop_1): Ditto.
53 (vect_analyze_loop_form): Ditto.
54 (vect_analyze_loop_operations): Ditto.
55 (vect_analyze_loop_2): Ditto.
56 (vect_analyze_loop): Ditto.
57 (report_vect_op): Ditto.
58 (vect_is_slp_reduction): Ditto.
59 (vect_is_simple_reduction_1): Ditto.
60 (vect_get_known_peeling_cost): Ditto.
61 (vect_estimate_min_profitable_iters): Ditto.
62 (vect_model_reduction_cost): Ditto.
63 (vect_model_induction_cost): Ditto.
64 (get_initial_def_for_induction): Ditto.
65 (vect_create_epilog_for_reduction): Ditto.
66 (vectorizable_reduction): Ditto.
67 (vectorizable_induction): Ditto.
68 (vectorizable_live_operation): Ditto.
69 (vect_loop_kill_debug_uses): Ditto.
70 (vect_transform_loop): Ditto.
71 * tree-vect-stmts.c (vect_mark_relevant): Ditto.
72 (vect_stmt_relevant_p): Ditto.
74 (vect_mark_stmts_to_be_vectorized): Ditto.
75 (vect_model_simple_cost): Ditto.
76 (vect_model_promotion_demotion_cost): Ditto.
77 (vect_model_store_cost): Ditto.
78 (vect_get_store_cost): Ditto.
79 (vect_model_load_cost): Ditto.
80 (vect_get_load_cost): Ditto.
81 (vect_init_vector_1): Ditto.
82 (vect_get_vec_def_for_operand): Ditto.
83 (vect_finish_stmt_generation): Ditto.
84 (vectorizable_call): Ditto.
85 (vectorizable_conversion): Ditto.
86 (vectorizable_assignment): Ditto.
87 (vectorizable_shift): Ditto.
88 (vectorizable_operation): Ditto.
89 (vectorizable_store): Ditto.
90 (vectorizable_load): Ditto.
91 (vectorizable_condition): Ditto.
92 (vect_analyze_stmt): Ditto.
93 (vect_transform_stmt): Ditto.
94 (vect_is_simple_use): Ditto.
95 * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
96 (vect_can_advance_ivs_p): Ditto.
97 (vect_update_ivs_after_vectorizer): Ditto.
98 (vect_do_peeling_for_loop_bound): Ditto.
99 (vect_gen_niters_for_prolog_loop): Ditto.
100 (vect_update_inits_of_drs): Ditto.
101 (vect_create_cond_for_alias_checks): Ditto.
102 * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
103 (vect_build_slp_tree_1): Ditto.
104 (vect_supported_load_permutation_p): Ditto.
105 (vect_analyze_slp_instance): Ditto.
106 (vect_analyze_slp): Ditto.
107 (vect_make_slp_decision): Ditto.
108 (vect_detect_hybrid_slp): Ditto.
109 (vect_bb_vectorization_profitable_p): Ditto.
110 (vect_slp_analyze_bb_1): Ditto.
111 (vect_update_slp_costs_according_to_vf): Ditto.
112 (vect_get_mask_element): Ditto.
113 (vect_transform_slp_perm_load): Ditto.
114 (vect_schedule_slp_instance): Ditto.
115 (vect_schedule_slp): Ditto.
116 (vect_slp_transform_bb): Ditto.
117 * profile.c (read_profile_edge_counts): Ditto.
118 (compute_branch_probabilities): Ditto.
119 * coverage.c (get_coverage_counts): Ditto.
121 2013-09-16 Diego Novillo <dnovillo@google.com>
123 * tree-core.h: Add missing comment lines from refactoring
126 2013-09-16 Jan Hubicka <jh@suse.cz>
128 * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
129 abstract functions; for static functions check the presence
132 2013-09-16 James Greenhalgh <james.greenhalgh@arm.com>
134 * config/aarch64/aarch64-simd-builtins.def (fma): New.
135 * config/aarch64/aarch64-simd.md
136 (aarch64_mla_elt<mode>): New.
137 (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
138 (aarch64_mls_elt<mode>): Likewise.
139 (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
140 (aarch64_fma4_elt<mode>): Likewise.
141 (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
142 (aarch64_fma4_elt_to_128v2df): Likewise.
143 (aarch64_fma4_elt_to_64df): Likewise.
144 (fnma<mode>4): Likewise.
145 (aarch64_fnma4_elt<mode>): Likewise.
146 (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
147 (aarch64_fnma4_elt_to_128v2df): Likewise.
148 (aarch64_fnma4_elt_to_64df): Likewise.
149 * config/aarch64/iterators.md (VDQSF): New.
150 * config/aarch64/arm_neon.h
151 (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
152 (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
154 2013-09-16 James Greenhalgh <james.greenhalgh@arm.com>
156 * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
157 (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
158 (aarch64_mul3_elt_to_128df): Likewise.
159 (aarch64_mul3_elt_to_64v2df): Likewise.
160 * config/aarch64/iterators.md (VEL): Also handle DFmode.
162 (VMUL_CHANGE_NLANES) Likewise.
165 * config/aarch64/arm_neon.h
166 (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
168 2013-09-16 James Greenhalgh <james.greenhalgh@arm.com>
170 * config/aarch64/arm_neon.h
171 (vcvtx_high_f32_f64): Fix parameters.
173 2013-09-16 Jan-Benedict Glaw <jbglaw@lug-owl.de>
175 * config/alpha.c: Include tree-ssa.h.
177 2013-09-16 Richard Biener <rguenther@suse.de>
179 * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
180 (dot_rdg_1): Handle control_dd.
181 (create_edge_for_control_dependence): New function.
182 (create_rdg_edges): Add control dependences if asked for.
183 (build_rdg): Likewise.
184 (generate_loops_for_partition): If there are not necessary
185 control stmts remove all their dependencies.
186 (collect_condition_stmts, rdg_flag_loop_exits): Remove.
187 (distribute_loop): Pass on control dependences.
188 (tree_loop_distribution): Compute control dependences and remove
189 restriction on number of loop nodes.
191 2013-09-16 Jakub Jelinek <jakub@redhat.com>
193 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
196 2013-09-16 Richard Sandiford <rdsandiford@googlemail.com>
198 * cse.c (try_const_anchors): Punt on CC modes.
200 2013-09-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
202 * config/vax/constraints.md (T): Add missing CONSTANT_P check.
204 2013-09-14 John David Anglin <danglin@gcc.gnu.org>
207 * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
210 2013-09-14 Iain Sandoe <iain@codesourcery.com>
213 * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
215 (darwin_objc1_section): Likewise.
216 (darwin_file_end): Emit Image Info section when required.
218 2013-09-14 Jan Hubicka <jh@suse.cz>
220 * tree-into-ssa.c (gate_into_ssa): New.
221 (pass_data_build_ssa): Use it.
222 * cgraph.h (expand_thunk): Update prototype.
223 * cgraphunit.c (analyze_function): Expand thunks early.
224 (expand_thunk): Fix DECL_CONTEXT of reust_decl;
225 build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
226 set lowered flag; do not add new function.
227 (assemble_thunks_and_aliases): Update.
228 * tree-ssa.c (gate_init_datastructures): New gate.
229 (pass_data_init_datastructures): Use it.
231 2013-09-14 Iain Sandoe <iain@codesourcery.com>
234 * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
235 xmm register set described in the psABI.
237 2013-09-13 Evgeny Gavrin <e.gavrin@samsung.com>
239 * dwarf2out.c (should_emit_struct_debug): Add check
240 for type_decl variable is not NULL.
242 2013-09-13 Jacek Caban <jacek@codeweavers.com>
244 * config.gcc: Use new winnt-c.c target hooks
245 * config/t-winnt: New file
246 * config/winnt-c.c: New file
247 * doc/tm.texi.in: Document new hook
248 * doc/tm.texi: Regenerated
250 2013-09-13 Jan Hubicka <jh@suse.cz>
253 * ipa-inline.c (check_callers): New function.
254 (check_caller_edge): Remove.
255 (want_inline_function_to_all_callers_p): Also permit alises that are
257 (inline_to_all_callers): Terminate the walk when devirtualization
260 2013-09-13 Jan Hubicka <jh@suse.cz>
262 * ipa-inline-analysis.c (struct growth_data): Add node.
263 (do_estimate_growth_1): Fix detection of recursion.
265 2013-09-13 Jakub Jelinek <jakub@redhat.com>
267 PR tree-optimization/58392
268 * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
269 to avoid shadowing of outer loop variable. If
270 saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
271 replace_by_duplicate_decl simduid of loops that have it set and
272 set dest_cfun->has_simduid_loops and/or
273 dest_cfun->has_force_vect_loops.
274 * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
275 instead of maybe_lookup_decl.
276 * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
277 Use id->blocks_to_copy instead of blocks_to_copy. Adjust recursive
278 call. Copy over force_vect and copy and remap simduid. Set
279 cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
280 (copy_cfg_body): Remove blocks_to_copy argument. Use
281 id->blocks_to_copy instead of blocks_to_copy. Adjust copy_loops
282 caller. Don't set cfun->has_simduid_loops and/or
283 cfun->has_force_vect_loops here.
284 (copy_body): Remove blocks_to_copy argument. Adjust copy_cfg_body
286 (expand_call_inline, tree_function_versioning): Adjust copy_body
289 2013-09-13 Martin Jambor <mjambor@suse.cz>
292 * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
293 the assert if the edge was a speculative one.
295 2013-09-13 Richard Biener <rguenther@suse.de>
297 * tree-data-ref.h (known_dependences_p): Move here ...
298 * tree-loop-distribution.c (known_dependences_p): ... from here.
299 (dump_rdg_component, debug_rdg_component): Remove.
301 (generate_loops_for_partition): Use gimple_uid instead of
302 relying on matching stmt visit order.
303 (rdg_build_partitions): Take starting stmt vector.
304 (ldist_gen): Merge into ...
305 (distribute_loop): ... this function. Do not compute starting
307 * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
309 2013-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
311 * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
312 Set type attribute correctly. Set predicable_short_it attribute.
313 (cmpsi_shiftsi): Remove %? from output template.
315 2013-09-13 Richard Biener <rguenther@suse.de>
317 * tree-loop-distribution.c (struct rdg_component,
318 rdg_defs_used_in_other_loops_p, free_rdg_components,
319 rdg_build_components): Remove.
320 (stmts_from_loop): Do not record virtual PHIs.
321 (generate_loops_for_partition): Skip virtual PHIs.
322 (build_rdg_partition_for_component): Rename to ...
323 (build_rdg_partition_for_vertex): ... this and adjust.
324 (rdg_build_partitions): Take a vector of starting vertices
325 instead of components. Remove unnecessary leftover handling.
326 (ldist_gen): Do not build components or record other stores.
327 (distribute_loop): Do not distribute loops containing stmts
330 2013-09-13 Christian Bruel <christian.bruel@st.com>
333 * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
335 2013-09-13 Kai Tietz <ktietz@redhat.com>
337 * config.gcc: Separate cases for mingw and cygwin targets,
338 and add 64-bit cygwin target case.
340 * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
341 dll-export inline-functions.
342 * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
345 2013-09-13 Jeff Law <law@redhat.com>
349 2013-09-06 Jeff Law <law@redhat.com>
351 * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
352 edge implied equivalences into successor phis.
354 2013-09-12 DJ Delorie <dj@redhat.com>
356 * config/rl78/rl78-virt.md: Change from | to \; for asm line
359 2013-09-12 Brooks Moses <bmoses@google.com>
362 * Makefile.in: Do not install driver binaries in $(target)/bin.
364 2013-09-12 DJ Delorie <dj@redhat.com>
366 * config/rl78/rl78.opt (mrelax): New.
367 * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
368 * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
370 * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
372 (rl78_expand_epilogue): Likewise.
374 2013-09-12 Vladimir Makarov <vmakarov@redhat.com>
377 * lra-eliminations.c (remove_reg_equal_offset_note): New.
378 (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
379 pointer elimination with using remove_reg_equal_offset_note.
381 2013-09-12 DJ Delorie <dj@redhat.com>
383 * config/msp430/: New port.
384 * config.gcc (msp430): Added.
385 * doc/invoke.texi: Document MSP430 options.
386 * doc/install.texi: Document msp430-elf
387 * doc/md.texi: Document msp430-elf
388 * doc/contrib.texi: Document msp430-elf
390 * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
393 2013-09-12 Martin Jambor <mjambor@suse.cz>
396 * ipa-prop.c (remove_described_reference): Give up if the edge in the
397 reference descriptor is NULL.
398 (ipa_edge_removal_hook): If owning a reference descriptor, set its
401 2013-09-12 Andrew MacLeod <amacleod@redhat.com>
403 * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
404 (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
405 num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
406 * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
407 make_temp_ssa_name): move to tree-ssanames.h
408 * tree-ssa-alias.h: Move prototype.
409 * tree-ssa.h: Include tree-ssanames.h.
410 * tree-ssanames.c (FREE_SSANAMES): Move to here.
411 * tree-ssanames.h: New. Move items from tree-flow*.h
412 * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
414 2013-09-12 Richard Biener <rguenther@suse.de>
416 PR tree-optimization/58404
417 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
418 propagate non-invariant addresses into dereferences wrapped
419 in component references.
421 2013-09-12 Richard Biener <rguenther@suse.de>
423 PR tree-optimization/58402
424 * passes.def: Move pass_late_warn_uninitialized later.
426 2013-09-12 Andrew MacLeod <amacleod@redhat.com>
428 * tree-ssa.h: New. Move content from tree-flow.h and
430 * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
431 Move prototypes belonging to tree-ssa.c.
432 * tree-flow-inline.h (redirect_edge_var_map_def,
433 redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
435 * gimple.h: Adjust prototypes.
436 * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
438 * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
439 * tree.h: Move prototype to tree-ssa.h.
440 * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
441 * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
442 * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
443 cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
444 cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
445 ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
446 gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
447 gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
448 graphite-blocking.c, graphite-clast-to-gimple.c,
449 graphite-dependences.c, graphite-interchange.c,
450 graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
451 graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
452 ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
453 ipa-reference.c, ipa-split.c, ipa-utils.c,
454 loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
455 lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
456 passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
457 tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
458 tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
459 tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
460 tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
461 tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
462 tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
463 tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
464 tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
465 tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
466 tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
467 tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
468 value-prof.c, var-tracking.c,
469 varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
471 2013-09-12 Richard Biener <rguenther@suse.de>
473 PR tree-optimization/58396
474 * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
475 (build_rdg): Take a loop-nest parameter, fix memleaks.
476 (distribute_loop): Compute loop-nest here and pass it to build_rdg.
478 2013-09-12 Yuri Rumyantsev <ysrumyan@gmail.com>
480 * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
483 2013-09-12 Cameron McInally <cameron.mcinally@nyu.edu>
485 * doc/extend.texi: Fix errors in x86 FMA builtin naming.
486 The FMA instruction names should have a 'v' prefix.
488 2013-09-12 Richard Biener <rguenther@suse.de>
490 * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
491 (dot_rdg): Use popen instead of system in optional code.
492 (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
493 (already_processed_vertex_p): Adjust.
494 (has_anti_or_output_dependence, predecessor_has_mem_write,
495 mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
496 rdg_flag_uses): Remove.
497 (rdg_flag_vertex): Simplify.
498 (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
500 (build_rdg_partition_for_component): Process the first vertex
502 (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
504 2013-09-12 Alan Modra <amodra@gmail.com>
506 * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
508 2013-09-11 DJ Delorie <dj@redhat.com>
509 Nick Clifton <nickc@redhat.com>
511 * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
512 (rl78_stack_based_mem): New.
513 * config/rl78/constraints.md (Iv08): New.
522 * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
524 (movsi): Change from expand to insn-and-split.
525 (ashrsi3): Clobber AX.
529 * config/rl78/rl78.md (CC_REG): Fix.
530 (addsi3): Allow memory and immediate operands.
531 (addsi3_internal): Split into...
532 (addsi3_internal_virt): ...new, and ...
533 (addsi3_internal_real): ...new.
535 (subsi3_internal_virt): New.
536 (subsi3_internal_real): New.
537 (mulsi3): Add memory operand.
538 (mulsi3_rl78): Likewise.
539 (mulsi3_g13): Likewise.
540 * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
541 (cbranchqi4_real): Add more constraint options.
542 (cbranchhi4_real): Expand pattern.
543 (cbranchhi4_real_signed): New.
544 (cbranchhi4_real_inverted): New.
545 (cbranchsi4_real_lt): New.
546 (cbranchsi4_real_ge): New.
547 (cbranchsi4_real_signed): New.
548 (cbranchsi4_real): New.
550 * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
552 (lshrsi3_virt): Likewise.
553 (ashlsi3_virt): Likewise.
554 (cbranchqi4_virt_signed): New.
555 (cbranchhi4_virt_signed): New.
556 (cbranchsi4_virt): New.
557 * config/rl78/rl78.c: Whitespace fixes throughout.
558 (move_elim_pass): New.
559 (pass_data_rl78_move_elim): New.
560 (pass_rl78_move_elim): New.
561 (make_pass_rl78_move_elim): New.
562 (rl78_devirt_info): Run devirt earlier.
563 (rl78_move_elim_info): New.
564 (rl78_asm_file_start): Register it.
565 (rl78_split_movsi): New.
566 (rl78_as_legitimate_address): Allow virtual base registers when
568 (rl78_addr_space_convert): Remove spurious debug stuff.
569 (rl78_print_operand_1): Add z,s,S,r,E modifiers.
570 (rl78_print_operand): More cases for not printing '#'.
571 (rl78_expand_compare): Remove most of the logic.
572 (content_memory): New.
573 (clear_content_memory): New.
574 (get_content_index): New.
575 (get_content_name): New.
576 (display_content_memory): New.
577 (update_content): New.
578 (record_content): New.
579 (already_contains): New.
580 (insn_ok_now): Re-recog insns with virtual registers.
581 (add_postponed_content_update): New.
582 (process_postponed_content_update): New.
583 (gen_and_emit_move): New.
584 (transcode_memory_rtx): Record new location content.
585 Use gen_and_emit_move.
586 (force_into_acc): New.
587 (move_to_acc): Use gen_and_emit_move.
588 (move_from_acc): Likewise.
589 (move_acc_to_reg): Likewise.
590 (move_to_x): Likewise.
591 (move_to_hl): Likewise.
592 (move_to_de): Likewise.
593 (rl78_alloc_physical_registers_op1): Record location content.
594 (has_constraint): New.
595 (rl78_alloc_physical_registers_op2): Record location content.
597 (rl78_alloc_physical_registers_ro1): Likewise.
598 (rl78_alloc_physical_registers_cmp): Likewise.
599 (rl78_alloc_physical_registers_umul): Likewise.
600 (rl78_alloc_address_registers_macax): New.
601 (rl78_alloc_physical_registers): Initialize and set location
602 content memory as needed.
603 (rl78_reorg): Make sure split2 is called.
604 (rl78_rtx_costs): New.
606 2013-09-11 Richard Sandiford <rdsandiford@googlemail.com>
608 * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
609 for (not (neg ...)) and (neg (not ...)) cases.
611 2013-09-11 Richard Biener <rguenther@suse.de>
614 * passes.def: Split critical edges before late uninit warning passes.
615 * tree-cfg.c (pass_split_crit_edges): Implement clone method.
617 2013-09-11 Jakub Jelinek <jakub@redhat.com>
619 PR tree-optimization/58385
620 * fold-const.c (build_range_check): If both low and high are NULL,
621 use omit_one_operand_loc to preserve exp side-effects.
623 2013-09-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
625 * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
627 2013-09-11 Richard Biener <rguenther@suse.de>
629 * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
630 dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
631 dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
632 create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
633 stmts_from_loop, known_dependences_p, build_empty_rdg,
634 build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
635 * tree-loop-distribution.c: ... here.
636 * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
637 RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
638 RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
639 struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
640 * tree-loop-distribution.c: ... here.
641 * tree-loop-distribution.c: Include gimple-pretty-print.h.
642 (struct partition_s): Add loops member.
643 (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
644 rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
645 build_rdg_partition_for_component, rdg_build_partitions): Adjust.
647 2013-09-11 Alexander Ivchenko <alexander.ivchenko@intel.com>
648 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
649 Sergey Lega <sergey.s.lega@intel.com>
650 Anna Tikhonova <anna.tikhonova@intel.com>
651 Ilya Tocar <ilya.tocar@intel.com>
652 Andrey Turetskiy <andrey.turetskiy@intel.com>
653 Ilya Verbin <ilya.verbin@intel.com>
654 Kirill Yukhin <kirill.yukhin@intel.com>
655 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
657 * config/i386/constraints.md (k): New.
659 * config/i386/i386.c (const regclass_map): Add new mask registers.
660 (dbx_register_map): Ditto.
661 (dbx64_register_map): Ditto.
662 (svr4_dbx_register_map): Ditto.
663 (ix86_conditional_register_usage): Squash mask registers if AVX512F is
665 (ix86_preferred_reload_class): Disable constants for mask registers.
666 (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
667 (ix86_hard_regno_mode_ok): Support new mask registers.
668 (x86_order_regs_for_local_alloc): Ditto.
669 * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
670 (FIXED_REGISTERS): Add new mask registers.
671 (CALL_USED_REGISTERS): Ditto.
672 (REG_ALLOC_ORDER): Ditto.
673 (VALID_MASK_REG_MODE): New.
674 (FIRST_MASK_REG): Ditto.
675 (LAST_MASK_REG): Ditto.
676 (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
677 (MAYBE_MASK_CLASS_P): New.
678 (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
679 (REG_CLASS_CONTENTS): Ditto.
681 (ANY_MASK_REG_P): Ditto.
682 (HI_REGISTER_NAMES): Add new mask registers.
683 * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
684 MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
686 (attribute "type"): Add mskmov, msklog.
687 (attribute "length_immediate"): Support them.
688 (attribute "memory"): Ditto.
689 (attribute "prefix_0f"): Ditto.
690 (*movhi_internal): Support new mask registers.
691 (*movqi_internal): Ditto.
692 (define_split): Split out clobber pattern is a logic
693 insn on mask registers.
694 (*k<logic><mode>): New.
695 (*andhi_1): Extend to support mask regs.
696 (*andqi_1): Extend to support mask regs.
698 (define_split): Split and-not to and and not if operands
700 (*<code><mode>_1): Separate HI mode to new pattern...
702 (*<code>qi_1): Extend to support mask regs.
707 (*one_cmpl<mode>2_1): Remove HImode and handle it...
708 (*one_cmplhi2_1): ...Here, now with mask registers support.
709 (*one_cmplqi2_1): Support new mask registers.
710 (HI/QImode arithmetics splitter): Don't split if mask registers
712 (HI/QImode not splitter): Ditto.
713 * config/i386/predicated.md (mask_reg_operand): New.
714 (general_reg_operand): Ditto.
716 2013-09-11 Alexander Ivchenko <alexander.ivchenko@intel.com>
718 * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
719 * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
721 2013-09-10 Jeff Law <law@redhat.com>
723 PR tree-optimization/58380
724 * tree-ssa-threadupdate.c (thread_block): Recognize another case
725 of threading through a buried loop header.
727 * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
728 return value for single successor case.
730 2013-09-10 Jan Hubicka <jh@suse.cz>
732 * ipa-devirt.c (ipa_devirt): Enable with LTO.
734 2013-09-10 Richard Earnshaw <rearnsha@arm.com>
737 * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
738 support conditional execution.
739 (combine_vcvt_f64_<FCVTI32typename>): Likewise.
741 2013-09-10 Vladimir Makarov <vmakarov@redhat.com>
743 * lra.c (lra): Clear lra_optional_reload_pseudos before every
745 * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
746 Check destination too to check move insn.
747 (undo_optional_reloads): Add check that the original peudo did not
748 changed its allocation and the optional reload was inherited on last
749 inheritance pass. Break loop after deciding to keep optional reload.
750 (lra_undo_inheritance): Add check that inherited pseudo still in
753 2013-09-10 James Greenhalgh <james.greenhalgh@arm.com>
755 * config/aarch64/aarch64.md (generic_sched): New.
756 * config/aarch64/aarch64-generic.md (load): Make conditional
757 on generic_sched attribute.
760 2013-09-10 Jan Hubicka <jh@suse.cz>
762 * lto-cgraph.c: Include ipa-utils.h.
763 (compute_ltrans_boundary): Also add possible targets into the boundary.
765 2013-09-10 Jan Hubicka <jh@suse.cz>
767 * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
768 VAR_DECL of vtable rather than full expression.
770 2013-09-10 Jan Hubicka <jh@suse.cz>
771 Paolo Carlini <paolo.carlini@oracle.com>
773 * cgraphunit.c (analyze_functions): Save input_location, set it
774 to UNKNOWN_LOCATION and restore it at the end.
776 2013-09-10 Martin Jambor <mjambor@suse.cz>
778 * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
779 represented by a thunk.
781 2013-09-10 Jeff Law <law@redhat.com>
783 PR tree-optimization/58343
784 * tree-ssa-threadupdate.c (thread_block): Identify and disable
785 jump threading requests through loop headers buried in the middle
786 of a jump threading path.
788 * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
789 in return value/type.
791 2013-09-10 Jakub Jelinek <jakub@redhat.com>
793 PR rtl-optimization/58365
794 * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
795 resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
798 2013-09-10 Richard Biener <rguenther@suse.de>
800 * tree-data-ref.h (build_rdg): Drop all parameters but loop.
801 * tree-data-ref.c (create_rdg_vertices): Collect all data
802 references, signal failure to the caller, use data-ref API.
803 (build_rdg): Compute data references only once. Maintain lifetime
804 of data references and data dependences from within RDG.
805 (free_rdg): Free dependence relations.
806 * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
807 inventing extra dependences.
808 (distribute_loop): Update for RDG API changes.
810 2013-09-10 Kai Tietz <ktietz@redhat.com>
812 * doc/invoke.texi (fms-extensions): Document changed
813 behavior for ms-abi targets.
814 * config/i386/i386.c (ix86_option_override_internal):
815 Set default value of option -fms-extension for ms-abi targets.
817 2013-09-10 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
819 * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
821 2013-09-10 Alan Modra <amodra@gmail.com>
824 * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
826 2013-09-10 Alan Modra <amodra@gmail.com>
828 * config/rs6000/predicates.md (add_cint_operand): New.
829 (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
830 * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
831 using add_cint_operand.
832 (largetoc_high_plus_aix): Likewise.
834 2013-09-09 Jakub Jelinek <jakub@redhat.com>
836 PR tree-optimization/58364
837 * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
838 BOOLEAN_TYPE, only invert in_p and continue with arg0 if
839 the current range can't be an unconditional true or false.
841 2013-09-09 James Greenhalgh <james.greenhalgh@arm.com>
843 * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
845 2013-09-09 Uros Bizjak <ubizjak@gmail.com>
847 * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
849 2013-09-09 Paolo Carlini <paolo.carlini@oracle.com>
852 * doc/invoke.texi (-Wdelete-incomplete): Document it.
854 2013-09-09 Ian Bolton <ian.bolton@arm.com>
856 * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
857 NO_REGS for immediate that can't be moved directly into FP_REGS.
859 2013-09-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
861 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
862 comparison with negated operand.
863 * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
866 2013-09-09 Richard Biener <rguenther@suse.de>
869 * cfgloopmanip.c (fix_bb_placements): When fixing the placement
870 of a subloop record all its block as affecting loop-closed SSA form.
872 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
874 * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
875 of an rtx/bitpos pair.
876 (store_fixed_bit_field): Update accordingly.
878 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
880 * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
882 * builtins.c (expand_errno_check): Likewise.
883 * dwarf2cfi.c (init_return_column_size): Likewise.
884 * except.c (sjlj_mark_call_sites): Likewise.
885 * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
886 * lra-constraints.c (emit_inc): Likewise.
887 * ree.c (combine_set_extension): Likewise.
888 * regmove.c (fixup_match_2): Likewise.
889 * reload1.c (inc_for_reload): Likewise.
891 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
893 * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
894 (force_to_mode, simplify_shift_const_1, simplify_comparison):
895 Use gen_int_mode with the mode of the associated simplify_* call.
896 * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
897 * expmed.c (expand_shift_1): Likewise.
898 * function.c (instantiate_virtual_regs_in_insn): Likewise.
899 * loop-iv.c (iv_number_of_iterations): Likewise.
900 * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
901 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
903 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
905 * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
906 of the associated expand_* call.
907 (asan_emit_stack_protection): Likewise.
908 * builtins.c (round_trampoline_addr): Likewise.
909 * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
910 * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
911 (emit_store_flag): Likewise.
912 * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
913 (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
915 * function.c (instantiate_virtual_regs_in_insn): Likewise.
916 * ifcvt.c (noce_try_store_flag_constants): Likewise.
917 * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
918 * modulo-sched.c (generate_prolog_epilog): Likewise.
919 * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
920 (expand_ctz, expand_ffs, expand_unop): Likewise.
922 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
924 * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
925 of the associated gen_rtx_* call.
926 * caller-save.c (init_caller_save): Likewise.
927 * combine.c (find_split_point, make_extraction): Likewise.
928 (make_compound_operation): Likewise.
929 * dwarf2out.c (mem_loc_descriptor): Likewise.
930 * explow.c (plus_constant, probe_stack_range): Likewise.
931 * expmed.c (expand_mult_const): Likewise.
932 * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
933 * reload1.c (init_reload): Likewise.
934 * valtrack.c (cleanup_auto_inc_dec): Likewise.
935 * var-tracking.c (adjust_mems): Likewise.
936 * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
937 rather than gen_rtx_fmt_ee.
939 2013-09-09 Jan Hubicka <jh@suse.cz>
942 * value-prof.c (gimple_ic): Copy also abnormal edges.
944 2013-09-09 Richard Sandiford <rdsandiford@googlemail.com>
946 * asan.c (asan_shadow_cst): Use gen_int_mode.
948 2013-09-08 Jan Hubicka <jh@suse.cz>
950 * ipa-profile.c: Add toplevel comment.
951 (ipa_propagate_frequency_1): Be more conservative when profile is read.
952 (contains_hot_call_p): New function.
953 (ipa_propagate_frequency): Set frequencies based on counts when
955 * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
956 profile; do not tamper with profile after inlining if it is read.
958 2013-09-08 Jan Hubicka <jh@suse.cz>
960 * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
963 2013-09-08 Jan Hubicka <jh@suse.cz>
965 * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
968 2013-09-08 Jeff Law <law@redhat.com>
971 * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
974 2013-09-08 Andi Kleen <ak@linux.intel.com>
976 * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
978 2013-09-08 Jan Hubicka <jh@suse.cz>
980 * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
982 2013-09-08 Richard Sandiford <rdsandiford@googlemail.com>
984 * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
986 * lra.c (new_insn_reg): Take the containing insn as a parameter.
987 Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
988 (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
991 2013-09-08 Jan Hubicka <jh@suse.cz>
993 * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
994 targets and devirtualize to BUILT_IN_UNREACHABLE.
995 * timevar.def (TV_IPA_UNREACHABLE): New timevar.
996 * ipa.c (walk_polymorphic_call_targets): New function.
997 (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
998 functions; use the new timevar.
999 * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
1000 was removed from the program.
1001 (record_binfo): If BINFO corresponds to an anonymous namespace, we may
1002 not consider it in the walk when its vtable is dead.
1003 (possible_polymorphic_call_targets_1): Pass anonymous flag to
1005 (devirt_variable_node_removal_hook): New function.
1006 (possible_polymorphic_call_targets): Also register
1007 devirt_variable_node_removal_hook.
1008 (ipa_devirt): Do not do non-speculative devirtualization.
1009 (gate_ipa_devirt): One execute if devirtualizing speculatively.
1011 2013-09-08 Jan Hubicka <jh@suse.cz>
1013 * cgraph.h (varpool_node_hook, varpool_node_hook_list,
1014 varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
1015 varpool_remove_variable_insertion_hook): Declare.
1016 * varpool.c (varpool_node_hook_list): New structure.
1017 (first_varpool_node_removal_hook,
1018 first_varpool_variable_insertion_hook): New variables.
1019 (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
1020 varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
1021 varpool_remove_variable_insertion_hook,
1022 varpool_call_variable_insertion_hooks): New functions.
1023 (varpool_remove_node): Use it.
1025 2013-09-08 Paolo Carlini <paolo.carlini@oracle.com>
1028 * diagnostic.c (diagnostic_build_prefix): When s.file is
1029 "<built-in>" don't output line and column numbers.
1031 2013-09-06 Jan Hubicka <jh@suse.cz>
1033 * cgraphunit.c (expand_thunk): Get body before touching arguments.
1034 * lto-streamer-out.c: Stream thunks, too.
1035 * lto-streamer-in.c (input_function): Pop cfun here
1036 (lto_read_body): Instead of here.
1038 2013-09-06 Caroline Tice <cmtice@google.com>
1040 * doc/install.texi: Add documentation for the --enable-vtable-verify
1041 and the --disable-libvtv configure options.
1043 2013-09-06 Jeff Law <law@redhat.com>
1045 * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
1046 edge implied equivalences into successor phis.
1048 2013-09-06 Joern Rennecke <joern.rennecke@embecosm.com>
1050 * resource.c (mark_referenced_resources): Handle COND_EXEC.
1052 2013-09-06 Claudiu Zissulescu <claziss@synopsys.com>
1054 * resource.c (mark_target_live_regs): Compute resources taking
1055 into account if a call is predicated or not.
1057 2013-09-06 Eric Botcazou <ebotcazou@adacore.com>
1059 * toplev.c (output_stack_usage): Be prepared for suffixes created by
1060 the compiler in the function names.
1062 2013-09-06 Jan Hubicka <jh@suse.cz>
1065 * ipa-inline.c (has_caller_p): New function.
1066 (want_inline_function_to_all_callers_p): Use it.
1067 (sum_callers, inline_to_all_callers): Break out from ...
1068 (ipa_inline): ... here.
1070 2013-09-06 Jan Hubicka <jh@suse.cz>
1072 * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
1073 only when AVX is enabled.
1075 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1077 * config/aarch64/aarch64.md
1078 (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
1079 is fpsimd_<load/store>2.
1080 (load_pair<mode>): Likewise.
1081 (store_pair<mode>): Likewise.
1083 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1085 * config/arm/types.md (type): Add "mrs" type.
1086 * config/aarch64/aarch64.md
1087 (aarch64_load_tp_hard): Make type "mrs".
1089 (load_tp_hard): Make type "mrs".
1090 * config/arm/cortex-a15.md: Update with new attributes.
1091 * config/arm/cortex-a5.md: Update with new attributes.
1092 * config/arm/cortex-a53.md: Update with new attributes.
1093 * config/arm/cortex-a7.md: Update with new attributes.
1094 * config/arm/cortex-a8.md: Update with new attributes.
1095 * config/arm/cortex-a9.md: Update with new attributes.
1096 * config/arm/cortex-m4.md: Update with new attributes.
1097 * config/arm/cortex-r4.md: Update with new attributes.
1098 * config/arm/fa526.md: Update with new attributes.
1099 * config/arm/fa606te.md: Update with new attributes.
1100 * config/arm/fa626te.md: Update with new attributes.
1101 * config/arm/fa726te.md: Update with new attributes.
1103 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1105 * config/aarch64/aarch64.md
1106 (*movti_aarch64): Use "multiple" for type where v8type is "move2".
1107 (*movtf_aarch64): Likewise.
1109 (thumb1_movdi_insn): Use "multiple" for type where more than one
1110 instruction is used for a move.
1111 (*arm32_movhf): Likewise.
1112 (*thumb_movdf_insn): Likewise.
1114 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1116 * config/arm/types.md (type): Rename fcpys to fmov.
1118 (*arm_movsi_vfp): Rename type fcpys as fmov.
1119 (*thumb2_movsi_vfp): Likewise
1120 (*movhf_vfp_neon): Likewise
1121 (*movhf_vfp): Likewise
1122 (*movsf_vfp): Likewise
1123 (*thumb2_movsf_vfp): Likewise
1124 (*movsfcc_vfp): Likewise
1125 (*thumb2_movsfcc_vfp): Likewise
1126 * config/aarch64/aarch64-simd.md
1127 (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
1128 * config/aarch64/aarch64.md
1129 (*movsi_aarch64): Replace type mov_reg with fmovs.
1130 (*movdi_aarch64): Likewise
1131 (*movsf_aarch64): Likewise
1132 (*movdf_aarch64): Likewise
1134 (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
1135 * config/arm/iwmmxt.md
1136 (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
1137 * config/arm/arm1020e.md: Update with new attributes.
1138 * config/arm/cortex-a15-neon.md: Update with new attributes.
1139 * config/arm/cortex-a5.md: Update with new attributes.
1140 * config/arm/cortex-a53.md: Update with new attributes.
1141 * config/arm/cortex-a7.md: Update with new attributes.
1142 * config/arm/cortex-a8-neon.md: Update with new attributes.
1143 * config/arm/cortex-a9.md: Update with new attributes.
1144 * config/arm/cortex-m4-fpu.md: Update with new attributes.
1145 * config/arm/cortex-r4f.md: Update with new attributes.
1146 * config/arm/marvell-pj4.md: Update with new attributes.
1147 * config/arm/vfp11.md: Update with new attributes.
1149 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1151 * config/aarch64/aarch64.md
1152 (*madd<mode>): Fix type attribute.
1153 (*maddsi_uxtw): Likewise.
1154 (*msub<mode>): Likewise.
1155 (*msubsi_uxtw): Likewise.
1156 (<su_optab>maddsidi4): Likewise.
1157 (<su_optab>msubsidi4): Likewise.
1159 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1161 * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
1162 * config/arm/arm.md (core_cycles): Remove fdiv.
1163 * config/arm/vfp.md:
1164 (*sqrtsf2_vfp): Update for attribute changes.
1165 (*sqrtdf2_vfp): Likewise.
1166 * config/aarch64/aarch64.md:
1167 (sqrt<mode>2): Update for attribute changes.
1168 * config/arm/arm1020e.md: Update with new attributes.
1169 * config/arm/cortex-a15-neon.md: Update with new attributes.
1170 * config/arm/cortex-a5.md: Update with new attributes.
1171 * config/arm/cortex-a53.md: Update with new attributes.
1172 * config/arm/cortex-a7.md: Update with new attributes.
1173 * config/arm/cortex-a8-neon.md: Update with new attributes.
1174 * config/arm/cortex-a9.md: Update with new attributes.
1175 * config/arm/cortex-m4-fpu.md: Update with new attributes.
1176 * config/arm/cortex-r4f.md: Update with new attributes.
1177 * config/arm/marvell-pj4.md: Update with new attributes.
1178 * config/arm/vfp11.md: Update with new attributes.
1180 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1182 * config/arm/types.md
1183 (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
1184 * config/aarch64/aarch64.md
1185 (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
1187 (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
1188 (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
1189 (float<GPI:mode><GPF:mode>2): Likewise.
1191 (*truncsisf2_vfp): Update with new attributes.
1192 (*truncsidf2_vfp): Likewise.
1193 (fixuns_truncsfsi2): Likewise.
1194 (fixuns_truncdfsi2): Likewise.
1195 (*floatsisf2_vfp): Likewise.
1196 (*floatsidf2_vfp): Likewise.
1197 (floatunssisf2): Likewise.
1198 (floatunssidf2): Likewise.
1199 (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
1200 (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
1201 * config/arm/arm1020e.md: Update with new attributes.
1202 * config/arm/cortex-a15-neon.md: Update with new attributes.
1203 * config/arm/cortex-a5.md: Update with new attributes.
1204 * config/arm/cortex-a53.md: Update with new attributes.
1205 * config/arm/cortex-a7.md: Update with new attributes.
1206 * config/arm/cortex-a8-neon.md: Update with new attributes.
1207 * config/arm/cortex-a9.md: Update with new attributes.
1208 * config/arm/cortex-m4-fpu.md: Update with new attributes.
1209 * config/arm/cortex-r4f.md: Update with new attributes.
1210 * config/arm/marvell-pj4.md: Update with new attributes.
1211 * config/arm/vfp11.md: Update with new attributes.
1213 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1215 * config/aarch64/arm_neon.h
1216 (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
1217 (vqtbx<1,2,3,4><q>_s8): Likewise.
1219 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1221 * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
1222 * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
1223 (add<mode>3): Add type attribute.
1224 (add<mode>3): Likewise.
1225 (usadd<mode>3): Likewise.
1226 (ssadd<mode>3): Likewise.
1227 (sub<mode>3): Likewise.
1228 (sub<mode>3): Likewise.
1229 (ussub<mode>3): Likewise.
1230 (sssub<mode>3): Likewise.
1231 (ssmulsa3): Likewise.
1232 (usmulusa3): Likewise.
1233 (arm_usatsihi): Likewise.
1235 (*movdi_vfp): Add types for all instructions.
1236 (*movdi_vfp_cortexa8): Likewise.
1237 (*movhf_vfp_neon): Likewise.
1238 (*movhf_vfp): Likewise.
1239 (*movdf_vfp): Likewise.
1240 (*thumb2_movdf_vfp): Likewise.
1241 (*thumb2_movdfcc_vfp): Likewise.
1242 * config/arm/arm.md: Add type attribute to all insn patterns.
1243 (*thumb1_adddi3): Add type attribute.
1244 (*arm_adddi3): Likewise.
1245 (*adddi_sesidi_di): Likewise.
1246 (*adddi_zesidi_di): Likewise.
1247 (*thumb1_addsi3): Likewise.
1248 (addsi3_compare0): Likewise.
1249 (*addsi3_compare0_scratch): Likewise.
1250 (*compare_negsi_si): Likewise.
1251 (cmpsi2_addneg): Likewise.
1252 (*addsi3_carryin_<optab>): Likewise.
1253 (*addsi3_carryin_alt2_<optab>): Likewise.
1254 (*addsi3_carryin_clobercc_<optab>): Likewise.
1255 (*subsi3_carryin): Likewise.
1256 (*subsi3_carryin_const): Likewise.
1257 (*subsi3_carryin_compare): Likewise.
1258 (*subsi3_carryin_compare_const): Likewise.
1259 (*arm_subdi3): Likewise.
1260 (*thumb_subdi3): Likewise.
1261 (*subdi_di_zesidi): Likewise.
1262 (*subdi_di_sesidi): Likewise.
1263 (*subdi_zesidi_di): Likewise.
1264 (*subdi_sesidi_di): Likewise.
1265 (*subdi_zesidi_ze): Likewise.
1266 (thumb1_subsi3_insn): Likewise.
1267 (*arm_subsi3_insn): Likewise.
1268 (*anddi3_insn): Likewise.
1269 (*anddi_zesidi_di): Likewise.
1270 (*anddi_sesdi_di): Likewise.
1271 (*ne_zeroextracts): Likewise.
1272 (*ne_zeroextracts): Likewise.
1273 (*ite_ne_zeroextr): Likewise.
1274 (*ite_ne_zeroextr): Likewise.
1275 (*anddi_notdi_di): Likewise.
1276 (*anddi_notzesidi): Likewise.
1277 (*anddi_notsesidi): Likewise.
1278 (andsi_notsi_si): Likewise.
1279 (thumb1_bicsi3): Likewise.
1280 (*iordi3_insn): Likewise.
1281 (*iordi_zesidi_di): Likewise.
1282 (*iordi_sesidi_di): Likewise.
1283 (*thumb1_iorsi3_insn): Likewise.
1284 (*xordi3_insn): Likewise.
1285 (*xordi_zesidi_di): Likewise.
1286 (*xordi_sesidi_di): Likewise.
1287 (*arm_xorsi3): Likewise.
1288 (*andsi_iorsi3_no): Likewise.
1289 (*smax_0): Likewise.
1290 (*smax_m1): Likewise.
1291 (*arm_smax_insn): Likewise.
1292 (*smin_0): Likewise.
1293 (*arm_smin_insn): Likewise.
1294 (*arm_umaxsi3): Likewise.
1295 (*arm_uminsi3): Likewise.
1296 (*minmax_arithsi): Likewise.
1297 (*minmax_arithsi_): Likewise.
1298 (*satsi_<SAT:code>): Likewise.
1299 (arm_ashldi3_1bit): Likewise.
1300 (arm_ashrdi3_1bit): Likewise.
1301 (arm_lshrdi3_1bit): Likewise.
1302 (*arm_negdi2): Likewise.
1303 (*thumb1_negdi2): Likewise.
1304 (*arm_negsi2): Likewise.
1305 (*thumb1_negsi2): Likewise.
1306 (*negdi_extendsid): Likewise.
1307 (*negdi_zero_extend): Likewise.
1308 (*arm_abssi2): Likewise.
1309 (*thumb1_abssi2): Likewise.
1310 (*arm_neg_abssi2): Likewise.
1311 (*thumb1_neg_abss): Likewise.
1312 (one_cmpldi2): Likewise.
1313 (extend<mode>di2): Likewise.
1314 (*compareqi_eq0): Likewise.
1315 (*arm_extendhisi2addsi): Likewise.
1316 (*arm_movdi): Likewise.
1317 (*thumb1_movdi_insn): Likewise.
1318 (*arm_movt): Likewise.
1319 (*thumb1_movsi_insn): Likewise.
1320 (pic_add_dot_plus_four): Likewise.
1321 (pic_add_dot_plus_eight): Likewise.
1322 (tls_load_dot_plus_eight): Likewise.
1323 (*thumb1_movhi_insn): Likewise.
1324 (*thumb1_movsf_insn): Likewise.
1325 (*movdf_soft_insn): Likewise.
1326 (*thumb_movdf_insn): Likewise.
1327 (cbranchsi4_insn): Likewise.
1328 (cbranchsi4_scratch): Likewise.
1329 (*negated_cbranchsi4): Likewise.
1330 (*tbit_cbranch): Likewise.
1331 (*tlobits_cbranch): Likewise.
1332 (*tstsi3_cbranch): Likewise.
1333 (*cbranchne_decr1): Likewise.
1334 (*addsi3_cbranch): Likewise.
1335 (*addsi3_cbranch_scratch): Likewise.
1336 (*arm_cmpdi_insn): Likewise.
1337 (*arm_cmpdi_unsig): Likewise.
1338 (*arm_cmpdi_zero): Likewise.
1339 (*thumb_cmpdi_zero): Likewise.
1340 (*deleted_compare): Likewise.
1341 (*mov_scc): Likewise.
1342 (*mov_negscc): Likewise.
1343 (*mov_notscc): Likewise.
1344 (*cstoresi_eq0_thumb1_insn): Likewise.
1345 (cstoresi_nltu_thumb1): Likewise.
1346 (cstoresi_ltu_thu): Likewise.
1347 (thumb1_addsi3_addgeu): Likewise.
1348 (*arm_jump): Likewise.
1349 (*thumb_jump): Likewise.
1350 (*check_arch2): Likewise.
1351 (arm_casesi_internal): Likewise.
1352 (thumb1_casesi_dispatch): Likewise.
1353 (*arm_indirect_jump): Likewise.
1354 (*thumb1_indirect_jump): Likewise.
1356 (*and_scc): Likewise.
1357 (*ior_scc): Likewise.
1358 (*compare_scc): Likewise.
1359 (*cond_move): Likewise.
1360 (*cond_arith): Likewise.
1361 (*cond_sub): Likewise.
1362 (*cmp_ite0): Likewise.
1363 (*cmp_ite1): Likewise.
1364 (*cmp_and): Likewise.
1365 (*cmp_ior): Likewise.
1366 (*ior_scc_scc): Likewise.
1367 (*ior_scc_scc_cmp): Likewise.
1368 (*and_scc_scc): Likewise.
1369 (*and_scc_scc_cmp): Likewise.
1370 (*and_scc_scc_nod): Likewise.
1371 (*negscc): Likewise.
1372 (movcond_addsi): Likewise.
1373 (movcond): Likewise.
1374 (*ifcompare_plus_move): Likewise.
1375 (*if_plus_move): Likewise.
1376 (*ifcompare_move_plus): Likewise.
1377 (*if_move_plus): Likewise.
1378 (*ifcompare_arith_arith): Likewise.
1379 (*if_arith_arith): Likewise.
1380 (*ifcompare_arith_move): Likewise.
1381 (*if_arith_move): Likewise.
1382 (*ifcompare_move_arith): Likewise.
1383 (*if_move_arith): Likewise.
1384 (*ifcompare_move_not): Likewise.
1385 (*if_move_not): Likewise.
1386 (*ifcompare_not_move): Likewise.
1387 (*if_not_move): Likewise.
1388 (*ifcompare_shift_move): Likewise.
1389 (*if_shift_move): Likewise.
1390 (*ifcompare_move_shift): Likewise.
1391 (*if_move_shift): Likewise.
1392 (*ifcompare_shift_shift): Likewise.
1393 (*ifcompare_not_arith): Likewise.
1394 (*ifcompare_arith_not): Likewise.
1395 (*if_arith_not): Likewise.
1396 (*ifcompare_neg_move): Likewise.
1397 (*if_neg_move): Likewise.
1398 (*ifcompare_move_neg): Likewise.
1399 (*if_move_neg): Likewise.
1400 (prologue_thumb1_interwork): Likewise.
1401 (*cond_move_not): Likewise.
1402 (*sign_extract_onebit): Likewise.
1403 (*not_signextract_onebit): Likewise.
1404 (stack_tie): Likewise.
1405 (align_4): Likewise.
1406 (align_8): Likewise.
1407 (consttable_end): Likewise.
1408 (consttable_1): Likewise.
1409 (consttable_2): Likewise.
1410 (consttable_4): Likewise.
1411 (consttable_8): Likewise.
1412 (consttable_16): Likewise.
1413 (*thumb1_tablejump): Likewise.
1414 (prefetch): Likewise.
1415 (force_register_use): Likewise.
1416 (thumb_eh_return): Likewise.
1417 (load_tp_hard): Likewise.
1418 (load_tp_soft): Likewise.
1419 (tlscall): Likewise.
1420 (*arm_movtas_ze): Likewise.
1421 (*arm_rev): Likewise.
1422 (*arm_revsh): Likewise.
1423 (*arm_rev16): Likewise.
1424 * config/arm/thumb2.md
1425 (*thumb2_smaxsi3): Likewise.
1426 (*thumb2_sminsi3): Likewise.
1427 (*thumb32_umaxsi3): Likewise.
1428 (*thumb2_uminsi3): Likewise.
1429 (*thumb2_negdi2): Likewise.
1430 (*thumb2_abssi2): Likewise.
1431 (*thumb2_neg_abss): Likewise.
1432 (*thumb2_movsi_insn): Likewise.
1433 (tls_load_dot_plus_four): Likewise.
1434 (*thumb2_movhi_insn): Likewise.
1435 (*thumb2_mov_scc): Likewise.
1436 (*thumb2_mov_negs): Likewise.
1437 (*thumb2_mov_negs): Likewise.
1438 (*thumb2_mov_nots): Likewise.
1439 (*thumb2_mov_nots): Likewise.
1440 (*thumb2_movsicc_): Likewise.
1441 (*thumb2_movsfcc_soft_insn): Likewise.
1442 (*thumb2_indirect_jump): Likewise.
1443 (*thumb2_and_scc): Likewise.
1444 (*thumb2_ior_scc): Likewise.
1445 (*thumb2_ior_scc_strict_it): Likewise.
1446 (*thumb2_cond_move): Likewise.
1447 (*thumb2_cond_arith): Likewise.
1448 (*thumb2_cond_ari): Likewise.
1449 (*thumb2_cond_sub): Likewise.
1450 (*thumb2_negscc): Likewise.
1451 (*thumb2_movcond): Likewise.
1452 (thumb2_casesi_internal): Likewise.
1453 (thumb2_casesi_internal_pic): Likewise.
1454 (*thumb2_alusi3_short): Likewise.
1455 (*thumb2_mov<mode>_shortim): Likewise.
1456 (*thumb2_addsi_short): Likewise.
1457 (*thumb2_subsi_short): Likewise.
1458 (thumb2_addsi3_compare0): Likewise.
1459 (*thumb2_cbz): Likewise.
1460 (*thumb2_cbnz): Likewise.
1461 (*thumb2_one_cmplsi2_short): Likewise.
1462 (*thumb2_negsi2_short): Likewise.
1463 (*orsi_notsi_si): Likewise.
1464 * config/arm/arm1020e.md: Update with new attributes.
1465 * config/arm/arm1026ejs.md: Update with new attributes.
1466 * config/arm/arm1136jfs.md: Update with new attributes.
1467 * config/arm/arm926ejs.md: Update with new attributes.
1468 * config/arm/cortex-a15.md: Update with new attributes.
1469 * config/arm/cortex-a5.md: Update with new attributes.
1470 * config/arm/cortex-a53.md: Update with new attributes.
1471 * config/arm/cortex-a7.md: Update with new attributes.
1472 * config/arm/cortex-a8.md: Update with new attributes.
1473 * config/arm/cortex-a9.md: Update with new attributes.
1474 * config/arm/cortex-m4.md: Update with new attributes.
1475 * config/arm/cortex-r4.md: Update with new attributes.
1476 * config/arm/fa526.md: Update with new attributes.
1477 * config/arm/fa606te.md: Update with new attributes.
1478 * config/arm/fa626te.md: Update with new attributes.
1479 * config/arm/fa726te.md: Update with new attributes.
1481 2013-09-06 James Greenhalgh <james.greenhalgh@arm.com>
1483 * config/aarch64/aarch64-simd.md
1484 (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
1485 <vwx> iterator to ensure correct register choice.
1486 (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1487 (aarch64_sqdmull_n<mode>): Likewise.
1488 (aarch64_sqdmull2_n<mode>_internal): Likewise.
1489 * config/aarch64/arm_neon.h
1490 (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
1491 (vml<as><q>_n_<su>16): Likewise.
1492 (vml<as>l_high_lane<q>_<su>16): Likewise.
1493 (vml<as>l_high_n_<su>16): Likewise.
1494 (vml<as>l_lane<q>_<su>16): Likewise.
1495 (vml<as>l_n_<su>16): Likewise.
1496 (vmul<q>_lane<q>_<su>16): Likewise.
1497 (vmul<q>_n_<su>16): Likewise.
1498 (vmull_lane<q>_<su>16): Likewise.
1499 (vmull_n_<su>16): Likewise.
1500 (vmull_high_lane<q>_<su>16): Likewise.
1501 (vmull_high_n_<su>16): Likewise.
1502 (vqrdmulh<q>_n_s16): Likewise.
1504 2013-09-06 Tejas Belagod <tejas.belagod@arm.com>
1506 * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
1507 have the correct lane parameter.
1509 2013-09-06 Richard Biener <rguenther@suse.de>
1511 * cfganal.c (control_dependences::~control_dependences):
1512 Properly free all of the vector.
1514 2013-09-06 Kirill Yukhin <kirill.yukhin@intel.com>
1517 * config/i386/i386.c (ix86_conditional_register_usage):
1518 Proper initialize extended SSE registers.
1520 2013-09-06 Jan Hubicka <jh@suse.cz>
1522 PR tree-optimization/58311
1523 * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
1525 2013-09-06 Jan Hubicka <jh@suse.cz>
1527 * Makefile.in (tree-sra.o): Update dependencies.
1528 * tree-sra.c: Include ipa-utils.h
1529 (scan_function): Use recursive_call_p.
1530 (has_caller_p): New function.
1531 (cgraph_for_node_and_aliases): Count also callers of aliases.
1533 2013-09-06 Jan Hubicka <jh@suse.cz>
1536 * cgraph.h (symtab_semantically_equivalent_p): Declare.
1537 * tree-tailcall.c: Include ipa-utils.h.
1538 (find_tail_calls): Use it.
1539 * ipa-pure-const.c (check_call): Likewise.
1540 * ipa-utils.c (recursive_call_p): New function.
1541 * ipa-utils.h (recursive_call_p): Dclare.
1542 * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
1543 (symtab_semantically_equivalent_p): New function.
1544 * Makefile.in (tree-tailcall.o): Update dependencies.
1546 2013-09-06 Eric Botcazou <ebotcazou@adacore.com>
1548 * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
1551 2013-09-06 Richard Biener <rguenther@suse.de>
1553 * lto-streamer.h (lto_global_var_decls): Remove.
1554 * Makefile.in (OBJS): Remove lto-symtab.o.
1555 (lto-symtab.o): Remove.
1556 (GTFILES): Remove lto-symtab.c
1557 * lto-symtab.c: Move to lto/
1559 2013-09-06 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1561 * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
1562 (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
1563 (UNSPEC_FPINT_RINT): New constant definitions.
1564 (FPINT, fpint_name, fpint_roundingmode): New integer iterator
1565 definition with 2 attributes.
1566 ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
1567 ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
1570 2013-09-06 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1572 * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
1573 ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
1574 Set the mnemonic attribute to "bcr_flush". Set the "z196prop"
1575 attribute to "z196_alone".
1576 * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
1579 2013-09-06 Richard Biener <rguenther@suse.de>
1581 * basic-block.h (class control_dependences): New.
1582 * tree-ssa-dce.c (control_dependence_map): Remove.
1584 (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
1585 (set_control_dependence_map_bit, clear_control_dependence_bitmap,
1586 find_pdom, find_control_dependence, find_all_control_dependences):
1588 (mark_control_dependent_edges_necessary,
1589 find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
1590 tree_dce_done, perform_tree_ssa_dce): Adjust.
1591 * cfganal.c (set_control_dependence_map_bit,
1592 clear_control_dependence_bitmap, find_pdom, find_control_dependence,
1593 find_all_control_dependences): Move from tree-ssa-dce.c and
1594 implement as methods of control_dependences class.
1595 (control_dependences::control_dependences): New.
1596 (control_dependences::~control_dependences): Likewise.
1597 (control_dependences::get_edges_dependent_on): Likewise.
1598 (control_dependences::get_edge): Likewise.
1600 2013-09-04 Jan Hubicka <jh@suse.cz>
1602 * tree.c (types_same_for_odr): Drop overactive check.
1603 * ipa-devirt.c (hash_type_name): Likewise.
1605 2013-09-04 Jan Hubicka <jh@suse.cz>
1607 * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
1608 (analyze_functions): ... here.
1610 2013-09-04 Jan Hubicka <jh@suse.cz>
1613 * cgraphunit.c (analyze_functions): Clear AUX fields
1614 after processing; initialize assembler name has.
1616 2013-09-05 Jeff Law <law@redhat.com>
1618 * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
1619 from thread_around_empty_block. Record threading path into PATH.
1620 Recurse if threading through the initial block is successful.
1621 (thread_across_edge): Corresponding changes to slightly simplify.
1623 2013-09-05 James Greenhalgh <james.greenhalgh@arm.com>
1625 * config/aarch64/aarch64.md
1626 (type): Remove frecpe, frecps, frecpx.
1627 (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
1628 fix to be a TARGET_SIMD instruction.
1629 (aarch64_frecps): Remove.
1630 * config/aarch64/aarch64-simd.md
1631 (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
1632 (aarch64_frecps<mode>): Handle all float/vector of float modes.
1634 2013-09-05 James Greenhalgh <james.greenhalgh@arm.com>
1635 Sofiane Naci <sofiane.naci@arm.com>
1637 * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
1638 into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
1639 Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
1640 "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
1641 "logic_reg", "logics_reg", "rev". Expand "arlo_shift" into
1642 "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
1643 "logics_shift_imm". Expand "arlo_shift_reg" into "alu_shift_reg",
1644 "alus_shift_reg", "logic_shift_reg", "logics_shift_reg". Expand "clz"
1645 into "clz, "rbit". Rename "shift" to "shift_imm".
1646 * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
1647 changes. Update for attribute changes all occurrences of arlo_* and
1649 * config/arm/arm-fixed.md: Update for attribute changes
1650 all occurrences of arlo_* types.
1651 * config/arm/thumb2.md: Update for attribute changes all occurrences
1653 * config/arm/arm.c (xscale_sched_adjust_cost): (rtx insn, rtx
1654 (cortexa7_older_only): Likewise.
1655 (cortexa7_younger): Likewise.
1656 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
1657 (1020alu_shift_op): Likewise.
1658 (1020alu_shift_reg_op): Likewise.
1659 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
1660 (alu_shift_op): Likewise.
1661 (alu_shift_reg_op): Likewise.
1662 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
1663 (11_alu_shift_op): Likewise.
1664 (11_alu_shift_reg_op): Likewise.
1665 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
1666 (9_alu_shift_reg_op): Likewise.
1667 * config/arm/cortex-a15.md (cortex_a15_alu): Update for
1669 (cortex_a15_alu_shift): Likewise.
1670 (cortex_a15_alu_shift_reg): Likewise.
1671 * config/arm/cortex-a5.md (cortex_a5_alu): Update for
1673 (cortex_a5_alu_shift): Likewise.
1674 * config/arm/cortex-a53.md (cortex_a53_alu): Update for
1676 (cortex_a53_alu_shift): Likewise.
1677 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
1679 (cortex_a7_alu_reg): Likewise.
1680 (cortex_a7_alu_shift): Likewise.
1681 * config/arm/cortex-a8.md (cortex_a8_alu): Update for
1683 (cortex_a8_alu_shift): Likewise.
1684 (cortex_a8_alu_shift_reg): Likewise.
1685 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
1686 (cortex_a9_dp_shift): Likewise.
1687 * config/arm/cortex-m4.md (cortex_m4_alu): Update for
1689 * config/arm/cortex-r4.md
1690 (cortex_r4_alu): Update for attribute changes.
1691 (cortex_r4_mov): Likewise.
1692 (cortex_r4_alu_shift_reg): Likewise.
1693 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
1694 (526_alu_shift_op): Likewise.
1695 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
1696 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
1697 (626te_alu_shift_op): Likewise.
1698 * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
1699 (726te_alu_shift_op): Likewise.
1700 (726te_alu_shift_reg_op): Likewise.
1701 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
1702 (mp626_alu_shift_op): Likewise.
1703 * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
1704 (pj4_alu_conds): Likewise.
1705 (pj4_shift): Likewise.
1706 (pj4_shift_conds): Likewise.
1707 (pj4_alu_shift): Likewise.
1708 (pj4_alu_shift_conds): Likewise.
1709 * config/aarch64/aarch64.md: Update for attribute change
1710 all occurrences of arlo_* and shift* types.
1712 2013-09-05 Mike Stump <mikestump@comcast.net>
1714 * tree.h: Move documentation for tree_function_decl to tree-core.h
1715 with the declaration.
1717 2013-09-05 Peter Bergner <bergner@vnet.ibm.com>
1720 * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
1721 looking for widest mode.
1723 2013-09-05 Eric Botcazou <ebotcazou@adacore.com>
1725 * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
1727 2013-09-05 Richard Biener <rguenther@suse.de>
1729 PR tree-optimization/58137
1730 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
1731 Do not create vectors of pointers.
1732 * tree-vect-loop.c (get_initial_def_for_induction): Use proper
1733 types for the components of the vector initializer.
1734 * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
1735 allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
1737 2013-09-05 Martin Jambor <mjambor@suse.cz>
1739 * ipa-prop.c (remove_described_reference): Accept missing references,
1740 return false if that hppens, otherwise return true.
1741 (cgraph_node_for_jfunc): New function.
1742 (try_decrement_rdesc_refcount): Likewise.
1743 (try_make_edge_direct_simple_call): Use them.
1744 (ipa_edge_removal_hook): Remove references from rdescs.
1745 (ipa_edge_duplication_hook): Clone rdescs and their references
1746 when the new edge has the same caller as the old one.
1747 * cgraph.c (cgraph_resolve_speculation): Remove speculative
1748 reference before removing any edges.
1750 2013-09-05 Richard Earnshaw <rearnsha@arm.com>
1752 * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
1754 * thumb2.md (thumb2_storewb_parisi): New pattern.
1756 2013-09-05 Yufeng Zhang <yufeng.zhang@arm.com>
1758 * config/aarch64/aarch64-option-extensions.def: Add
1759 AARCH64_OPT_EXTENSION of 'crc'.
1760 * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
1761 (AARCH64_ISA_CRC): Ditto.
1762 * doc/invoke.texi (-march and -mcpu feature modifiers): Add
1763 description of the CRC extension.
1765 2013-09-05 Alexander Ivchenko <alexander.ivchenko@intel.com>
1767 * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
1768 * config/rs6000/linux.h: Ditto.
1769 * alpha/linux.h: Ditto.
1770 * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
1771 no_c99_libc_has_function.
1772 * config/c6x/uclinux-elf.h: Ditto.
1773 * config/lm32/uclinux-elf.h: Ditto.
1774 * config/m68k/uclinux.h: Ditto.
1775 * config/moxie/uclinux.h: Ditto.
1776 * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
1777 (crisv32-*-linux*, cris-*-linux*): Ditto.
1778 * config/bfin/bfin.c: Include "tm_p.h".
1780 2013-09-05 Richard Biener <rguenther@suse.de>
1782 * tree-vect-loop.c (vect_analyze_loop_operations): Properly
1783 check for a definition without a basic-block.
1785 2013-09-05 James Greenhalgh <james.greenhalgh@arm.com>
1786 Sofiane Naci <sofiane.naci@arm.com>
1788 * config/aarch64/aarch64.md
1789 (*movti_aarch64): Rename r_2_f and f_2_r.
1790 (*movsf_aarch64): Likewise.
1791 (*movdf_aarch64): Likewise.
1792 (*movtf_aarch64): Likewise.
1793 (aarch64_movdi_<mode>low): Likewise.
1794 (aarch64_movdi_<mode>high): Likewise.
1795 (aarch64_mov<mode>high_di): Likewise.
1796 (aarch64_mov<mode>low_di): Likewise.
1797 (aarch64_movtilow_tilow): Likewise.
1798 * config/arm/arm.md (attribute "neon_type"): Delete. Move attribute
1799 values to config/arm/types.md
1800 (attribute "conds"): Update for attribute change.
1801 (anddi3_insn): Likewise.
1802 (iordi3_insn): Likewise.
1803 (xordi3_insn): Likewise.
1804 (one_cmpldi2): Likewise.
1805 * config/arm/types.md (type): Add Neon types.
1806 * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
1807 use "type" attribute.
1808 (movmisalign<mode>_neon_store): Likewise.
1809 (movmisalign<mode>_neon_load): Likewise.
1810 (vec_set<mode>_internal): Likewise.
1811 (vec_setv2di_internal): Likewise.
1812 (vec_extract<mode>): Likewise.
1813 (vec_extractv2di): Likewise.
1814 (add<mode>3_neon): Likewise.
1815 (adddi3_neon): Likewise.
1816 (sub<mode>3_neon): Likewise.
1817 (subdi3_neon): Likewise.
1818 (mul<mode>3_neon): Likewise.
1819 (mul<mode>3add<mode>_neon): Likewise.
1820 (mul<mode>3neg<mode>add<mode>_neon): Likewise.
1821 (fma<VCVTF:mode>4)): Likewise.
1822 (fma<VCVTF:mode>4_intrinsic): Likewise.
1823 (fmsub<VCVTF:mode>4)): Likewise.
1824 (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1825 (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1826 (ior<mode>3): Likewise.
1827 (and<mode>3): Likewise.
1828 (anddi3_neon): Likewise.
1829 (orn<mode>3_neon): Likewise.
1830 (orndi3_neon): Likewise.
1831 (bic<mode>3_neon): Likewise.
1832 (bicdi3_neon): Likewise.
1833 (xor<mode>3): Likewise.
1834 (one_cmpl<mode>2): Likewise.
1835 (abs<mode>2): Likewise.
1836 (neg<mode>2): Likewise.
1837 (umin<mode>3_neon): Likewise.
1838 (umax<mode>3_neon): Likewise.
1839 (smin<mode>3_neon): Likewise.
1840 (smax<mode>3_neon): Likewise.
1841 (vashl<mode>3): Likewise.
1842 (vashr<mode>3_imm): Likewise.
1843 (vlshr<mode>3_imm): Likewise.
1844 (ashl<mode>3_signed): Likewise.
1845 (ashl<mode>3_unsigned): Likewise.
1846 (neon_load_count): Likewise.
1847 (ashldi3_neon_noclobber): Likewise.
1848 (signed_shift_di3_neon): Likewise.
1849 (unsigned_shift_di3_neon): Likewise.
1850 (ashrdi3_neon_imm_noclobber): Likewise.
1851 (lshrdi3_neon_imm_noclobber): Likewise.
1852 (widen_ssum<mode>3): Likewise.
1853 (widen_usum<mode>3): Likewise.
1854 (quad_halves_<code>v4si): Likewise.
1855 (quad_halves_<code>v4sf): Likewise.
1856 (quad_halves_<code>v8hi): Likewise.
1857 (quad_halves_<code>v16qi): Likewise.
1858 (reduc_splus_v2di): Likewise.
1859 (neon_vpadd_internal<mode>): Likewise.
1860 (neon_vpsmin<mode>): Likewise.
1861 (neon_vpsmax<mode>): Likewise.
1862 (neon_vpumin<mode>): Likewise.
1863 (neon_vpumax<mode>): Likewise.
1864 (ss_add<mode>_neon): Likewise.
1865 (us_add<mode>_neon): Likewise.
1866 (ss_sub<mode>_neon): Likewise.
1867 (us_sub<mode>_neon): Likewise.
1868 (neon_vadd<mode>_unspec): Likewise.
1869 (neon_vaddl<mode>): Likewise.
1870 (neon_vaddw<mode>): Likewise.
1871 (neon_vhadd<mode>): Likewise.
1872 (neon_vqadd<mode>): Likewise.
1873 (neon_vaddhn<mode>): Likewise.
1874 (neon_vmul<mode>): Likewise.
1875 (neon_vmla<mode>): Likewise.
1876 (neon_vmlal<mode>): Likewise.
1877 (neon_vmls<mode>): Likewise.
1878 (neon_vmlsl<mode>): Likewise.
1879 (neon_vqdmulh<mode>): Likewise.
1880 (neon_vqdmlal<mode>): Likewise.
1881 (neon_vqdmlsl<mode>): Likewise.
1882 (neon_vmull<mode>): Likewise.
1883 (neon_vqdmull<mode>): Likewise.
1884 (neon_vsub<mode>_unspec): Likewise.
1885 (neon_vsubl<mode>): Likewise.
1886 (neon_vsubw<mode>): Likewise.
1887 (neon_vqsub<mode>): Likewise.
1888 (neon_vhsub<mode>): Likewise.
1889 (neon_vsubhn<mode>): Likewise.
1890 (neon_vceq<mode>): Likewise.
1891 (neon_vcge<mode>): Likewise.
1892 (neon_vcgeu<mode>): Likewise.
1893 (neon_vcgt<mode>): Likewise.
1894 (neon_vcgtu<mode>): Likewise.
1895 (neon_vcle<mode>): Likewise.
1896 (neon_vclt<mode>): Likewise.
1897 (neon_vcage<mode>): Likewise.
1898 (neon_vcagt<mode>): Likewise.
1899 (neon_vtst<mode>): Likewise.
1900 (neon_vabd<mode>): Likewise.
1901 (neon_vabdl<mode>): Likewise.
1902 (neon_vaba<mode>): Likewise.
1903 (neon_vabal<mode>): Likewise.
1904 (neon_vmax<mode>): Likewise.
1905 (neon_vmin<mode>): Likewise.
1906 (neon_vpaddl<mode>): Likewise.
1907 (neon_vpadal<mode>): Likewise.
1908 (neon_vpmax<mode>): Likewise.
1909 (neon_vpmin<mode>): Likewise.
1910 (neon_vrecps<mode>): Likewise.
1911 (neon_vrsqrts<mode>): Likewise.
1912 (neon_vqabs<mode>): Likewise.
1913 (neon_vqneg<mode>): Likewise.
1914 (neon_vcls<mode>): Likewise.
1915 (clz<mode>2): Likewise.
1916 (popcount<mode>2): Likewise.
1917 (neon_vrecpe): Likewise.
1918 (neon_vrsqrte): Likewise.
1919 (neon_vget_lane<mode>_sext_internal): Likewise.
1920 (neon_vget_lane<mode>_zext_internal): Likewise.
1921 (neon_vdup_n<mode>): Likewise.
1922 (neon_vdup_nv2di): Likewise.
1923 (neon_vdpu_lane<mode>_internal): Likewise.
1924 (neon_vswp<mode>): Likewise.
1925 (float<mode><V_cvtto>2): Likewise.
1926 (floatuns<mode><V_cvtto>2): Likewise.
1927 (fix_trunc<mode><V_cvtto>)2): Likewise
1928 (fixuns_trunc<mode><V_cvtto)2): Likewise.
1929 (neon_vcvt<mode>): Likewise.
1930 (neon_vcvtv4sfv4hf): Likewise.
1931 (neon_vcvtv4hfv4sf): Likewise.
1932 (neon_vcvt_n<mode>): Likewise.
1933 (neon_vmovn<mode>): Likewise.
1934 (neon_vqmovn<mode>): Likewise.
1935 (neon_vqmovun<mode>): Likewise.
1936 (neon_vmovl<mode>): Likewise.
1937 (neon_vmul_lane<mode>): Likewise.
1938 (neon_vmull_lane<mode>): Likewise.
1939 (neon_vqdmull_lane<mode>): Likewise.
1940 (neon_vqdmulh_lane<mode>): Likewise.
1941 (neon_vmla_lane<mode>): Likewise.
1942 (neon_vmlal_lane<mode>): Likewise.
1943 (neon_vqdmlal_lane<mode>): Likewise.
1944 (neon_vmls_lane<mode>): Likewise.
1945 (neon_vmlsl_lane<mode>): Likewise.
1946 (neon_vqdmlsl_lane<mode>): Likewise.
1947 (neon_vext<mode>): Likewise.
1948 (neon_vrev64<mode>): Likewise.
1949 (neon_vrev32<mode>): Likewise.
1950 (neon_vrev16<mode>): Likewise.
1951 (neon_vbsl<mode>_internal): Likewise.
1952 (neon_vshl<mode>): Likewise.
1953 (neon_vqshl<mode>): Likewise.
1954 (neon_vshr_n<mode>): Likewise.
1955 (neon_vshrn_n<mode>): Likewise.
1956 (neon_vqshrn_n<mode>): Likewise.
1957 (neon_vqshrun_n<mode>): Likewise.
1958 (neon_vshl_n<mode>): Likewise.
1959 (neon_vqshl_n<mode>): Likewise.
1960 (neon_vqshlu_n<mode>): Likewise.
1961 (neon_vshll_n<mode>): Likewise.
1962 (neon_vsra_n<mode>): Likewise.
1963 (neon_vsri_n<mode>): Likewise.
1964 (neon_vsli_n<mode>): Likewise.
1965 (neon_vtbl1v8qi): Likewise.
1966 (neon_vtbl2v8qi): Likewise.
1967 (neon_vtbl3v8qi): Likewise.
1968 (neon_vtbl4v8qi): Likewise.
1969 (neon_vtbx1v8qi): Likewise.
1970 (neon_vtbx2v8qi): Likewise.
1971 (neon_vtbx3v8qi): Likewise.
1972 (neon_vtbx4v8qi): Likewise.
1973 (neon_vtrn<mode>_internal): Likewise.
1974 (neon_vzip<mode>_internal): Likewise.
1975 (neon_vuzp<mode>_internal): Likewise.
1976 (neon_vld1<mode>): Likewise.
1977 (neon_vld1_lane<mode>): Likewise.
1978 (neon_vld1_dup<mode>): Likewise.
1979 (neon_vld1_dupv2di): Likewise.
1980 (neon_vst1<mode>): Likewise.
1981 (neon_vst1_lane<mode>): Likewise.
1982 (neon_vld2<mode>): Likewise.
1983 (neon_vld2_lane<mode>): Likewise.
1984 (neon_vld2_dup<mode>): Likewise.
1985 (neon_vst2<mode>): Likewise.
1986 (neon_vst2_lane<mode>): Likewise.
1987 (neon_vld3<mode>): Likewise.
1988 (neon_vld3qa<mode>): Likewise.
1989 (neon_vld3qb<mode>): Likewise.
1990 (neon_vld3_lane<mode>): Likewise.
1991 (neon_vld3_dup<mode>): Likewise.
1992 (neon_vst3<mode>): Likewise.
1993 (neon_vst3qa<mode>): Likewise.
1994 (neon_vst3qb<mode>): Likewise.
1995 (neon_vst3_lane<mode>): Likewise.
1996 (neon_vld4<mode>): Likewise.
1997 (neon_vld4qa<mode>): Likewise.
1998 (neon_vld4qb<mode>): Likewise.
1999 (neon_vld4_lane<mode>): Likewise.
2000 (neon_vld4_dup<mode>): Likewise.
2001 (neon_vst4<mode>): Likewise.
2002 (neon_vst4qa<mode>): Likewise.
2003 (neon_vst4qb<mode>): Likewise.
2004 (neon_vst4_lane<mode>): Likewise.
2005 (neon_vec_unpack<US>_lo_<mode>): Likewise.
2006 (neon_vec_unpack<US>_hi_<mode>): Likewise.
2007 (neon_vec_<US>mult_lo_<mode>): Likewise.
2008 (neon_vec_<US>mult_hi_<mode>): Likewise.
2009 (neon_vec_<US>shiftl_<mode>): Likewise.
2010 (neon_unpack<US>_<mode>): Likewise.
2011 (neon_vec_<US>mult_<mode>): Likewise.
2012 (vec_pack_trunc_<mode>): Likewise.
2013 (neon_vec_pack_trunk_<mode>): Likewise.
2014 (neon_vabd<mode>_2): Likewise.
2015 (neon_vabd<mode>_3): Likewise.
2016 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
2017 (thumb2_movsi_vfp): Likewise.
2018 (movdi_vfp): Likewise.
2019 (movdi_vfp_cortexa8): Likewise.
2020 (movhf_vfp_neon): Likewise.
2021 (movhf_vfp): Likewiwse.
2022 (movsf_vfp): Likewiwse.
2023 (thumb2_movsf_vfp): Likewiwse.
2024 (movdf_vfp): Likewise.
2025 (thumb2_movdf_vfp): Likewise.
2026 (movsfcc_vfp): Likewise.
2027 (thumb2_movsfcc_vfp): Likewise.
2028 (movdfcc_vfp): Likewise.
2029 (thumb2_movdfcc_vfp): Likewise.
2030 * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
2031 * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
2032 (v10_v2c): Likewise.
2033 * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
2035 (cortex_a15_neon_int_2): Likewise.
2036 (cortex_a15_neon_int_3): Likewise.
2037 (cortex_a15_neon_int_4): Likewise.
2038 (cortex_a15_neon_int_5): Likewise.
2039 (cortex_a15_neon_vqneg_vqabs): Likewise.
2040 (cortex_a15_neon_vmov): Likewise.
2041 (cortex_a15_neon_vaba): Likewise.
2042 (cortex_a15_neon_vaba_qqq): Likewise.
2043 (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2044 (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2045 (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
2046 scalar_64_32_long_scalar): Likewise.
2047 (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2048 (cortex_a15_neon_mla_qqq_8_16): Likewise.
2049 (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2050 lotype_qdd_64_32_long): Likewise.
2051 (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2052 (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2053 (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
2054 (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2055 (cortex_a15_neon_shift_1): Likewise.
2056 (cortex_a15_neon_shift_2): Likewise.
2057 (cortex_a15_neon_shift_3): Likewise.
2058 (cortex_a15_neon_vshl_ddd): Likewise.
2059 (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2060 (cortex_a15_neon_vsra_vrsra): Likewise.
2061 (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
2062 (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
2063 (cortex_a15_neon_fp_vmul_ddd): Likewise.
2064 (cortex_a15_neon_fp_vmul_qqd): Likewise.
2065 (cortex_a15_neon_fp_vmla_ddd): Likewise.
2066 (cortex_a15_neon_fp_vmla_qqq): Likewise.
2067 (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
2068 (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
2069 (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2070 (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2071 (cortex_a15_neon_bp_simple): Likewise.
2072 (cortex_a15_neon_bp_2cycle): Likewise.
2073 (cortex_a15_neon_bp_3cycle): Likewise.
2074 (cortex_a15_neon_vld1_1_2_regs): Likewise.
2075 (cortex_a15_neon_vld1_3_4_regs): Likewise.
2076 (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2077 (cortex_a15_neon_vld2_4_regs): Likewise.
2078 (cortex_a15_neon_vld3_vld4): Likewise.
2079 (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2080 (cortex_a15_neon_vst1_3_4_regs): Likewise.
2081 (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
2082 (cortex_a15_neon_vst3_vst4): Likewise.
2083 (cortex_a15_neon_vld1_vld2_lane): Likewise.
2084 (cortex_a15_neon_vld3_vld4_lane" 10
2085 (cortex_a15_neon_vst1_vst2_lane): Likewise.
2086 (cortex_a15_neon_vst3_vst4_lane): Likewise.
2087 (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
2088 (cortex_a15_neon_ldm_2): Likewise.0
2089 (cortex_a15_neon_stm_2): Likewise.
2090 (cortex_a15_neon_mcr): Likewise.
2091 (cortex_a15_neon_mcr_2_mcrr): Likewise.
2092 (cortex_a15_neon_mrc): Likewise.
2093 (cortex_a15_neon_mrrc): Likewise.
2094 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
2096 (cortex_a15_alu_shift): Likewise.
2097 (cortex_a15_alu_shift_reg): Likewise.
2098 (cortex_a15_mult32): Likewise.
2099 (cortex_a15_mult64): Likewise.
2100 (cortex_a15_block): Likewise.
2101 (cortex_a15_branch): Likewise.
2102 (cortex_a15_load1): Likewise.
2103 (cortex_a15_load3): Likewise.
2104 (cortex_a15_store1): Likewise.
2105 (cortex_a15_store3): Likewise.
2106 (cortex_a15_call): Likewise.
2107 * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
2108 (cortex_a5_f2r): Likewise.
2109 * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
2111 (cortex_a53_f2r): Likewise.
2112 * config/arm/cortex-a7.md
2113 (cortex_a7_branch): Update for attribute change.
2114 (cortex_a7_call): Likewise.
2115 (cortex_a7_alu_imm): Likewise.
2116 (cortex_a7_alu_reg): Likewise.
2117 (cortex_a7_alu_shift): Likewise.
2118 (cortex_a7_mul): Likewise.
2119 (cortex_a7_load1): Likewise.
2120 (cortex_a7_store1): Likewise.
2121 (cortex_a7_load2): Likewise.
2122 (cortex_a7_store2): Likewise.
2123 (cortex_a7_load3): Likewise.
2124 (cortex_a7_store3): Likewise.
2125 (cortex_a7_load4): Likewise.
2126 (cortex_a7_store4): Likewise.
2127 (cortex_a7_fpalu): Likewise.
2128 (cortex_a7_fconst): Likewise.
2129 (cortex_a7_fpmuls): Likewise.
2130 (cortex_a7_neon_mul): Likewise.
2131 (cortex_a7_fpmacs): Likewise.
2132 (cortex_a7_neon_mla: Likewise.
2133 (cortex_a7_fpmuld: Likewise.
2134 (cortex_a7_fpmacd: Likewise.
2135 (cortex_a7_fpfmad: Likewise.
2136 (cortex_a7_fdivs: Likewise.
2137 (cortex_a7_fdivd: Likewise.
2138 (cortex_a7_r2f: Likewise.
2139 (cortex_a7_f2r: Likewise.
2140 (cortex_a7_f_flags: Likewise.
2141 (cortex_a7_f_loads: Likewise.
2142 (cortex_a7_f_loadd: Likewise.
2143 (cortex_a7_f_stores: Likewise.
2144 (cortex_a7_f_stored: Likewise.
2145 (cortex_a7_neon): Likewise.
2146 * config/arm/cortex-a8-neon.md
2147 (cortex_a8_neon_mrc): Update for attribute change.
2148 (cortex_a8_neon_mrrc): Likewise.
2149 (cortex_a8_neon_int_1): Likewise.
2150 (cortex_a8_neon_int_2): Likewise.
2151 (cortex_a8_neon_int_3): Likewise.
2152 (cortex_a8_neon_int_4): Likewise.
2153 (cortex_a8_neon_int_5): Likewise.
2154 (cortex_a8_neon_vqneg_vqabs): Likewise.
2155 (cortex_a8_neon_vmov): Likewise.
2156 (cortex_a8_neon_vaba): Likewise.
2157 (cortex_a8_neon_vaba_qqq): Likewise.
2158 (cortex_a8_neon_vsma): Likewise.
2159 (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2160 (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2161 (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
2162 long_scalar): Likewise.
2163 (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2164 (cortex_a8_neon_mla_qqq_8_16): Likewise.
2165 (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2166 long_scalar_qdd_64_32_long): Likewise.
2167 (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2168 (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2169 (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
2170 (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2171 (cortex_a8_neon_shift_1): Likewise.
2172 (cortex_a8_neon_shift_2): Likewise.
2173 (cortex_a8_neon_shift_3): Likewise.
2174 (cortex_a8_neon_vshl_ddd): Likewise.
2175 (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2176 (cortex_a8_neon_vsra_vrsra): Likewise.
2177 (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
2178 (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
2179 (cortex_a8_neon_fp_vsum): Likewise.
2180 (cortex_a8_neon_fp_vmul_ddd): Likewise.
2181 (cortex_a8_neon_fp_vmul_qqd): Likewise.
2182 (cortex_a8_neon_fp_vmla_ddd): Likewise.
2183 (cortex_a8_neon_fp_vmla_qqq): Likewise.
2184 (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
2185 (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
2186 (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2187 (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2188 (cortex_a8_neon_bp_simple): Likewise.
2189 (cortex_a8_neon_bp_2cycle): Likewise.
2190 (cortex_a8_neon_bp_3cycle): Likewise.
2191 (cortex_a8_neon_ldr): Likewise.
2192 (cortex_a8_neon_str): Likewise.
2193 (cortex_a8_neon_vld1_1_2_regs): Likewise.
2194 (cortex_a8_neon_vld1_3_4_regs): Likewise.
2195 (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2196 (cortex_a8_neon_vld2_4_regs): Likewise.
2197 (cortex_a8_neon_vld3_vld4): Likewise.
2198 (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2199 (cortex_a8_neon_vst1_3_4_regs): Likewise.
2200 (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
2201 (cortex_a8_neon_vst3_vst4): Likewise.
2202 (cortex_a8_neon_vld1_vld2_lane): Likewise.
2203 (cortex_a8_neon_vld3_vld4_lane): Likewise.
2204 (cortex_a8_neon_vst1_vst2_lane): Likewise.
2205 (cortex_a8_neon_vst3_vst4_lane): Likewise.
2206 (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
2207 (cortex_a8_neon_mcr): Likewise.
2208 (cortex_a8_neon_mcr_2_mcrr): Likewise.
2209 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
2210 * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
2212 (ca9_neon_mrrc): Likewise.
2213 (cortex_a9_neon_int_1): Likewise.
2214 (cortex_a9_neon_int_2): Likewise.
2215 (cortex_a9_neon_int_3): Likewise.
2216 (cortex_a9_neon_int_4): Likewise.
2217 (cortex_a9_neon_int_5): Likewise.
2218 (cortex_a9_neon_vqneg_vqabs): Likewise.
2219 (cortex_a9_neon_vmov): Likewise.
2220 (cortex_a9_neon_vaba): Likewise.
2221 (cortex_a9_neon_vaba_qqq): Likewise.
2222 (cortex_a9_neon_vsma): Likewise.
2223 (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2224 (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
2225 (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
2226 long_scalar): Likewise.
2227 (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
2228 (cortex_a9_neon_mla_qqq_8_16): Likewise.
2229 (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
2230 long_scalar_qdd_64_32_long): Likewise.
2231 (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
2232 (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
2233 (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
2234 (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
2235 (cortex_a9_neon_shift_1): Likewise.
2236 (cortex_a9_neon_shift_2): Likewise.
2237 (cortex_a9_neon_shift_3): Likewise.
2238 (cortex_a9_neon_vshl_ddd): Likewise.
2239 (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
2240 (cortex_a9_neon_vsra_vrsra): Likewise.
2241 (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
2242 (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
2243 (cortex_a9_neon_fp_vsum): Likewise.
2244 (cortex_a9_neon_fp_vmul_ddd): Likewise.
2245 (cortex_a9_neon_fp_vmul_qqd): Likewise.
2246 (cortex_a9_neon_fp_vmla_ddd): Likewise.
2247 (cortex_a9_neon_fp_vmla_qqq): Likewise.
2248 (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
2249 (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
2250 (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
2251 (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
2252 (cortex_a9_neon_bp_simple): Likewise.
2253 (cortex_a9_neon_bp_2cycle): Likewise.
2254 (cortex_a9_neon_bp_3cycle): Likewise.
2255 (cortex_a9_neon_ldr): Likewise.
2256 (cortex_a9_neon_str): Likewise.
2257 (cortex_a9_neon_vld1_1_2_regs): Likewise.
2258 (cortex_a9_neon_vld1_3_4_regs): Likewise.
2259 (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
2260 (cortex_a9_neon_vld2_4_regs): Likewise.
2261 (cortex_a9_neon_vld3_vld4): Likewise.
2262 (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
2263 (cortex_a9_neon_vst1_3_4_regs): Likewise.
2264 (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
2265 (cortex_a9_neon_vst3_vst4): Likewise.
2266 (cortex_a9_neon_vld1_vld2_lane): Likewise.
2267 (cortex_a9_neon_vld3_vld4_lane): Likewise.
2268 (cortex_a9_neon_vst1_vst2_lane): Likewise.
2269 (cortex_a9_neon_vst3_vst4_lane): Likewise.
2270 (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
2271 (cortex_a9_neon_mcr): Likewise.
2272 (cortex_a9_neon_mcr_2_mcrr): Likewise.
2273 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
2274 (cortex_a9_fps): Likewise.
2275 * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
2277 (cortex_m4_fmuls): Likewise.
2278 * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
2280 (cortex_r4_mrc): Likewise.
2281 * config/arm/iterators.md: Update comment referring to neon_type.
2282 * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
2283 (iwmmxt_movsi_insn): Likewise.
2284 * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
2286 (pj4_core_to_vfp): Likewise.
2287 * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
2289 * config/arm/vfp11.md (vfp_fload): Update for attribute change.
2290 (vfp_fstore): Likewise.
2291 * doc/md.texi: Change references to neon_type to refer to type.
2293 2013-09-04 Dodji Seketeli <dodji@redhat.com>
2295 * tree.h (DECL_BUILT_IN): Fix typo in comment.
2297 2013-09-04 David Edelsohn <dje.gcc@gmail.com>
2299 * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
2302 2013-09-04 Easwaran Raman <eraman@google.com>
2305 * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
2306 (build_and_add_sum): Use it.
2307 (appears_later_in_bb): Simplify code.
2309 2013-09-04 Teresa Johnson <tejohnson@google.com>
2311 * dumpfile.c (dump_finish): Don't close stderr/stdout.
2313 2013-09-04 James Greenhalgh <james.greenhalgh@arm.com>
2315 * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
2317 2013-09-04 Jan Hubicka <jh@suse.cz>
2319 * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
2320 * ipa-devirt.c: Include diganostic.h
2321 (odr_type_d): Add types and types_set.
2322 (hash_type_name): Work for types with vtables during LTO.
2323 (odr_hasher::remove): Fix comment; destroy types_set.
2324 (add_type_duplicate): New function,
2325 (get_odr_type): Use it.
2326 (dump_type_inheritance_graph): Dump type duplicates.
2327 * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
2329 * tree.c (types_same_for_odr): Give exact answers on types with
2332 2013-09-04 Dodji Seketeli <dodji@redhat.com>
2334 * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
2335 explaining their differences.
2337 2013-09-04 Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
2339 * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
2341 2013-09-03 Jeff Law <law@redhat.com>
2343 * tree-ssa-threadedge.c (thread_across_edge): Record entire path
2344 when not threading through a joiner block. Pass joiner/no joiner
2345 state to register_jump_thread.
2346 * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
2347 state from argument rather than implying on path length.
2348 Dump the entire jump thread path into debugging dump.
2349 * tree-flow.h (register_jump_thread): Update prototype.
2351 2013-08-29 Xinliang David Li <davidxl@google.com>
2353 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
2354 Remove a trivial gcc_assert.
2356 2013-08-29 Xinliang David Li <davidxl@google.com>
2358 * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
2359 * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
2360 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
2361 Delay base decl alignment adjustment.
2362 * tree-vectorizer.c (vect_destroy_datarefs): New function.
2363 * tree-vectorizer.h: New data structure.
2364 (set_dr_misalignment): New function.
2365 (dr_misalignment): Ditto.
2366 * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
2367 (vectorizable_load): Ditto.
2368 (ensure_base_align): New function.
2369 (vectorize_loops): Add dbg_cnt support.
2370 (execute_vect_slp): Ditto.
2371 * dbgcnt.def: New debug counter.
2372 * Makefile: New dependency.
2374 2013-09-03 Meador Inge <meadori@codesourcery.com>
2378 2013-08-30 Meador Inge <meadori@codesourcery.com>
2380 * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2382 2013-09-03 David Edelsohn <dje.gcc@gmail.com>
2384 * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
2385 function descriptor.
2387 2013-09-03 Richard Biener <rguenther@suse.de>
2389 * tree-affine.c (add_elt_to_tree): Fix association issue,
2390 avoid useless converts and make sure to always return a
2391 properly typed result.
2393 2013-09-03 Richard Biener <rguenther@suse.de>
2396 * fold-const.c (negate_expr_p): Fix division case.
2397 (negate_expr): Likewise.
2399 2013-09-03 Richard Biener <rguenther@suse.de>
2402 * tree-streamer-out.c: Include tm.h.
2403 * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
2405 2013-09-03 Jan Hubicka <jh@suse.cz>
2407 * tree-profile.c (tree_profiling): Cleanup CFG when done.
2409 2013-09-03 Alan Modra <amodra@gmail.com>
2411 * config.gcc (powerpc*-*-linux*): Add support for little-endian
2412 multilibs to big-endian target and vice versa.
2413 * config/rs6000/t-linux64: Use := assignment on all vars.
2414 (MULTILIB_EXTRA_OPTS): Remove fPIC.
2415 (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
2416 * config/rs6000/t-linux64le: New file.
2417 * config/rs6000/t-linux64bele: New file.
2418 * config/rs6000/t-linux64lebe: New file.
2420 2013-09-02 Jan Hubicka <jh@suse.cz>
2422 * ipa-inline-transform.c (inline_transform): Do not
2423 optimize_inline_calls when not optimizing.
2425 2013-09-02 Jan Hubicka <jh@suse.cz>
2427 * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
2428 duplicated nodes for assembler names.
2429 * symtab.c (symtab_unregister_node): Do not attempt to unlink
2430 hard registers from assembler name hash.
2432 2013-09-02 Jan Hubicka <jh@suse.cz>
2434 * ipa-split.c (execute_split_functions): Split externally visible
2435 functions called once.
2437 2013-09-02 Martin Jambor <mjambor@suse.cz>
2440 * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
2441 linked list. When finding the correct duplicate, also consider also
2442 the caller in additon to its inlined_to node.
2444 2013-09-02 James Greenhalgh <james.greenhalgh@arm.com>
2446 * config/aarch64/aarch64-simd-builtins.def
2447 (dup_lane_scalar): Remove.
2448 * config/aarch64/aarch64-simd.md
2449 (aarch64_simd_dup): Add 'w->w' alternative.
2450 (aarch64_dup_lane<mode>): Allow for VALL.
2451 (aarch64_dup_lane_scalar<mode>): Remove.
2452 (aarch64_dup_lane_<vswap_width_name><mode>): New.
2453 (aarch64_get_lane_signed<mode>): Add w->w altenative.
2454 (aarch64_get_lane_unsigned<mode>): Likewise.
2455 (aarch64_get_lane<mode>): Likewise.
2456 * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
2457 (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
2458 * config/aarch64/iterators.md (VSWAP_WIDTH): New.
2459 (VCON): Change container of V2SF.
2460 (vswap_width_name): Likewise.
2461 * config/aarch64/arm_neon.h
2462 (__aarch64_vdup_lane_any): New.
2463 (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2464 (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
2465 (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
2467 2013-09-02 Eric Botcazou <ebotcazou@adacore.com>
2470 * expr.c (emit_move_complex): Do not move complex FP values as parts if
2471 the source or the destination is a single hard register.
2473 2013-09-02 Richard Biener <rguenther@suse.de>
2476 * tree-scalar-evolution.c (instantiate_scev_name): Allow
2479 2013-09-02 Richard Biener <rguenther@suse.de>
2481 * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
2482 arithmetic to sizetype.
2484 2013-09-02 Bin Cheng <bin.cheng@arm.com>
2486 * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
2487 Find auto-increment use both before and after candidate.
2489 2013-09-02 Marek Polacek <polacek@redhat.com>
2491 * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
2493 2013-09-01 Jan Hubicka <jh@suse.cz>
2495 * Makefile.in: Add ipa-profile.o
2496 (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
2497 * cgraph.c (struct cgraph_propagate_frequency_data,
2498 cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
2499 ipa-profile.c; replace cgraph_ by ipa_ prefix.
2500 * cgraph.h (cgraph_propagate_frequency): Remove.
2501 * ipa-inline-analysis.c: Include ipa-utils.h;
2502 drop duplicated cfgloop.h.
2503 (inline_update_callee_summaries): Update.
2504 * ipa-profile.c: New file.
2505 * ipa-utils.h (ipa_propagate_frequency): Declare.
2506 * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
2507 data-streamer.h, value-prof.h.
2508 (symtab_remove_unreachable_nodes): Update profile.
2509 (struct histogram_entry, histogram, histogram_pool, histogram_hash,
2510 account_time_size, cmp_counts, dump_histogram,
2511 ipa_profile_generate_summary, ipa_profile_write_summary,
2512 ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
2513 pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
2514 Move to ipa-profile.c.
2516 2013-09-01 John David Anglin <danglin@gcc.gnu.org>
2518 * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
2520 2013-09-01 Jan Hubicka <jh@suse.cz>
2522 * common.opt (fdevirtualize-speculatively): New function.
2523 * invoke.texi (fdevirtualize-speculatively): Document.
2524 * ipa-devirt.c: Include ipa-inline.h
2525 (likely_target_p): New function.
2526 (ipa_devirt): New function.
2527 (gate_ipa_devirt): New function.
2528 (pass_data_ipa_devirt): New static var.
2529 (pass_ipa_devirt): Likewise.
2530 (make_pass_ipa_devirt): New function.
2531 * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
2532 (common_handle_option): Disable devirtualization when
2533 value range profiling is available.
2534 * passes.def (pass_ipa_devirt): Add.
2535 * timever.def (TV_IPA_DEVIRT): New timevar.
2536 * tree-pass.h (make_pass_ipa_devirt):
2538 2013-09-01 Iain Sandoe <iain@codesourcery.com>
2540 * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
2541 include sanitize(undefined).
2543 2013-08-31 Diego Novillo <dnovillo@google.com>
2545 * Makefile.in (TREE_CORE_H): Define.
2547 (GTFILES): Add tree-core.h.
2548 * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
2550 * tree-core.h: New file.
2551 Move all data structures, enum, typedefs, global
2552 declarations and constants from ...
2555 2013-08-31 Jan Hubicka <jh@suse.cz>
2557 * bulitins.c (expand_builtin): Do not early exit for gcov
2558 instrumented functions.
2560 2013-08-31 Marek Polacek <polacek@redhat.com>
2562 * ubsan.c: Include tm_p.h.
2564 2013-08-31 Jan Hubicka <jh@suse.cz>
2566 * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
2569 * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
2570 * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
2571 * tree-cfg.c (verify_gimple_label): ... here.
2572 * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
2573 (ipa_merge_profiles): New function.
2574 * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
2575 (lto_input_function_body): Likewise.
2576 * ipa-utils.h (ipa_merge_profiles): Declare.
2577 * lto-streamer.h (lto_input_function_body): Update prototype.
2578 (emit_label_in_global_context_p): Remove.
2579 * lto-symtab.c: Include ipa-utils.h
2580 (lto_cgraph_replace_node): Use ipa_merge_profiles.
2582 2013-08-31 Jan Hubicka <jh@suse.cz>
2584 * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
2586 2013-08-31 Jan Hubicka <jh@suse.cz>
2588 * basic-block.h (apply_scale): Make scale parmeter gcov_type.
2590 2013-08-31 Uros Bizjak <ubizjak@gmail.com>
2592 * config/alpha/alpha.c (alpha_emit_conditional_move): Update
2593 "cmp" RTX before signed_comparison_operator check to account
2596 2013-08-30 Jan Hubicka <jh@suse.cz>
2598 * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
2599 (detect_type_change_1): Rename to ...
2600 (detect_type_change): ... this one; early return on non-polymorphic
2602 (detect_type_change_ssa): Add comp_type parameter; update
2603 use of detect_type_change.
2604 (compute_complex_assign_jump_func): Add param_type parameter;
2605 update use of detect_type_change_ssa.
2606 (compute_complex_ancestor_jump_func): Likewise.
2607 (ipa_get_callee_param_type): New function.
2608 (ipa_compute_jump_functions_for_edge): Compute parameter type;
2609 update calls to the jump function computation functions.
2611 2013-08-30 Teresa Johnson <tejohnson@google.com>
2612 Steven Bosscher <steven@gcc.gnu.org>
2614 * cfgrtl.c (fixup_new_cold_bb): New routine.
2615 (commit_edge_insertions): Invoke fixup_partitions.
2616 (find_partition_fixes): New routine.
2617 (fixup_partitions): Ditto.
2618 (verify_hot_cold_block_grouping): Update comments.
2619 (rtl_verify_edges): Invoke find_partition_fixes.
2620 (rtl_verify_bb_pointers): Update comments.
2621 (rtl_verify_bb_layout): Ditto.
2622 * basic-block.h (probably_never_executed_edge_p): Declare.
2623 (fixup_partitions): Ditto.
2624 * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
2625 * bb-reorder.c (sanitize_hot_paths): New function.
2626 (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
2628 * predict.c (probably_never_executed_edge_p): New routine.
2629 * cfg.c (check_bb_profile): Add partition insanity warnings.
2631 2013-08-30 Meador Inge <meadori@codesourcery.com>
2633 * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
2635 2013-08-30 Marek Polacek <polacek@redhat.com>
2637 * Makefile.in (ubsan.o): Add.
2638 (c-family/c-ubsan.o): Add.
2639 (builtins.o): Add ubsan.h dependency.
2640 * ubsan.h: New file.
2641 * ubsan.c: New file.
2642 * common.opt: Add -fsanitize=undefined option.
2643 (flag_sanitize): Add variable.
2644 (fsanitize=): Add option. Add Driver.
2645 (fsanitize=thread): Remove option.
2646 (fsanitize=address): Likewise.
2647 (static-libubsan): New option.
2648 * doc/invoke.texi: Document the new flag and -static-libubsan.
2649 * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
2650 (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
2651 * builtin-attrs.def (ATTR_COLD): Define.
2652 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2653 * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
2654 BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
2655 * flag-types.h (sanitize_code): New enum.
2656 * opts.c (common_handle_option): Parse command line arguments
2657 of -fsanitize=. Add -fsanitize=unreachable option.
2658 * varasm.c (get_variable_section): Adjust.
2659 (assemble_noswitch_variable): Likewise.
2660 (assemble_variable): Likewise.
2661 (output_constant_def_contents): Likewise.
2662 (categorize_decl_for_section): Likewise.
2663 (place_block_symbol): Likewise.
2664 (output_object_block): Likewise.
2665 * builtins.def: Likewise.
2666 * toplev.c (compile_file): Likewise.
2667 (process_options): Likewise.
2668 * cppbuiltin.c: Likewise.
2669 * tsan.c (tsan_pass): Likewise.
2670 (tsan_gate): Likewise.
2671 (tsan_gate_O0): Likewise.
2672 * cfgexpand.c (partition_stack_vars): Likewise.
2673 (expand_stack_vars): Likewise.
2674 (defer_stack_allocation): Likewise.
2675 (expand_used_vars): Likewise.
2676 * cfgcleanup.c (old_insns_match_p): Likewise.
2677 * asan.c (asan_finish_file): Likewise.
2678 (asan_instrument): Likewise.
2679 (gate_asan): Likewise.
2680 (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
2681 (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
2682 (asan_global_struct): Use pointer_sized_int_node instead
2683 calling build_nonstandard_integer_type.
2684 (initialize_sanitizer_builtins): Likewise.
2685 (asan_finish_file): Likewise.
2686 * gcc.c: Document %{%:function(args):X}.
2687 (static_spec_functions): Add sanitize.
2688 (handle_spec_function): Add retval_nonnull argument and if non-NULL,
2689 store funcval != NULL there.
2690 (do_spec_1): Adjust handle_spec_function caller.
2691 (handle_braces): Allow %:function(args) as condition.
2692 (sanitize_spec_function): New function.
2693 (ADD_STATIC_LIBUBSAN_LIBS): Define.
2694 (LIBUBSAN_SPEC): Likewise.
2695 (LIBUBSAN_EARLY_SPEC): Likewise.
2696 (SANITIZER_SPEC): Handle libubsan.
2697 (SANITIZER_EARLY_SPEC): Likewise.
2698 * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
2699 instead of fsanitize=address.
2700 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
2701 instead of fsanitize=address*.
2702 * builtins.c: Include ubsan.h.
2703 (fold_builtin_0): Instrument __builtin_unreachable.
2704 * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
2705 instead of flag_asan.
2706 * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
2707 (pointer_sized_int_node): Define.
2708 * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
2710 2013-08-30 Mike Stump <mikestump@comcast.net>
2712 * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
2715 2013-08-29 Jan Hubicka <jh@suse.cz>
2717 * cgraph.c (cgraph_function_body_availability): Handle weakref
2719 * passes.def: Remove pass_fixup_cfg.
2720 * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
2721 track when we need to remove functions.
2722 (gate_ipa_inline): Execute inlining always; add comment why.
2723 (pass_data_ipa_inline): Remove TODO_remove_functions.
2724 * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
2725 do not produce summaries.
2726 * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
2727 (symtab_nonoverwritable_alias): Assert we are not called on weakref.
2728 * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
2729 constant pool and vtable.
2731 2013-08-30 Tejas Belagod <tejas.belagod@arm.com>
2733 * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
2734 New arm_neon.h's internal macros to specify 64-bit constants.
2735 Avoid using stdint.h's macros.
2737 2013-08-30 Joern Rennecke <joern.rennecke@embecosm.com>
2739 * recog.c (verify_changes): Verify that changes[i].old is non-zero
2740 before applying REG_P.
2742 2013-08-30 Jakub Jelinek <jakub@redhat.com>
2744 PR tree-optimization/58277
2745 * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
2746 after seeing too many stmts with vdef in between dombb and current
2747 bb, invalidate everything.
2749 2013-08-30 Richard Biener <rguenther@suse.de>
2751 * fold-const.c (fold_single_bit_test): Fix overflow test.
2753 2013-08-30 Eric Botcazou <ebotcazou@adacore.com>
2755 * function.c (assign_parm_setup_reg): For a parameter passed by pointer
2756 and which can live in a register, always retrieve the value on entry.
2757 * var-tracking.c (add_stores): Treat the copy on entry for a parameter
2758 passed by invisible reference specially.
2759 (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
2760 (vt_add_function_parameter): Correctly deal with a parameter passed by
2761 invisible reference.
2763 2013-08-30 Jan Hubicka <jh@suse.cz>
2765 * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
2767 2013-08-30 Richard Biener <rguenther@suse.de>
2769 PR tree-optimization/58228
2770 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
2771 allow invariant loads in nested loop vectorization.
2773 2013-08-30 Richard Biener <rguenther@suse.de>
2775 PR tree-optimization/58223
2776 * tree-loop-distribution.c (has_anti_dependence): Rename to ...
2777 (has_anti_or_output_dependence): ... this and adjust to also
2778 look for output dependences.
2779 (mark_nodes_having_upstream_mem_writes): Adjust.
2780 (rdg_flag_uses): Likewise.
2782 2013-08-30 Richard Biener <rguenther@suse.de>
2784 PR tree-optimization/58010
2785 * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
2786 assert that we have a loop-closed PHI.
2788 2013-08-29 Jan Hubicka <jh@suse.cz>
2790 * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
2791 * cgraph.c (cgraph_release_function_body): Free decl_in_state.
2792 * lto-section-in.c (lto_free_function_in_decl_state): New function.
2793 (lto_free_function_in_decl_state_for_node): New function.
2795 2013-08-29 Xinliang David Li <davidxl@google.com>
2797 * loop-unroll.c (report_unroll_peel): Minor message change.
2798 * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
2799 Emit alignment peeling message with default -fopt-info.
2800 (vect_loop_versioning): Emit loop version info message.
2801 * tree-vectorizer.c (vectorize_loops): Minor message change.
2802 (execute_vect_slp): Ditto.
2804 2013-08-29 Eric Botcazou <ebotcazou@adacore.com>
2806 * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
2807 of the clone from the DECL_NAME of the original function.
2809 2013-08-29 Oleg Endo <olegendo@gcc.gnu.org>
2811 * passes.c (register_pass): Add overload.
2812 * tree-pass.h (register_pass): Forward declare it. Add comment.
2814 2013-08-29 Jan Hubicka <jh@suse.cz>
2816 * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
2817 DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
2818 * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
2819 DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2820 (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
2821 * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
2822 Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
2823 (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
2825 2013-08-29 Teresa Johnson <tejohnson@google.com>
2827 * dumpfile.c (dump_loc): Output column number.
2828 * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
2829 * doc/invoke.texi: Document optall -fopt-info flag.
2830 * profile.c (read_profile_edge_counts): Use new dump framework.
2831 (compute_branch_probabilities): Ditto.
2832 * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
2833 when pass not in any opt group.
2834 * pass_manager.h (pass_manager::get_pass_profile): New method.
2835 * value-prof.c (check_counter): Use new dump framework.
2836 (check_ic_target): Ditto.
2837 * coverage.c (get_coverage_counts): Ditto.
2838 (coverage_init): Setup new dump framework.
2840 2013-08-29 Richard Biener <rguenther@suse.de>
2842 PR tree-optimization/58246
2843 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
2844 handle the dominance check inside a basic-block.
2846 2013-08-29 Richard Biener <rguenther@suse.de>
2849 * tree-ssa-copy.c (may_propagate_copy): Allow propagating
2850 of default defs that appear in abnormal PHI nodes.
2852 2013-08-29 Richard Biener <rguenther@suse.de>
2854 PR tree-optimization/57685
2855 * tree-vrp.c (register_edge_assert_for_1): Recurse only for
2856 single-use operands to avoid exponential complexity.
2858 2013-08-28 Dehao Chen <dehao@google.com>
2860 * ipa-inline.c (edge_badness): Fix integer underflow.
2862 2013-08-28 Uros Bizjak <ubizjak@gmail.com>
2864 * gtm-builtins.def (_ITM_free): Declare leaf.
2866 2013-08-28 Jakub Jelinek <jakub@redhat.com>
2869 * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
2870 (*tls_local_dynamic_base_64_largepic): Likewise.
2871 (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
2872 Remove predicate from call operand.
2873 * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
2874 return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
2876 2013-08-28 Jeff Law <law@redhat.com>
2878 * tree-ssa-threadedge.c (thread_around_empty_block): Remove
2879 checks for the number of predecessors and successors allowed.
2880 * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
2881 which require copying a joiner block if there is a request which
2882 is a subpath that requires no joiner block copying.
2884 2013-08-28 Jan Hubicka <jh@suse.cz>
2886 * lto-streamer-out.c (DFS_write_tree_body): Drop
2887 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
2888 (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
2889 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
2890 * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
2891 Do not read DECL_ERROR_ISSUED.
2892 (unpack_ts_decl_with_vis_value_fields): Do not read
2894 (lto_input_ts_binfo_tree_pointers): Do not read
2895 BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
2896 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
2897 write DECL_ERROR_ISSUED..
2898 (pack_ts_decl_with_vis_value_fields): Do not write
2900 (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
2901 BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
2902 * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
2903 * tree.h (tree_decl_common): Update comment.
2904 (DECL_ERROR_ISSUED): Remove.
2906 2013-08-28 Jakub Jelinek <jakub@redhat.com>
2909 * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
2911 2013-08-28 Jan Hubicka <jh@suse.cz>
2913 * builtins.def (free): Declare leaf.
2915 2013-08-27 David Malcolm <dmalcolm@redhat.com>
2918 * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
2919 * configure: Regenerate.
2921 2013-08-27 Martin Jambor <mjambor@suse.cz>
2923 * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
2924 (ipa_ancestor_jf_data): Likewise.
2925 (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
2926 (ipa_get_jf_pass_through_type_preserved): New function.
2927 (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
2928 (ipa_get_jf_ancestor_type_preserved): New function.
2929 * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
2930 (ipa_get_jf_ancestor_result): Likewise.
2931 (propagate_vals_accross_pass_through): Use
2932 ipa_get_jf_pass_through_result to do all the value mappings.
2933 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
2934 type_preserved flag.
2935 (ipa_set_jf_cst_copy): New function.
2936 (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
2937 (ipa_set_jf_arith_pass_through): Likewise.
2938 (ipa_set_ancestor_jf): Likewise.
2939 (compute_complex_assign_jump_func): Set type_preserved instead of
2941 (ipa_compute_jump_functions_for_edge): Likewise.
2942 (combine_known_type_and_ancestor_jfs): Honor type_preserved.
2943 (update_jump_functions_after_inlining): Update type_preserved.
2944 Explicitely create jump functions when combining one with pass_through.
2945 (ipa_write_jump_function): Stream the type_preserved flags.
2946 (ipa_read_jump_function): Likewise.
2948 2013-08-27 Jakub Jelinek <jakub@redhat.com>
2949 Aldy Hernandez <aldyh@redhat.com>
2951 * Makefile.in (omp-low.o): Depend on $(TARGET_H).
2952 * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
2953 * function.h (struct function): Add has_force_vect_loops and
2955 * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
2956 * gimple.c (gimple_build_omp_critical): Add KIND argument and
2958 * gimple.def: Update CLAUSES comments.
2959 * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
2960 (gimple_build_omp_for): Add argument to prototype.
2961 (gimple_omp_for_kind): New.
2962 (gimple_omp_for_set_kind): New.
2963 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
2964 GOVD_DATA_SHARE_CLASS.
2965 (enum omp_region_type): Add ORT_SIMD.
2966 (gimple_add_tmp_var): Handle ORT_SIMD.
2967 (gimplify_var_or_parm_decl): Same.
2968 (is_gimple_stmt): Same.
2969 (omp_firstprivatize_variable): Same.
2970 (omp_add_variable): Only use splay_tree_insert if lookup failed.
2971 (omp_notice_variable): Handle ORT_SIMD.
2972 (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
2973 (omp_check_private): Handle ORT_SIMD.
2974 (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2976 (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
2977 Handle OMP_CLAUSE_LASTPRIVATE.
2978 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
2980 (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
2981 (gimplify_expr): Handle OMP_SIMD.
2982 * internal-fn.c (expand_GOMP_SIMD_LANE): New.
2983 (expand_GOMP_SIMD_VF): New.
2984 (expand_GOMP_SIMD_LAST_LANE): New.
2985 * internal-fn.def (GOMP_SIMD_LANE): New.
2986 (GOMP_SIMD_VF): New.
2987 (GOMP_SIMD_LAST_LANE): New.
2988 * omp-low.c: Include target.h.
2989 (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
2991 (check_omp_nesting_restrictions): Same.
2993 (lower_rec_simd_input_clauses): New.
2994 (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
2996 (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
2997 GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
2998 (expand_omp_build_assign): New.
2999 (expand_omp_for_init_counts): New.
3000 (expand_omp_for_init_vars): New.
3001 (extract_omp_for_update_vars): New.
3002 (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
3003 and rewrite accordingly.
3004 (expand_omp_simd): New.
3005 (expand_omp_for): Use expand_omp_simd.
3006 (lower_omp_for_lastprivate): Unshare vinit when appropriate.
3007 (lower_omp_for): Do not lower the body.
3008 * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
3010 * tree-flow.h (find_omp_clause): Remove prototype.
3011 * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
3012 forcing vectorization of the loop, or if flag_tree_vectorize.
3013 (gate_tree_if_conversion): Similarly.
3014 * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
3015 gimple_build_omp_for.
3016 (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
3017 * tree-parloops (create_parallel_loop): Pass kind argument to
3018 gimple_build_omp_for.
3019 * tree-pretty-print.c (dump_omp_clause): Add cases for
3020 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
3021 OMP_CLAUSE__SIMDUID_.
3022 (dump_generic_node): Handle OMP_SIMD.
3023 * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
3024 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
3025 unroll OMP_SIMD loops here.
3026 * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
3027 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
3029 (vect_analyze_data_refs): Handle simd loops.
3030 * tree-vect-loop.c (vectorizable_live_operation): Handle
3032 * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
3033 (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
3034 (vectorizable_load): Same.
3035 * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
3036 (struct simduid_to_vf): New.
3037 (simduid_to_vf::hash): New.
3038 (simduid_to-vf::equal): New.
3039 (struct simd_array_to_simduid): New.
3040 (simd_array_to_simduid::hash): New.
3041 (simd_array_to_simduid::equal): New.
3042 (adjust_simduid_builtins): New.
3043 (struct note_simd_array_uses_struct): New.
3044 (note_simd_array_uses_cb): New.
3045 (note_simd_array_uses): New.
3046 (vectorize_loops): Handle simd hints and adjust simd builtins
3048 * tree-vectorizer.h (struct _stmt_vec_info): Add
3049 simd_lane_access_p field.
3050 (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
3051 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
3052 OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
3053 (omp_clause_code_name): Same.
3054 (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
3055 OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
3056 * tree.def (OMP_SIMD): New entry.
3057 * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
3058 OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
3059 (OMP_CLAUSE_DECL): Adjust range for new clauses.
3060 (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
3061 (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
3062 (OMP_CLAUSE_LINEAR_STEP): New.
3063 (OMP_CLAUSE_SAFELEN_EXPR): New.
3064 (OMP_CLAUSE__SIMDUID__DECL): New.
3065 (find_omp_clause): New prototype.
3067 2013-08-27 H.J. Lu <hongjiu.lu@intel.com>
3069 * config/i386/driver-i386.c (host_detect_local_cpu): Update
3070 Haswell processor detection.
3072 2013-08-27 Christian Widmer <shadow@umbrox.de>
3075 * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
3076 of Ivy Bridge and Haswell processors. Assume core-avx2 for unknown
3077 AVX2 capable processors.
3079 2013-08-27 Tejas Belagod <tejas.belagod@arm.com>
3081 * config/aarch64/arm_neon.h: Replace all inline asm implementations
3082 of vget_low_* with implementations in terms of other intrinsics.
3084 2013-08-27 Marc Glisse <marc.glisse@inria.fr>
3087 * doc/extend.texi (__builtin_isinf_sign): Restrict the return
3088 values to -1, 0 and 1.
3090 2013-08-27 Vidya Praveen <vidyapraveen@arm.com>
3092 * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
3093 UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
3094 (<optab><mode>3_insn): Remove.
3095 (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
3096 (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
3097 (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
3098 (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
3099 (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
3100 (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
3101 (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
3102 (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
3103 (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
3104 (ror<mode>3_insn): Likewise.
3105 * config/aarch64/predicates.md (aarch64_simd_register): New.
3107 2013-08-27 Richard Biener <rguenther@suse.de>
3109 PR tree-optimization/57521
3110 * tree-if-conv.c (if_convertible_bb_p): Verify that at least
3111 one edge is non-critical.
3112 (find_phi_replacement_condition): Make sure to use a non-critical
3113 edge. Cleanup and remove old bug workarounds.
3114 (bb_postdominates_preds): Remove.
3115 (if_convertible_loop_p_1): Do not compute post-dominators.
3116 (combine_blocks): Do not free post-dominators.
3117 (main_tree_if_conversion): Likewise.
3118 (pass_data_if_conversion): Add TODO_verify_ssa.
3120 2013-08-27 DJ Delorie <dj@redhat.com>
3122 * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
3124 2013-08-27 Yufeng Zhang <yufeng.zhang@arm.com>
3126 * function.c (assign_parm_find_data_types): Set passed_mode and
3127 nominal_mode to the TYPE_MODE of nominal_type for the built
3128 pointer type in case of the struct-pass-by-reference.
3130 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
3132 * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
3133 (UINT16_TYPE): Change default to "unsigned int".
3135 * config/avr/avr.opt (mfract-convert-truncate): New option.
3136 * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
3137 is set, round negative fractional integers according to n1169
3138 when converting to integer types.
3140 2013-08-26 Jan Hubicka <jh@suse.cz>
3142 * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
3143 methods can not be called indirectly when their address is not taken.
3145 2013-08-26 Jan Hubicka <jh@suse.cz>
3147 * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
3150 2013-08-26 Jan Hubicka <jh@suse.cz>
3152 * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
3155 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
3157 * reload.c (find_valid_class): Allow classes that do not include
3158 FIRST_PSEUDO_REGISTER - 1.
3160 2013-08-26 Jan Hubicka <jh@suse.cz>
3162 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
3163 fix edge count/frequency when speculation failed; fix type check
3164 for the direct call.
3166 2013-08-26 Jan Hubicka <jh@suse.cz>
3168 * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
3170 2013-08-26 Jan Hubicka <jh@suse.cz>
3172 * ipa-inline-transform.c (inline_transform): Be ready for basic block
3173 to be changed by edge redirection.
3175 2013-08-26 Jan Hubicka <jh@suse.cz>
3177 * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
3178 formating; add sanity check.
3179 (cgraph_resolve_speculation): Add FIXME about scaling profiles.
3180 (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
3181 * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
3182 (resolve_noninline_speculation): Update callee keys, too.
3184 2013-08-26 Jan Hubicka <jh@suse.cz>
3186 * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
3187 (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
3189 2013-08-26 Joern Rennecke <joern.rennecke@embecosm.com>
3191 * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
3194 2013-08-26 Uros Bizjak <ubizjak@gmail.com>
3196 * config/i386/i386.c (ix86_debug_options): Remove prototype.
3197 (x86_64_elf_select_section): Ditto.
3198 (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
3200 (ix86_pass_by_reference): Ditto.
3201 (output_set_got): Ditto.
3202 (ix86_unary_operator_ok): Ditto.
3203 (ix86_expand_builtin): Ditto.
3205 2013-08-23 Jan Hubicka <jh@suse.cz>
3207 * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
3209 2013-08-23 Jan Hubicka <jh@suse.cz>
3211 * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
3212 (tree_decl_with_vis): Add FINAL field.
3214 2013-08-23 Jeff Law <law@redhat.com>
3216 * tree-ssa-pre.c (do_regular_insertion): Include the expression in
3217 the debugging dump when the expression is fully redundant.
3219 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
3221 * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
3222 * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
3223 * pretty-print.c (pp_formatted_text_data): Likewise.
3224 (pp_write_text_to_stream): Likewise.
3225 (pp_write_text_as_dot_label_to_stream): Likewise.
3226 (pp_append_r): Likewise.
3227 (pp_format): Likewise.
3228 (pp_flush): Likewise.
3229 (pp_clear_output_area): Likewise.
3230 (pp_append_text): Likewise.
3231 (pp_formatted_text): Likewise.
3232 (pp_remaining_character_count_for_line): Likewise.
3233 (pp_newline): Likewise.
3234 (pp_character): Likewise.
3235 (output_buffer::~output_buffer): Define.
3236 (pretty_printer::~pretty_printer): Destruct output buffer.
3237 * pretty-print.h (output_buffer::~output_buffer): Declare.
3238 (pretty_printer::~pretty_printer): Declare virtual.
3240 2013-08-24 Marc Glisse <marc.glisse@inria.fr>
3243 * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
3244 HOST_WIDE_INT_M1U): New macros.
3245 * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
3246 fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
3247 unsigned -1 for lshift.
3248 * cse.c (cse_insn): Likewise.
3249 * double-int.c (rshift_double, lshift_double): Likewise.
3250 * builtins.c (fold_builtin_bitop): Likewise.
3251 * combine.c (force_to_mode): Likewise.
3252 * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
3253 * simplify-rtx.c (simplify_const_unary_operation,
3254 simplify_const_binary_operation): Likewise.
3255 * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
3256 check_va_list_escapes): Likewise.
3257 * rtlanal.c (nonzero_bits1): Likewise.
3258 * expmed.c (expand_smod_pow2): Likewise.
3259 * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
3261 2013-08-23 Jan Hubicka <jh@suse.cz>
3263 * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
3264 as having address taken.
3266 2013-08-23 Jan Hubicka <jh@suse.cz>
3268 * ipa-utils.h (method_class_type): Declare.
3269 * ipa-devirt.c (method_class_type): Export.
3271 * cgraphunit.c (analyze_functions): Do basic devirtualization;
3272 do not walk base classes of anonymous types.
3274 2013-08-23 Kaz Kojima <kkojima@gcc.gnu.org>
3276 PR rtl-optimization/58220
3278 * final.c (reemit_insn_block_notes): Use NEXT_INSN to
3279 handle SEQUENCE insns properly.
3281 2013-08-23 Gabriel Dos Reis <gdr@integrable-solutions.net>
3283 * pretty-print.h (pp_newline_and_flush): Declare. Remove macro
3285 (pp_newline_and_indent): Likewise.
3286 (pp_separate_with): Likewise.
3287 * pretty-print.c (pp_newline_and_flush): Define.
3288 (pp_newline_and_indent): Likewise.
3289 (pp_separate_with): Likewise.
3291 2013-08-23 Jakub Jelinek <jakub@redhat.com>
3294 * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
3295 * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
3297 2013-08-23 Kirill Yukhin <kirill.yukhin@intel.com>
3299 * config/i386/predicates.md (ext_sse_reg_operand): New.
3300 * config/i386/i386.md (*movti_internal): Use
3301 predicate to determine if EVEX is needed.
3302 (*movsi_internal): Ditto.
3303 (*movdf_internal): Ditto.
3304 (*movsf_internal): Ditto.
3305 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
3307 2013-08-23 Jakub Jelinek <jakub@redhat.com>
3309 PR tree-optimization/58209
3310 * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
3311 (find_tail_calls): Give up for pointer result types if m is non-NULL.
3312 (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
3313 emit POINTER_PLUS_EXPR.
3314 (create_tailcall_accumulator): For pointer result type accumulate in
3317 2013-08-22 Paolo Carlini <paolo.carlini@oracle.com>
3319 * configure.ac: Add backslashes missing from the last change.
3320 * configure: Regenerate.
3322 2013-08-22 Jan Hubicka <jh@susue.cz>
3324 * ipa.c (function_and_variable_visibility): First remember function
3325 was global and then make it local.
3327 2013-08-22 Julian Brown <julian@codesourcery.com>
3329 * configure.ac: Add aarch64 to list of arches which use "nop" in
3331 * configure: Regenerate.
3333 2013-08-22 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3335 * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
3336 gnu_libc_has_function.
3337 * config/s390/tpf.h: Likewise.
3339 2013-08-22 Jan Hubicka <jh@susue.cz>
3341 * timevar.c (validate_phases): Add cast.
3343 2013-08-22 Jan Hubicka <jh@susue.cz>
3345 * timevar.c (validate_phases): Use size_t for memory.
3346 * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
3348 2013-08-22 Gabriel Dos Reis <gdr@integrable-solutions.net>
3350 * pretty-print.h (output_buffer::output_buffer): Declare.
3351 (pretty_printer::pretty_printer): Likewise.
3352 (pp_construct): Remove.
3353 * pretty-print.c (output_buffer::output_buffer): Define.
3354 (pretty_printer::pretty_printer): Rename from pp_construct. Simplify.
3355 * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
3356 (print_gimple_expr): Likewise.
3357 (print_gimple_seq): Likewise.
3358 (gimple_dump_bb): Likewise.
3359 * sched-vis.c (dump_value_slim): Likewise.
3360 (dump_insn_slim): Likewise.
3361 (dump_rtl_slim): Likewise.
3362 (str_pattern_slim): Likewise.
3363 * tree-mudflap.c (mf_varname_tree): Likewise.
3364 * graph.c (print_graph_cfg): Likewise.
3365 (start_graph_dump): Likewise.
3366 * tree-pretty-print.c (maybe_init_pretty_print): Likewise. Use
3368 * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
3369 pretty printer initialization.
3370 * coretypes.h (diagnostic_context): Remove superflous type alias
3372 (pretty_printer): Likewise. Declare directly as a class.
3373 (pretty_print_info): Remove declaration as class.
3374 * asan.c (asan_emit_stack_protection): Remove call to pp_construct
3375 and pp_clear_output_area.
3376 (asan_add_global): Likewise.
3378 2013-08-22 Jan Hubicka <jh@suse.cz>
3380 * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
3381 * ipa-utils.h (update_type_inheritance_graph): Declare.
3382 (possible_polymorphic_call_target_p): Declare.
3383 (possible_polymorphic_call_target_p): New.
3384 * ipa-devirt.c: Update toplevel comments.
3385 (cached_polymorphic_call_targets): Move up.
3386 (odr_type_d): Move ID down.
3387 (polymorphic_type_binfo_p): Update comment.
3388 (odr_hasher::remove): Likewise;
3389 (get_odr_type): Set anonymous_namespace.
3390 (dump_odr_type): Dump it.
3391 (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
3392 (maybe_record_node): Record node in cached_polymorphic_call_targets.
3393 (record_binfo): Add comment.
3394 (free_polymorphic_call_targets_hash): Do not ICE when cache is not
3396 (devirt_node_removal_hook): Do not iCE when cache is freed.
3397 (possible_polymorphic_call_target_p): New predicate.
3398 (update_type_inheritance_graph): New function.
3400 2013-08-22 Alexander Ivchenko <alexander.ivchenko@intel.com>
3401 Maxim Kuznetsov <maxim.kuznetsov@intel.com>
3402 Sergey Lega <sergey.s.lega@intel.com>
3403 Anna Tikhonova <anna.tikhonova@intel.com>
3404 Ilya Tocar <ilya.tocar@intel.com>
3405 Andrey Turetskiy <andrey.turetskiy@intel.com>
3406 Ilya Verbin <ilya.verbin@intel.com>
3407 Kirill Yukhin <kirill.yukhin@intel.com>
3408 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
3410 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
3411 (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
3412 (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
3413 (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
3414 (OPTION_MASK_ISA_AVX2_UNSET): Update.
3415 (OPTION_MASK_ISA_AVX512F_UNSET): New.
3416 (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
3417 (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
3418 (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
3419 (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
3420 OPT_mavx512pf, OPT_mavx512er cases.
3421 * config/i386/constraints.md (v): New constraint.
3422 (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
3423 * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
3424 (bit_AVX512CD): New.
3425 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
3426 AVX512F, AVX512ER, AVX512PF, AVX512CD features.
3427 * config/i386/i386-c.c (ix86_target_macros_internal):
3428 Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
3430 * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
3431 (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
3432 * config/i386/i386.c (regclass_map, dbx_register_map)
3433 (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
3434 (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
3435 (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
3437 (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
3438 PTA_AVX512PF, PTA_AVX512CD. Handle -mavx512f, -mavx512er, -mavx512cd,
3439 -mavx512pf options. Fix formatting.
3440 (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
3441 targets. Squash EVEX_SSE_REGS if AVX512F is disabled.
3442 (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
3443 -mavx512cd, -mavx512pf options.
3444 (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
3445 (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
3446 (ix86_preferred_output_reload_class): Replace SSE_REGS with
3448 (ix86_hard_regno_mode_ok): Support 512-bit registers.
3449 (ix86_set_reg_reg_cost): Ditto.
3450 (x86_order_regs_for_local_alloc): Ditto.
3451 (MAX_VECT_LEN): Extend to 64-byte.
3452 (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
3453 * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
3454 (TARGET_AVX512ER, TARGET_AVX512CD): New.
3455 (BIGGEST_ALIGNMENT): Extend to 512-bits.
3456 (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
3457 (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
3458 (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
3459 (SSE_REG_MODE_P): Support new modes.
3460 (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
3461 (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
3462 (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
3463 (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
3464 (REG_CLASS_CONTENTS): Add new registers.
3465 (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
3466 (EXT_REX_SSE_REGNO_P): New.
3467 (HI_REGISTER_NAMES): Add new registers.
3468 * config/i386/i386.md: Define constants for new registers.
3469 (mode): Add new 512-bit modes.
3470 (prefix): Support evex prefix.
3471 (isa): Support avx512f, noavx512f, fma_avx512f.
3472 (ssemodesuffix): Add new 512-bit modes.
3474 (*movxi_internal_avx512f): Ditto.
3475 (*movdi_internal): Replace constraint "x" with the new constraint "v".
3477 (*movsi_internal): Likewise.
3478 (*movdf_internal): Likewise.
3479 (*movsf_internal): Likewise.
3480 (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
3481 (<code><mode>3): Likewise.
3482 * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
3484 * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
3485 with the new constraint "v".
3486 * config/i386/sse.md (*mov<mode>_internal): Support new registers and
3488 (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
3489 with the new constraint "v".
3490 (<sse2>_loaddqu<avxsizesuffix>): Likewise.
3491 (<sse2>_storedqu<avxsizesuffix>): Likewise.
3492 (*<plusminus_insn><mode>3): Likewise.
3493 (<sse>_vm<plusminus_insn><mode>3): Likewise.
3494 (*mul<mode>3): Likewise.
3495 (<sse>_vmmul<mode>3): Likewise.
3496 (<sse>_div<mode>3): Likewise.
3497 (<sse>_vmdiv<mode>3): Likewise.
3498 (<sse>_sqrt<mode>2): Likewise.
3499 (<sse>_vmsqrt<mode>2): Likewise.
3500 (*<code><mode>3_finite): Likewise.
3501 (*<code><mode>3) <smaxmin>: Likewise.
3502 (<sse>_vm<code><mode>3): Likewise.
3503 (*<code><mode>3) <any_logic>: Likewise.
3504 (*fma_fmadd_<mode>): Likewise.
3505 (*fma_fmsub_<mode>): Likewise.
3506 (*fma_fnmadd_<mode>): Likewise.
3507 (*fma_fnmsub_<mode>): Likewise.
3508 (*fma_fmaddsub_<mode>): Likewise.
3509 (*fma_fmsubadd_<mode>): Likewise.
3510 (*fmai_fmadd_<mode>): Likewise.
3511 (*fmai_fmsub_<mode>): Likewise.
3512 (*fmai_fnmadd_<mode>): Likewise.
3513 (*fmai_fnmsub_<mode>): Likewise.
3514 (sse_cvtsi2ss): Likewise.
3515 (sse_cvtsi2ssq): Likewise.
3516 (sse_cvtss2si): Likewise.
3517 (sse_cvtss2si_2): Likewise.
3518 (sse_cvtss2siq): Likewise.
3519 (sse_cvtss2siq_2): Likewise.
3520 (sse_cvttss2si): Likewise.
3521 (sse_cvtss2siq_2): Likewise.
3522 (float<sseintvecmodelower><mode>2): Likewise.
3523 (sse2_cvtsd2si_2): Likewise.
3524 (sse2_cvtsd2siq_2): Likewise.
3525 (*<plusminus_insn><mode>3): Likewise.
3526 (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
3527 (*<sse4_1_avx2>_mul<mode>3): Likewise.
3528 (ashr<mode>3): Likewise.
3529 (<shift_insn><mode>3): Likewise.
3530 (avx2_<code><mode>3): Likewise.
3531 (*avx2_<code><mode>3): Likewise.
3532 (*andnot<mode>3): Likewise.
3533 (*<code><mode>3) <any_logic>: Likewise.
3534 (abs<mode>2): Likewise.
3535 (avx2_permvar<mode>): Likewise.
3536 (avx2_perm<mode>_1): Likewise.
3537 (*avx_vpermilp<mode>): Likewise.
3538 (avx_vpermilvar<mode>3): Likewise.
3539 (avx2_ashrv<mode>): Likewise.
3540 (avx2_<shift_insn>v<mode>): Likewise.
3541 * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
3543 * doc/rtl.texi: Document XImode.
3545 2013-08-21 Jeff Law <law@redhat.com>
3547 * tree-flow.h (register_jump_thread): Pass vector of edges
3548 instead of each important edge.
3549 * tree-ssa-threadedge.c (thread_across_edge): Build the jump
3550 thread path into a vector and pass that to register_jump_thread.
3551 * tree-ssa-threadupdate.c (register_jump_thread): Conver the
3552 passed in edge vector to the current 3-edge form.
3555 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
3557 * dce.c (fini_dce): Call df_analyze again just in case
3558 delete_unmarked_insns removed anything.
3560 2013-08-21 Joern Rennecke <joern.rennecke@embecosm.com>
3562 * reload.h (struct reg_equivs): Rename to ..
3563 (struct reg_equivs_s): .. this.
3565 2013-08-20 Martin Liska <marxin.liska@gmail.com>
3567 * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
3569 2013-08-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3571 * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
3573 2013-08-21 Jeff Law <law@redhat.com>
3575 * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
3576 simplify assignments too. If the RHS collapses to a singleton
3577 range, then return the value for the range.
3579 2013-08-21 Kirill Yukhin <kirill.yukhin@intel.com>
3581 * config/i386/sse.md (V16): Rename to...
3583 (mov<mode>): Update iterator name.
3584 (*mov<mode>_internal): Ditto.
3585 (push<mode>1): Ditto.
3586 (movmisalign<mode>): Ditto.
3588 2013-08-20 Jan Hubicka <jh@suse.cz>
3591 * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
3592 entry for direct edges.
3593 (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
3595 2013-08-20 David Malcolm <dmalcolm@redhat.com>
3597 Revert my last two changes, r201865 and r201864:
3600 2013-08-20 David Malcolm <dmalcolm@redhat.com>
3602 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3603 instances can own GC refs.
3605 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3606 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3607 (gcc::context::gt_pch_nx): Likewise.
3608 (gcc::context::gt_pch_nx): Likewise.
3609 * ggc.h (gt_ggc_mx <T>): New.
3610 (gt_pch_nx_with_op <T>): New.
3611 (gt_pch_nx <T>): New.
3612 * passes.c (opt_pass::gt_ggc_mx): New.
3613 (opt_pass::gt_pch_nx): New.
3614 (opt_pass::gt_pch_nx_with_op): New.
3615 (pass_manager::gt_ggc_mx): New.
3616 (pass_manager::gt_pch_nx): New.
3617 (pass_manager::gt_pch_nx_with_op): New.
3618 (pass_manager::operator new): Use
3619 ggc_internal_cleared_alloc_stat rather than xcalloc.
3620 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3621 (pass_manager::gt_ggc_mx): New.
3622 (pass_manager::gt_pch_nx): New.
3623 (pass_manager::gt_pch_nx_with_op): New.
3624 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3625 (opt_pass::operator new): New.
3626 (opt_pass::gt_ggc_mx): New.
3627 (opt_pass::gt_pch_nx): New.
3628 (opt_pass::gt_pch_nx_with_op): New.
3631 2013-08-20 David Malcolm <dmalcolm@redhat.com>
3633 * Makefile.in (GTFILES): Add context.h.
3634 * context.c (gcc::context::operator new): New.
3635 (gcc::context::gt_ggc_mx): New.
3636 (gcc::context::gt_pch_nx): New.
3637 (gcc::context::gt_pch_nx): New.
3638 * context.h (gcc::context): Add GTY((user)) marking.
3639 (gcc::context::operator new): New.
3640 (gcc::context::gt_ggc_mx): New.
3641 (gcc::context::gt_pch_nx): New.
3642 (gcc::context::gt_pch_nx): New.
3643 (g): Add GTY marking.
3644 (gt_ggc_mx (gcc::context *)): New.
3645 (gt_pch_nx (gcc::context *)): New.
3646 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3647 void *cookie)): New.
3648 * gengtype.c (open_base_files) <ifiles>: Add context.h.
3650 2013-08-20 Alexey Makhalov <makhaloff@gmail.com>
3652 * dce.c (fini_dce): Call df_analyze again just in case
3653 delete_unmarked_insns removed anything.
3655 2013-08-20 Teresa Johnson <tejohnson@google.com>
3657 PR rtl-optimizations/57451
3658 * final.c (reemit_insn_block_notes): Prevent lexical blocks
3659 from crossing split section boundaries.
3661 2013-08-20 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
3663 * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
3664 * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
3665 with MULTLIB_DEFAULTS.
3667 2013-08-20 Nick Clifton <nickc@redhat.com>
3669 * target.def (narrow_volatile_bitfield): Note that the default
3670 value is false, not !TARGET_STRICT_ALIGN.
3671 * doc/tm.texi: Regenerate.
3673 2013-08-20 Pavel Chupin <pavel.v.chupin@intel.com>
3675 Fix LIB_SPEC for systems without libpthread.
3677 * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
3678 * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
3680 * config/i386/linux-common.h: Likewise.
3681 * config/mips/linux-common.h: Likewise.
3683 2013-08-20 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
3685 * tree-ssa-ccp.c (get_default_value): Remove redundant condition
3688 2013-08-20 David Malcolm <dmalcolm@redhat.com>
3690 Make opt_pass and gcc::pass_manager be GC-managed, so that pass
3691 instances can own GC refs.
3693 * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
3694 * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
3695 (gcc::context::gt_pch_nx): Likewise.
3696 (gcc::context::gt_pch_nx): Likewise.
3697 * ggc.h (gt_ggc_mx <T>): New.
3698 (gt_pch_nx_with_op <T>): New.
3699 (gt_pch_nx <T>): New.
3700 * passes.c (opt_pass::gt_ggc_mx): New.
3701 (opt_pass::gt_pch_nx): New.
3702 (opt_pass::gt_pch_nx_with_op): New.
3703 (pass_manager::gt_ggc_mx): New.
3704 (pass_manager::gt_pch_nx): New.
3705 (pass_manager::gt_pch_nx_with_op): New.
3706 (pass_manager::operator new): Use
3707 ggc_internal_cleared_alloc_stat rather than xcalloc.
3708 * pass_manager.h (class pass_manager): Add GTY((user)) marking.
3709 (pass_manager::gt_ggc_mx): New.
3710 (pass_manager::gt_pch_nx): New.
3711 (pass_manager::gt_pch_nx_with_op): New.
3712 * tree-pass.h (class opt_pass): Add GTY((user)) marking.
3713 (opt_pass::operator new): New.
3714 (opt_pass::gt_ggc_mx): New.
3715 (opt_pass::gt_pch_nx): New.
3716 (opt_pass::gt_pch_nx_with_op): New.
3718 2013-08-20 David Malcolm <dmalcolm@redhat.com>
3720 * Makefile.in (GTFILES): Add context.h.
3721 * context.c (gcc::context::operator new): New.
3722 (gcc::context::gt_ggc_mx): New.
3723 (gcc::context::gt_pch_nx): New.
3724 (gcc::context::gt_pch_nx): New.
3725 * context.h (gcc::context): Add GTY((user)) marking.
3726 (gcc::context::operator new): New.
3727 (gcc::context::gt_ggc_mx): New.
3728 (gcc::context::gt_pch_nx): New.
3729 (gcc::context::gt_pch_nx): New.
3730 (g): Add GTY marking.
3731 (gt_ggc_mx (gcc::context *)): New.
3732 (gt_pch_nx (gcc::context *)): New.
3733 (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
3734 void *cookie)): New.
3735 * gengtype.c (open_base_files) <ifiles>: Add context.h.
3737 2013-08-20 Alan Modra <amodra@gmail.com>
3740 * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
3741 (rs6000_emit_epilogue): Likewise.
3743 2013-08-19 Dehao Chen <dehao@google.com>
3745 * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
3747 2013-08-19 Peter Bergner <bergner@vnet.ibm.com>
3748 Jakub Jelinek <jakub@redhat.com>
3750 * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
3751 (BUILT_IN_FABSD64): Likewise.
3752 (BUILT_IN_FABSD128): Likewise.
3753 * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
3754 (fold_builtin_1): Likewise.
3755 * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
3756 destination and source operands.
3757 (*abstd2_fpr): Likewise.
3758 (*nabstd2_fpr): Likewise.
3760 2013-08-19 Richard Sandiford <rdsandiford@googlemail.com>
3762 * config/mips/mips.c (mips_adjust_insn_length): Add checks for
3765 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
3767 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
3770 2013-08-19 Alexander Ivchenko <alexander.ivchenko@intel.com>
3772 * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
3773 * builtins.c (default_libc_has_function): New.
3774 (gnu_libc_has_function): Ditto.
3775 (no_c99_libc_has_function): Ditto.
3776 (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
3777 instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
3778 (fold_builtin_sincos): Likewise.
3779 (fold_builtin_cexp): Likewise.
3780 * builtins.def (DEF_C94_BUILTIN): Likewise.
3781 (DEF_C99_BUILTIN): Likewise.
3782 (DEF_C99_C90RES_BUILTIN): Likewise.
3783 (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
3784 definitions to using this define.
3785 * config/darwin-protos.h (darwin_libc_has_function): New.
3786 * config/darwin.c (darwin_libc_has_function): Ditto.
3787 * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
3788 TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
3789 * config/darwin.h: Ditto.
3790 * config/elfos.h: Ditto.
3791 * config/freebsd.h: Ditto.
3792 * config/i386/cygming.h: Ditto.
3793 * config/i386/djgpp.h: Ditto.
3794 * config/i386/i386-interix.h: Ditto.
3795 * config/microblaze/microblaze.h: Ditto.
3796 * config/mmix/mmix.h: Ditto.
3797 * config/gnu-user.h: Ditto.
3798 * config/ia64/hpux.h: Ditto.
3799 * config/pa/pa-hpux.h: Ditto.
3800 * config/pdp11/pdp11.h: Ditto.
3801 * config/picochip/picochip.h: Ditto.
3802 * config/linux.h: Ditto.
3803 * config/netbsd.h: Ditto.
3804 * config/openbsd.h: Ditto.
3805 * config/rs6000/aix43.h: Ditto.
3806 * config/rs6000/aix51.h: Ditto.
3807 * config/rs6000/aix52.h: Ditto.
3808 * config/rs6000/aix53.h: Ditto.
3809 * config/rs6000/aix61.h: Ditto.
3810 * config/rs6000/darwin.h: Ditto.
3811 * config/rs6000/linux.h: Ditto.
3812 * config/rs6000/linux64.h: Ditto.
3813 * config/s390/tpf.h: Ditto.
3814 * config/sol2-10.h: Ditto.
3815 * config/sol2.h: Ditto.
3816 * config/vms/vms.h: Ditto.
3817 * config/vxworks.h: Ditto.
3818 * config/linux-android.c (linux_android_libc_has_function):
3819 New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
3820 * config/linux-protos.h (linux_android_libc_has_function):
3822 * config/i386/i386.c (ix86_libc_has_function): New.
3823 * config/i386/i386-protos.h
3824 (ix86_libc_has_function): New declaration.
3825 * config/i386/i386.md
3826 ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
3827 ("isinf<mode>2): Likewise.
3828 * convert.c (convert_to_integer): Using new target hook
3829 TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
3830 TARGET_C99_FUNCTIONS.
3831 * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
3832 * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
3833 * coretypes.h (function_class): New enum for different
3834 classes of functions.
3835 * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
3836 * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
3837 (TARGET_HAS_SINCOS): Likewise.
3838 (TARGET_LIBC_HAS_FUNCTION): New.
3839 * doc/tm.texi: Regenerated.
3840 * targhooks.h (default_libc_has_function): New declaration.
3841 (no_c99_libc_has_function): Ditto.
3842 (gnu_libc_has_function): Ditto.
3843 * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
3844 and TARGET_HAS_SINCOS.
3846 2013-08-18 Jan Hubicka <jh@suse.cz>
3848 * Makeifle-in (ipa-devirt.o): New.
3849 (GTFILES): Add ipa-utils.h and ipa-devirt.c
3850 * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
3851 (analyze_functions): Look into possible targets of polymorphic call.
3852 * dumpfile.c (dump_files): Add type-inheritance dump.
3853 * dumpfile.h (TDI_inheritance): New.
3854 * ipa-devirt.c: New file.
3855 * ipa-utils.h (odr_type_d): Forward declare.
3856 (odr_type): New type.
3857 (build_type_inheritance_graph): Declare.
3858 (possible_polymorphic_call_targets): Declare and introduce inline
3859 variant when only edge is pased.
3860 (dump_possible_polymorphic_call_targets): Likewise.
3861 * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
3862 * tree.c (type_in_anonymous_namespace_p): Break out from ...
3863 (types_same_for_odr): ... here.
3864 * tree.h (type_in_anonymous_namespace_p): Declare.
3866 2013-08-18 Jakub Jelinek <jakub@redhat.com>
3868 PR tree-optimization/58006
3869 * tree-parloops.c (take_address_of): Don't ICE if get_name
3871 (eliminate_local_variables_stmt): Remove clobber stmts.
3873 2013-08-18 Eric Botcazou <ebotcazou@adacore.com>
3875 * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
3876 error message is issued for an alias to undefined symbol.
3878 2013-08-18 Jan Hubicka <jh@suse.cz>
3880 * cgraph.c (cgraph_create_indirect_edge): Discover
3881 polymorphic calls and record basic info into indirect_info.
3882 * gimple-fold.c (gimple_fold_call): When doing BINFO based
3883 devirtualization, ignore objc function calls.
3884 * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
3885 call with no parm index info.
3886 * ipa-prop.c (ipa_analyze_call_uses): Likewise.
3887 * tree.c (virtual_method_call_p): New function.
3888 * tree.h (virtual_method_call_p): Declare.
3890 2013-08-16 Jan Hubicka <jh@suse.cz>
3893 * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
3895 2013-08-16 David Edelsohn <dje.gcc@gmail.com>
3897 * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
3900 2013-08-16 David Malcolm <dmalcolm@redhat.com>
3902 * gengtype.c (type_for_name): Add special-case support for
3903 locating types within the "gcc::" namespace.
3904 (open_base_files): Emit a "using namespace gcc" directive.
3906 2013-08-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3909 * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
3910 memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
3912 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
3913 array instead of each individual operand as a separate argument.
3914 (emit_fusion_gpr_load): Likewise.
3915 (expand_fusion_gpr_load): Add new function declaration.
3917 * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
3918 signature to have the operands passed as an array, instead of as
3919 separate arguments. Allow ZERO_EXTEND to be in the memory
3920 address, and also SIGN_EXTEND if -mpower8-fusion-sign. Do not
3921 depend on the register live/dead flags when peepholes are run.
3922 (expand_fusion_gpr_load): New function to be called from the
3923 peephole2 pass, to change the register that addis sets to be the
3925 (emit_fusion_gpr_load): Change the calling signature to have the
3926 operands passed as an array, instead of as separate arguments.
3927 Allow ZERO_EXTEND to be in the memory address, and also
3928 SIGN_EXTEND if -mpower8-fusion-sign.
3930 * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
3932 (power8 fusion peephole/peephole2): Rework the fusion peepholes to
3933 adjust the register addis loads up in the peephole2 pass. Do not
3934 depend on the register live/dead state when the peephole pass is done.
3936 2013-08-16 David Malcolm <dmalcolm@redhat.com>
3938 * gengtype.c (create_user_defined_type): Ensure that the kind
3939 is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
3940 declaration is seen before the GTY((user)) marking.
3942 2013-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
3945 * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
3947 2013-08-16 Jan Hubicka <jh@suse.cz>
3949 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
3950 arugment expected_type.
3951 (gimple_fold_call): Use it.
3952 * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
3953 * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
3954 * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
3955 (try_make_edge_direct_virtual_call): Likewise.
3956 * tree.c (obj_type_ref_class): New.
3957 * tree.h (obj_type_ref_class): Use it.
3959 2013-08-16 Gabriel Dos Reis <gdr@integrable-solutions.net>
3961 * sched-vis.c (rtl_slim_pp_initialized): Remove.
3962 (rtl_slim_pp): Likewise.
3963 (init_rtl_slim_pretty_print): Likewise.
3964 (dump_value_slim): Don't call it. Use local pretty printer.
3965 (dump_insn_slim): Likewise.
3966 (dump_rtl_slim): Likewise.
3967 (str_pattern_slim): Likewise.
3968 * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
3971 2013-08-16 Jakub Jelinek <jakub@redhat.com>
3973 PR tree-optimization/58164
3974 * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
3975 walk gimple_goto_dest of GIMPLE_GOTO.
3977 PR tree-optimization/58165
3978 * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
3979 bi_call must be the last stmt in a bb, don't split_block, instead
3980 use fallthru edge from it and give up if there is none.
3981 Release conds vector when returning early.
3983 2013-08-14 Xinliang David Li <davidxl@google.com>
3985 * config/i386/i386.c (ix86_option_override_internal):
3986 Remove unused variable and field.
3988 2013-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3991 * doc/invoke.texi: Add documentation of mcompat-align-parm option.
3992 * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
3993 * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
3994 and Linux, correct BLKmode alignment when 128-bit alignment is
3995 required and compatibility flag is not set.
3996 (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
3997 for zero-size arguments when compatibility flag is not set.
3999 2013-08-14 Jakub Jelinek <jakub@redhat.com>
4001 PR tree-optimization/58145
4002 * tree-sra.c (build_ref_for_offset): If prev_base has
4003 TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
4005 2013-08-14 Xinliang David Li <davidxl@google.com>
4007 * config/i386/i386.c (ix86_option_override_internal):
4008 Fix uninitialized variable error.
4010 2013-08-14 Xinliang David Li <davidxl@google.com>
4012 * config/i386/i386.opt: Define two new options.
4013 * config/i386/x86-tune.def: Add arch selector field in macros.
4014 * config/i386/i386.h: Adjust macro definition.
4015 * config/i386/i386.c (ix86_option_override_internal):
4017 (parse_mtune_ctrl_str): New function.
4018 (set_ix86_tune_features): New function.
4019 (ix86_function_specific_restore): Call the new helper function.
4021 2013-08-14 Andrey Belevantsev <abel@ispras.ru>
4023 PR rtl-optimization/57662
4024 * sel-sched.c (code_motion_process_successors): When the current insn
4025 is removed after the recursive traversal, break from the loop.
4026 Add comments and debug printouts.
4028 2013-08-14 Jakub Jelinek <jakub@redhat.com>
4029 Alexandre Oliva <aoliva@redhat.com>
4032 * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
4033 and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
4034 there also UNSPEC_PLTOFF.
4036 2013-08-14 Marek Polacek <polacek@redhat.com>
4038 * ipa-inline-analysis.c (add_clause): Avoid shifting integer
4039 NUM_CONDITIONS bit positions.
4041 2013-08-13 Cary Coutant <ccoutant@google.com>
4043 * dwarf2out.c (CHECKSUM_BLOCK): New macro.
4044 (attr_checksum): Hash vector contents instead of pointer.
4045 (attr_checksum_ordered): Likewise.
4047 2013-08-13 Uros Bizjak <ubizjak@gmail.com>
4049 * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
4050 when Pmode != word_mode. Add length_address attribute.
4051 (sse3_monitor_<mode>): Merge from sse3_monitor and
4052 sse3_monitor64_<mode> insn patterns. Emit addr32 prefix when
4053 Pmode != word_mode. Update insn length attribute.
4054 * config/i386/i386.c (ix86_option_override_internal): Update
4055 ix86_gen_monitor selection for merged sse3_monitor insn.
4057 2013-08-13 Julian Brown <julian@codesourcery.com>
4059 * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
4060 perform invalid legitimization on greater-than-word-size modes for
4063 2013-08-13 Vladimir Makarov <vmakarov@redhat.com>
4065 * ira.c (setup_class_translate_array): Use aclass instead of cl
4066 for classes not fully covered by allocno classes.
4068 2013-08-13 Jakub Jelinek <jakub@redhat.com>
4070 PR tree-optimization/57661
4071 * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
4072 * tree-inline.c (tree_function_versioning): Initialize it.
4073 (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
4074 if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
4075 that is not being copied.
4078 * asan.c (instrument_strlen_call): Fix typo in comment.
4079 Use char * type even for the lhs of POINTER_PLUS_EXPR.
4081 2013-08-13 Steve Ellcey <sellcey@mips.com>
4083 * config/mips/mips.md (prefetch): Use lw instead of ld on
4084 loongson in 32bit mode.
4086 2013-08-13 Nick Clifton <nickc@redhat.com>
4088 * config.gcc: (avr-linux): Allow for tmake_file not being empty.
4090 2013-08-13 Jan Hubicka <jh@suse.cz>
4092 * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
4093 introduced edge; fix typo in sanity check.
4094 (cgraph_resolve_speculation): Export; improve diagnostic.
4095 (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
4096 speculation at type mismatch.
4097 * cgraph.h (cgraph_turn_edge_to_speculative): Update.
4098 (cgraph_resolve_speculation): Declare.
4099 (symtab_can_be_discarded): New function.
4100 * value-prof.c (gimple_ic_transform): Remove actual transform code.
4101 * ipa-inline-transform.c (speculation_removed): New global var.
4102 (clone_inlined_nodes): See if speculation can be removed.
4103 (inline_call): If speculations was removed, we growths may not match.
4104 * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
4105 (speculation_useful_p): New function.
4106 (resolve_noninline_speculation): New function.
4107 (inline_small_functions): Resolve useless speculations.
4108 * ipa-inline.h (speculation_useful_p): Declare
4109 * ipa.c (can_replace_by_local_alias): Simplify.
4110 (ipa_profile): Produce speculative calls in non-lto, too;
4111 add simple cost model; produce local aliases.
4113 2013-08-13 David Malcolm <dmalcolm@redhat.com>
4115 * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
4118 2013-08-12 Paolo Carlini <paolo.carlini@oracle.com>
4120 * config/i386/i386.c (ix86_function_versions): Use error + inform.
4122 2013-08-12 Uros Bizjak <ubizjak@gmail.com>
4124 * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
4125 iterator instead of X87MODEF.
4127 2013-08-12 Perez Read <netfirewall@gmail.com>
4130 * config/i386/i386.md (*movabs<mode>_1): Add <ptrsize> PTR before
4131 operand 0 for intel asm alternative.
4132 (*movabs<mode>_2): Ditto for operand 1.
4134 2013-08-12 James Greenhalgh <james.greenhalgh@arm.com>
4136 * config/aarch64/arm_none.h
4137 (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
4139 2013-08-12 Nick Clifton <nickc@redhat.com>
4141 * config.gcc (m32r-linux): Allow for tmake_file not being empty.
4143 2013-08-12 Yuri Rumyantsev <ysrumyan@gmail.com>
4145 * config/i386/i386.md (floatunssi<mode>2 expand): Add new
4146 expand for QI/HImode operand to produce more effictive code for
4147 unsigned char(short) --> float(double) conversion.
4149 2013-08-12 Alexander Monakov <amonakov@ispras.ru>
4151 * doc/invoke.texi: Mention that -ftls-model does not force the final
4154 2013-08-12 Marek Polacek <polacek@redhat.com>
4155 Marc Glisse <marc.glisse@inria.fr>
4157 PR tree-optimization/57980
4158 * tree-tailcall.c (process_assignment): Call build_minus_one_cst
4159 when creating -1 constant.
4161 2013-08-10 Jan Hubicka <jh@suse.cz>
4163 Workaround binutils PR14342.
4164 * tree-profile.c (init_ic_make_global_vars): Add LTO path.
4165 (gimple_init_edge_profiler): Likewise.
4166 (gimple_gen_ic_func_profiler): Likewise.
4168 2013-08-09 Jan Hubicka <jh@suse.cz>
4170 * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
4172 2013-08-09 Xinliang David Li <davidxl@google.com>
4174 * config/i386/stringop.def: New file.
4175 * config/i386/stringop.opt: New file.
4176 * config/i386/i386-opts.h: Include stringopt.def.
4177 * config/i386/i386.opt: Include stringopt.opt.
4178 * config/i386/i386.c (ix86_option_override_internal):
4179 Override default size based stringop inline strategies with options.
4180 * config/i386/i386.c (ix86_parse_stringop_strategy_string):
4183 2013-08-09 Jan Hubicka <jh@suse.cz>
4185 * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
4187 2013-08-09 Jan Hubicka <jh@suse.cz>
4189 * cgraph.c (cgraph_resolve_speculation): Cut frequency to
4191 (dump_cgraph_node): Dump profile-id.
4192 * cgraph.h (cgraph_indirect_call_info): Add common_target_id
4193 and common_target_probability.
4194 * lto-cgraph.c (lto_output_edge): Stream common targets.
4195 (lto_output_node): Stream profile ids.
4196 (input_node): Stream profile ids.
4197 (input_edge): Stream common targets.
4198 * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
4199 * ipa.c: Include value-prof.h
4200 (ipa_profile_generate_summary): Turn indirect call statement histograms
4201 into common targets.
4202 (ipa_profile): Turn common targets into speculative edges.
4204 2013-08-09 Jan Hubicka <jh@suse.cz>
4206 * cgraph.h (cgraph_node): Add profile_id.
4207 * value-prof.c (cgraph_node_map): Turn into pointer_map.
4208 (init_node_map): Rewrite to handle hashes increas of incremental IDs.
4209 (del_node_map): Update.
4210 (find_func_by_funcdef_no): Replace by ...
4211 (find_func_by_profile_id): ... this one.
4212 (gimple_ic_transform): Do not remove useful histograms when
4213 speculation is not done; dump info when indirect call removal
4215 * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
4216 * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
4217 (__gcov_indirect_call_profiler_v2): .. this one.
4218 * profile.h (init_node_map): Update.
4219 * coverage.c (coverage_compute_profile_id): New function.
4220 * coverage.h (coverage_compute_profile_id): Declare.
4221 * tree-profile.c (init_ic_make_global_vars): Make
4222 __gcov_indirect_call_callee and __gcov_indirect_call_counters global.
4223 (gimple_init_edge_profiler): Update prototype of
4224 __gcov_indirect_call_profiler.
4225 (gimple_gen_ic_func_profiler): Simplify.
4226 (tree_profiling): Use init_node_map
4228 2013-08-09 Jan Hubicka <jh@suse.cz>
4230 * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
4231 non-speculative refs.
4232 * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
4233 (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
4234 (cgraph_set_call_stmt): Likewise.
4235 (cgraph_create_edge_1): Fix release checking compilatoin;
4237 (cgraph_free_edge): Free indirect info.
4238 (cgraph_turn_edge_to_speculative): New function.
4239 (cgraph_speculative_call_info): New function.
4240 (cgraph_make_edge_direct): Return direct edge; handle speculation.
4241 (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
4242 (dump_cgraph_node): Dump speculation.
4243 (verify_edge_count_and_frequency): Accept speculative edges.
4244 (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
4245 (verify_cgraph_node): Handle speculation.
4246 * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
4247 (cgraph_set_call_stmt): Update prototype.
4248 (cgraph_make_edge_direct): Update prototype.
4249 (cgraph_speculative_call_info): Declare.
4250 * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
4251 to change; update call of ipa_find_references.
4252 * ipa-ref.c (ipa_record_reference): Fix return value; clear
4253 lto_stmt_uid and speculative flags.
4254 (ipa_dump_references): Dump speculation.
4255 (ipa_clone_references): Clone speculative flag.
4256 (ipa_clone_referring): Likewise.
4257 (ipa_clone_ref): New function.
4258 (ipa_find_reference): Look into lto_stmt_uids
4259 (ipa_clear_stmts_in_references): Do not clear speculative calls.
4260 * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
4261 (ipa_find_reference): Update declaration.
4262 (ipa_clone_ref): Declare.
4263 * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
4264 stream speculative flag.
4265 (lto_output_ref): Stream statements uids and speculation.
4266 (input_ref): Likewise.
4267 (input_edge): Stream speuclation.
4268 * cgraphclones.c (cgraph_clone_edge): Clone speculation.
4269 (cgraph_set_call_stmt_including_clones): Handle speculation.
4270 * ipa-inline.c (heap_edge_removal_hook): New function.
4271 (inline_small_functions): Register it.
4272 * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
4273 also initialize refs.
4274 * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
4276 (try_make_edge_direct_simple_call): Likewise.
4277 (try_make_edge_direct_simple_call): Likewise.
4278 (update_indirect_edges_after_inlining): Likewise.
4279 (remove_described_reference): Look proper lto_stmt_uid.
4280 (propagate_controlled_uses): Likewise.
4281 (propagate_controlled_uses): Liekwise.
4282 * tree-inline.c (copy_bb): Copy speculative edges.
4283 (redirect_all_calls): New function.
4284 (copy_cfg_body): Do redirection after loop info is updated.
4285 (delete_unreachable_blocks_update_callgraph): Updadte speculation.
4287 2013-08-09 Jan Hubicka <jh@suse.cz>
4289 * lto-streamer-out.c (output_function): Renumber PHIs.
4290 * lto-streamer-in.c (input_function): Likewise.
4292 2013-08-09 James Greenhalgh <james.greenhalgh@arm.com>
4294 * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
4295 (get_lane_unsigned): Likewise.
4296 (dup_lane_scalar): Likewise.
4297 (get_lane): enable for VALL.
4298 * config/aarch64/aarch64-simd.md
4299 (aarch64_dup_lane_scalar<mode>): Remove.
4300 (aarch64_get_lane_signed<mode>): Likewise.
4301 (aarch64_get_lane_unsigned<mode>): Likewise.
4302 (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
4303 (aarch64_get_lane_zero_extendsi<mode>): Likewise.
4304 (aarch64_get_lane<mode>): Enable for all vector modes.
4305 (aarch64_get_lanedi): Remove misleading constraints.
4306 * config/aarch64/arm_neon.h
4307 (__aarch64_vget_lane_any): Define.
4308 (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
4309 (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
4310 (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
4311 * config/aarch64/iterators.md (VDQQH): New.
4315 2013-08-09 Eric Botcazou <ebotcazou@adacore.com>
4317 * configure.ac: Add GAS check for LEON instructions on SPARC.
4318 * configure: Regenerate.
4319 * config.in: Likewise.
4320 * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
4322 * config/sparc/sparc.opt (LEON, LEON3): New masks.
4323 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
4325 (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
4326 (AS_LEON_FLAG): New macro.
4327 * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
4328 and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
4329 Deal with LEON and LEON3 for the memory model.
4330 * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
4331 (atomic_compare_and_swap<mode>_1): Likewise.
4332 (*atomic_compare_and_swap<mode>_1): Likewise.
4334 2013-08-09 Zhenqiang Chen <zhenqiang.chen@linaro.org>
4336 * config/arm/neon.md (vcond): Fix floating-point vector
4337 comparisons against 0.
4339 2013-08-08 Vladimir Makarov <vmakarov@redhat.com>
4341 * lra-constraints.c (emit_spill_move): Remove assert.
4342 (process_alt_operands): Add more debugging
4343 output. Increase reject for spilling into memory. Decrease
4344 reject for reloading scratch.
4345 (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
4347 2013-08-08 Steve Ellcey <sellcey@mips.com>
4349 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
4350 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
4351 micromips incompatible. Add nan2008.
4352 (MULTILIB_DIRNAMES): Add nan2008.
4353 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
4354 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
4355 and micromips incompatible. Add nan2008.
4356 (MULTILIB_DIRNAMES): Add nan2008.
4357 (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
4359 2013-08-08 Richard Sandiford <rdsandiford@googlemail.com>
4361 PR rtl-optimization/58079
4362 * combine.c (combine_simplify_rtx): Avoid using SUBST if
4363 simplify_comparison has widened a comparison with an integer.
4365 2013-08-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4367 * config/arm/neon.md (movmisalign<mode>): Disable when we
4368 don't allow unaligned accesses.
4369 (*movmisalign<mode>_neon_store): Likewise.
4370 (*movmisalign<mode>_neon_load): Likewise.
4371 (*movmisalign<mode>_neon_store): Likewise.
4372 (*movmisalign<mode>_neon_load): Likewise.
4374 2013-08-08 Jan Hubicka <jh@suse.cz>
4376 * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
4377 (make_pass_rebuild_cgraph_edges): Also clear references.
4378 * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
4379 * ipa-inline-transform.c (inline_transform): Remove all references
4381 * cgraphunit.c (expand_function): Remove all references after
4383 * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
4384 (ipa_find_reference): Rewrite to iterator.
4385 (remove_stmt_references): Likewise.
4386 (ipa_clear_stmts_in_references): New function.
4387 * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
4388 * cgraphclones.c (cgraph_materialize_all_clones): Remove or
4390 * ipa-split.c (split_function): Remove references in split function.
4392 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
4395 * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
4396 (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
4398 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
4401 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
4402 suggested mode for the assignment isn't compatible with the
4405 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4408 * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
4410 2013-08-07 Xinliang David Li <davidxl@google.com>
4412 * config/i386/i386.opt: New option -mtune-ctrl=.
4413 * config/i386/x86-tune.def: New file.
4414 * config/i386/i386.h: include x86-tune.def.
4415 * config/i386/i386.c (ix86_option_override_internal):
4416 Parsing -mtune-ctrl= option and set tune features.
4418 2013-08-07 Oleg Endo <olegendo@gcc.gnu.org>
4421 * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
4422 (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
4425 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
4427 * rtl.h (update_alignments): Declare.
4428 * final.c (grow_label_align): New function extracted from...
4429 (shorten_branches): ...here. Call it.
4430 (update_alignments): New function.
4431 * reorg.c (sibling_labels): New variable.
4432 (get_label_before): Add SIBLING parameter. If it is non-zero, push
4433 the new label along with it onto the sibling_labels vector.
4434 (fill_simple_delay_slots): Adjust call to get_label_before.
4435 (fill_slots_from_thread): Likewise.
4436 (relax_delay_slots): Likewise.
4437 (make_return_insns): Likewise.
4438 (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
4440 2013-08-07 Eric Botcazou <ebotcazou@adacore.com>
4442 * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
4443 document its semantics.
4444 (diagnostic_report_diagnostic): Adjust accordingly.
4446 2013-08-07 David Malcolm <dmalcolm@redhat.com>
4448 * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
4449 (sparc_option_override): ...and port to new C++ pass API.
4450 * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
4452 2013-08-07 Peter Bergner <bergner@vnet.ibm.com>
4454 * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
4456 2013-08-06 Caroline Tice <cmtice@google.com>
4458 * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
4459 (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
4460 * tree-pass.h: Add pass_vtable_verify.
4461 * varasm.c (assemble_variable): Add code to properly set the comdat
4462 section and name for the .vtable_map_vars section.
4463 (assemble_vtyv_preinit_initializer): New function.
4464 (default_sectin_type_flags): Make sure .vtable_map_vars section has
4466 * output.h: Add function decl for assemble_vtv_preinit_initializer.
4467 * vtable-verify.c: New file.
4468 * vtable-verify.h: New file.
4469 * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
4470 initialiation levels.
4471 * timevar.def (TV_VTABLE_VERIFICATION): New definition.
4472 * passes.def: Insert pass_vtable_verify.
4473 * aclocal.m4: Reorder includes.
4474 * doc/invoke.texi: Document the -fvtable-verify=, -fvtv-debug, and
4475 -fvtv-counts options.
4476 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
4477 as appropriate, if -fvtable-verify=... is used.
4478 (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
4479 -fvtable-verify=... is used.
4480 * Makefile.in (OBJS): Add vtable-verify.o to list.
4481 (vtable-verify.o): Add new build rule.
4482 (GTFILES): Add vtable-verify.c to list.
4483 * common.opt (fvtable-verify=): New flag.
4484 (vtv_priority): Values for fvtable-verify= flag.
4485 (fvtv-counts): New flag.
4486 (fvtv-debug): New flag.
4487 * tree.h (save_vtable_map_decl): New extern function decl.
4489 2013-08-07 David Malcolm <dmalcolm@redhat.com>
4491 * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
4492 (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
4493 (pass_data_rl78_devirt): ...new pass_data instance and...
4494 (make_pass_rl78_devirt): ...new function.
4495 (rl78_asm_file_start): Port pass registration to new C++ API.
4497 2013-08-07 David Malcolm <dmalcolm@redhat.com>
4499 * coretypes.h (rtl_opt_pass): Add.
4500 (gcc::context): Add.
4501 * config/epiphany/epiphany.c (pass_mode_switch_use): New.
4502 (epiphany_init): Port to new C++ pass API.
4503 (epiphany_optimize_mode_switching): Likewise.
4504 * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
4505 (pass_manager::get_pass_mode_switching): New.
4506 (pass_manager::get_pass_peephole2): New.
4507 * mode-switching.c (pass_mode_switching): Add clone method.
4508 * recog.c (pass_peephole2): Add clone method.
4509 (pass_split_all_insns): Add clone method.
4511 2013-08-06 David Malcolm <dmalcolm@redhat.com>
4513 * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
4514 (mips_option_override): ...here, porting to new C++ API for passes.
4516 2013-08-06 Jan Hubicka <jh@suse.cz>
4518 * cgraph.c (cgraph_get_body): New function based on lto.c
4520 * cgraph.h (cgraph_get_body): Declare.
4521 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
4523 * cgraphunit.c (expand_function): Get body prior expanding.
4524 * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
4525 * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
4527 * passes.c (do_per_function_toporder): Get body.
4528 * tree-inline.c (expand_call_inline): Get body prior inlining it.
4529 * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
4531 2013-08-06 Martin Jambor <mjambor@suse.cz>
4534 * cgraphunit.c (cgraph_finalize_function): Assert that nested function
4535 is not re-finalized. Rename second parameter to no_collect.
4537 2013-08-06 Martin Jambor <mjambor@suse.cz>
4540 * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
4541 MEM_REF has proper alignment information.
4543 2013-08-05 Oleg Endo <olegendo@gcc.gnu.org>
4546 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
4548 * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
4549 rtx (*) (rtx, ...) with insn_gen_fn.
4550 * genoutput.c (output_insn_data): Cast gen_? function pointers to
4551 insn_gen_fn::stored_funcptr. Add initializer braces.
4553 2013-08-05 David Malcolm <dmalcolm@redhat.com>
4555 Rewrite how instances of passes are cloned to remove assumptions
4556 about their sizes (thus allowing pass subclasses to have
4557 additional data fields, albeit non-GC-managed ones at this point).
4559 * passes.c (make_pass_instance): Now that passes have clone
4560 methods, rewrite this function to eliminate XNEW and memcpy
4561 calls that used hardcoded sizes. Since this function no longer
4562 creates pass instances, rename it to...
4563 (add_pass_instance): ...this. Document the old way that passes were
4564 numbered and flagged, and rework this function to continue using it.
4565 (next_pass_1): Add an initial_pass argument for use by
4567 (position_pass): When adding multiple instances of a pass, use
4568 the pass's clone method, rather than relying on the XNEW/memcpy
4569 within the former make_pass_instance (now add_pass_instance).
4570 (pass_manager::pass_manager): When invoking next_pass_1, also supply
4571 the initial instance of the current pass within the pass manager.
4573 2013-08-05 David Malcolm <dmalcolm@redhat.com>
4575 This is the automated part of the conversion of passes from C
4576 structs to C++ classes.
4578 Patch autogenerated by refactor_passes.py from
4579 https://github.com/davidmalcolm/gcc-refactoring-scripts
4580 revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
4582 * asan.c (pass_asan): Convert from a global struct to a subclass of
4583 gimple_opt_pass along with...
4584 (pass_data_asan): ...new pass_data instance and...
4585 (make_pass_asan): ...new function.
4586 (pass_asan_O0): Convert from a global struct to a subclass of
4587 gimple_opt_pass along with...
4588 (pass_data_asan_O0): ...new pass_data instance and...
4589 (make_pass_asan_O0): ...new function.
4590 * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
4591 subclass of rtl_opt_pass along with...
4592 (pass_data_inc_dec): ...new pass_data instance and...
4593 (make_pass_inc_dec): ...new function.
4594 * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
4595 a subclass of rtl_opt_pass along with...
4596 (pass_data_reorder_blocks): ...new pass_data instance and...
4597 (make_pass_reorder_blocks): ...new function.
4598 (pass_duplicate_computed_gotos): Convert from a global struct to a
4599 subclass of rtl_opt_pass along with...
4600 (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
4601 (make_pass_duplicate_computed_gotos): ...new function.
4602 (pass_partition_blocks): Convert from a global struct to a subclass of
4603 rtl_opt_pass along with...
4604 (pass_data_partition_blocks): ...new pass_data instance and...
4605 (make_pass_partition_blocks): ...new function.
4606 * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
4607 struct to a subclass of rtl_opt_pass along with...
4608 (pass_data_branch_target_load_optimize1): ...new pass_data instance
4610 (make_pass_branch_target_load_optimize1): ...new function.
4611 (pass_branch_target_load_optimize2): Convert from a global struct to a
4612 subclass of rtl_opt_pass along with...
4613 (pass_data_branch_target_load_optimize2): ...new pass_data instance
4615 (make_pass_branch_target_load_optimize2): ...new function.
4616 * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
4617 of rtl_opt_pass along with...
4618 (pass_data_jump): ...new pass_data instance and...
4619 (make_pass_jump): ...new function.
4620 (pass_jump2): Convert from a global struct to a subclass of
4621 rtl_opt_pass along with...
4622 (pass_data_jump2): ...new pass_data instance and...
4623 (make_pass_jump2): ...new function.
4624 * cfgexpand.c (pass_expand): Convert from a global struct to a
4625 subclass of rtl_opt_pass along with...
4626 (pass_data_expand): ...new pass_data instance and...
4627 (make_pass_expand): ...new function.
4628 * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
4629 of rtl_opt_pass along with...
4630 (pass_data_free_cfg): ...new pass_data instance and...
4631 (make_pass_free_cfg): ...new function.
4632 (pass_into_cfg_layout_mode): Convert from a global struct to a
4633 subclass of rtl_opt_pass along with...
4634 (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
4635 (make_pass_into_cfg_layout_mode): ...new function.
4636 (pass_outof_cfg_layout_mode): Convert from a global struct to a
4637 subclass of rtl_opt_pass along with...
4638 (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
4639 (make_pass_outof_cfg_layout_mode): ...new function.
4640 * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
4641 struct to a subclass of gimple_opt_pass along with...
4642 (pass_data_build_cgraph_edges): ...new pass_data instance and...
4643 (make_pass_build_cgraph_edges): ...new function.
4644 (pass_rebuild_cgraph_edges): Convert from a global struct to a
4645 subclass of gimple_opt_pass along with...
4646 (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
4647 (make_pass_rebuild_cgraph_edges): ...new function.
4648 (pass_remove_cgraph_callee_edges): Convert from a global struct to a
4649 subclass of gimple_opt_pass along with...
4650 (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
4652 (make_pass_remove_cgraph_callee_edges): ...new function.
4653 * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
4654 struct to a subclass of rtl_opt_pass along with...
4655 (pass_data_stack_adjustments): ...new pass_data instance and...
4656 (make_pass_stack_adjustments): ...new function.
4657 * combine.c (pass_combine): Convert from a global struct to a subclass
4658 of rtl_opt_pass along with...
4659 (pass_data_combine): ...new pass_data instance and...
4660 (make_pass_combine): ...new function.
4661 * compare-elim.c (pass_compare_elim_after_reload): Convert from a
4662 global struct to a subclass of rtl_opt_pass along with...
4663 (pass_data_compare_elim_after_reload): ...new pass_data instance
4665 (make_pass_compare_elim_after_reload): ...new function.
4666 * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
4667 of rtl_opt_pass along with...
4668 (pass_data_rtl_cprop): ...new pass_data instance and...
4669 (make_pass_rtl_cprop): ...new function.
4670 * cse.c (pass_cse): Convert from a global struct to a subclass of
4671 rtl_opt_pass along with...
4672 (pass_data_cse): ...new pass_data instance and...
4673 (make_pass_cse): ...new function.
4674 (pass_cse2): Convert from a global struct to a subclass of
4675 rtl_opt_pass along with...
4676 (pass_data_cse2): ...new pass_data instance and...
4677 (make_pass_cse2): ...new function.
4678 (pass_cse_after_global_opts): Convert from a global struct to a
4679 subclass of rtl_opt_pass along with...
4680 (pass_data_cse_after_global_opts): ...new pass_data instance and...
4681 (make_pass_cse_after_global_opts): ...new function.
4682 * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
4683 of rtl_opt_pass along with...
4684 (pass_data_ud_rtl_dce): ...new pass_data instance and...
4685 (make_pass_ud_rtl_dce): ...new function.
4686 (pass_fast_rtl_dce): Convert from a global struct to a subclass of
4687 rtl_opt_pass along with...
4688 (pass_data_fast_rtl_dce): ...new pass_data instance and...
4689 (make_pass_fast_rtl_dce): ...new function.
4690 * df-core.c (pass_df_initialize_opt): Convert from a global struct to
4691 a subclass of rtl_opt_pass along with...
4692 (pass_data_df_initialize_opt): ...new pass_data instance and...
4693 (make_pass_df_initialize_opt): ...new function.
4694 (pass_df_initialize_no_opt): Convert from a global struct to a
4695 subclass of rtl_opt_pass along with...
4696 (pass_data_df_initialize_no_opt): ...new pass_data instance and...
4697 (make_pass_df_initialize_no_opt): ...new function.
4698 (pass_df_finish): Convert from a global struct to a subclass of
4699 rtl_opt_pass along with...
4700 (pass_data_df_finish): ...new pass_data instance and...
4701 (make_pass_df_finish): ...new function.
4702 * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
4703 rtl_opt_pass along with...
4704 (pass_data_rtl_dse1): ...new pass_data instance and...
4705 (make_pass_rtl_dse1): ...new function.
4706 (pass_rtl_dse2): Convert from a global struct to a subclass of
4707 rtl_opt_pass along with...
4708 (pass_data_rtl_dse2): ...new pass_data instance and...
4709 (make_pass_rtl_dse2): ...new function.
4710 * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
4711 subclass of rtl_opt_pass along with...
4712 (pass_data_dwarf2_frame): ...new pass_data instance and...
4713 (make_pass_dwarf2_frame): ...new function.
4714 * except.c (pass_set_nothrow_function_flags): Convert from a global
4715 struct to a subclass of rtl_opt_pass along with...
4716 (pass_data_set_nothrow_function_flags): ...new pass_data instance
4718 (make_pass_set_nothrow_function_flags): ...new function.
4719 (pass_convert_to_eh_region_ranges): Convert from a global struct to a
4720 subclass of rtl_opt_pass along with...
4721 (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
4723 (make_pass_convert_to_eh_region_ranges): ...new function.
4724 * final.c (pass_compute_alignments): Convert from a global struct to a
4725 subclass of rtl_opt_pass along with...
4726 (pass_data_compute_alignments): ...new pass_data instance and...
4727 (make_pass_compute_alignments): ...new function.
4728 (pass_final): Convert from a global struct to a subclass of
4729 rtl_opt_pass along with...
4730 (pass_data_final): ...new pass_data instance and...
4731 (make_pass_final): ...new function.
4732 (pass_shorten_branches): Convert from a global struct to a subclass of
4733 rtl_opt_pass along with...
4734 (pass_data_shorten_branches): ...new pass_data instance and...
4735 (make_pass_shorten_branches): ...new function.
4736 (pass_clean_state): Convert from a global struct to a subclass of
4737 rtl_opt_pass along with...
4738 (pass_data_clean_state): ...new pass_data instance and...
4739 (make_pass_clean_state): ...new function.
4740 * function.c (pass_instantiate_virtual_regs): Convert from a global
4741 struct to a subclass of rtl_opt_pass along with...
4742 (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
4743 (make_pass_instantiate_virtual_regs): ...new function.
4744 (pass_leaf_regs): Convert from a global struct to a subclass of
4745 rtl_opt_pass along with...
4746 (pass_data_leaf_regs): ...new pass_data instance and...
4747 (make_pass_leaf_regs): ...new function.
4748 (pass_thread_prologue_and_epilogue): Convert from a global struct to a
4749 subclass of rtl_opt_pass along with...
4750 (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
4752 (make_pass_thread_prologue_and_epilogue): ...new function.
4753 (pass_match_asm_constraints): Convert from a global struct to a
4754 subclass of rtl_opt_pass along with...
4755 (pass_data_match_asm_constraints): ...new pass_data instance and...
4756 (make_pass_match_asm_constraints): ...new function.
4757 * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
4758 subclass of rtl_opt_pass along with...
4759 (pass_data_rtl_fwprop): ...new pass_data instance and...
4760 (make_pass_rtl_fwprop): ...new function.
4761 (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
4762 rtl_opt_pass along with...
4763 (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
4764 (make_pass_rtl_fwprop_addr): ...new function.
4765 * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
4766 rtl_opt_pass along with...
4767 (pass_data_rtl_pre): ...new pass_data instance and...
4768 (make_pass_rtl_pre): ...new function.
4769 (pass_rtl_hoist): Convert from a global struct to a subclass of
4770 rtl_opt_pass along with...
4771 (pass_data_rtl_hoist): ...new pass_data instance and...
4772 (make_pass_rtl_hoist): ...new function.
4773 * gimple-low.c (pass_lower_cf): Convert from a global struct to a
4774 subclass of gimple_opt_pass along with...
4775 (pass_data_lower_cf): ...new pass_data instance and...
4776 (make_pass_lower_cf): ...new function.
4777 * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
4778 from a global struct to a subclass of gimple_opt_pass along with...
4779 (pass_data_strength_reduction): ...new pass_data instance and...
4780 (make_pass_strength_reduction): ...new function.
4781 * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
4782 of rtl_opt_pass along with...
4783 (pass_data_rtl_ifcvt): ...new pass_data instance and...
4784 (make_pass_rtl_ifcvt): ...new function.
4785 (pass_if_after_combine): Convert from a global struct to a subclass of
4786 rtl_opt_pass along with...
4787 (pass_data_if_after_combine): ...new pass_data instance and...
4788 (make_pass_if_after_combine): ...new function.
4789 (pass_if_after_reload): Convert from a global struct to a subclass of
4790 rtl_opt_pass along with...
4791 (pass_data_if_after_reload): ...new pass_data instance and...
4792 (make_pass_if_after_reload): ...new function.
4793 * init-regs.c (pass_initialize_regs): Convert from a global struct to
4794 a subclass of rtl_opt_pass along with...
4795 (pass_data_initialize_regs): ...new pass_data instance and...
4796 (make_pass_initialize_regs): ...new function.
4797 * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
4798 of ipa_opt_pass_d along with...
4799 (pass_data_ipa_cp): ...new pass_data instance and...
4800 (make_pass_ipa_cp): ...new function.
4801 * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
4802 global struct to a subclass of gimple_opt_pass along with...
4803 (pass_data_inline_parameters): ...new pass_data instance and...
4804 (make_pass_inline_parameters): ...new function.
4805 * ipa-inline.c (pass_early_inline): Convert from a global struct to a
4806 subclass of gimple_opt_pass along with...
4807 (pass_data_early_inline): ...new pass_data instance and...
4808 (make_pass_early_inline): ...new function.
4809 (pass_ipa_inline): Convert from a global struct to a subclass of
4810 ipa_opt_pass_d along with...
4811 (pass_data_ipa_inline): ...new pass_data instance and...
4812 (make_pass_ipa_inline): ...new function.
4813 * ipa-pure-const.c (pass_local_pure_const): Convert from a global
4814 struct to a subclass of gimple_opt_pass along with...
4815 (pass_data_local_pure_const): ...new pass_data instance and...
4816 (make_pass_local_pure_const): ...new function.
4817 (pass_ipa_pure_const): Convert from a global struct to a subclass of
4818 ipa_opt_pass_d along with...
4819 (pass_data_ipa_pure_const): ...new pass_data instance and...
4820 (make_pass_ipa_pure_const): ...new function.
4821 * ipa-reference.c (pass_ipa_reference): Convert from a global struct
4822 to a subclass of ipa_opt_pass_d along with...
4823 (pass_data_ipa_reference): ...new pass_data instance and...
4824 (make_pass_ipa_reference): ...new function.
4825 * ipa-split.c (pass_split_functions): Convert from a global struct to
4826 a subclass of gimple_opt_pass along with...
4827 (pass_data_split_functions): ...new pass_data instance and...
4828 (make_pass_split_functions): ...new function.
4829 (pass_feedback_split_functions): Convert from a global struct to a
4830 subclass of gimple_opt_pass along with...
4831 (pass_data_feedback_split_functions): ...new pass_data instance and...
4832 (make_pass_feedback_split_functions): ...new function.
4833 * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
4834 global struct to a subclass of simple_ipa_opt_pass along with...
4835 (pass_data_ipa_function_and_variable_visibility): ...new pass_data
4837 (make_pass_ipa_function_and_variable_visibility): ...new function.
4838 (pass_ipa_free_inline_summary): Convert from a global struct to a
4839 subclass of simple_ipa_opt_pass along with...
4840 (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
4841 (make_pass_ipa_free_inline_summary): ...new function.
4842 (pass_ipa_whole_program_visibility): Convert from a global struct to a
4843 subclass of ipa_opt_pass_d along with...
4844 (pass_data_ipa_whole_program_visibility): ...new pass_data instance
4846 (make_pass_ipa_whole_program_visibility): ...new function.
4847 (pass_ipa_profile): Convert from a global struct to a subclass of
4848 ipa_opt_pass_d along with...
4849 (pass_data_ipa_profile): ...new pass_data instance and...
4850 (make_pass_ipa_profile): ...new function.
4851 (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
4852 ipa_opt_pass_d along with...
4853 (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
4854 (make_pass_ipa_cdtor_merge): ...new function.
4855 * ira.c (pass_ira): Convert from a global struct to a subclass of
4856 rtl_opt_pass along with...
4857 (pass_data_ira): ...new pass_data instance and...
4858 (make_pass_ira): ...new function.
4859 (pass_reload): Convert from a global struct to a subclass of
4860 rtl_opt_pass along with...
4861 (pass_data_reload): ...new pass_data instance and...
4862 (make_pass_reload): ...new function.
4863 * jump.c (pass_cleanup_barriers): Convert from a global struct to a
4864 subclass of rtl_opt_pass along with...
4865 (pass_data_cleanup_barriers): ...new pass_data instance and...
4866 (make_pass_cleanup_barriers): ...new function.
4867 * loop-init.c (pass_loop2): Convert from a global struct to a subclass
4868 of rtl_opt_pass along with...
4869 (pass_data_loop2): ...new pass_data instance and...
4870 (make_pass_loop2): ...new function.
4871 (pass_rtl_loop_init): Convert from a global struct to a subclass of
4872 rtl_opt_pass along with...
4873 (pass_data_rtl_loop_init): ...new pass_data instance and...
4874 (make_pass_rtl_loop_init): ...new function.
4875 (pass_rtl_loop_done): Convert from a global struct to a subclass of
4876 rtl_opt_pass along with...
4877 (pass_data_rtl_loop_done): ...new pass_data instance and...
4878 (make_pass_rtl_loop_done): ...new function.
4879 (pass_rtl_move_loop_invariants): Convert from a global struct to a
4880 subclass of rtl_opt_pass along with...
4881 (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
4882 (make_pass_rtl_move_loop_invariants): ...new function.
4883 (pass_rtl_unswitch): Convert from a global struct to a subclass of
4884 rtl_opt_pass along with...
4885 (pass_data_rtl_unswitch): ...new pass_data instance and...
4886 (make_pass_rtl_unswitch): ...new function.
4887 (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
4888 subclass of rtl_opt_pass along with...
4889 (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
4891 (make_pass_rtl_unroll_and_peel_loops): ...new function.
4892 (pass_rtl_doloop): Convert from a global struct to a subclass of
4893 rtl_opt_pass along with...
4894 (pass_data_rtl_doloop): ...new pass_data instance and...
4895 (make_pass_rtl_doloop): ...new function.
4896 * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
4897 a subclass of rtl_opt_pass along with...
4898 (pass_data_lower_subreg): ...new pass_data instance and...
4899 (make_pass_lower_subreg): ...new function.
4900 (pass_lower_subreg2): Convert from a global struct to a subclass of
4901 rtl_opt_pass along with...
4902 (pass_data_lower_subreg2): ...new pass_data instance and...
4903 (make_pass_lower_subreg2): ...new function.
4904 * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
4905 struct to a subclass of ipa_opt_pass_d along with...
4906 (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
4907 (make_pass_ipa_lto_gimple_out): ...new function.
4908 (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
4909 of ipa_opt_pass_d along with...
4910 (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
4911 (make_pass_ipa_lto_finish_out): ...new function.
4912 * mode-switching.c (pass_mode_switching): Convert from a global struct
4913 to a subclass of rtl_opt_pass along with...
4914 (pass_data_mode_switching): ...new pass_data instance and...
4915 (make_pass_mode_switching): ...new function.
4916 * modulo-sched.c (pass_sms): Convert from a global struct to a
4917 subclass of rtl_opt_pass along with...
4918 (pass_data_sms): ...new pass_data instance and...
4919 (make_pass_sms): ...new function.
4920 * omp-low.c (pass_expand_omp): Convert from a global struct to a
4921 subclass of gimple_opt_pass along with...
4922 (pass_data_expand_omp): ...new pass_data instance and...
4923 (make_pass_expand_omp): ...new function.
4924 (pass_lower_omp): Convert from a global struct to a subclass of
4925 gimple_opt_pass along with...
4926 (pass_data_lower_omp): ...new pass_data instance and...
4927 (make_pass_lower_omp): ...new function.
4928 (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
4929 of gimple_opt_pass along with...
4930 (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
4931 (make_pass_diagnose_omp_blocks): ...new function.
4932 * passes.c (pass_early_local_passes): Convert from a global struct to
4933 a subclass of simple_ipa_opt_pass along with...
4934 (pass_data_early_local_passes): ...new pass_data instance and...
4935 (make_pass_early_local_passes): ...new function.
4936 (pass_all_early_optimizations): Convert from a global struct to a
4937 subclass of gimple_opt_pass along with...
4938 (pass_data_all_early_optimizations): ...new pass_data instance and...
4939 (make_pass_all_early_optimizations): ...new function.
4940 (pass_all_optimizations): Convert from a global struct to a subclass
4941 of gimple_opt_pass along with...
4942 (pass_data_all_optimizations): ...new pass_data instance and...
4943 (make_pass_all_optimizations): ...new function.
4944 (pass_all_optimizations_g): Convert from a global struct to a subclass
4945 of gimple_opt_pass along with...
4946 (pass_data_all_optimizations_g): ...new pass_data instance and...
4947 (make_pass_all_optimizations_g): ...new function.
4948 (pass_rest_of_compilation): Convert from a global struct to a subclass
4949 of rtl_opt_pass along with...
4950 (pass_data_rest_of_compilation): ...new pass_data instance and...
4951 (make_pass_rest_of_compilation): ...new function.
4952 (pass_postreload): Convert from a global struct to a subclass of
4953 rtl_opt_pass along with...
4954 (pass_data_postreload): ...new pass_data instance and...
4955 (make_pass_postreload): ...new function.
4956 * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
4957 subclass of rtl_opt_pass along with...
4958 (pass_data_gcse2): ...new pass_data instance and...
4959 (make_pass_gcse2): ...new function.
4960 * postreload.c (pass_postreload_cse): Convert from a global struct to
4961 a subclass of rtl_opt_pass along with...
4962 (pass_data_postreload_cse): ...new pass_data instance and...
4963 (make_pass_postreload_cse): ...new function.
4964 * predict.c (pass_profile): Convert from a global struct to a subclass
4965 of gimple_opt_pass along with...
4966 (pass_data_profile): ...new pass_data instance and...
4967 (make_pass_profile): ...new function.
4968 (pass_strip_predict_hints): Convert from a global struct to a subclass
4969 of gimple_opt_pass along with...
4970 (pass_data_strip_predict_hints): ...new pass_data instance and...
4971 (make_pass_strip_predict_hints): ...new function.
4972 * recog.c (pass_peephole2): Convert from a global struct to a subclass
4973 of rtl_opt_pass along with...
4974 (pass_data_peephole2): ...new pass_data instance and...
4975 (make_pass_peephole2): ...new function.
4976 (pass_split_all_insns): Convert from a global struct to a subclass of
4977 rtl_opt_pass along with...
4978 (pass_data_split_all_insns): ...new pass_data instance and...
4979 (make_pass_split_all_insns): ...new function.
4980 (pass_split_after_reload): Convert from a global struct to a subclass
4981 of rtl_opt_pass along with...
4982 (pass_data_split_after_reload): ...new pass_data instance and...
4983 (make_pass_split_after_reload): ...new function.
4984 (pass_split_before_regstack): Convert from a global struct to a
4985 subclass of rtl_opt_pass along with...
4986 (pass_data_split_before_regstack): ...new pass_data instance and...
4987 (make_pass_split_before_regstack): ...new function.
4988 (pass_split_before_sched2): Convert from a global struct to a subclass
4989 of rtl_opt_pass along with...
4990 (pass_data_split_before_sched2): ...new pass_data instance and...
4991 (make_pass_split_before_sched2): ...new function.
4992 (pass_split_for_shorten_branches): Convert from a global struct to a
4993 subclass of rtl_opt_pass along with...
4994 (pass_data_split_for_shorten_branches): ...new pass_data instance
4996 (make_pass_split_for_shorten_branches): ...new function.
4997 * ree.c (pass_ree): Convert from a global struct to a subclass of
4998 rtl_opt_pass along with...
4999 (pass_data_ree): ...new pass_data instance and...
5000 (make_pass_ree): ...new function.
5001 * reg-stack.c (pass_stack_regs): Convert from a global struct to a
5002 subclass of rtl_opt_pass along with...
5003 (pass_data_stack_regs): ...new pass_data instance and...
5004 (make_pass_stack_regs): ...new function.
5005 (pass_stack_regs_run): Convert from a global struct to a subclass of
5006 rtl_opt_pass along with...
5007 (pass_data_stack_regs_run): ...new pass_data instance and...
5008 (make_pass_stack_regs_run): ...new function.
5009 * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
5010 subclass of rtl_opt_pass along with...
5011 (pass_data_cprop_hardreg): ...new pass_data instance and...
5012 (make_pass_cprop_hardreg): ...new function.
5013 * reginfo.c (pass_reginfo_init): Convert from a global struct to a
5014 subclass of rtl_opt_pass along with...
5015 (pass_data_reginfo_init): ...new pass_data instance and...
5016 (make_pass_reginfo_init): ...new function.
5017 * regmove.c (pass_regmove): Convert from a global struct to a subclass
5018 of rtl_opt_pass along with...
5019 (pass_data_regmove): ...new pass_data instance and...
5020 (make_pass_regmove): ...new function.
5021 * regrename.c (pass_regrename): Convert from a global struct to a
5022 subclass of rtl_opt_pass along with...
5023 (pass_data_regrename): ...new pass_data instance and...
5024 (make_pass_regrename): ...new function.
5025 * reorg.c (pass_delay_slots): Convert from a global struct to a
5026 subclass of rtl_opt_pass along with...
5027 (pass_data_delay_slots): ...new pass_data instance and...
5028 (make_pass_delay_slots): ...new function.
5029 (pass_machine_reorg): Convert from a global struct to a subclass of
5030 rtl_opt_pass along with...
5031 (pass_data_machine_reorg): ...new pass_data instance and...
5032 (make_pass_machine_reorg): ...new function.
5033 * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
5034 of rtl_opt_pass along with...
5035 (pass_data_sched): ...new pass_data instance and...
5036 (make_pass_sched): ...new function.
5037 (pass_sched2): Convert from a global struct to a subclass of
5038 rtl_opt_pass along with...
5039 (pass_data_sched2): ...new pass_data instance and...
5040 (make_pass_sched2): ...new function.
5041 * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
5042 to a subclass of rtl_opt_pass along with...
5043 (pass_data_stack_ptr_mod): ...new pass_data instance and...
5044 (make_pass_stack_ptr_mod): ...new function.
5045 * store-motion.c (pass_rtl_store_motion): Convert from a global struct
5046 to a subclass of rtl_opt_pass along with...
5047 (pass_data_rtl_store_motion): ...new pass_data instance and...
5048 (make_pass_rtl_store_motion): ...new function.
5049 * tracer.c (pass_tracer): Convert from a global struct to a subclass
5050 of gimple_opt_pass along with...
5051 (pass_data_tracer): ...new pass_data instance and...
5052 (make_pass_tracer): ...new function.
5053 * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
5054 to a subclass of gimple_opt_pass along with...
5055 (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
5056 (make_pass_diagnose_tm_blocks): ...new function.
5057 (pass_lower_tm): Convert from a global struct to a subclass of
5058 gimple_opt_pass along with...
5059 (pass_data_lower_tm): ...new pass_data instance and...
5060 (make_pass_lower_tm): ...new function.
5061 (pass_tm_init): Convert from a global struct to a subclass of
5062 gimple_opt_pass along with...
5063 (pass_data_tm_init): ...new pass_data instance and...
5064 (make_pass_tm_init): ...new function.
5065 (pass_tm_mark): Convert from a global struct to a subclass of
5066 gimple_opt_pass along with...
5067 (pass_data_tm_mark): ...new pass_data instance and...
5068 (make_pass_tm_mark): ...new function.
5069 (pass_tm_edges): Convert from a global struct to a subclass of
5070 gimple_opt_pass along with...
5071 (pass_data_tm_edges): ...new pass_data instance and...
5072 (make_pass_tm_edges): ...new function.
5073 (pass_tm_memopt): Convert from a global struct to a subclass of
5074 gimple_opt_pass along with...
5075 (pass_data_tm_memopt): ...new pass_data instance and...
5076 (make_pass_tm_memopt): ...new function.
5077 (pass_ipa_tm): Convert from a global struct to a subclass of
5078 simple_ipa_opt_pass along with...
5079 (pass_data_ipa_tm): ...new pass_data instance and...
5080 (make_pass_ipa_tm): ...new function.
5081 * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
5082 subclass of gimple_opt_pass along with...
5083 (pass_data_call_cdce): ...new pass_data instance and...
5084 (make_pass_call_cdce): ...new function.
5085 * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
5086 subclass of gimple_opt_pass along with...
5087 (pass_data_build_cfg): ...new pass_data instance and...
5088 (make_pass_build_cfg): ...new function.
5089 (pass_split_crit_edges): Convert from a global struct to a subclass of
5090 gimple_opt_pass along with...
5091 (pass_data_split_crit_edges): ...new pass_data instance and...
5092 (make_pass_split_crit_edges): ...new function.
5093 (pass_warn_function_return): Convert from a global struct to a
5094 subclass of gimple_opt_pass along with...
5095 (pass_data_warn_function_return): ...new pass_data instance and...
5096 (make_pass_warn_function_return): ...new function.
5097 (pass_warn_function_noreturn): Convert from a global struct to a
5098 subclass of gimple_opt_pass along with...
5099 (pass_data_warn_function_noreturn): ...new pass_data instance and...
5100 (make_pass_warn_function_noreturn): ...new function.
5101 (pass_warn_unused_result): Convert from a global struct to a subclass
5102 of gimple_opt_pass along with...
5103 (pass_data_warn_unused_result): ...new pass_data instance and...
5104 (make_pass_warn_unused_result): ...new function.
5105 * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
5106 a subclass of gimple_opt_pass along with...
5107 (pass_data_merge_phi): ...new pass_data instance and...
5108 (make_pass_merge_phi): ...new function.
5109 * tree-complex.c (pass_lower_complex): Convert from a global struct to
5110 a subclass of gimple_opt_pass along with...
5111 (pass_data_lower_complex): ...new pass_data instance and...
5112 (make_pass_lower_complex): ...new function.
5113 (pass_lower_complex_O0): Convert from a global struct to a subclass of
5114 gimple_opt_pass along with...
5115 (pass_data_lower_complex_O0): ...new pass_data instance and...
5116 (make_pass_lower_complex_O0): ...new function.
5117 * tree-eh.c (pass_lower_eh): Convert from a global struct to a
5118 subclass of gimple_opt_pass along with...
5119 (pass_data_lower_eh): ...new pass_data instance and...
5120 (make_pass_lower_eh): ...new function.
5121 (pass_refactor_eh): Convert from a global struct to a subclass of
5122 gimple_opt_pass along with...
5123 (pass_data_refactor_eh): ...new pass_data instance and...
5124 (make_pass_refactor_eh): ...new function.
5125 (pass_lower_resx): Convert from a global struct to a subclass of
5126 gimple_opt_pass along with...
5127 (pass_data_lower_resx): ...new pass_data instance and...
5128 (make_pass_lower_resx): ...new function.
5129 (pass_lower_eh_dispatch): Convert from a global struct to a subclass
5130 of gimple_opt_pass along with...
5131 (pass_data_lower_eh_dispatch): ...new pass_data instance and...
5132 (make_pass_lower_eh_dispatch): ...new function.
5133 (pass_cleanup_eh): Convert from a global struct to a subclass of
5134 gimple_opt_pass along with...
5135 (pass_data_cleanup_eh): ...new pass_data instance and...
5136 (make_pass_cleanup_eh): ...new function.
5137 * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
5138 to a subclass of simple_ipa_opt_pass along with...
5139 (pass_data_ipa_lower_emutls): ...new pass_data instance and...
5140 (make_pass_ipa_lower_emutls): ...new function.
5141 * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
5142 a subclass of gimple_opt_pass along with...
5143 (pass_data_if_conversion): ...new pass_data instance and...
5144 (make_pass_if_conversion): ...new function.
5145 * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
5146 subclass of gimple_opt_pass along with...
5147 (pass_data_build_ssa): ...new pass_data instance and...
5148 (make_pass_build_ssa): ...new function.
5149 * tree-loop-distribution.c (pass_loop_distribution): Convert from a
5150 global struct to a subclass of gimple_opt_pass along with...
5151 (pass_data_loop_distribution): ...new pass_data instance and...
5152 (make_pass_loop_distribution): ...new function.
5153 * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
5154 subclass of gimple_opt_pass along with...
5155 (pass_data_mudflap_1): ...new pass_data instance and...
5156 (make_pass_mudflap_1): ...new function.
5157 (pass_mudflap_2): Convert from a global struct to a subclass of
5158 gimple_opt_pass along with...
5159 (pass_data_mudflap_2): ...new pass_data instance and...
5160 (make_pass_mudflap_2): ...new function.
5161 * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
5162 subclass of gimple_opt_pass along with...
5163 (pass_data_mudflap_1): ...new pass_data instance and...
5164 (make_pass_mudflap_1): ...new function.
5165 (pass_mudflap_2): Convert from a global struct to a subclass of
5166 gimple_opt_pass along with...
5167 (pass_data_mudflap_2): ...new pass_data instance and...
5168 (make_pass_mudflap_2): ...new function.
5169 * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
5170 gimple_opt_pass along with...
5171 (pass_data_nrv): ...new pass_data instance and...
5172 (make_pass_nrv): ...new function.
5173 (pass_return_slot): Convert from a global struct to a subclass of
5174 gimple_opt_pass along with...
5175 (pass_data_return_slot): ...new pass_data instance and...
5176 (make_pass_return_slot): ...new function.
5177 * tree-object-size.c (pass_object_sizes): Convert from a global struct
5178 to a subclass of gimple_opt_pass along with...
5179 (pass_data_object_sizes): ...new pass_data instance and...
5180 (make_pass_object_sizes): ...new function.
5181 * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
5182 global struct to a subclass of gimple_opt_pass along with...
5183 (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
5185 (make_pass_cleanup_cfg_post_optimizing): ...new function.
5186 (pass_fixup_cfg): Convert from a global struct to a subclass of
5187 gimple_opt_pass along with...
5188 (pass_data_fixup_cfg): ...new pass_data instance and...
5189 (make_pass_fixup_cfg): ...new function.
5190 * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
5191 (make_pass_mudflap_1): ...new function.
5192 (pass_mudflap_2): Replace declaration with that of...
5193 (make_pass_mudflap_2): ...new function.
5194 (pass_asan): Replace declaration with that of...
5195 (make_pass_asan): ...new function.
5196 (pass_asan_O0): Replace declaration with that of...
5197 (make_pass_asan_O0): ...new function.
5198 (pass_tsan): Replace declaration with that of...
5199 (make_pass_tsan): ...new function.
5200 (pass_tsan_O0): Replace declaration with that of...
5201 (make_pass_tsan_O0): ...new function.
5202 (pass_lower_cf): Replace declaration with that of...
5203 (make_pass_lower_cf): ...new function.
5204 (pass_refactor_eh): Replace declaration with that of...
5205 (make_pass_refactor_eh): ...new function.
5206 (pass_lower_eh): Replace declaration with that of...
5207 (make_pass_lower_eh): ...new function.
5208 (pass_lower_eh_dispatch): Replace declaration with that of...
5209 (make_pass_lower_eh_dispatch): ...new function.
5210 (pass_lower_resx): Replace declaration with that of...
5211 (make_pass_lower_resx): ...new function.
5212 (pass_build_cfg): Replace declaration with that of...
5213 (make_pass_build_cfg): ...new function.
5214 (pass_early_tree_profile): Replace declaration with that of...
5215 (make_pass_early_tree_profile): ...new function.
5216 (pass_cleanup_eh): Replace declaration with that of...
5217 (make_pass_cleanup_eh): ...new function.
5218 (pass_sra): Replace declaration with that of...
5219 (make_pass_sra): ...new function.
5220 (pass_sra_early): Replace declaration with that of...
5221 (make_pass_sra_early): ...new function.
5222 (pass_early_ipa_sra): Replace declaration with that of...
5223 (make_pass_early_ipa_sra): ...new function.
5224 (pass_tail_recursion): Replace declaration with that of...
5225 (make_pass_tail_recursion): ...new function.
5226 (pass_tail_calls): Replace declaration with that of...
5227 (make_pass_tail_calls): ...new function.
5228 (pass_tree_loop): Replace declaration with that of...
5229 (make_pass_tree_loop): ...new function.
5230 (pass_tree_loop_init): Replace declaration with that of...
5231 (make_pass_tree_loop_init): ...new function.
5232 (pass_lim): Replace declaration with that of...
5233 (make_pass_lim): ...new function.
5234 (pass_tree_unswitch): Replace declaration with that of...
5235 (make_pass_tree_unswitch): ...new function.
5236 (pass_predcom): Replace declaration with that of...
5237 (make_pass_predcom): ...new function.
5238 (pass_iv_canon): Replace declaration with that of...
5239 (make_pass_iv_canon): ...new function.
5240 (pass_scev_cprop): Replace declaration with that of...
5241 (make_pass_scev_cprop): ...new function.
5242 (pass_empty_loop): Replace declaration with that of...
5243 (make_pass_empty_loop): ...new function.
5244 (pass_record_bounds): Replace declaration with that of...
5245 (make_pass_record_bounds): ...new function.
5246 (pass_graphite): Replace declaration with that of...
5247 (make_pass_graphite): ...new function.
5248 (pass_graphite_transforms): Replace declaration with that of...
5249 (make_pass_graphite_transforms): ...new function.
5250 (pass_if_conversion): Replace declaration with that of...
5251 (make_pass_if_conversion): ...new function.
5252 (pass_loop_distribution): Replace declaration with that of...
5253 (make_pass_loop_distribution): ...new function.
5254 (pass_vectorize): Replace declaration with that of...
5255 (make_pass_vectorize): ...new function.
5256 (pass_slp_vectorize): Replace declaration with that of...
5257 (make_pass_slp_vectorize): ...new function.
5258 (pass_complete_unroll): Replace declaration with that of...
5259 (make_pass_complete_unroll): ...new function.
5260 (pass_complete_unrolli): Replace declaration with that of...
5261 (make_pass_complete_unrolli): ...new function.
5262 (pass_parallelize_loops): Replace declaration with that of...
5263 (make_pass_parallelize_loops): ...new function.
5264 (pass_loop_prefetch): Replace declaration with that of...
5265 (make_pass_loop_prefetch): ...new function.
5266 (pass_iv_optimize): Replace declaration with that of...
5267 (make_pass_iv_optimize): ...new function.
5268 (pass_tree_loop_done): Replace declaration with that of...
5269 (make_pass_tree_loop_done): ...new function.
5270 (pass_ch): Replace declaration with that of...
5271 (make_pass_ch): ...new function.
5272 (pass_ccp): Replace declaration with that of...
5273 (make_pass_ccp): ...new function.
5274 (pass_phi_only_cprop): Replace declaration with that of...
5275 (make_pass_phi_only_cprop): ...new function.
5276 (pass_build_ssa): Replace declaration with that of...
5277 (make_pass_build_ssa): ...new function.
5278 (pass_build_alias): Replace declaration with that of...
5279 (make_pass_build_alias): ...new function.
5280 (pass_build_ealias): Replace declaration with that of...
5281 (make_pass_build_ealias): ...new function.
5282 (pass_dominator): Replace declaration with that of...
5283 (make_pass_dominator): ...new function.
5284 (pass_dce): Replace declaration with that of...
5285 (make_pass_dce): ...new function.
5286 (pass_dce_loop): Replace declaration with that of...
5287 (make_pass_dce_loop): ...new function.
5288 (pass_cd_dce): Replace declaration with that of...
5289 (make_pass_cd_dce): ...new function.
5290 (pass_call_cdce): Replace declaration with that of...
5291 (make_pass_call_cdce): ...new function.
5292 (pass_merge_phi): Replace declaration with that of...
5293 (make_pass_merge_phi): ...new function.
5294 (pass_split_crit_edges): Replace declaration with that of...
5295 (make_pass_split_crit_edges): ...new function.
5296 (pass_pre): Replace declaration with that of...
5297 (make_pass_pre): ...new function.
5298 (pass_profile): Replace declaration with that of...
5299 (make_pass_profile): ...new function.
5300 (pass_strip_predict_hints): Replace declaration with that of...
5301 (make_pass_strip_predict_hints): ...new function.
5302 (pass_lower_complex_O0): Replace declaration with that of...
5303 (make_pass_lower_complex_O0): ...new function.
5304 (pass_lower_complex): Replace declaration with that of...
5305 (make_pass_lower_complex): ...new function.
5306 (pass_lower_vector): Replace declaration with that of...
5307 (make_pass_lower_vector): ...new function.
5308 (pass_lower_vector_ssa): Replace declaration with that of...
5309 (make_pass_lower_vector_ssa): ...new function.
5310 (pass_lower_omp): Replace declaration with that of...
5311 (make_pass_lower_omp): ...new function.
5312 (pass_diagnose_omp_blocks): Replace declaration with that of...
5313 (make_pass_diagnose_omp_blocks): ...new function.
5314 (pass_expand_omp): Replace declaration with that of...
5315 (make_pass_expand_omp): ...new function.
5316 (pass_expand_omp_ssa): Replace declaration with that of...
5317 (make_pass_expand_omp_ssa): ...new function.
5318 (pass_object_sizes): Replace declaration with that of...
5319 (make_pass_object_sizes): ...new function.
5320 (pass_strlen): Replace declaration with that of...
5321 (make_pass_strlen): ...new function.
5322 (pass_fold_builtins): Replace declaration with that of...
5323 (make_pass_fold_builtins): ...new function.
5324 (pass_stdarg): Replace declaration with that of...
5325 (make_pass_stdarg): ...new function.
5326 (pass_early_warn_uninitialized): Replace declaration with that of...
5327 (make_pass_early_warn_uninitialized): ...new function.
5328 (pass_late_warn_uninitialized): Replace declaration with that of...
5329 (make_pass_late_warn_uninitialized): ...new function.
5330 (pass_cse_reciprocals): Replace declaration with that of...
5331 (make_pass_cse_reciprocals): ...new function.
5332 (pass_cse_sincos): Replace declaration with that of...
5333 (make_pass_cse_sincos): ...new function.
5334 (pass_optimize_bswap): Replace declaration with that of...
5335 (make_pass_optimize_bswap): ...new function.
5336 (pass_optimize_widening_mul): Replace declaration with that of...
5337 (make_pass_optimize_widening_mul): ...new function.
5338 (pass_warn_function_return): Replace declaration with that of...
5339 (make_pass_warn_function_return): ...new function.
5340 (pass_warn_function_noreturn): Replace declaration with that of...
5341 (make_pass_warn_function_noreturn): ...new function.
5342 (pass_cselim): Replace declaration with that of...
5343 (make_pass_cselim): ...new function.
5344 (pass_phiopt): Replace declaration with that of...
5345 (make_pass_phiopt): ...new function.
5346 (pass_forwprop): Replace declaration with that of...
5347 (make_pass_forwprop): ...new function.
5348 (pass_phiprop): Replace declaration with that of...
5349 (make_pass_phiprop): ...new function.
5350 (pass_tree_ifcombine): Replace declaration with that of...
5351 (make_pass_tree_ifcombine): ...new function.
5352 (pass_dse): Replace declaration with that of...
5353 (make_pass_dse): ...new function.
5354 (pass_nrv): Replace declaration with that of...
5355 (make_pass_nrv): ...new function.
5356 (pass_rename_ssa_copies): Replace declaration with that of...
5357 (make_pass_rename_ssa_copies): ...new function.
5358 (pass_sink_code): Replace declaration with that of...
5359 (make_pass_sink_code): ...new function.
5360 (pass_fre): Replace declaration with that of...
5361 (make_pass_fre): ...new function.
5362 (pass_check_data_deps): Replace declaration with that of...
5363 (make_pass_check_data_deps): ...new function.
5364 (pass_copy_prop): Replace declaration with that of...
5365 (make_pass_copy_prop): ...new function.
5366 (pass_vrp): Replace declaration with that of...
5367 (make_pass_vrp): ...new function.
5368 (pass_uncprop): Replace declaration with that of...
5369 (make_pass_uncprop): ...new function.
5370 (pass_return_slot): Replace declaration with that of...
5371 (make_pass_return_slot): ...new function.
5372 (pass_reassoc): Replace declaration with that of...
5373 (make_pass_reassoc): ...new function.
5374 (pass_rebuild_cgraph_edges): Replace declaration with that of...
5375 (make_pass_rebuild_cgraph_edges): ...new function.
5376 (pass_remove_cgraph_callee_edges): Replace declaration with that of...
5377 (make_pass_remove_cgraph_callee_edges): ...new function.
5378 (pass_build_cgraph_edges): Replace declaration with that of...
5379 (make_pass_build_cgraph_edges): ...new function.
5380 (pass_local_pure_const): Replace declaration with that of...
5381 (make_pass_local_pure_const): ...new function.
5382 (pass_tracer): Replace declaration with that of...
5383 (make_pass_tracer): ...new function.
5384 (pass_warn_unused_result): Replace declaration with that of...
5385 (make_pass_warn_unused_result): ...new function.
5386 (pass_diagnose_tm_blocks): Replace declaration with that of...
5387 (make_pass_diagnose_tm_blocks): ...new function.
5388 (pass_lower_tm): Replace declaration with that of...
5389 (make_pass_lower_tm): ...new function.
5390 (pass_tm_init): Replace declaration with that of...
5391 (make_pass_tm_init): ...new function.
5392 (pass_tm_mark): Replace declaration with that of...
5393 (make_pass_tm_mark): ...new function.
5394 (pass_tm_memopt): Replace declaration with that of...
5395 (make_pass_tm_memopt): ...new function.
5396 (pass_tm_edges): Replace declaration with that of...
5397 (make_pass_tm_edges): ...new function.
5398 (pass_split_functions): Replace declaration with that of...
5399 (make_pass_split_functions): ...new function.
5400 (pass_feedback_split_functions): Replace declaration with that of...
5401 (make_pass_feedback_split_functions): ...new function.
5402 (pass_strength_reduction): Replace declaration with that of...
5403 (make_pass_strength_reduction): ...new function.
5404 (pass_ipa_lower_emutls): Replace declaration with that of...
5405 (make_pass_ipa_lower_emutls): ...new function.
5406 (pass_ipa_function_and_variable_visibility): Replace declaration with
5408 (make_pass_ipa_function_and_variable_visibility): ...new function.
5409 (pass_ipa_tree_profile): Replace declaration with that of...
5410 (make_pass_ipa_tree_profile): ...new function.
5411 (pass_early_local_passes): Replace declaration with that of...
5412 (make_pass_early_local_passes): ...new function.
5413 (pass_ipa_whole_program_visibility): Replace declaration with that
5415 (make_pass_ipa_whole_program_visibility): ...new function.
5416 (pass_ipa_lto_gimple_out): Replace declaration with that of...
5417 (make_pass_ipa_lto_gimple_out): ...new function.
5418 (pass_ipa_increase_alignment): Replace declaration with that of...
5419 (make_pass_ipa_increase_alignment): ...new function.
5420 (pass_ipa_inline): Replace declaration with that of...
5421 (make_pass_ipa_inline): ...new function.
5422 (pass_ipa_free_lang_data): Replace declaration with that of...
5423 (make_pass_ipa_free_lang_data): ...new function.
5424 (pass_ipa_free_inline_summary): Replace declaration with that of...
5425 (make_pass_ipa_free_inline_summary): ...new function.
5426 (pass_ipa_cp): Replace declaration with that of...
5427 (make_pass_ipa_cp): ...new function.
5428 (pass_ipa_reference): Replace declaration with that of...
5429 (make_pass_ipa_reference): ...new function.
5430 (pass_ipa_pure_const): Replace declaration with that of...
5431 (make_pass_ipa_pure_const): ...new function.
5432 (pass_ipa_pta): Replace declaration with that of...
5433 (make_pass_ipa_pta): ...new function.
5434 (pass_ipa_lto_finish_out): Replace declaration with that of...
5435 (make_pass_ipa_lto_finish_out): ...new function.
5436 (pass_ipa_tm): Replace declaration with that of...
5437 (make_pass_ipa_tm): ...new function.
5438 (pass_ipa_profile): Replace declaration with that of...
5439 (make_pass_ipa_profile): ...new function.
5440 (pass_ipa_cdtor_merge): Replace declaration with that of...
5441 (make_pass_ipa_cdtor_merge): ...new function.
5442 (pass_cleanup_cfg_post_optimizing): Replace declaration with that
5444 (make_pass_cleanup_cfg_post_optimizing): ...new function.
5445 (pass_init_datastructures): Replace declaration with that of...
5446 (make_pass_init_datastructures): ...new function.
5447 (pass_fixup_cfg): Replace declaration with that of...
5448 (make_pass_fixup_cfg): ...new function.
5449 (pass_expand): Replace declaration with that of...
5450 (make_pass_expand): ...new function.
5451 (pass_instantiate_virtual_regs): Replace declaration with that of...
5452 (make_pass_instantiate_virtual_regs): ...new function.
5453 (pass_rtl_fwprop): Replace declaration with that of...
5454 (make_pass_rtl_fwprop): ...new function.
5455 (pass_rtl_fwprop_addr): Replace declaration with that of...
5456 (make_pass_rtl_fwprop_addr): ...new function.
5457 (pass_jump): Replace declaration with that of...
5458 (make_pass_jump): ...new function.
5459 (pass_jump2): Replace declaration with that of...
5460 (make_pass_jump2): ...new function.
5461 (pass_lower_subreg): Replace declaration with that of...
5462 (make_pass_lower_subreg): ...new function.
5463 (pass_cse): Replace declaration with that of...
5464 (make_pass_cse): ...new function.
5465 (pass_fast_rtl_dce): Replace declaration with that of...
5466 (make_pass_fast_rtl_dce): ...new function.
5467 (pass_ud_rtl_dce): Replace declaration with that of...
5468 (make_pass_ud_rtl_dce): ...new function.
5469 (pass_rtl_dce): Replace declaration with that of...
5470 (make_pass_rtl_dce): ...new function.
5471 (pass_rtl_dse1): Replace declaration with that of...
5472 (make_pass_rtl_dse1): ...new function.
5473 (pass_rtl_dse2): Replace declaration with that of...
5474 (make_pass_rtl_dse2): ...new function.
5475 (pass_rtl_dse3): Replace declaration with that of...
5476 (make_pass_rtl_dse3): ...new function.
5477 (pass_rtl_cprop): Replace declaration with that of...
5478 (make_pass_rtl_cprop): ...new function.
5479 (pass_rtl_pre): Replace declaration with that of...
5480 (make_pass_rtl_pre): ...new function.
5481 (pass_rtl_hoist): Replace declaration with that of...
5482 (make_pass_rtl_hoist): ...new function.
5483 (pass_rtl_store_motion): Replace declaration with that of...
5484 (make_pass_rtl_store_motion): ...new function.
5485 (pass_cse_after_global_opts): Replace declaration with that of...
5486 (make_pass_cse_after_global_opts): ...new function.
5487 (pass_rtl_ifcvt): Replace declaration with that of...
5488 (make_pass_rtl_ifcvt): ...new function.
5489 (pass_into_cfg_layout_mode): Replace declaration with that of...
5490 (make_pass_into_cfg_layout_mode): ...new function.
5491 (pass_outof_cfg_layout_mode): Replace declaration with that of...
5492 (make_pass_outof_cfg_layout_mode): ...new function.
5493 (pass_loop2): Replace declaration with that of...
5494 (make_pass_loop2): ...new function.
5495 (pass_rtl_loop_init): Replace declaration with that of...
5496 (make_pass_rtl_loop_init): ...new function.
5497 (pass_rtl_move_loop_invariants): Replace declaration with that of...
5498 (make_pass_rtl_move_loop_invariants): ...new function.
5499 (pass_rtl_unswitch): Replace declaration with that of...
5500 (make_pass_rtl_unswitch): ...new function.
5501 (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
5502 (make_pass_rtl_unroll_and_peel_loops): ...new function.
5503 (pass_rtl_doloop): Replace declaration with that of...
5504 (make_pass_rtl_doloop): ...new function.
5505 (pass_rtl_loop_done): Replace declaration with that of...
5506 (make_pass_rtl_loop_done): ...new function.
5507 (pass_web): Replace declaration with that of...
5508 (make_pass_web): ...new function.
5509 (pass_cse2): Replace declaration with that of...
5510 (make_pass_cse2): ...new function.
5511 (pass_df_initialize_opt): Replace declaration with that of...
5512 (make_pass_df_initialize_opt): ...new function.
5513 (pass_df_initialize_no_opt): Replace declaration with that of...
5514 (make_pass_df_initialize_no_opt): ...new function.
5515 (pass_reginfo_init): Replace declaration with that of...
5516 (make_pass_reginfo_init): ...new function.
5517 (pass_inc_dec): Replace declaration with that of...
5518 (make_pass_inc_dec): ...new function.
5519 (pass_stack_ptr_mod): Replace declaration with that of...
5520 (make_pass_stack_ptr_mod): ...new function.
5521 (pass_initialize_regs): Replace declaration with that of...
5522 (make_pass_initialize_regs): ...new function.
5523 (pass_combine): Replace declaration with that of...
5524 (make_pass_combine): ...new function.
5525 (pass_if_after_combine): Replace declaration with that of...
5526 (make_pass_if_after_combine): ...new function.
5527 (pass_ree): Replace declaration with that of...
5528 (make_pass_ree): ...new function.
5529 (pass_partition_blocks): Replace declaration with that of...
5530 (make_pass_partition_blocks): ...new function.
5531 (pass_match_asm_constraints): Replace declaration with that of...
5532 (make_pass_match_asm_constraints): ...new function.
5533 (pass_regmove): Replace declaration with that of...
5534 (make_pass_regmove): ...new function.
5535 (pass_split_all_insns): Replace declaration with that of...
5536 (make_pass_split_all_insns): ...new function.
5537 (pass_fast_rtl_byte_dce): Replace declaration with that of...
5538 (make_pass_fast_rtl_byte_dce): ...new function.
5539 (pass_lower_subreg2): Replace declaration with that of...
5540 (make_pass_lower_subreg2): ...new function.
5541 (pass_mode_switching): Replace declaration with that of...
5542 (make_pass_mode_switching): ...new function.
5543 (pass_sms): Replace declaration with that of...
5544 (make_pass_sms): ...new function.
5545 (pass_sched): Replace declaration with that of...
5546 (make_pass_sched): ...new function.
5547 (pass_ira): Replace declaration with that of...
5548 (make_pass_ira): ...new function.
5549 (pass_reload): Replace declaration with that of...
5550 (make_pass_reload): ...new function.
5551 (pass_clean_state): Replace declaration with that of...
5552 (make_pass_clean_state): ...new function.
5553 (pass_branch_prob): Replace declaration with that of...
5554 (make_pass_branch_prob): ...new function.
5555 (pass_value_profile_transformations): Replace declaration with that
5557 (make_pass_value_profile_transformations): ...new function.
5558 (pass_postreload_cse): Replace declaration with that of...
5559 (make_pass_postreload_cse): ...new function.
5560 (pass_gcse2): Replace declaration with that of...
5561 (make_pass_gcse2): ...new function.
5562 (pass_split_after_reload): Replace declaration with that of...
5563 (make_pass_split_after_reload): ...new function.
5564 (pass_branch_target_load_optimize1): Replace declaration with that
5566 (make_pass_branch_target_load_optimize1): ...new function.
5567 (pass_thread_prologue_and_epilogue): Replace declaration with that
5569 (make_pass_thread_prologue_and_epilogue): ...new function.
5570 (pass_stack_adjustments): Replace declaration with that of...
5571 (make_pass_stack_adjustments): ...new function.
5572 (pass_peephole2): Replace declaration with that of...
5573 (make_pass_peephole2): ...new function.
5574 (pass_if_after_reload): Replace declaration with that of...
5575 (make_pass_if_after_reload): ...new function.
5576 (pass_regrename): Replace declaration with that of...
5577 (make_pass_regrename): ...new function.
5578 (pass_cprop_hardreg): Replace declaration with that of...
5579 (make_pass_cprop_hardreg): ...new function.
5580 (pass_reorder_blocks): Replace declaration with that of...
5581 (make_pass_reorder_blocks): ...new function.
5582 (pass_branch_target_load_optimize2): Replace declaration with that
5584 (make_pass_branch_target_load_optimize2): ...new function.
5585 (pass_leaf_regs): Replace declaration with that of...
5586 (make_pass_leaf_regs): ...new function.
5587 (pass_split_before_sched2): Replace declaration with that of...
5588 (make_pass_split_before_sched2): ...new function.
5589 (pass_compare_elim_after_reload): Replace declaration with that of...
5590 (make_pass_compare_elim_after_reload): ...new function.
5591 (pass_sched2): Replace declaration with that of...
5592 (make_pass_sched2): ...new function.
5593 (pass_stack_regs): Replace declaration with that of...
5594 (make_pass_stack_regs): ...new function.
5595 (pass_stack_regs_run): Replace declaration with that of...
5596 (make_pass_stack_regs_run): ...new function.
5597 (pass_df_finish): Replace declaration with that of...
5598 (make_pass_df_finish): ...new function.
5599 (pass_compute_alignments): Replace declaration with that of...
5600 (make_pass_compute_alignments): ...new function.
5601 (pass_duplicate_computed_gotos): Replace declaration with that of...
5602 (make_pass_duplicate_computed_gotos): ...new function.
5603 (pass_variable_tracking): Replace declaration with that of...
5604 (make_pass_variable_tracking): ...new function.
5605 (pass_free_cfg): Replace declaration with that of...
5606 (make_pass_free_cfg): ...new function.
5607 (pass_machine_reorg): Replace declaration with that of...
5608 (make_pass_machine_reorg): ...new function.
5609 (pass_cleanup_barriers): Replace declaration with that of...
5610 (make_pass_cleanup_barriers): ...new function.
5611 (pass_delay_slots): Replace declaration with that of...
5612 (make_pass_delay_slots): ...new function.
5613 (pass_split_for_shorten_branches): Replace declaration with that of...
5614 (make_pass_split_for_shorten_branches): ...new function.
5615 (pass_split_before_regstack): Replace declaration with that of...
5616 (make_pass_split_before_regstack): ...new function.
5617 (pass_convert_to_eh_region_ranges): Replace declaration with that
5619 (make_pass_convert_to_eh_region_ranges): ...new function.
5620 (pass_shorten_branches): Replace declaration with that of...
5621 (make_pass_shorten_branches): ...new function.
5622 (pass_set_nothrow_function_flags): Replace declaration with that of...
5623 (make_pass_set_nothrow_function_flags): ...new function.
5624 (pass_dwarf2_frame): Replace declaration with that of...
5625 (make_pass_dwarf2_frame): ...new function.
5626 (pass_final): Replace declaration with that of...
5627 (make_pass_final): ...new function.
5628 (pass_rtl_seqabstr): Replace declaration with that of...
5629 (make_pass_rtl_seqabstr): ...new function.
5630 (pass_release_ssa_names): Replace declaration with that of...
5631 (make_pass_release_ssa_names): ...new function.
5632 (pass_early_inline): Replace declaration with that of...
5633 (make_pass_early_inline): ...new function.
5634 (pass_inline_parameters): Replace declaration with that of...
5635 (make_pass_inline_parameters): ...new function.
5636 (pass_update_address_taken): Replace declaration with that of...
5637 (make_pass_update_address_taken): ...new function.
5638 (pass_convert_switch): Replace declaration with that of...
5639 (make_pass_convert_switch): ...new function.
5640 * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
5641 to a subclass of simple_ipa_opt_pass along with...
5642 (pass_data_ipa_tree_profile): ...new pass_data instance and...
5643 (make_pass_ipa_tree_profile): ...new function.
5644 * tree-sra.c (pass_sra_early): Convert from a global struct to a
5645 subclass of gimple_opt_pass along with...
5646 (pass_data_sra_early): ...new pass_data instance and...
5647 (make_pass_sra_early): ...new function.
5648 (pass_sra): Convert from a global struct to a subclass of
5649 gimple_opt_pass along with...
5650 (pass_data_sra): ...new pass_data instance and...
5651 (make_pass_sra): ...new function.
5652 (pass_early_ipa_sra): Convert from a global struct to a subclass of
5653 gimple_opt_pass along with...
5654 (pass_data_early_ipa_sra): ...new pass_data instance and...
5655 (make_pass_early_ipa_sra): ...new function.
5656 * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
5657 subclass of gimple_opt_pass along with...
5658 (pass_data_ccp): ...new pass_data instance and...
5659 (make_pass_ccp): ...new function.
5660 (pass_fold_builtins): Convert from a global struct to a subclass of
5661 gimple_opt_pass along with...
5662 (pass_data_fold_builtins): ...new pass_data instance and...
5663 (make_pass_fold_builtins): ...new function.
5664 * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
5665 subclass of gimple_opt_pass along with...
5666 (pass_data_copy_prop): ...new pass_data instance and...
5667 (make_pass_copy_prop): ...new function.
5668 * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
5669 global struct to a subclass of gimple_opt_pass along with...
5670 (pass_data_rename_ssa_copies): ...new pass_data instance and...
5671 (make_pass_rename_ssa_copies): ...new function.
5672 * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
5673 subclass of gimple_opt_pass along with...
5674 (pass_data_dce): ...new pass_data instance and...
5675 (make_pass_dce): ...new function.
5676 (pass_dce_loop): Convert from a global struct to a subclass of
5677 gimple_opt_pass along with...
5678 (pass_data_dce_loop): ...new pass_data instance and...
5679 (make_pass_dce_loop): ...new function.
5680 (pass_cd_dce): Convert from a global struct to a subclass of
5681 gimple_opt_pass along with...
5682 (pass_data_cd_dce): ...new pass_data instance and...
5683 (make_pass_cd_dce): ...new function.
5684 * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
5685 subclass of gimple_opt_pass along with...
5686 (pass_data_dominator): ...new pass_data instance and...
5687 (make_pass_dominator): ...new function.
5688 (pass_phi_only_cprop): Convert from a global struct to a subclass of
5689 gimple_opt_pass along with...
5690 (pass_data_phi_only_cprop): ...new pass_data instance and...
5691 (make_pass_phi_only_cprop): ...new function.
5692 * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
5693 subclass of gimple_opt_pass along with...
5694 (pass_data_dse): ...new pass_data instance and...
5695 (make_pass_dse): ...new function.
5696 * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
5697 a subclass of gimple_opt_pass along with...
5698 (pass_data_forwprop): ...new pass_data instance and...
5699 (make_pass_forwprop): ...new function.
5700 * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
5701 struct to a subclass of gimple_opt_pass along with...
5702 (pass_data_tree_ifcombine): ...new pass_data instance and...
5703 (make_pass_tree_ifcombine): ...new function.
5704 * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
5705 subclass of gimple_opt_pass along with...
5706 (pass_data_ch): ...new pass_data instance and...
5707 (make_pass_ch): ...new function.
5708 * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
5709 subclass of gimple_opt_pass along with...
5710 (pass_data_tree_loop): ...new pass_data instance and...
5711 (make_pass_tree_loop): ...new function.
5712 (pass_tree_loop_init): Convert from a global struct to a subclass of
5713 gimple_opt_pass along with...
5714 (pass_data_tree_loop_init): ...new pass_data instance and...
5715 (make_pass_tree_loop_init): ...new function.
5716 (pass_lim): Convert from a global struct to a subclass of
5717 gimple_opt_pass along with...
5718 (pass_data_lim): ...new pass_data instance and...
5719 (make_pass_lim): ...new function.
5720 (pass_tree_unswitch): Convert from a global struct to a subclass of
5721 gimple_opt_pass along with...
5722 (pass_data_tree_unswitch): ...new pass_data instance and...
5723 (make_pass_tree_unswitch): ...new function.
5724 (pass_predcom): Convert from a global struct to a subclass of
5725 gimple_opt_pass along with...
5726 (pass_data_predcom): ...new pass_data instance and...
5727 (make_pass_predcom): ...new function.
5728 (pass_vectorize): Convert from a global struct to a subclass of
5729 gimple_opt_pass along with...
5730 (pass_data_vectorize): ...new pass_data instance and...
5731 (make_pass_vectorize): ...new function.
5732 (pass_graphite): Convert from a global struct to a subclass of
5733 gimple_opt_pass along with...
5734 (pass_data_graphite): ...new pass_data instance and...
5735 (make_pass_graphite): ...new function.
5736 (pass_graphite_transforms): Convert from a global struct to a subclass
5737 of gimple_opt_pass along with...
5738 (pass_data_graphite_transforms): ...new pass_data instance and...
5739 (make_pass_graphite_transforms): ...new function.
5740 (pass_check_data_deps): Convert from a global struct to a subclass of
5741 gimple_opt_pass along with...
5742 (pass_data_check_data_deps): ...new pass_data instance and...
5743 (make_pass_check_data_deps): ...new function.
5744 (pass_iv_canon): Convert from a global struct to a subclass of
5745 gimple_opt_pass along with...
5746 (pass_data_iv_canon): ...new pass_data instance and...
5747 (make_pass_iv_canon): ...new function.
5748 (pass_scev_cprop): Convert from a global struct to a subclass of
5749 gimple_opt_pass along with...
5750 (pass_data_scev_cprop): ...new pass_data instance and...
5751 (make_pass_scev_cprop): ...new function.
5752 (pass_record_bounds): Convert from a global struct to a subclass of
5753 gimple_opt_pass along with...
5754 (pass_data_record_bounds): ...new pass_data instance and...
5755 (make_pass_record_bounds): ...new function.
5756 (pass_complete_unroll): Convert from a global struct to a subclass of
5757 gimple_opt_pass along with...
5758 (pass_data_complete_unroll): ...new pass_data instance and...
5759 (make_pass_complete_unroll): ...new function.
5760 (pass_complete_unrolli): Convert from a global struct to a subclass of
5761 gimple_opt_pass along with...
5762 (pass_data_complete_unrolli): ...new pass_data instance and...
5763 (make_pass_complete_unrolli): ...new function.
5764 (pass_parallelize_loops): Convert from a global struct to a subclass
5765 of gimple_opt_pass along with...
5766 (pass_data_parallelize_loops): ...new pass_data instance and...
5767 (make_pass_parallelize_loops): ...new function.
5768 (pass_loop_prefetch): Convert from a global struct to a subclass of
5769 gimple_opt_pass along with...
5770 (pass_data_loop_prefetch): ...new pass_data instance and...
5771 (make_pass_loop_prefetch): ...new function.
5772 (pass_iv_optimize): Convert from a global struct to a subclass of
5773 gimple_opt_pass along with...
5774 (pass_data_iv_optimize): ...new pass_data instance and...
5775 (make_pass_iv_optimize): ...new function.
5776 (pass_tree_loop_done): Convert from a global struct to a subclass of
5777 gimple_opt_pass along with...
5778 (pass_data_tree_loop_done): ...new pass_data instance and...
5779 (make_pass_tree_loop_done): ...new function.
5780 * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
5781 struct to a subclass of gimple_opt_pass along with...
5782 (pass_data_cse_reciprocals): ...new pass_data instance and...
5783 (make_pass_cse_reciprocals): ...new function.
5784 (pass_cse_sincos): Convert from a global struct to a subclass of
5785 gimple_opt_pass along with...
5786 (pass_data_cse_sincos): ...new pass_data instance and...
5787 (make_pass_cse_sincos): ...new function.
5788 (pass_optimize_bswap): Convert from a global struct to a subclass of
5789 gimple_opt_pass along with...
5790 (pass_data_optimize_bswap): ...new pass_data instance and...
5791 (make_pass_optimize_bswap): ...new function.
5792 (pass_optimize_widening_mul): Convert from a global struct to a
5793 subclass of gimple_opt_pass along with...
5794 (pass_data_optimize_widening_mul): ...new pass_data instance and...
5795 (make_pass_optimize_widening_mul): ...new function.
5796 * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
5797 subclass of gimple_opt_pass along with...
5798 (pass_data_phiopt): ...new pass_data instance and...
5799 (make_pass_phiopt): ...new function.
5800 (pass_cselim): Convert from a global struct to a subclass of
5801 gimple_opt_pass along with...
5802 (pass_data_cselim): ...new pass_data instance and...
5803 (make_pass_cselim): ...new function.
5804 * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
5805 subclass of gimple_opt_pass along with...
5806 (pass_data_phiprop): ...new pass_data instance and...
5807 (make_pass_phiprop): ...new function.
5808 * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
5809 subclass of gimple_opt_pass along with...
5810 (pass_data_pre): ...new pass_data instance and...
5811 (make_pass_pre): ...new function.
5812 (pass_fre): Convert from a global struct to a subclass of
5813 gimple_opt_pass along with...
5814 (pass_data_fre): ...new pass_data instance and...
5815 (make_pass_fre): ...new function.
5816 * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
5817 subclass of gimple_opt_pass along with...
5818 (pass_data_reassoc): ...new pass_data instance and...
5819 (make_pass_reassoc): ...new function.
5820 * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
5821 subclass of gimple_opt_pass along with...
5822 (pass_data_sink_code): ...new pass_data instance and...
5823 (make_pass_sink_code): ...new function.
5824 * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
5825 subclass of gimple_opt_pass along with...
5826 (pass_data_strlen): ...new pass_data instance and...
5827 (make_pass_strlen): ...new function.
5828 * tree-ssa-structalias.c (pass_build_alias): Convert from a global
5829 struct to a subclass of gimple_opt_pass along with...
5830 (pass_data_build_alias): ...new pass_data instance and...
5831 (make_pass_build_alias): ...new function.
5832 (pass_build_ealias): Convert from a global struct to a subclass of
5833 gimple_opt_pass along with...
5834 (pass_data_build_ealias): ...new pass_data instance and...
5835 (make_pass_build_ealias): ...new function.
5836 (pass_ipa_pta): Convert from a global struct to a subclass of
5837 simple_ipa_opt_pass along with...
5838 (pass_data_ipa_pta): ...new pass_data instance and...
5839 (make_pass_ipa_pta): ...new function.
5840 * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
5841 subclass of gimple_opt_pass along with...
5842 (pass_data_uncprop): ...new pass_data instance and...
5843 (make_pass_uncprop): ...new function.
5844 * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
5845 global struct to a subclass of gimple_opt_pass along with...
5846 (pass_data_late_warn_uninitialized): ...new pass_data instance and...
5847 (make_pass_late_warn_uninitialized): ...new function.
5848 * tree-ssa.c (pass_init_datastructures): Convert from a global struct
5849 to a subclass of gimple_opt_pass along with...
5850 (pass_data_init_datastructures): ...new pass_data instance and...
5851 (make_pass_init_datastructures): ...new function.
5852 (pass_early_warn_uninitialized): Convert from a global struct to a
5853 subclass of gimple_opt_pass along with...
5854 (pass_data_early_warn_uninitialized): ...new pass_data instance and...
5855 (make_pass_early_warn_uninitialized): ...new function.
5856 (pass_update_address_taken): Convert from a global struct to a
5857 subclass of gimple_opt_pass along with...
5858 (pass_data_update_address_taken): ...new pass_data instance and...
5859 (make_pass_update_address_taken): ...new function.
5860 * tree-ssanames.c (pass_release_ssa_names): Convert from a global
5861 struct to a subclass of gimple_opt_pass along with...
5862 (pass_data_release_ssa_names): ...new pass_data instance and...
5863 (make_pass_release_ssa_names): ...new function.
5864 * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
5865 subclass of gimple_opt_pass along with...
5866 (pass_data_stdarg): ...new pass_data instance and...
5867 (make_pass_stdarg): ...new function.
5868 * tree-switch-conversion.c (pass_convert_switch): Convert from a
5869 global struct to a subclass of gimple_opt_pass along with...
5870 (pass_data_convert_switch): ...new pass_data instance and...
5871 (make_pass_convert_switch): ...new function.
5872 * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
5873 to a subclass of gimple_opt_pass along with...
5874 (pass_data_tail_recursion): ...new pass_data instance and...
5875 (make_pass_tail_recursion): ...new function.
5876 (pass_tail_calls): Convert from a global struct to a subclass of
5877 gimple_opt_pass along with...
5878 (pass_data_tail_calls): ...new pass_data instance and...
5879 (make_pass_tail_calls): ...new function.
5880 * tree-vect-generic.c (pass_lower_vector): Convert from a global
5881 struct to a subclass of gimple_opt_pass along with...
5882 (pass_data_lower_vector): ...new pass_data instance and...
5883 (make_pass_lower_vector): ...new function.
5884 (pass_lower_vector_ssa): Convert from a global struct to a subclass of
5885 gimple_opt_pass along with...
5886 (pass_data_lower_vector_ssa): ...new pass_data instance and...
5887 (make_pass_lower_vector_ssa): ...new function.
5888 * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
5889 to a subclass of gimple_opt_pass along with...
5890 (pass_data_slp_vectorize): ...new pass_data instance and...
5891 (make_pass_slp_vectorize): ...new function.
5892 (pass_ipa_increase_alignment): Convert from a global struct to a
5893 subclass of simple_ipa_opt_pass along with...
5894 (pass_data_ipa_increase_alignment): ...new pass_data instance and...
5895 (make_pass_ipa_increase_alignment): ...new function.
5896 * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
5897 gimple_opt_pass along with...
5898 (pass_data_vrp): ...new pass_data instance and...
5899 (make_pass_vrp): ...new function.
5900 * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
5901 subclass of simple_ipa_opt_pass along with...
5902 (pass_data_ipa_free_lang_data): ...new pass_data instance and...
5903 (make_pass_ipa_free_lang_data): ...new function.
5904 * tsan.c (pass_tsan): Convert from a global struct to a subclass of
5905 gimple_opt_pass along with...
5906 (pass_data_tsan): ...new pass_data instance and...
5907 (make_pass_tsan): ...new function.
5908 (pass_tsan_O0): Convert from a global struct to a subclass of
5909 gimple_opt_pass along with...
5910 (pass_data_tsan_O0): ...new pass_data instance and...
5911 (make_pass_tsan_O0): ...new function.
5912 * var-tracking.c (pass_variable_tracking): Convert from a global
5913 struct to a subclass of rtl_opt_pass along with...
5914 (pass_data_variable_tracking): ...new pass_data instance and...
5915 (make_pass_variable_tracking): ...new function.
5916 * web.c (pass_web): Convert from a global struct to a subclass of
5917 rtl_opt_pass along with...
5918 (pass_data_web): ...new pass_data instance and...
5919 (make_pass_web): ...new function.
5920 * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
5921 declaration with that of...
5922 (make_pass_mode_switch_use): ...new function.
5923 (pass_resolve_sw_modes): Replace declaration with that of...
5924 (make_pass_resolve_sw_modes): ...new function.
5925 * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
5926 from a global struct to a subclass of rtl_opt_pass along with...
5927 (pass_data_mode_switch_use): ...new pass_data instance and...
5928 (make_pass_mode_switch_use): ...new function.
5929 * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
5930 from a global struct to a subclass of rtl_opt_pass along with...
5931 (pass_data_resolve_sw_modes): ...new pass_data instance and...
5932 (make_pass_resolve_sw_modes): ...new function.
5933 * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
5934 struct to a subclass of rtl_opt_pass along with...
5935 (pass_data_insert_vzeroupper): ...new pass_data instance and...
5936 (make_pass_insert_vzeroupper): ...new function.
5937 * config/sparc/sparc.c (pass_work_around_errata): Convert from a
5938 global struct to a subclass of rtl_opt_pass along with...
5939 (pass_data_work_around_errata): ...new pass_data instance and...
5940 (make_pass_work_around_errata): ...new function.
5941 * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
5942 struct to a subclass of rtl_opt_pass along with...
5943 (pass_data_mips_machine_reorg2): ...new pass_data instance and...
5944 (make_pass_mips_machine_reorg2): ...new function.
5946 2013-08-05 David Malcolm <dmalcolm@redhat.com>
5948 * passes.c (pass_manager::operator new): New.
5950 2013-08-05 David Malcolm <dmalcolm@redhat.com>
5952 Handwritten part of conversion of passes to C++ classes.
5954 * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
5955 (toplev.o): Add dep on PASS_MANAGER_H.
5956 * cgraphunit.c (cgraph_process_new_functions): Rework invocation
5957 of early local pases to reflect this moving from a global to a
5958 member of gcc::pass_manager.
5959 (cgraph_add_new_function): Likewise.
5960 * lto-cgraph.c (lto_output_node): Update for conversion of
5961 struct ipa_opt_pass_d to a C++ subclass of opt_pass.
5962 * passes.c (opt_pass::clone): New.
5963 (opt_pass::gate): New.
5964 (opt_pass::execute): New.
5965 (opt_pass::opt_pass): New.
5966 (pass_manager::execute_early_local_passes): New.
5967 (pass_manager::execute_pass_mode_switching): new.
5968 (finish_optimization_passes): Convert to...
5969 (pass_manager::finish_optimization_passes): ...this.
5970 (finish_optimization_passes): Update for conversion of passes to
5972 (register_dump_files_1): Use has_gate since we cannot portably
5973 check a vtable entry against NULL.
5974 (dump_one_pass): Likewise.
5975 (ipa_write_summaries_2): Likewise.
5976 (ipa_write_optimization_summaries_1): Likewise.
5977 (ipa_read_summaries_1): Likewise.
5978 (ipa_read_optimization_summaries_1): Likewise.
5979 (execute_ipa_stmt_fixups): Likewise.
5980 (pass_manager::pass_manager): Rewrite pass-creation, invoking
5981 pass-creation functions rather than wiring up globals, and
5982 storing the results in fields of pass_manager generated using
5984 (pass_manager::dump_profile_report): Update for conversion of
5985 passes to C++ classes.
5986 (pass_manager::execute_ipa_summary_passes): Likewise.
5987 (execute_one_ipa_transform_pass): Likewise.
5988 (execute_one_pass): Use has_gate and has_execute since we cannot
5989 portably check a vtable entry against NULL.
5990 * pass_manager.h (pass_manager::finish_optimization_passes): New.
5991 (pass_manager): Use pass-instances.def to add fields for the
5992 various pass instances.
5993 * toplev.c (finalize): Update for move of
5994 finish_optimization_passes to a method of gcc::pass_manager.
5995 * toplev.h (finish_optimization_passes): Move to method of class
5997 * tree-pass.h (struct pass_data): New.
5998 (opt_pass): Convert to C++ class, make it a subclass of pass_data.
5999 (opt_pass::gate): Convert to virtual function.
6000 (opt_pass::~opt_pass): New.
6001 (opt_pass::clone): New.
6002 (opt_pass::execute): Convert to virtual function.
6003 (opt_pass::opt_pass): New.
6004 (opt_pass::ctxt_): new.
6005 (gimple_opt_pass): Convert to subclass of opt_pass.
6006 (gimple_opt_pass::gimple_opt_pass): New.
6007 (rtl_opt_pass): Convert to subclass of opt_pass.
6008 (rtl_opt_pass::rtl_opt_pass): New.
6009 (ipa_opt_pass_d): Convert to subclass of opt_pass.
6010 (ipa_opt_pass_d::ipa_opt_pass_d): New.
6011 (simple_ipa_opt_pass): Convert to subclass of opt_pass.
6012 (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
6013 * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
6014 invocation of pass_mode_switching to reflect this moving from a
6015 global to a member of gcc::pass_manager.
6016 (ix86_option_override): Rework how pass_insert_vzeroupper is
6017 added to the pass_manager to reflect autogenerated changes.
6018 * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
6020 2013-08-05 Richard Earnshaw <rearnsha@arm.com>
6022 PR rtl-optimization/57708
6023 * recog.c (peep2_find_free_register): Validate all regs in a
6026 2013-08-05 Jan Hubicka <jh@suse.cz>
6029 * cgraph.c (verify_cgraph_node): Accept local flags from other
6031 * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
6032 (function_and_variable_visibility): Likewise.
6033 * trans-mem.c (ipa_tm_create_version): TM versions are not local.
6035 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
6037 * graph.c (init_graph_slim_pretty_print): Remove.
6038 (print_graph_cfg): Do not call it. Use local pretty printer.
6039 (start_graph_dump): Likewise.
6041 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
6043 * gimple-pretty-print.c (buffer): Remove.
6044 (initialized): Likewise.
6045 (maybe_init_pretty_print): Likewise.
6046 (print_gimple_stmt): Do not call it. Use non-static local
6047 pretty_printer variable.
6048 (print_gimple_expr): Likewise.
6049 (print_gimple_seq): Likewise.
6050 (gimple_dump_bb): Likewise.
6052 2013-08-05 Gabriel Dos Reis <gdr@integrable-solutions.net>
6054 * asan.c (asan_pp): Remove.
6055 (asan_pp_initialized): Likewise.
6056 (asan_pp_initialize): Likewise.
6057 (asan_pp_string): Take a pretty_printer parameter. Adjust callers.
6058 (asan_emit_stack_protection): Tidy. Use local pretty printer.
6059 (asan_add_global): Likewise.
6061 2013-08-04 Gabriel Dos Reis <gdr@integrable-solutions.net>
6063 * pretty-print.h (pp_base): Remove. Adjust dependent macros.
6064 * diagnostic.h (diagnostic_flush_buffer): Adjust.
6065 * pretty-print.c (pp_formatted_text_data): Likewise.
6066 (pp_indent): Rename from pp_base_indent.
6067 (pp_format): Rename from pp_base_format.
6068 (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
6069 (pp_format_verbatim): Rename from pp_base_format_verbatim.
6070 (pp_flush): Rename from pp_base_flush.
6071 (pp_set_line_maximum_length): Rename from
6072 pp_base_set_line_maximum_length.
6073 (pp_clear_output_area): Rename from pp_base_clear_output_area.
6074 (pp_set_prefix): Rename from pp_base_set_prefix.
6075 (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
6076 (pp_emit_prefix): Rename from pp_base_emit_prefix.
6077 (pp_append_text): Rename from pp_base_append_text.
6078 (pp_formatted_text): Rename from pp_base_formatted_text.
6079 (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
6080 (pp_remaining_character_count_for_line): Rename from
6081 pp_base_remaining_character_count_for_line.
6082 (pp_newline): Rename from pp_base_newline.
6083 (pp_character): Rename from pp_base_character.
6084 (pp_string): Rename from pp_base_string.
6085 (pp_maybe_space): Rename from pp_base_maybe_space.
6086 * asan.c (asan_pp_string): Adjust.
6087 (asan_emit_stack_protection): Likewise.
6088 (asan_add_global): Likewise.
6089 * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
6090 * tree-mudflap.c (mf_varname_tree): Likewise.
6091 * tree-pretty-print.c (pp_tree_identifier): Rename from
6092 pp_base_tree_identifier.
6093 * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
6094 Declare as function.
6096 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
6098 * pretty-print.h (pp_bar_bar): New.
6099 (pp_ampersand_ampersand): Likewise.
6100 (pp_less_equal): Likewise.
6101 (pp_greater_equal): Likewise.
6102 * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
6103 printer functions instead of pp_string or operators and punctuators.
6104 (dump_gimple_call): Likewise.
6105 (dump_gimple_omp_for): Likewise.
6106 (dump_gimple_transaction): Likewise.
6107 (dump_gimple_phi): Likewise.
6108 (pp_gimple_stmt_1): Likewise.
6109 * sched-vis.c (print_insn): Likewise.
6110 * tree-mudflap.c (mf_varname_tree): Likewise.
6111 * tree-pretty-print.c (dump_block_node): Likewise.
6112 (dump_generic_node): Likewise.
6114 2013-08-02 Jan Hubicka <jh@suse.cz>
6116 * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
6118 * lto-streamer-out.c (tree_is_indexable): Results decls and
6119 parm decls are not indexable.
6120 (DFS_write_tree_body): Do not follow args and results.
6121 (hash_tree): Likewise.
6122 (output_functions): Rearrange so struct function is needed
6123 only when real body is output; be able to also ouptut abstract
6124 functions; output DECL_ARGUMENTS and DECL_RESULT.
6125 (lto_output): When not in WPA, ale store abstract functions.
6126 (write_symbol): Do not care about RESULT_DECL.
6127 (output_symbol_p): Handle correctly sbtract decls.
6128 * lto-streamer-in.c (input_function): Rearrange so struct
6129 function can be NULL at entry; allow streaming of
6130 functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
6131 * ipa.c (symtab_remove_unreachable_nodes): Silence confused
6132 sanity check during LTO.
6133 * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
6134 RESULT_DECl and DECL_ARGUMENTS.
6135 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6138 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
6140 * pretty-print.h (pp_underscore): New.
6142 * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
6143 printer functions instead of pp_character.
6144 (dump_binary_rhs): Likewise.
6145 (dump_ternary_rhs): Likewise.
6146 (dump_gimple_call_args): Likewise.
6147 (pp_points_to_solution): Likewise.
6148 (dump_gimple_call): Likewise.
6149 (dump_gimple_switch): Likewise.
6150 (dump_gimple_cond): Likewise.
6151 (dump_gimple_bind): Likewise.
6152 (dump_gimple_try): Likewise.
6153 (dump_gimple_omp_for): Likewise.
6154 (dump_gimple_omp_continue): Likewise.
6155 (dump_gimple_omp_single): Likewise.
6156 (dump_gimple_omp_sections): Likewise.
6157 (dump_gimple_omp_block): Likewise.
6158 (dump_gimple_omp_critical): Likewise.
6159 (dump_gimple_transaction): Likewise.
6160 (dump_gimple_asm): Likewise.
6161 (dump_gimple_phi): Likewise.
6162 (dump_gimple_omp_parallel): Likewise.
6163 (dump_gimple_omp_task): Likewise.
6164 (dump_gimple_omp_atomic_load): Likewise.
6165 (dump_gimple_omp_atomic_store): Likewise.
6166 (dump_gimple_mem_ops): Likewise.
6167 (pp_gimple_stmt_1): Likewise.
6168 (pp_cfg_jump): Likewise.
6169 (dump_implicit_edges): Likewise.
6170 (gimple_dump_bb_for_graph): Likewise.
6171 * graph.c (draw_cfg_node): Likewise.
6172 * langhooks.c (lhd_print_error_function): Likewise.
6173 * sched-vis.c (print_exp): Likewise.
6174 (print_value): Likewise.
6175 (print_pattern): Likewise.
6176 (print_insn): Likewise.
6177 (rtl_dump_bb_for_graph): Likewise.
6178 * tree-pretty-print.c (dump_function_declaration): Likewise.
6179 (dump_array_domain): Likewise.
6180 (dump_omp_clause): Likewise.
6181 (dump_location): Likewise.
6182 (dump_generic_node): Likewise.
6183 (print_struct_decl): Likewise.
6184 * diagnostic.c (diagnostic_show_locus): Use pp_space.
6186 2013-08-03 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
6188 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
6189 candidate table when replacing a candidate statement.
6190 (replace_rhs_if_not_dup): Likewise.
6191 (replace_one_candidate): Likewise.
6193 2013-08-02 Jan Hubicka <jh@suse.cz>
6194 Martin Liska <marxin.liska@gmail.com>
6196 * cgraphunit.c (add_new_function): Fix logic when adding from
6198 (assemble_thunk): Rename to ...
6199 (expand_thunk); .. this one; export; get it working with
6200 general functions; make produced gimple valid.
6201 * cgraph.h (expand_thunk): Declare.
6203 2013-08-02 Jan Hubicka <jh@suse.cz>
6205 * ipa-cp.c (gather_context_independent_values): Use
6206 ipa_get_param_move_cost.
6207 (get_replacement_map): Remove PARAM; move parameter folding
6209 (create_specialized_node): Update.
6210 * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
6211 assert that we have gimple body; update move_cost.
6212 (count_formal_params): Assert that we have gimple body.
6213 (ipa_dump_param): New function.
6214 (ipa_alloc_node_params): Break out from ...
6215 (ipa_initialize_node_params): ... here.
6216 (ipa_get_vector_of_formal_parms): ICE when used in WPA.
6217 (ipa_write_node_info): Stream move costs.
6218 (ipa_read_node_info): Read move costs.
6219 (ipa_update_after_lto_read): Do not recompute node params.
6220 * ipa-prop.h (ipa_param_descriptor): Add move_cost.
6221 (ipa_get_param): Check we are not in WPA.
6222 (ipa_get_param_move_cost): New.
6223 * tree-inline.c (tree_function_versioning): Fold replacement as needed.
6224 * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
6225 parm numbers to be present.
6227 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
6229 PR rtl-optimization/58048
6230 * lra-constraints.c (process_alt_operands): Don't check asm
6231 operand on register.
6233 2013-08-02 Eric Botcazou <ebotcazou@adacore.com>
6235 * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
6236 the implied StoreLoad barrier for atomic operations if before.
6238 2013-08-02 Jan Hubicka <jh@suse.cz>
6239 Martin Liska <marxin.liska@gmail.com>
6241 * cgraph.c (cgraph_function_body_availability): Do not check
6243 * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
6244 symtab_node_availability): Declare.
6245 * ipa.c (can_replace_by_local_alias): New.
6246 (function_and_variable_visibility): Use it.
6247 * symtab.c (symtab_for_node_and_aliases,
6248 symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
6250 2013-08-02 Vladimir Makarov <vmakarov@redhat.com>
6252 PR rtl-optimization/57963
6253 * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
6254 (lra_constraints): Use them.
6256 2013-08-02 Sofiane Naci <sofiane.naci@arm.com>
6258 * config/arm/types.md (define_attr "type"): Add "load_acq"
6260 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
6262 (cortex_a53_store1): Likewise.
6264 2013-08-01 Jan Hubicka <jh@suse.cz>
6266 * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
6267 partitions are not needed.
6269 2013-08-01 Uros Bizjak <ubizjak@gmail.com>
6271 * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
6272 * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
6273 MAYBE_NON_Q_CLASS_P where appropriate.
6275 2013-08-01 Jan Hubicka <jh@suse.cz>
6277 * cgraph.h (release_function_body): Declare.
6278 * tree.c (free_lang_data_in_decl): Free, parameters and return values
6279 of unused delcarations.
6281 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6283 * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
6284 RTL form when subtracting a constant.
6286 2013-08-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6288 * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
6289 Generate canonical plus rtx with negated immediate instead of minus
6291 * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
6293 2013-08-01 Jan Hubicka <jh@suse.cz>
6295 * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
6296 (cgraph_release_function_body): Likewise.
6297 (cgraph_can_remove_if_no_direct_calls_p): Likewise.
6298 * cgraph.h (cgrpah_node): Rename abstract_and_needed
6299 to used_as_abstract_origin.
6300 * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
6301 symbols used as abstract origins.
6302 * cgraphunit.c (analyze_functions): Update.
6303 * ipa.c (symtab_remove_unreachable_nodes): Recompute
6304 used_as_abstract_origin.
6305 * tree-inline.c (tree_function_versioning): Update
6306 used_as_abstract_origin; be ready for DECL_RESULT and
6307 DECL_ARGUMENTS to be NULL.
6309 * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
6310 for abstract functions.
6311 * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
6314 2013-08-01 Jan Hubicka <jh@suse.cz>
6316 * profile.c (compute_value_histograms): Fix thinko.
6318 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
6320 * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs. Add
6321 aarch-common-protos.h to extra_headers.
6322 (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6323 * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
6324 * config/aarch64/t-aarch64 (aarch-common.o): Define.
6326 2013-08-01 Sofiane Naci <sofiane.naci@arm.com>
6328 * config/aarch64/aarch64.md (define_attr "type"): Delete.
6329 Include "../arm/types.md". Define "type" attribute for all patterns.
6330 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
6333 2013-07-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6335 * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
6336 to support power8 load fusion.
6337 (fusion_gpr_mem_load): Likewise.
6339 * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
6341 * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
6342 declarations for power8 load fusion.
6343 (emit_fusion_gpr_load): Likewise.
6345 * config/rs6000/rs6000.c (rs6000_option_override_internal): If
6346 tuning for power8, turn on fusion mode by default. Turn on sign
6347 extending fusion mode if normal fusion mode is on, and we are at
6349 (fusion_gpr_load_p): New function, return true if we can fuse an
6350 addis instruction with a dependent load to a GPR.
6351 (emit_fusion_gpr_load): Emit the instructions for power8 load
6354 * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
6355 (VSX load fusion peepholes): New peepholes to fuse together an
6356 addi instruction with a VSX load instruction.
6358 * config/rs6000/rs6000.md (GPR load fusion peepholes): New
6359 peepholes to fuse an addis instruction with a load to a GPR base
6360 register. If we are supporting sign extending fusions, convert
6361 sign extending loads to zero extending loads and add an explicit
6364 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
6366 * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs. Add
6367 aarch-common-protos.h to extra_headers.
6368 (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
6369 * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
6370 (arm_early_store_addr_dep): Likewise.
6371 (arm_no_early_alu_shift_dep): Likewise.
6372 (arm_no_early_alu_shift_value_dep): Likewise.
6373 (arm_no_early_mul_dep): Likewise.
6374 (arm_no_early_store_addr_dep): Likewise.
6375 (arm_mac_accumulator_is_mul_result): Likewise.
6376 (arm_mac_accumulator_is_result): Likewise.
6377 * config/arm/aarch-common.c: ... here. New file.
6378 * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
6380 (arm_early_store_addr_dep): Likewise.
6381 (arm_no_early_alu_shift_dep): Likewise.
6382 (arm_no_early_alu_shift_value_dep): Likewise.
6383 (arm_no_early_mul_dep): Likewise.
6384 (arm_no_early_store_addr_dep): Likewise.
6385 (arm_mac_accumulator_is_mul_result): Likewise.
6386 (arm_mac_accumulator_is_result): Likewise.
6387 * config/arm/aarch-common-protos.h: ... here. New file.
6388 * config/arm/t-arm (aarch-common.o): Define.
6390 2013-07-31 Sofiane Naci <sofiane.naci@arm.com>
6392 * config/arm/arm.md: Include new file "types.md".
6393 (define_attr "type"): Move from here to ...
6394 (define_attr "mul32"): Likewise.
6395 (define_attr "mul64"): Likewise.
6396 * config/arm/types.md: ... here. New file.
6398 2013-07-31 Sebastian Huber <sebastian.huber@embedded-brains.de>
6400 * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
6401 * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
6403 2013-07-31 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6405 * gen-pass-instances.awk: Fix offset of substr().
6407 2013-07-31 David Malcolm <dmalcolm@redhat.com>
6409 * Makefile.in (pass-instances.def): New.
6410 (passes.o): Replace dependency on passes.def with one on
6413 * gen-pass-instances.awk: New.
6415 * passes.c (pass_manager::pass_manager): Use pass-instances.def
6416 rather than passes.def, updating local definition of NEXT_PASS
6417 macro to add an extra NUM parameter (currently unused).
6419 2013-07-30 David Malcolm <dmalcolm@redhat.com>
6421 * Makefile.in (PASS_MANAGER_H): New.
6422 (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
6423 (passes.o): Likewise.
6424 (statistics.o): Likewise.
6425 (cgraphunit.o): Likewise.
6426 (context.o): Depend on PASS_MANAGER_H.
6428 * pass_manager.h: New.
6430 * cgraphunit.c (cgraph_add_new_function): Update for moves
6431 of globals to fields of pass_manager.
6432 (analyze_function): Likewise.
6433 (expand_function): Likewise.
6434 (ipa_passes): Likewise.
6435 (compile): Likewise.
6437 * context.c (context::context): New.
6438 * context.h (context::context): New.
6439 (context::get_passes): New.
6440 (context::passes_): New.
6442 * lto-cgraph.c (input_node): Update for moves of globals to
6443 fields of pass_manager.
6445 * passes.c (all_passes): Remove, in favor of a field of the
6446 same name within the new class pass_manager.
6447 (all_small_ipa_passes): Likewise.
6448 (all_lowering_passes): Likewise.
6449 (all_regular_ipa_passes): Likewise.
6450 (all_late_ipa_passes): Likewise.
6451 (all_lto_gen_passes): Likewise.
6452 (passes_by_id): Likewise.
6453 (passes_by_id_size): Likewise.
6454 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6455 the new class pass_manager.
6456 (set_pass_for_id): Convert to...
6457 (pass_manager::set_pass_for_id): ...method.
6458 (get_pass_for_id): Convert to...
6459 (pass_manager::get_pass_for_id): ...method.
6460 (register_one_dump_file): Move body of implementation into...
6461 (pass_manager::register_one_dump_file): ...here.
6462 (register_dump_files_1): Convert to...
6463 (pass_manager::register_dump_files_1): ...method.
6464 (register_dump_files): Convert to...
6465 (pass_manager::register_dump_files): ...method.
6466 (create_pass_tab): Update for moves of globals to fields of
6468 (dump_passes): Move body of implementation into...
6469 (pass_manager::dump_passes): ...here.
6470 (register_pass): Move body of implementation into...
6471 (pass_manager::register_pass): ...here.
6472 (init_optimization_passes): Convert into...
6473 (pass_manager::pass_manager): ...constructor for new
6474 pass_manager class, and initialize the pass_lists array.
6475 (check_profile_consistency): Update for moves of globals to
6476 fields of pass_manager.
6477 (dump_profile_report): Move body of implementation into...
6478 (pass_manager::dump_profile_report): ...here.
6479 (ipa_write_summaries_1): Update for moves of pass lists from
6480 being globals to fields of pass_manager.
6481 (ipa_write_optimization_summaries): Likewise.
6482 (ipa_read_summaries): Likewise.
6483 (ipa_read_optimization_summaries): Likewise.
6484 (execute_all_ipa_stmt_fixups): Likewise.
6486 * statistics.c (statistics_fini): Update for moves of globals to
6487 fields of pass_manager.
6489 * toplev.c (general_init): Replace call to
6490 init_optimization_passes with construction of the pass_manager
6493 * tree-pass.h (all_passes): Remove, in favor of a field of the
6494 same name within the new class pass_manager.
6495 (all_small_ipa_passes): Likewise.
6496 (all_lowering_passes): Likewise.
6497 (all_regular_ipa_passes): Likewise.
6498 (all_lto_gen_passes): Likewise.
6499 (all_late_ipa_passes): Likewise.
6500 (passes_by_id): Likewise.
6501 (passes_by_id_size): Likewise.
6502 (gcc_pass_lists): Remove, in favor of "pass_lists" field within
6503 the new class pass_manager.
6504 (get_pass_for_id): Remove.
6506 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
6508 * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
6511 2013-07-30 Richard Earnshaw <rearnsha@arm.com>
6513 * arm.md (mulhi3): New expand pattern.
6515 2013-07-30 Jan Hubicka <jh@suse.cz>
6516 Martin Liska <marxin.liska@gmail.com>
6518 * profile.c (compute_value_histograms): Do not ICE when
6519 there is mismatch only on some counters.
6521 2013-07-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
6523 PR rtl-optimization/57637
6524 * function.c (move_insn_for_shrink_wrap): Also check the
6525 GEN set of the LIVE problem for the liveness analysis
6526 if it exists, otherwise give up.
6528 2013-07-29 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
6530 PR tree-optimization/57993
6531 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
6532 replaced statement in the candidate table.
6533 (phi_add_costs): Return infinite cost when the hidden basis does
6534 not dominate all phis on which the candidate is dependent.
6535 (replace_one_candidate): Record replaced statement in the
6538 2013-07-29 Joern Rennecke <joern.rennecke@embecosm.com>
6540 * config/epiphany/epiphany.md (*isub_i+2): New peephole.
6541 (ashlv2si3): New expander.
6542 (*ashlv2si3_i): New define_insn_and_split.
6543 * predicates.md (float_operation): Allow patterns with three
6546 PR rtl-optimization/58021
6547 * mode-switching.c (create_pre_exit): Always split off preceding
6548 insns if we are not at the basic block head.
6550 2013-07-29 Maciej W. Rozycki <macro@codesourcery.com>
6552 * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
6553 (UCLIBC_DYNAMIC_LINKER): New macro.
6554 * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
6556 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
6557 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
6559 (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
6560 (UCLIBC_DYNAMIC_LINKERN32): Likewise.
6561 * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
6562 for SF and DF modes. Use ieee_quad_format for TF mode.
6563 * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
6564 * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
6565 (mips_option_override): Handle `-mnan=legacy'.
6566 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
6567 `-mabs=2008' and `-mnan=2008'.
6568 (OPTION_DEFAULT_SPECS): Add "nan" default.
6569 (ASM_SPEC): Handle `-mnan='.
6570 [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
6571 * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
6572 comment accordingly.
6573 (neg<mode>2): Likewise.
6574 * config/mips/mips.opt (mabs, mnan): New options.
6575 * doc/install.texi (Configuration): Document `--with-nan=' option.
6576 * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
6578 (MIPS Options): Document them.
6579 * config.gcc <mips*-*-*>: Handle `--with-nan='.
6580 * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
6581 * configure: Regenerate.
6582 * config.in: Regenerate.
6584 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
6586 * config/i386/i386.md (float post-reload splitters): Do not check
6587 for subregs of SSE registers.
6589 2013-07-29 Uros Bizjak <ubizjak@gmail.com>
6590 H.J. Lu <hongjiu.lu@intel.com>
6594 * config/i386/i386.md (post-reload splitter
6595 to avoid partial SSE reg dependency stalls): New pattern.
6597 2013-07-29 Dominik Vogt <vogt@linux.vnet.ibm.com>
6599 * config/s390/s390.md ("movcc"): Swap load and store instructions.
6601 2013-07-27 Joern Rennecke <joern.rennecke@embecosm.com>
6603 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
6604 Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
6606 2013-07-26 Cary Coutant <ccoutant@google.com>
6608 * dwarf2out.c (die_checksum_ordered): Don't include template
6609 instantiations in signature.
6610 (is_template_parameter): New function.
6611 (is_template_instantiation): New function.
6612 (generate_skeleton_bottom_up): Don't include template instantiations
6614 (generate_skeleton): Likewise.
6615 (break_out_comdat_types): Move recursive call to break out nested
6617 (prune_unused_types_mark_generic_parms_dies): Call
6618 is_template_parameter.
6620 2013-07-26 Ian Bolton <ian.bolton@arm.com>
6622 * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
6623 uses vector registers.
6624 * config/aarch64/iterators.md: Add attributes rtn and vas.
6626 2013-07-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6627 Richard Earnshaw <richard.earnshaw@arm.com>
6629 * combine.c (simplify_comparison): Re-canonicalize operands
6631 * config/arm/arm.md (movcond_addsi): New splitter.
6633 2013-07-25 Sterling Augustine <saugustine@google.com>
6635 * dwarf2out.c (size_of_pubnames): Move code to...
6636 (include_pubname_in_output): ...here. New.
6637 (want_pubnames): Rearrange.
6638 (output_pubnames): Call include_pubname_in_output. Move assertion.
6640 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
6642 * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
6644 2013-07-25 Cameron McInally <cameron.mcinally@nyu.edu>
6647 * doc/extend.texi: Remove obsolete builtins. Fix
6648 typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
6650 2013-07-25 Jan Hubicka <jh@suse.cz>
6652 * cgraph.c (release_function_body): Break out from ...
6653 (cgraph_release_function_body): ... this one; also release DECL_RESULT
6655 * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
6656 old_tree in the map.
6657 (create_specialized_node): Update.
6658 * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
6660 * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
6661 DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
6662 * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
6663 * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
6666 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6668 * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
6669 addsi3_carryin_alt2_<optab>): Correct output template.
6671 2013-07-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6673 * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
6674 Adjust for arm_restrict_it.
6675 Remove trailing whitespace.
6677 2013-07-25  Mark Kettenis  <kettenis@openbsd.org>
6679 * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
6680 libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
6682 * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
6684 2013-07-25 Vladimir Makarov <vmakarov@redhat.com>
6686 PR rtl-optimization/57960
6687 * lra-constraints.c (process_alt_operands): Use the right mode
6688 when checking strict_low.
6690 2013-07-25 Jan Hubicka <jh@suse.cz>
6692 * lto-symtab.c (lto_cgraph_replace_node): Release function body.
6693 * cgraph.c (cgraph_remove_node): Do not release function body
6694 when in cgraph streaming.
6695 * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
6696 in other partitions are not considered reachable; fix handling of
6699 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6701 * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
6703 2013-07-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6708 * config/arm/arm.md ("*sibcall_insn): Replace use of
6709 Ss with US. Adjust output for v5 and v4t.
6710 (*sibcall_value_insn): Likewise and loosen predicate on operand0.
6712 * config/arm/constraints.md ("Ss"): Rename to US.
6714 2013-07-25 Terry Guo <terry.guo@arm.com>
6716 * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
6717 shift_add/shift_sub0/shift_sub1 RTXs.
6719 2013-07-24 Bill Schmidt <wschmidt@linux.ibm.com>
6720 Anton Blanchard <anton@au1.ibm.com>
6722 * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
6723 (altivec_vpks<VI_char>ss): Likewise.
6724 (altivec_vpks<VI_char>us): Likewise.
6725 (altivec_vpku<VI_char>us): Likewise.
6726 (altivec_vpku<VI_char>um): Likewise.
6728 2013-07-24 David Malcolm <dmalcolm@redhat.com>
6730 Introduce context class.
6732 * Makefile.in (CONTEXT_H): New.
6733 (OBJS): Add context.o.
6734 (toplev.o): Add CONTEXT_H to dependencies.
6737 * toplev.c (general_init): Create the singleton gcc::context instance.
6743 2013-07-24 Joern Rennecke <joern.rennecke@embecosm.com>
6745 PR rtl-optimization/57968
6746 * mode-switching.c (create_pre_exit): Allow instructions that
6747 don't set a return register to need a non-exit mode.
6749 2013-07-24 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
6750 Anton Blanchard <anton@au1.ibm.com>
6752 * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
6753 operands to vperm for little endian.
6754 * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
6755 of lvsl to create the control mask for a vperm for little endian.
6757 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6758 Anton Blanchard <anton@au1.ibm.com>
6760 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
6761 two operands for little-endian.
6763 2013-07-23 Steve Ellcey <sellcey@mips.com>
6765 * config/mips/mips.c (mips_case_values_threshold): New.
6766 (TARGET_CASE_VALUES_THRESHOLD): Define.
6768 2013-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6769 Anton Blanchard <anton@au1.ibm.com>
6771 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
6772 selection of field for vector splat in little endian mode.
6774 2013-07-23 Michael Meissner <meissner@linux.vnet.ibm.com>
6776 * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
6777 expanders to rs6000.md.
6778 (ior<mode>3): Likewise.
6779 (and<mode>3): Likewise.
6780 (one_cmpl<mode>2): Likewise.
6781 (nor<mode>3): Likewise.
6782 (andc<mode>3): Likewise.
6783 (eqv<mode>3): Likewise.
6784 (nand<mode>3): Likewise.
6785 (orc<mode>3): Likewise.
6787 * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
6790 * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
6791 to split multi-word logical operations.
6792 (rs6000_split_logical_di): Likewise.
6793 (rs6000_split_logical): Likewise.
6795 * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
6796 (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
6797 and allow TImode operations in 32-bit.
6798 (vsx_and<mode>3_64bit): Likewise.
6799 (vsx_ior<mode>3_32bit): Likewise.
6800 (vsx_ior<mode>3_64bit): Likewise.
6801 (vsx_xor<mode>3_32bit): Likewise.
6802 (vsx_xor<mode>3_64bit): Likewise.
6803 (vsx_one_cmpl<mode>2_32bit): Likewise.
6804 (vsx_one_cmpl<mode>2_64bit): Likewise.
6805 (vsx_nor<mode>3_32bit): Likewise.
6806 (vsx_nor<mode>3_64bit): Likewise.
6807 (vsx_andc<mode>3_32bit): Likewise.
6808 (vsx_andc<mode>3_64bit): Likewise.
6809 (vsx_eqv<mode>3_32bit): Likewise.
6810 (vsx_eqv<mode>3_64bit): Likewise.
6811 (vsx_nand<mode>3_32bit): Likewise.
6812 (vsx_nand<mode>3_64bit): Likewise.
6813 (vsx_orc<mode>3_32bit): Likewise.
6814 (vsx_orc<mode>3_64bit): Likewise.
6816 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
6817 logical types in GPRs.
6819 * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
6820 logical insns to rs6000.md, and allow TImode operations in
6822 (altivec_ior<mode>3): Likewise.
6823 (altivec_xor<mode>3): Likewise.
6824 (altivec_one_cmpl<mode>2): Likewise.
6825 (altivec_nor<mode>3): Likewise.
6826 (altivec_andc<mode>3): Likewise.
6828 * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
6829 attributes for moving the 128-bit logical operations into
6831 (BOOL_REGS_OUTPUT): Likewise.
6832 (BOOL_REGS_OP1): Likewise.
6833 (BOOL_REGS_OP2): Likewise.
6834 (BOOL_REGS_UNARY): Likewise.
6835 (BOOL_REGS_AND_CR0): Likewise.
6836 (one_cmpl<mode>2): Add support for DI logical operations on
6837 32-bit, splitting the operations to 32-bit.
6841 (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
6842 changes to combine the 32/64-bit code, allow logical operations on
6843 TI mode in 32-bit, and to use similar match_operator patterns like
6844 scalar mode uses. Combine the Altivec and VSX code for logical
6845 operations, and move it here.
6846 (ior<mode>3, 128-bit types): Likewise.
6847 (xor<mode>3, 128-bit types): Likewise.
6848 (one_cmpl<mode>3, 128-bit types): Likewise.
6849 (nor<mode>3, 128-bit types): Likewise.
6850 (andc<mode>3, 128-bit types): Likewise.
6851 (eqv<mode>3, 128-bit types): Likewise.
6852 (nand<mode>3, 128-bit types): Likewise.
6853 (orc<mode>3, 128-bit types): Likewise.
6854 (and<mode>3_internal): Likewise.
6855 (bool<mode>3_internal): Likewise.
6856 (boolc<mode>3_internal1): Likewise.
6857 (boolc<mode>3_internal2): Likewise.
6858 (boolcc<mode>3_internal1): Likewise.
6859 (boolcc<mode>3_internal2): Likewise.
6860 (eqv<mode>3_internal1): Likewise.
6861 (eqv<mode>3_internal2): Likewise.
6862 (one_cmpl1<mode>3_internal): Likewise.
6864 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
6866 * config/microblaze/microblaze.c (microblaze_expand_prologue):
6867 Rename flag_stack_usage to flag_stack_usage_info.
6869 2013-07-23 David Holsgrove <david.holsgrove@xilinx.com>
6871 * config/microblaze/sync.md: New file.
6872 * config/microblaze/microblaze.md: Include sync.md
6873 * config/microblaze/microblaze.c: Add print_operand 'y'.
6874 * config/microblaze/constraints.md: Add memory_contraint
6875 'Q' which is a single register.
6877 2013-07-23 Eric Botcazou <ebotcazou@adacore.com>
6879 * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
6881 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
6883 * reload.c (find_reloads): Exit loop once we find this operand
6884 cannot be reloaded somehow for this alternative.
6886 * reload.c (find_reloads): Exit loop once we find a hard register.
6888 * rtlanal.c (computed_jump_p): Exit loop once we find label
6891 * i386.c (ix86_pad_returns): Exit loop after setting replace.
6893 * cfgloopmanip.c (remove_path): Exit loop after setting
6896 * gensupport.c (subst_dup): Avoid loop if code is not
6897 MATCH_DUP nor MATCH_OP_DUP.
6899 2013-07-23 Nicklas Bo Jensen <nbjensen@gmail.com>
6901 * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
6903 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
6905 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
6906 true for SP_REGNUM if mode == ptr_mode.
6907 * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
6908 with value R0_REGNUM + 31.
6910 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
6912 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
6913 pad pointer-typed argument downward.
6915 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
6917 * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
6918 and __ILP32__ when the ILP32 model is in use.
6920 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
6922 * config/aarch64/aarch64.c (POINTER_BYTES): New define.
6923 (aarch64_load_symref_appropriately): In the case of
6924 SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
6925 to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
6926 (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
6927 change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
6928 if 'mode' doesn't equal to 'ptr_mode'.
6929 (aarch64_output_mi_thunk): Add an assertion on the alignment of
6930 'vcall_offset'; change to call aarch64_emit_move differently depending
6931 on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
6932 to calculate the upper bound of 'vcall_offset'.
6933 (aarch64_cannot_force_const_mem): Change to also return true if
6935 (aarch64_legitimize_reload_address): In the case of large
6936 displacements, add new local variable 'xmode' and an assertion
6937 based on it; change to use 'xmode' to generate the new rtx and
6939 (aarch64_asm_trampoline_template): Change to generate the template
6940 differently depending on TARGET_ILP32 or not; change to use
6941 'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
6942 (aarch64_trampoline_size): Removed.
6943 (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
6944 and replace immediate literals with it. Change to use 'ptr_mode'
6945 instead of 'DImode' and call convert_memory_address if the mode
6946 of 'fnaddr' doesn't equal to 'ptr_mode'.
6947 (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
6949 (aarch64_elf_asm_destructor): Likewise.
6950 * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
6951 on TARGET_ILP32 instead of aarch64_trampoline_size.
6952 * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
6953 of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
6954 (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
6955 (loadwb_pair<GPI:mode>_<P:mode>): ... this. Replace PTR with P.
6956 (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
6957 (storewb_pair<GPI:mode>_<P:mode>): ... this.
6958 (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
6959 depending on the value of 'mode'.
6960 (add_losym_<mode>): New.
6961 (ldr_got_small_<mode>): New, based on ldr_got_small.
6962 (ldr_got_small): Remove.
6963 (ldr_got_small_sidi): New.
6964 * config/aarch64/iterators.md (P): New.
6965 (PTR): Change to 'ptr_mode' in the condition.
6967 2013-07-23 Yufeng Zhang <yufeng.zhang@arm.com>
6969 * config.gcc (aarch64*-*-*): Support --with-abi.
6970 (aarch64*-*-elf): Support --with-multilib-list.
6971 (aarch64*-*-linux*): Likewise.
6972 (supported_defaults): Add abi to aarch64*-*-*.
6973 * configure.ac: Mention AArch64 for --with-multilib-list.
6974 * configure: Re-generated.
6975 * config/aarch64/biarchilp32.h: New file.
6976 * config/aarch64/biarchlp64.h: New file.
6977 * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
6979 (MULTILIB_DEFAULTS): Ditto.
6980 (DRIVER_SELF_SPECS): Ditto.
6981 (ASM_SPEC): Update to also substitute -mabi.
6982 * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
6983 file whose name depends on -mabi= and -mbig-endian.
6984 * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
6986 (POINTER_SIZE): New define.
6987 (POINTERS_EXTEND_UNSIGNED): Ditto.
6988 (enum aarch64_abi_type): New enumeration tag.
6989 (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
6990 (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
6991 (TARGET_ILP32): New define.
6992 * config/aarch64/aarch64.opt (mabi): New.
6994 (ilp32, lp64): New values for -mabi.
6995 * config/aarch64/t-aarch64 (comma): New define.
6996 (MULTILIB_OPTIONS): Ditto.
6997 (MULTILIB_DIRNAMES): Ditto.
6998 * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
6999 * doc/invoke.texi: Document -mabi for AArch64.
7001 2013-07-23 Georg-Johann Lay <avr@gjlay.de>
7003 * config/avr/avr.md: Explain asm print modifier 'r' for REG.
7005 2013-07-22 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
7006 Anton Blanchard <anton@au1.ibm.com>
7008 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
7009 endianness when selecting field to splat.
7011 2013-07-22 Eric Christopher <echristo@gmail.com>
7013 * dwarf2out.c (die_odr_checksum): New function to use
7014 CHECKSUM_ macros and ULEB128 for DIE tag.
7015 (generate_type_signature): Use.
7017 2013-07-22 Eric Botcazou <ebotcazou@adacore.com>
7019 * config.gcc (sparc*-*-*): Accept leon3 processor.
7020 (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
7021 * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
7022 * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
7023 * config/sparc/sparc.opt (enum processor_type): Add leon3.
7024 (mfix-ut699): Adjust comment.
7025 * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
7026 (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
7027 (CPP_CPU_SPEC): Likewise.
7028 (ASM_CPU_SPEC): Likewise.
7029 * config/sparc/sparc.c (leon3_cost): New constant.
7030 (sparc_option_override): Add leon3 support.
7031 (mem_ref): New function.
7032 (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
7033 (sparc_do_work_around_errata): Look into the instruction in the delay
7034 slot and adjust accordingly. Add fix for the data cache nullify issues
7035 of the UT699. Change insertion position for the NOP.
7036 * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
7037 (leon3_load): New reservation.
7038 (leon_store): Bump latency to 2.
7039 (grfpu): New automaton.
7040 (grfpu_alu): New unit.
7041 (grfpu_ds): Likewise.
7042 (leon_fp_alu): Adjust.
7043 (leon_fp_mult): Delete.
7044 (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
7045 (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
7046 * config/sparc/sparc.md (cpu): Add leon3.
7047 * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
7049 (atomic_test_and_set): Likewise.
7052 2013-07-22 Jürgen Urban <JuergenUrban@gmx.de>
7054 * config.gcc (mips*-*-*): Add --with-fpu support. Make single the
7055 default for R5900 targets.
7056 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
7057 (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
7058 * config/mips/mips.c (mips_option_override): Report an error for
7059 -march=r5900 -mhard-float -mdouble-float. Use spu_single_format
7060 for -march=r5900 -mhard-float.
7062 2013-07-22 Po-Chun Chang <pchang9@cs.wisc.edu>
7064 * df-problems.c (can_move_insns_across): Exit loop once we
7065 find a non-fixed, non-global register.
7067 * ipa-pure-const.c (propagate_nothrow): Exit loop after
7070 * omega.c (omega_eliminate_red): Break after setting red_found.
7071 (omega_problem_has_red_equations): Similarly after setting found.
7072 (omega_query_variable): Similarly after setting coupled.
7074 2013-07-22 Marek Polacek <polacek@redhat.com>
7076 * gimplify.c: Don't include gimple.h twice.
7078 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7080 * config/arm/constraints.md (Pd): Allow TARGET_THUMB
7081 instead of TARGET_THUMB1.
7082 (Pz): New constraint.
7083 * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
7085 (compare_negsi_si): Likewise.
7086 (compare_addsi2_op0): Likewise.
7087 (compare_addsi2_op1): Likewise.
7088 (addsi3_carryin_<optab>): Likewise.
7089 (addsi3_carryin_alt2_<optab>): Likewise.
7090 (addsi3_carryin_shift_<optab>): Disable cond_exec variant
7091 for arm_restrict_it.
7092 (subsi3_carryin): Likewise.
7093 (arm_subsi3_insn): Add alternatives for 16-bit encoding.
7094 (minmax_arithsi): Disable for arm_restrict_it.
7095 (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
7096 (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
7097 (satsi_<SAT:code>_shift): Likewise.
7098 (arm_shiftsi3): Add alternative for 16-bit encoding.
7099 (arm32_movhf): Disable for arm_restrict_it.
7100 (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
7101 (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
7103 2013-07-22 Sofiane Naci <sofiane.naci@arm.com>
7105 * config/arm/arm.md (attribute "insn"): Delete.
7106 (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
7107 "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
7108 (not_shiftsi): Update for attribute change.
7109 (not_shiftsi_compare0): Likewise.
7110 (not_shiftsi_compare0_scratch): Likewise.
7111 (arm_one_cmplsi2): Likewise.
7112 (thumb1_one_cmplsi2): Likewise.
7113 (notsi_compare0): Likewise.
7114 (notsi_compare0_scratch): Likewise.
7115 (thumb1_movdi_insn): Likewise.
7116 (arm_movsi_insn): Likewise.
7117 (movhi_insn_arch4): Likewise.
7118 (movhi_bytes): Likewise.
7119 (arm_movqi_insn): Likewise.
7120 (thumb1_movqi_insn): Likewise.
7121 (arm32_movhf): Likewise.
7122 (thumb1_movhf): Likewise.
7123 (arm_movsf_soft_insn): Likewise.
7124 (thumb1_movsf_insn): Likewise.
7125 (thumb_movdf_insn): Likewise.
7126 (movsicc_insn): Likewise.
7127 (movsfcc_soft_insn): Likewise.
7128 (and_scc): Likewise.
7129 (cond_move): Likewise.
7130 (if_move_not): Likewise.
7131 (if_not_move): Likewise.
7132 (if_shift_move): Likewise.
7133 (if_move_shift): Likewise.
7134 (if_shift_shift): Likewise.
7135 (if_not_arith): Likewise.
7136 (if_arith_not): Likewise.
7137 (cond_move_not): Likewise.
7138 * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
7139 (neon_mov<mode>): Likewise.
7140 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
7141 (thumb2_movsi_vfp): Likewise.
7142 (movsf_vfp): Likewise.
7143 (thumb2_movsf_vfp): Likewise.
7144 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7146 (cortexa7_older_only): Likewise.
7147 (cortexa7_younger): Likewise.
7148 * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
7149 (1020alu_shift_op): Likewise.
7150 (1020alu_shift_reg_op): Likewise.
7151 * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
7152 (alu_shift_op): Likewise.
7153 (alu_shift_reg_op): Likewise.
7154 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
7155 (11_alu_shift_op): Likewise.
7156 (11_alu_shift_reg_op): Likewise.
7157 * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
7158 (9_alu_shift_reg_op): Likewise.
7159 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7161 (cortex_a15_alu_shift): Likewise.
7162 (cortex_a15_alu_shift_reg): Likewise.
7163 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
7164 (cortex_a5_alu_shift): Likewise.
7165 * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
7167 (cortex_a53_alu_shift): Likewise.
7168 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7170 (cortex_a7_alu_reg): Likewise.
7171 (cortex_a7_alu_shift): Likewise.
7172 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7173 (cortex_a8_alu_shift): Likewise.
7174 (cortex_a8_alu_shift_reg): Likewise.
7175 (cortex_a8_mov): Likewise.
7176 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7177 (cortex_a9_dp_shift): Likewise.
7178 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
7179 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
7180 (cortex_r4_mov): Likewise.
7181 (cortex_r4_alu_shift): Likewise.
7182 (cortex_r4_alu_shift_reg): Likewise.
7183 * config/arm/fa526.md (526_alu_op): Update for attribute change.
7184 (526_alu_shift_op): Likewise.
7185 * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
7186 * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
7187 (626te_alu_shift_op): Likewise.
7188 * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
7189 (726te_alu_op): Likewise.
7190 (726te_alu_shift_op): Likewise.
7191 (726te_alu_shift_reg_op): Likewise.
7192 * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
7193 (mp626_alu_shift_op): Likewise.
7194 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
7195 (pj4_alu_e1_conds): Likewise.
7196 (pj4_alu): Likewise.
7197 (pj4_alu_conds): Likewise.
7198 (pj4_shift): Likewise.
7199 (pj4_shift_conds): Likewise.
7200 (pj4_alu_shift): Likewise.
7201 (pj4_alu_shift_conds): Likewise.
7203 2013-07-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7205 * config/arm/predicates.md (shiftable_operator_strict_it):
7207 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
7208 Disable cond_exec version for arm_restrict_it.
7209 (thumb2_smaxsi3): Convert to generate cond_exec.
7210 (thumb2_sminsi3): Likewise.
7211 (thumb32_umaxsi3): Likewise.
7212 (thumb2_uminsi3): Likewise.
7213 (thumb2_abssi2): Adjust constraints for arm_restrict_it.
7214 (thumb2_neg_abssi2): Likewise.
7215 (thumb2_mov_scc): Add alternative for 16-bit encoding.
7216 (thumb2_movsicc_insn): Adjust alternatives.
7217 (thumb2_mov_negscc): Disable for arm_restrict_it.
7218 (thumb2_mov_negscc_strict_it): New pattern.
7219 (thumb2_mov_notscc_strict_it): New pattern.
7220 (thumb2_mov_notscc): Disable for arm_restrict_it.
7221 (thumb2_ior_scc): Likewise.
7222 (thumb2_ior_scc_strict_it): New pattern.
7223 (thumb2_cond_move): Adjust for arm_restrict_it.
7224 (thumb2_cond_arith): Disable for arm_restrict_it.
7225 (thumb2_cond_arith_strict_it): New pattern.
7226 (thumb2_cond_sub): Adjust for arm_restrict_it.
7227 (thumb2_movcond): Likewise.
7228 (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
7229 (thumb2_zero_extendhisi2_v6): Likewise.
7230 (thumb2_zero_extendqisi2_v6): Likewise.
7231 (orsi_notsi_si): Likewise.
7232 (orsi_not_shiftsi_si): Likewise.
7234 2013-07-22 Georg-Johann Lay <avr@gjlay.de>
7236 * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
7237 instruction sequence is 1 byte shorter.
7239 2013-07-22 Uros Bizjak <ubizjak@gmail.com>
7241 * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
7242 it is not needed after split.
7244 2013-07-20 Iain Sandoe <iain@codesourcery.com>
7247 * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
7248 second label for nonlocal goto receivers. Don't output pic base labels
7249 unless we're producing PIC; mark that action unreachable().
7250 (ix86_save_reg): If the function contains a nonlocal label, save the
7252 * config/darwin-protos.h (machopic_should_output_picbase_label): New.
7253 * config/darwin.c (emitted_pic_label_num): New GTY.
7254 (update_pic_label_number_if_needed): New.
7255 (machopic_output_function_base_name): Adjust for nonlocal receiver
7257 (machopic_should_output_picbase_label): New.
7258 * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
7259 (nonlocal_goto_receiver): New insn and split.
7261 2013-07-20 James Greenhalgh <james.greenhalgh@arm.com>
7263 * config/aarch64/aarch64-builtins.c
7264 (aarch64_fold_builtin): Fold abs in all modes.
7265 * config/aarch64/aarch64-simd-builtins.def
7266 (abs): Enable for all modes.
7267 * config/aarch64/arm_neon.h
7268 (vabs<q>_s<8,16,32,64): Rewrite using builtins.
7269 (vabs_f64): Add missing intrinsic.
7271 2013-07-19 Ian Bolton <ian.bolton@arm.com>
7273 * config/aarch64/arm_neon.h (vabs_s64): New function
7275 2013-07-19 Georg-Johann Lay <avr@gjlay.de>
7278 * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
7279 * config/avr/avr.md (adjust_len): Add `round'.
7280 * config/avr/avr-protos.h (avr_out_round): New prototype.
7281 (avr_out_plus): Add `out_label' argument.
7282 * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
7283 (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
7284 Handle the case where `insn' is just a pattern.
7285 (avr_out_bitop): Handle the case where `insn' is just a pattern.
7286 (avr_out_round): New function.
7287 (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
7289 2013-07-18 David Holsgrove <david.holsgrove@xilinx.com>
7291 * config/microblaze/microblaze.c (microblaze_expand_prologue):
7292 Add check for flag_stack_usage to handle -fstack-usage support
7294 2013-07-18 Pat Haugen <pthaugen@us.ibm.com>
7296 * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
7297 interaction for new Power8 flags and VSX.
7299 2013-07-18 Sriraman Tallam <tmsriram@google.com>
7302 * tree-inline.c (expand_call_inline): Emit errors during
7303 early_inlining only if optimization is not turned on.
7305 2013-07-18 David Malcolm <dmalcolm@redhat.com>
7309 * passes.c (init_optimization_passes): Move the construction of
7310 the pass hierarchy into a new passes.def file.
7312 * Makefile.in (passes.o): Add dependency on passes.def.
7314 2013-07-18 David Malcolm <dmalcolm@redhat.com>
7316 * passes.c (init_optimization_passes): Introduce macros for
7317 constructing the tree of passes (INSERT_PASSES_AFTER,
7318 PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
7319 TERMINATE_PASS_LIST).
7321 2013-07-18 Vladimir Makarov <vmakarov@redhat.com>
7322 Wei Mi <wmi@google.com>
7324 PR rtl-optimization/57878
7325 * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
7327 (reload_pseudo_compare_func): Check nregs first for reload
7330 2013-07-18 David Malcolm <dmalcolm@redhat.com>
7332 * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
7334 2013-07-18 Po-Chun Chang <pchang9@cs.wisc.edu>
7336 * read-rtl.c (validate_const_int): Once an invalid character is
7337 seen, quit the loop.
7339 * gengtype.c (write_roots): Similarly once we find the "deletable"
7340 or "if_marked" option.
7342 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
7344 * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
7345 "xtab" and "sat". Move value "clz" from here to ...
7346 (attriubte "type"): ... here.
7347 (satsi_<SAT:code>): Delete "insn" attribute.
7348 (satsi_<SAT:code>_shift): Likewise.
7349 (arm_zero_extendqisi2addsi): Likewise.
7350 (arm_extendqisi2addsi): Likewise.
7351 (clzsi2): Update for attribute changes.
7352 (rbitsi2): Likewise.
7353 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
7355 (arm_usatsihi): Likewise.
7356 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7358 2013-07-18 Sofiane Naci <sofiane.naci@arm.com>
7360 * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
7361 "arlo_imm". Rename "alu_reg" to "arlo_reg". Rename "simple_alu_shift"
7362 to "extend". Split "alu_shift" into "shift" and "arlo_shift". Split
7363 "alu_shift_reg" into "shift_reg" and "arlo_shift_reg". List types
7364 in alphabetical order.
7365 (attribute "core_cycles"): Update for attribute changes.
7366 (arm_addsi3): Likewise.
7367 (addsi3_compare0): Likewise.
7368 (addsi3_compare0_scratch): Likewise.
7369 (addsi3_compare_op1): Likewise.
7370 (addsi3_compare_op2): Likewise.
7371 (compare_addsi2_op0): Likewise.
7372 (compare_addsi2_op1): Likewise.
7373 (addsi3_carryin_shift_<optab>): Likewise.
7374 (subsi3_carryin_shift): Likewise.
7375 (rsbsi3_carryin_shift): Likewise.
7376 (arm_subsi3_insn): Likewise.
7377 (subsi3_compare0): Likewise.
7378 (subsi3_compare): Likewise.
7379 (arm_andsi3_insn): Likewise.
7380 (thumb1_andsi3_insn): Likewise.
7381 (andsi3_compare0): Likewise.
7382 (andsi3_compare0_scratch): Likewise.
7383 (zeroextractsi_compare0_scratch
7384 (andsi_not_shiftsi_si): Likewise.
7385 (iorsi3_insn): Likewise.
7386 (iorsi3_compare0): Likewise.
7387 (iorsi3_compare0_scratch): Likewise.
7388 (arm_xorsi3): Likewise.
7389 (thumb1_xorsi3_insn): Likewise.
7390 (xorsi3_compare0): Likewise.
7391 (xorsi3_compare0_scratch): Likewise.
7392 (satsi_<SAT:code>_shift): Likewise.
7394 (arm_shiftsi3): Likewise.
7395 (shiftsi3_compare0): Likewise.
7396 (not_shiftsi): Likewise.
7397 (not_shiftsi_compare0): Likewise.
7398 (not_shiftsi_compare0_scratch): Likewise.
7399 (arm_one_cmplsi2): Likewise.
7400 (thumb_one_complsi2): Likewise.
7401 (notsi_compare0): Likewise.
7402 (notsi_compare0_scratch): Likewise.
7403 (thumb1_zero_extendhisi2): Likewise.
7404 (arm_zero_extendhisi2): Likewise.
7405 (arm_zero_extendhisi2_v6): Likewise.
7406 (arm_zero_extendhisi2addsi): Likewise.
7407 (thumb1_zero_extendqisi2): Likewise.
7408 (thumb1_zero_extendqisi2_v6): Likewise.
7409 (arm_zero_extendqisi2): Likewise.
7410 (arm_zero_extendqisi2_v6): Likewise.
7411 (arm_zero_extendqisi2addsi): Likewise.
7412 (thumb1_extendhisi2): Likewise.
7413 (arm_extendhisi2): Likewise.
7414 (arm_extendhisi2_v6): Likewise.
7415 (arm_extendqisi): Likewise.
7416 (arm_extendqisi_v6): Likewise.
7417 (arm_extendqisi2addsi): Likewise.
7418 (thumb1_extendqisi2): Likewise.
7419 (thumb1_movdi_insn): Likewise.
7420 (arm_movsi_insn): Likewise.
7421 (movsi_compare0): Likewise.
7422 (movhi_insn_arch4): Likewise.
7423 (movhi_bytes): Likewise.
7424 (arm_movqi_insn): Likewise.
7425 (thumb1_movqi_insn): Likewise.
7426 (arm32_movhf): Likewise.
7427 (thumb1_movhf): Likewise.
7428 (arm_movsf_soft_insn): Likewise.
7429 (thumb1_movsf_insn): Likewise.
7430 (movdf_soft_insn): Likewise.
7431 (thumb_movdf_insn): Likewise.
7432 (arm_cmpsi_insn): Likewise.
7433 (cmpsi_shiftsi): Likewise.
7434 (cmpsi_shiftsi_swp): Likewise.
7435 (arm_cmpsi_negshiftsi_si): Likewise.
7436 (movsicc_insn): Likewise.
7437 (movsfcc_soft_insn): Likewise.
7438 (arith_shiftsi): Likewise.
7439 (arith_shiftsi_compare0
7440 (arith_shiftsi_compare0_scratch
7441 (sub_shiftsi): Likewise.
7442 (sub_shiftsi_compare0
7443 (sub_shiftsi_compare0_scratch
7444 (and_scc): Likewise.
7445 (cond_move): Likewise.
7446 (if_plus_move): Likewise.
7447 (if_move_plus): Likewise.
7448 (if_move_not): Likewise.
7449 (if_not_move): Likewise.
7450 (if_shift_move): Likewise.
7451 (if_move_shift): Likewise.
7452 (if_shift_shift): Likewise.
7453 (if_not_arith): Likewise.
7454 (if_arith_not): Likewise.
7455 (cond_move_not): Likewise.
7456 (thumb1_ashlsi3): Set type attribute.
7457 (thumb1_ashrsi3): Likewise.
7458 (thumb1_lshrsi3): Likewise.
7459 (thumb1_rotrsi3): Likewise.
7460 (shiftsi3_compare0_scratch): Likewise.
7461 * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
7462 (neon_mov<mode>): Likewise.
7463 * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
7465 (thumb2_movsi_insn): Likewise.
7466 (thumb2_cmpsi_neg_shiftsi): Likewise.
7467 (thumb2_extendqisi_v6): Likewise.
7468 (thumb2_zero_extendhisi2_v6): Likewise.
7469 (thumb2_zero_extendqisi2_v6): Likewise.
7470 (thumb2_shiftsi3_short): Likewise.
7471 (thumb2_addsi3_compare0_scratch): Likewise.
7472 (orsi_not_shiftsi_si): Likewise.
7473 * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7474 * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
7476 * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
7477 (1020alu_shift_op): Likewise.
7478 (1020alu_shift_reg_op): Likewise.
7479 * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
7480 (alu_shift_op): Likewise.
7481 (alu_shift_reg_op): Likewise.
7482 * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
7483 (11_alu_shift_op): Likewise.
7484 (11_alu_shift_reg_op): Likewise.
7485 * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
7486 (9_alu_shift_reg_op): Likewise.
7487 * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7489 (cortex_a15_alu_shift): Likewise.
7490 (cortex_a15_alu_shift_reg): Likewise.
7491 * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
7493 (cortex_a5_alu_shift): Likewise.
7494 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
7496 (cortex_a53_alu_shift): Likewise.
7497 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
7499 (cortex_a7_alu_reg): Likewise.
7500 (cortex_a7_alu_shift): Likewise.
7501 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
7503 (cortex_a8_alu_shift): Likewise.
7504 (cortex_a8_alu_shift_reg): Likewise.
7505 (cortex_a8_mov): Likewise.
7506 * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
7507 (cortex_a9_dp_shift): Likewise.
7508 * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
7510 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
7512 (cortex_r4_mov): Likewise.
7513 (cortex_r4_alu_shift): Likewise.
7514 (cortex_r4_alu_shift_reg): Likewise.
7515 * config/arm/fa526.md (526_alu_op): Update for attribute changes.
7516 (526_alu_shift_op): Likewise.
7517 * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
7518 * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
7519 (626te_alu_shift_op): Likewise.
7520 * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
7521 (726te_alu_op): Likewise.
7522 (726te_alu_shift_op): Likewise.
7523 (726te_alu_shift_reg_op): Likewise.
7524 * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
7525 (mp626_alu_shift_op): Likewise.
7526 * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
7527 (pj4_alu_e1_conds): Likewise.
7528 (pj4_alu): Likewise.
7529 (pj4_alu_conds): Likewise.
7530 (pj4_shift): Likewise.
7531 (pj4_shift_conds): Likewise.
7532 (pj4_alu_shift): Likewise.
7533 (pj4_alu_shift_conds): Likewise.
7534 * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
7536 (cortexa7_older_only): Likewise.
7537 (cortexa7_younger): Likewise.
7539 2013-07-18 David Malcolm <dmalcolm@redhat.com>
7541 * ipa-pure-const.c (generate_summary): Rename to...
7542 (pure_const_generate_summary): ... this.
7544 2013-07-17 Iain Sandoe <iain@codesourcery.com>
7546 * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
7548 2013-07-17 Yvan Roux <yvan.roux@linaro.org>
7551 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
7554 2013-07-17 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
7556 * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
7557 enabled without -march=zEC12.
7558 * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
7561 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
7563 * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
7564 (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
7565 (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
7566 terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
7567 (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
7568 * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
7569 ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
7571 <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
7572 and ISA_HAS_NMADD3_NMSUB3.
7573 * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
7574 (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
7575 (nmsub4<mode>, nmsub3<mode>): Likewise.
7576 (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
7578 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
7580 * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
7581 TARGET_MIPS5400 checking.
7583 2013-07-16 Jakub Jelinek <jakub@redhat.com>
7584 Peter Bergner <bergner@vnet.ibm.com>
7586 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
7587 registers in the comment.
7588 (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
7589 (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
7590 rather than FIRST_PSEUDO_REGISTERS.
7592 2013-07-16 Peter Bergner <bergner@vnet.ibm.com>
7594 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
7595 enable extra ISA flags with TARGET_HTM.
7597 2013-07-16 Maciej W. Rozycki <macro@codesourcery.com>
7599 * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
7602 2013-07-15 Cong Hou <congh@google.com>
7604 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
7605 in compare function for sorting.
7607 2013-07-15 Peter Bergner <bergner@vnet.ibm.com>
7609 * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
7610 * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
7611 * config/rs6000/rs6000.opt: Add -mhtm option.
7612 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
7613 (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
7614 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
7615 __HTM__ if the HTM instructions are available.
7616 * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
7617 htm_spr_reg_operand): New define_predicates.
7618 * config/rs6000/rs6000.md (define_attr "type"): Add htm.
7619 (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
7621 * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
7622 BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
7623 HTM builtin functions.
7624 * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
7625 (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
7626 (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
7627 (rs6000_builtin_mask_calculate): Likewise.
7628 (rs6000_option_override_internal): Likewise.
7629 (bdesc_htm): Add new HTM builtin support.
7630 (htm_spr_num): New function.
7631 (htm_spr_regno): Likewise.
7632 (rs6000_htm_spr_icode): Likewise.
7633 (htm_expand_builtin): Likewise.
7634 (htm_init_builtins): Likewise.
7635 (rs6000_expand_builtin): Add support for HTM builtin functions.
7636 (rs6000_init_builtins): Likewise.
7637 (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
7639 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
7640 (TARGET_HTM, MASK_HTM): Define macros.
7641 (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
7642 (FIXED_REGISTERS): Likewise.
7643 (CALL_USED_REGISTERS): Likewise.
7644 (CALL_REALLY_USED_REGISTERS): Likewise.
7645 (REG_ALLOC_ORDER): Likewise.
7646 (enum reg_class): Likewise.
7647 (REG_CLASS_NAMES): Likewise.
7648 (REG_CLASS_CONTENTS): Likewise.
7649 (REGISTER_NAMES): Likewise.
7650 (ADDITIONAL_REGISTER_NAMES): Likewise.
7651 (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
7652 RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
7653 (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
7654 * config/rs6000/htm.md: New file.
7655 * config/rs6000/htmintrin.h: New file.
7656 * config/rs6000/htmxlintrin.h: New file.
7658 2013-07-15 Marcus Shawcroft <marcus.shawcroft@arm.com>
7660 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
7661 Define SYMBOL_TINY_GOT, update comment.
7662 * config/aarch64/aarch64.c
7663 (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
7664 (aarch64_expand_mov_immediate): Likewise.
7665 (aarch64_print_operand): Likewise.
7666 (aarch64_classify_symbol): Likewise.
7667 * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
7668 (ldr_got_tiny): Define.
7670 2013-07-13 Tobias Grosser <tobias@grosser.es>
7672 PR tree-optimization/54094
7673 * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
7674 scheduling dimension for the parallelism check from the polyhedral
7675 information in the AST.
7676 * graphite-dependences.c (carries_deps): Do not assume the schedule is
7679 2013-07-13 Jason Merrill <jason@redhat.com>
7681 * print-tree.c (debug_vec_tree): Use debug_raw.
7682 (debug_raw (vec<tree, va_gc> &)): New.
7683 (debug_raw (vec<tree, va_gc> *)): New.
7684 * tree.h: Declare them.
7686 2013-07-13 Bin Cheng <bin.cheng@arm.com>
7688 * ifcvt.c (ifcvt_after_combine): New static variable.
7689 (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
7691 (if_convert): New parameter after_combine. Set ifcvt_after_combine.
7692 (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
7693 rest_of_handle_if_after_reload): Pass new argument for if_convert.
7695 2013-07-12 Maciej W. Rozycki <macro@codesourcery.com>
7697 * config/mips/mips.c (mips_expand_call): Remove empty statement.
7699 2013-07-12 Michael Matz <matz@suse.de>
7702 * convert.c (convert_to_real): Reject non-float inner types.
7704 2013-07-12 Tejas Belagod <tejas.belagod@arm.com>
7706 * config/aarch64/aarch64-protos.h
7707 (aarch64_simd_immediate_valid_for_move): Remove.
7708 * config/aarch64/aarch64.c (simd_immediate_info): New member.
7709 (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
7711 (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
7713 2013-07-11 Steve Ellcey <sellcey@mips.com>
7715 * config/mips/mips.c (mips_conditional_register_usage): Do not
7716 use t[0-7] registers in MIPS16 mode when optimizing for size.
7718 2013-07-11 Sriraman Tallam <tmsriram@google.com>
7720 * config/i386/i386.c (dispatch_function_versions): Fix array
7721 indexing of function_version_info to match actual_versions.
7723 2013-07-11 Teresa Johnson <tejohnson@google.com>
7725 * vec.h (struct va_gc): Move release out-of-line.
7726 (va_gc::release): Call ggc_free on released vec.
7728 2013-07-11 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
7730 * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
7731 Require GOT register as additional operand in UNSPEC.
7732 ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
7733 ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7734 ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
7735 ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
7736 ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
7737 ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
7738 ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
7740 2013-07-11 Georg-Johann Lay <avr@gjlay.de>
7743 * config/avr/avr.c (avr_set_current_function): Sanity-check signal
7744 name seen by assembler/linker rather if available.
7746 2013-07-11 Andreas Schwab <schwab@suse.de>
7748 * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
7750 2013-07-10 Vladimir Makarov <vmakarov@redhat.com>
7752 * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
7754 2013-07-10 Joseph Myers <joseph@codesourcery.com>
7756 * doc/tm.texi.in: Move hook documentation to ....
7757 * target.def: ... here.
7759 * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
7761 * doc/tm.texi: Regenerate.
7763 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
7766 * doc/invoke.texi: Document -Wconditionally-supported.
7768 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
7771 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
7774 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
7777 * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
7778 (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
7779 (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
7780 Remove duplicate devices.
7781 * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
7782 * config/avr/t-multilib: Regenerate.
7783 * config/avr/avr-tables.opt: Regenerate.
7784 * doc/avr-mmcu.texi: Regenerate.
7786 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
7789 * config/avr/avr.opt (Waddr-space-convert): Fix typo.
7791 2013-07-10 Graham Stott <graham.stott@btinternet.com>
7793 * config/mips/mips.c (mips_rtx_costs): Very slightly increase
7794 the cost of MULT when optimizing for size.
7796 2013-07-10 Jan-Benedict Glaw <jbglaw@lug-owl.de>
7798 * config/cr16/cr16-protos.h: Don't include target.h.
7800 2013-07-09 Joseph Myers <joseph@codesourcery.com>
7802 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
7803 adjust register size for TDmode and TFmode for VSX registers.
7805 2013-07-08 Kai Tietz <ktietz@redhat.com>
7808 * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
7809 hook_bool_const_tree_true.
7811 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
7813 * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
7814 * config/s390/s390.h: Remove F*_REGNUM macro definitions.
7815 * config/s390/s390.md: Define FPR*_REGNUM constants.
7816 Fix FPR2_REGNUM constant (18 -> 17).
7817 ("*trunc<BFP:mode><DFP_ALL:mode>2")
7818 ("*trunc<DFP_ALL:mode><BFP:mode>2")
7819 ("trunc<BFP:mode><DFP_ALL:mode>2")
7820 ("trunc<DFP_ALL:mode><BFP:mode>2")
7821 ("*extend<BFP:mode><DFP_ALL:mode>2")
7822 ("*extend<DFP_ALL:mode><BFP:mode>2")
7823 ("extend<BFP:mode><DFP_ALL:mode>2")
7824 ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
7827 2013-07-08 Graham Stott <graham.stott@btinternet.com>
7829 * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
7831 2013-07-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
7833 * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
7834 and cfun_fpr_bit_p to cfun_fpr_save_p.
7835 (s390_frame_area, s390_register_info, s390_frame_info)
7836 (s390_emit_prologue, s390_emit_epilogue)
7837 (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
7839 * config/s390/s390.h: Define *_REGNUM macros for floating point
7842 2013-07-08 Eric Botcazou <ebotcazou@adacore.com>
7844 * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
7846 2013-07-08 Po-Chun Chang <pchang9@cs.wisc.edu>
7848 PR rtl-optimization/57786
7849 * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
7850 and break out of the loop when it is set to false.
7852 2013-07-08 Jakub Jelinek <jakub@redhat.com>
7855 * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
7856 Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
7857 (const_int 63)) 0)).
7858 * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
7859 using simplify_gen_unary instead of gen_rtx_*_EXTEND.
7860 * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
7862 PR rtl-optimization/57829
7863 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
7864 mask bits outside of mode are just sign-extension from mode to HWI.
7866 2013-07-08 Michael Zolotukhin <michael.v.zolotukhin@gmail.com>
7868 * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
7869 * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
7870 adjust_address instead of change_address to keep info about alignment.
7871 (emit_strmov): Remove.
7872 (emit_memmov): New function.
7873 (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
7874 (expand_movmem_epilogue): Likewise and return updated rtx for
7876 (expand_constant_movmem_prologue): Likewise and return updated rtx for
7877 destination and source.
7878 (decide_alignment): Refactor, handle vector_loop.
7879 (ix86_expand_movmem): Likewise.
7880 (ix86_expand_setmem): Likewise.
7881 * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
7883 2013-07-07 Uros Bizjak <ubizjak@gmail.com>
7885 * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
7886 signature_TM2_ebx, it interferes with signature_INTEL_ebx.
7888 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
7890 * config/i386/sse.md (sse_movlhps): Change alternative 3
7891 of operand 2 to "m".
7893 2013-07-06 Uros Bizjak <ubizjak@gmail.com>
7896 * config/i386/sse.md (iptr): New mode attribute.
7897 (sse2_movq128): Add pointer size overrides for Intel asm dialect.
7898 (<sse>_vm<plusminus_insn><mode>3): Ditto.
7899 (<sse>_vmmul<mode>3): Ditto.
7900 (<sse>_vmdiv<mode>3): Ditto.
7901 (sse_vmrcpv4sf2): Ditto.
7902 (<sse>_vmsqrt<mode>2): Ditto.
7903 (sse_vmrsqrtv4sf2): Ditto.
7904 (<sse>_vm<code><mode>3): Ditto.
7905 (avx_vmcmp<mode>3): Ditto.
7906 (<sse>_vmmaskcmp<mode>3): Ditto.
7907 (<sse>_comi): Ditto.
7908 (<sse>_ucomi): Ditto.
7909 (*xop_vmfrcz_<mode>): Ditto.
7910 (*fmai_fmadd_<mode>): Ditto.
7911 (*fmai_fmsub_<mode>): Ditto.
7912 (*fmai_fnmadd_<mode>): Ditto.
7913 (*fmai_fnmsub_<mode>): Ditto.
7914 (*fma4i_vmfmadd_<mode>): Ditto.
7915 (*fma4i_vmfmsub_<mode>): Ditto.
7916 (*fma4i_vmfnmadd_<mode>): Ditto.
7917 (*fma4i_vmfnmsub_<mode>): Ditto.
7918 (*xop_vmfrcz_<mode>): Ditto.
7919 (sse_cvtps2pi): Ditto.
7920 (sse_cvttps2pi): Ditto.
7921 (sse_cvtss2si): Ditto.
7922 (sse_cvtss2si_2): Ditto.
7923 (sse_cvtss2siq_2): Ditto.
7924 (sse_cvttss2si): Ditto.
7925 (sse_cvttss2siq): Ditto.
7926 (sse_cvtsd2si): Ditto.
7927 (sse_cvtsd2si_2): Ditto.
7928 (sse_cvtsd2siq_2): Ditto.
7929 (sse_cvttsd2si): Ditto.
7930 (sse_cvttsd2siq): Ditto.
7931 (sse_cvtsd2ss): Ditto.
7932 (sse_cvtss2sd): Ditto.
7933 (avx2_pbroadcast<mode>): Ditto.
7934 (avx2_pbroadcast<mode>_1): Ditto.
7935 (*avx_vperm_broadcast_v4sf): Ditto.
7937 (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
7938 (sse_movlhps): Ditto.
7939 (sse_storehps): Ditto.
7940 (sse_loadhps): Ditto.
7941 (sse_storelps): Ditto.
7942 (sse_loadlps): Ditto.
7943 (*vec_concatv4sf): Ditto.
7944 (*vec_interleave_highv2df): Ditto.
7945 (*vec_interleave_lowv2df): Ditto.
7946 (*vec_extractv2df_1_sse): Ditto.
7947 (*vec_extractv2df_0_sse): Ditto.
7948 (sse2_storelpd): Ditto.
7949 (sse2_loadlpd): Ditto.
7950 (sse2_movsd): Ditto.
7951 (*vec_concatv4si): Ditto.
7952 (vec_concatv2di): Ditto.
7954 * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
7955 for Intel asm dialect.
7956 (mmx_punpcklwd): Ditto.
7957 (mmx_punpckldq): Ditto.
7959 * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
7960 for intel assembler dialect.
7962 2013-07-06 Jakub Jelinek <jakub@redhat.com>
7965 * fold-const.c (tree_call_nonnegative_warnv_p): Return true
7966 for BUILT_IN_C{LZ,LRSB}*.
7967 * tree.h (CASE_INT_FN): Add FN##IMAX case.
7968 * tree-vrp.c (extract_range_basic): Handle
7969 BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*. For
7970 BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
7971 fall thru to code calling set_value*.
7972 * builtins.c (expand_builtin): Remove *IMAX cases.
7973 (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
7974 if width is bigger than 2*HWI.
7976 2013-07-05 Vladimir Makarov <vmakarov@redhat.com>
7978 PR rtl-optimization/55342
7979 * lra-int.h (lra_subreg_reload_pseudos): New.
7980 * lra.c: Add undoing optional reloads to the block diagram.
7981 (lra_subreg_reload_pseudos): New.
7982 (lra_optional_reload_pseudos): Change comments.
7983 (lra): Init and clear lra_subreg_reload_pseudos. Clear
7984 lra_optional_reload_pseudos after undo transformations.
7985 * lra-assigns.c (pseudo_prefix_title): New.
7986 (lra_setup_reg_renumber): Use it.
7987 (spill_for): Ditto. Check subreg reload pseudos too.
7988 (assign_by_spills): Consider subreg reload pseudos too.
7989 * lra-constraints.c (simplify_operand_subreg): Use
7990 lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
7991 (curr_insn_transform): Recognize and do optional reloads.
7992 (undo_optional_reloads): New.
7993 (lra_undo_inheritance): Call undo_optional_reloads.
7995 2013-07-05 Thomas Quinot <quinot@adacore.com>
7997 * tree-complex.c (expand_complex_operations_1): Fix typo.
7999 2013-07-04 Tejas Belagod <tejas.belagod@arm.com>
8001 * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
8002 (tune_params): New member 'const vec_costs'.
8003 * config/aarch64/aarch64.c (generic_vector_cost): New.
8004 (generic_tunings): New member 'generic_vector_cost'.
8005 (aarch64_builtin_vectorization_cost): New.
8006 (aarch64_add_stmt_cost): New.
8007 (TARGET_VECTORIZE_ADD_STMT_COST): New.
8008 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
8010 2013-07-03 Jakub Jelinek <jakub@redhat.com>
8013 * config/i386/predicates.md (vsib_address_operand): Disallow
8014 SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
8016 2013-07-03 Hans-Peter Nilsson <hp@bitrange.com>
8019 * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
8020 expand_builtin_setjmp_receiver.
8021 (expand_label): Adjust, call expand_builtin_setjmp_receiver
8022 with NULL for the label parameter.
8023 * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
8024 the frame-pointer. Adjust comments.
8025 [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
8026 only if LABEL is non-NULL.
8028 2013-07-03 Yufeng Zhang <yufeng.zhang@arm.com>
8030 * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
8031 (ARM_ABI_AAPCS64): Ditto.
8033 (ARM_DEFAULT_ABI): Ditto.
8035 2013-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8037 * config/aarch64/aarch64-builtins.c
8038 (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
8039 * config/aarch64/aarch64-simd-builtins.def (ld1): New.
8041 * config/aarch64/aarch64-simd.md
8042 (aarch64_ld1<VALL:mode>): New.
8043 (aarch64_st1<VALL:mode>): Likewise.
8044 * config/aarch64/arm_neon.h
8045 (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
8047 2013-07-02 Sriraman Tallam <tmsriram@google.com>
8049 * config/i386/i386.c (gate_insert_vzeroupper): Check if
8051 (ix86_option_override_internal):Turn on all -mavx target flags by
8052 default as they are dependent on AVX anyway.
8054 2013-07-02 Cary Coutant <ccoutant@google.com>
8056 * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
8058 (loc_checksum_ordered): Likewise.
8059 (hash_loc_operands): Remove inline keyword.
8061 2013-07-02 Jakub Jelinek <jakub@redhat.com>
8063 PR tree-optimization/57741
8064 * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
8065 non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
8066 or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
8067 Allow REAL_CST step_exprs if flag_associative_math.
8068 (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
8070 2013-07-02 Ian Bolton <ian.bolton@arm.com>
8072 * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
8074 2013-07-02 Ian Bolton <ian.bolton@arm.com>
8076 * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
8078 2013-07-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8080 * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
8082 (iorsi3_insn): Likewise.
8083 (arm_xorsi3): Likewise.
8085 2013-07-01 Sofiane Naci <sofiane.naci@arm.com>
8087 * arm.md (attribute "wtype"): Delete. Move attribute values from here
8089 (attribute "type"): ... here, and prefix with "wmmx_".
8090 (attribute "core_cycles"): Update for attribute changes.
8091 * iwmmxt.md (tbcstv8qi): Update for attribute changes.
8092 (tbcstv4hi): Likewise.
8093 (tbcstv2si): Likewise.
8094 (iwmmxt_iordi3): Likewise.
8095 (iwmmxt_xordi3): Likewise.
8096 (iwmmxt_anddi3): Likewise.
8097 (iwmmxt_nanddi3): Likewise.
8098 (iwmmxt_arm_movdi): Likewise.
8099 (iwmmxt_movsi_insn): Likewise.
8100 (mov<mode>_internal): Likewise.
8101 (and<mode>3_iwmmxt): Likewise.
8102 (ior<mode>3_iwmmxt): Likewise.
8103 (xor<mode>3_iwmmxt): Likewise.
8104 (add<mode>3_iwmmxt): Likewise.
8105 (ssaddv8qi3): Likewise.
8106 (ssaddv4hi3): Likewise.
8107 (ssaddv2si3): Likewise.
8108 (usaddv8qi3): Likewise.
8109 (usaddv4hi3): Likewise.
8110 (usaddv2si3): Likewise.
8111 (sub<mode>3_iwmmxt): Likewise.
8112 (sssubv8qi3): Likewise.
8113 (sssubv4hi3): Likewise.
8114 (sssubv2si3): Likewise.
8115 (ussubv8qi3): Likewise.
8116 (ussubv4hi3): Likewise.
8117 (ussubv2si3): Likewise.
8118 (mulv4hi3_iwmmxt): Likewise.
8119 (smulv4hi3_highpart): Likewise.
8120 (umulv4hi3_highpart): Likewise.
8121 (iwmmxt_wmacs): Likewise.
8122 (iwmmxt_wmacsz): Likewise.
8123 (iwmmxt_wmacu): Likewise.
8124 (iwmmxt_wmacuz): Likewise.
8125 (iwmmxt_clrdi): Likewise.
8126 (iwmmxt_clrv8qi): Likewise.
8127 (iwmmxt_clr4hi): Likewise.
8128 (iwmmxt_clr2si): Likewise.
8129 (iwmmxt_uavgrndv8qi3): Likewise.
8130 (iwmmxt_uavgrndv4hi3): Likewise.
8131 (iwmmxt_uavgv8qi3): Likewise.
8132 (iwmmxt_uavgv4hi3): Likewise.
8133 (iwmmxt_tinsrb): Likewise.
8134 (iwmmxt_tinsrh): Likewise.
8135 (iwmmxt_tinsrw): Likewise.
8136 (iwmmxt_textrmub): Likewise.
8137 (iwmmxt_textrmsb): Likewise.
8138 (iwmmxt_textrmuh): Likewise.
8139 (iwmmxt_textrmsh): Likewise.
8140 (iwmmxt_textrmw): Likewise.
8141 (iwmxxt_wshufh): Likewise.
8142 (eqv8qi3): Likewise.
8143 (eqv4hi3): Likewise.
8144 (eqv2si3): Likewise.
8145 (gtuv8qi3): Likewise.
8146 (gtuv4hi3): Likewise.
8147 (gtuv2si3): Likewise.
8148 (gtv8qi3): Likewise.
8149 (gtv4hi3): Likewise.
8150 (gtv2si3): Likewise.
8151 (smax<mode>3_iwmmxt): Likewise.
8152 (umax<mode>3_iwmmxt): Likewise.
8153 (smin<mode>3_iwmmxt): Likewise.
8154 (umin<mode>3_iwmmxt): Likewise.
8155 (iwmmxt_wpackhss): Likewise.
8156 (iwmmxt_wpackwss): Likewise.
8157 (iwmmxt_wpackdss): Likewise.
8158 (iwmmxt_wpackhus): Likewise.
8159 (iwmmxt_wpackwus): Likewise.
8160 (iwmmxt_wpackdus): Likewise.
8161 (iwmmxt_wunpckihb): Likewise.
8162 (iwmmxt_wunpckihh): Likewise.
8163 (iwmmxt_wunpckihw): Likewise.
8164 (iwmmxt_wunpckilb): Likewise.
8165 (iwmmxt_wunpckilh): Likewise.
8166 (iwmmxt_wunpckilw): Likewise.
8167 (iwmmxt_wunpckehub): Likewise.
8168 (iwmmxt_wunpckehuh): Likewise.
8169 (iwmmxt_wunpckehuw): Likewise.
8170 (iwmmxt_wunpckehsb): Likewise.
8171 (iwmmxt_wunpckehsh): Likewise.
8172 (iwmmxt_wunpckehsw): Likewise.
8173 (iwmmxt_wunpckelub): Likewise.
8174 (iwmmxt_wunpckeluh): Likewise.
8175 (iwmmxt_wunpckeluw): Likewise.
8176 (iwmmxt_wunpckelsb): Likewise.
8177 (iwmmxt_wunpckelsh): Likewise.
8178 (iwmmxt_wunpckelsw): Likewise.
8179 (ror<mode>3): Likewise.
8180 (ashr<mode>3_iwmmxt): Likewise.
8181 (lshr<mode>3_iwmmxt): Likewise.
8182 (ashl<mode>3_iwmmxt): Likewise.
8183 (ror<mode>3_di): Likewise.
8184 (ashr<mode>3_di): Likewise.
8185 (lshr<mode>3_di): Likewise.
8186 (ashl<mode>3_di): Likewise.
8187 (iwmmxt_wmadds): Likewise.
8188 (iwmmxt_wmaddu): Likewise.
8189 (iwmmxt_tmia): Likewise.
8190 (iwmmxt_tmiaph): Likewise.
8191 (iwmmxt_tmiabb): Likewise.
8192 (iwmmxt_tmiatb): Likewise.
8193 (iwmmxt_tmiabt): Likewise.
8194 (iwmmxt_tmiatt): Likewise.
8195 (iwmmxt_tmovmskb): Likewise.
8196 (iwmmxt_tmovmskh): Likewise.
8197 (iwmmxt_tmovmskw): Likewise.
8198 (iwmmxt_waccb): Likewise.
8199 (iwmmxt_wacch): Likewise.
8200 (iwmmxt_waccw): Likewise.
8201 (iwmmxt_waligni): Likewise.
8202 (iwmmxt_walignr): Likewise.
8203 (iwmmxt_walignr0): Likewise.
8204 (iwmmxt_walignr1): Likewise.
8205 (iwmmxt_walignr2): Likewise.
8206 (iwmmxt_walignr3): Likewise.
8207 (iwmmxt_wsadb): Likewise.
8208 (iwmmxt_wsadh): Likewise.
8209 (iwmmxt_wsadbz): Likewise.
8210 (iwmmxt_wsadhz): Likewise.
8211 * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
8212 (iwmmxt_wabsdiffb): Likewise.
8213 (iwmmxt_wabsdiffh): Likewise.
8214 (iwmmxt_wabsdiffw): Likewise.
8215 (iwmmxt_waddsubhx): Likewise
8216 (iwmmxt_wsubaddhx): Likewise.
8217 (addc<mode>3): Likewise.
8218 (iwmmxt_avg4): Likewise.
8219 (iwmmxt_avg4r): Likewise.
8220 (iwmmxt_wmaddsx): Likewise.
8221 (iwmmxt_wmaddux): Likewise.
8222 (iwmmxt_wmaddsn): Likewise.
8223 (iwmmxt_wmaddun): Likewise.
8224 (iwmmxt_wmulwsm): Likewise.
8225 (iwmmxt_wmulwum): Likewise.
8226 (iwmmxt_wmulsmr): Likewise.
8227 (iwmmxt_wmulumr): Likewise.
8228 (iwmmxt_wmulwsmr): Likewise.
8229 (iwmmxt_wmulwumr): Likewise.
8230 (iwmmxt_wmulwl): Likewise.
8231 (iwmmxt_wqmulm): Likewise.
8232 (iwmmxt_wqmulwm): Likewise.
8233 (iwmmxt_wqmulmr): Likewise.
8234 (iwmmxt_wqmulwmr): Likewise.
8235 (iwmmxt_waddbhusm): Likewise.
8236 (iwmmxt_waddbhusl): Likewise.
8237 (iwmmxt_wqmiabb): Likewise.
8238 (iwmmxt_wqmiabt): Likewise.
8239 (iwmmxt_wqmiatb): Likewise.
8240 (iwmmxt_wqmiatt): Likewise.
8241 (iwmmxt_wqmiabbn): Likewise.
8242 (iwmmxt_wqmiabtn): Likewise.
8243 (iwmmxt_wqmiatbn): Likewise.
8244 (iwmmxt_wqmiattn): Likewise.
8245 (iwmmxt_wmiabb): Likewise.
8246 (iwmmxt_wmiabt): Likewise.
8247 (iwmmxt_wmiatb): Likewise.
8248 (iwmmxt_wmiatt): Likewise.
8249 (iwmmxt_wmiabbn): Likewise.
8250 (iwmmxt_wmiabtn): Likewise.
8251 (iwmmxt_wmiatbn): Likewise.
8252 (iwmmxt_wmiattn): Likewise.
8253 (iwmmxt_wmiawbb): Likewise.
8254 (iwmmxt_wmiawbt): Likewise.
8255 (iwmmxt_wmiawtb): Likewise.
8256 (iwmmxt_wmiawtt): Likewise.
8257 (iwmmxt_wmiawbbn): Likewise.
8258 (iwmmxt_wmiawbtn): Likewise.
8259 (iwmmxt_wmiawtbn): Likewise.
8260 (iwmmxt_wmiawttn): Likewise.
8261 (iwmmxt_wmerge): Likewise.
8262 (iwmmxt_tandc<mode>3): Likewise.
8263 (iwmmxt_torc<mode>3): Likewise.
8264 (iwmmxt_torvsc<mode>3): Likewise.
8265 (iwmmxt_textrc<mode>3): Likewise.
8266 * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
8267 (wmmxt_pack): Likewise.
8268 (wmmxt_mult_c1): Likewise.
8269 (wmmxt_mult_c2): Likewise.
8270 (wmmxt_alu_c1): Likewise.
8271 (wmmxt_alu_c2): Likewise.
8272 (wmmxt_alu_c3): Likewise.
8273 (wmmxt_transfer_c1): Likewise.
8274 (wmmxt_transfer_c2): Likewise.
8275 (wmmxt_transfer_c3): Likewise.
8276 (marvell_f_iwmmxt_wstr): Likewise.
8277 (marvell_f_iwmmxt_wldr): Likewise.
8279 2013-06-29 Yufeng Zhang <yufeng.zhang@arm.com>
8281 * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
8283 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
8286 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
8287 * lra-constraints.c (need_for_split_p): Check call used hard regs
8288 living through calls.
8290 * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
8291 call used regs for call insn.
8293 2013-06-28 Jakub Jelinek <jakub@redhat.com>
8296 * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
8297 mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
8299 2013-06-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
8301 * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
8303 2013-06-28 Vladimir Makarov <vmakarov@redhat.com>
8305 * lra-constraints.c (need_for_split_p): Check call used hard regs
8306 living through calls.
8308 2013-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8311 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
8312 to tie with any other modes. Eliminate Altivec vector mode tests,
8313 since these are a subset of ALTIVEC or VSX vector modes. Simplify
8314 code, to return 0 if testing MODE2 for a condition, if we've
8315 already tested MODE1 for the same condition.
8317 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
8319 * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
8322 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
8324 * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
8325 Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
8327 2013-06-28 Marcus Shawcroft <marcus.shawcroft@arm.com>
8329 * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
8331 (aarch64_symbolic_constant_p): Remove.
8332 * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
8333 static. Fix line length and white space.
8334 (aarch64_symbolic_constant_p): Remove.
8335 * config/aarch64/predicates.md (aarch64_valid_symref):
8336 Use aarch64_classify_symbol_expression.
8338 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8340 * config/arm/constraints.md (Ts): New constraint.
8341 * config/arm/arm.md (arm_movqi_insn): Add alternatives for
8343 (compare_scc): Use "Ts" constraint for operand 0.
8344 (ior_scc_scc): Likewise.
8345 (and_scc_scc): Likewise.
8346 (and_scc_scc_nodom): Likewise.
8347 (ior_scc_scc_cmp): Likewise for operand 7.
8348 (and_scc_scc_cmp): Likewise.
8349 * config/arm/thumb2.md (thumb2_movsi_insn):
8350 Add alternatives for 16-bit encodings.
8351 (thumb2_movhi_insn): Likewise.
8352 (thumb2_movsicc_insn): Likewise.
8353 (thumb2_and_scc): Take 'and' outside cond_exec. Use "Ts" constraint.
8354 (thumb2_negscc): Use "Ts" constraint.
8355 Move mvn instruction outside cond_exec block.
8356 * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
8357 for 16-bit encodings.
8359 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8361 * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
8363 (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
8364 (mulsi3subsi): Likewise.
8365 (mulsidi3adddi): Likewise.
8366 (mulsidi3_v6): Likewise.
8367 (umulsidi3_v6): Likewise.
8368 (umulsidi3adddi_v6): Likewise.
8369 (smulsi3_highpart_v6): Likewise.
8370 (umulsi3_highpart_v6): Likewise.
8371 (mulhisi3tb): Likewise.
8372 (mulhisi3bt): Likewise.
8373 (mulhisi3tt): Likewise.
8374 (maddhisi4): Likewise.
8375 (maddhisi4tb): Likewise.
8376 (maddhisi4tt): Likewise.
8377 (maddhidi4): Likewise.
8378 (maddhidi4tb): Likewise.
8379 (maddhidi4tt): Likewise.
8380 (zeroextractsi_compare0_scratch): Likewise.
8381 (insv_zero): Likewise.
8382 (insv_t2): Likewise.
8383 (anddi_notzesidi_di): Likewise.
8384 (anddi_notsesidi_di): Likewise.
8385 (andsi_notsi_si): Likewise.
8386 (iordi_zesidi_di): Likewise.
8387 (xordi_zesidi_di): Likewise.
8388 (andsi_iorsi3_notsi): Likewise.
8390 (smax_m1): Likewise.
8392 (not_shiftsi): Likewise.
8393 (unaligned_loadsi): Likewise.
8394 (unaligned_loadhis): Likewise.
8395 (unaligned_loadhiu): Likewise.
8396 (unaligned_storesi): Likewise.
8397 (unaligned_storehi): Likewise.
8398 (extv_reg): Likewise.
8399 (extzv_t2): Likewise.
8401 (udivsi3): Likewise.
8402 (arm_zero_extendhisi2addsi): Likewise.
8403 (arm_zero_extendqisi2addsi): Likewise.
8404 (compareqi_eq0): Likewise.
8405 (arm_extendhisi2_v6): Likewise.
8406 (arm_extendqisi2addsi): Likewise.
8407 (arm_movt): Likewise.
8408 (thumb2_ldrd): Likewise.
8409 (thumb2_ldrd_base): Likewise.
8410 (thumb2_ldrd_base_neg): Likewise.
8411 (thumb2_strd): Likewise.
8412 (thumb2_strd_base): Likewise.
8413 (thumb2_strd_base_neg): Likewise.
8414 (arm_negsi2): Add alternative for 16-bit encoding.
8415 (arm_one_cmplsi2): Likewise.
8417 2013-06-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8419 * config/arm/predicates.md (arm_cond_move_operator): New predicate.
8420 * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
8421 (movdfcc): Likewise.
8422 * config/arm/vfp.md (*thumb2_movsf_vfp):
8423 Disable predication for arm_restrict_it.
8424 (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
8425 (*thumb2_movdfcc_vfp): Likewise.
8426 (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
8427 *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
8428 *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
8429 *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
8430 *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
8431 *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
8432 *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
8433 *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
8434 *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
8435 *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
8436 *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
8437 *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
8438 Disable predication for arm_restrict_it.
8440 2013-06-28 Kirill Yukhin <kirill.yukhin@intel.com>
8442 * config/i386/bmiintrin.h (_bextr_u32): New.
8443 (_bextr_u64): Ditto.
8445 2013-06-27 Richard Sandiford <rdsandiford@googlemail.com>
8447 * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
8448 (mips64r5900el-*-elf*): Include mips/n32-elf.h.
8449 * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
8450 (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
8451 * config/mips/n32-elf.h: ...this new file.
8453 2013-06-27 Marc Glisse <marc.glisse@inria.fr>
8456 * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
8457 IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
8459 2013-06-27 Catherine Moore <clm@codesourcery.com>
8461 * config/mips/mips-tables.opt: Regenerate.
8462 * config/mips/mips-cpus.def: Add m14ke and m14kec.
8463 * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
8464 * doc/invoke.texi: Add -m14kc.
8466 2013-06-27 Jakub Jelinek <jakub@redhat.com>
8469 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
8470 constraints of operand 1 and 2.
8473 * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
8474 to match RTL canonicalization. Swap predicates and
8475 constraints of operand 1 and 2.
8477 2013-06-27 Vladimir Makarov <vmakarov@redhat.com>
8479 * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
8480 Process OP_INOUT regs for splitting too.
8482 2013-06-27 Jakub Jelinek <jakub@redhat.com>
8484 * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
8485 decl before the loop, initialize to NULL.
8486 (vectorizable_load): Initialize ptr_incr to NULL.
8488 2013-06-27 Martin Jambor <mjambor@suse.cz>
8491 * ipa-ref.h (ipa_maybe_record_reference): Declare.
8492 * ipa-ref.c (ipa_maybe_record_reference): New function.
8493 * cgraphclones.c (cgraph_create_virtual_clone): Use it.
8494 * ipa-cp.c (create_specialized_node): Record potential references from
8496 * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
8498 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
8500 * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
8501 parameter 'mode' of type 'enum machine_mode mode'; change to pass
8502 'mode' to force_reg.
8503 (aarch64_add_offset): Update calls to aarch64_force_temporary.
8504 (aarch64_expand_mov_immediate): Likewise.
8506 2013-06-27 Yufeng Zhang <yufeng.zhang@arm.com>
8508 * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
8509 'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
8511 2013-06-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8513 * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
8514 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8515 (struct machine_function): Add tbegin_p.
8516 (s390_canonicalize_comparison): Fold CC mode compares to
8517 conditional jump if possible.
8518 (s390_emit_jump): Return the emitted jump.
8519 (s390_branch_condition_mask, s390_branch_condition_mnemonic):
8520 Handle CCRAWmode compares.
8521 (s390_option_override): Default to -mhtm if available.
8522 (s390_reg_clobbered_rtx): Handle floating point regs as well.
8523 (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
8524 FPRs instead of df_regs_ever_live_p.
8525 (s390_optimize_nonescaping_tx): New function.
8526 (s390_init_frame_layout): Extend clobbered_regs array to cover
8528 (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
8529 (s390_expand_tbegin): New function.
8530 (enum s390_builtin): New enum definition.
8531 (code_for_builtin): New array definition.
8532 (s390_init_builtins): New function.
8533 (s390_expand_builtin): New function.
8534 (TARGET_INIT_BUILTINS): Define.
8535 (TARGET_EXPAND_BUILTIN): Define.
8536 * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
8537 * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
8538 (s390_alc_comparison): Likewise.
8539 * config/s390/s390-modes.def: Add CCRAWmode.
8540 * config/s390/s390.h (processor_flags): Add PF_TX.
8541 (TARGET_CPU_HTM): Define macro.
8542 (TARGET_HTM): Define macro.
8543 (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
8544 * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
8545 UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
8546 (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
8547 (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
8548 (TBEGIN_MASK, TBEGINC_MASK): New constants.
8549 ("*cc_to_int"): Move up.
8550 ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
8551 constants other than 0.
8552 ("*ccraw_to_int"): New insn and splitter definition.
8553 ("tbegin", "tbegin_nofloat", "tbegin_retry")
8554 ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
8555 ("tx_assist"): New expander.
8556 ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
8557 ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
8558 * config/s390/s390.opt: Add -mhtm option.
8559 * config/s390/s390-protos.h (s390_emit_jump): Add return type.
8560 * config/s390/htmxlintrin.h: New file.
8561 * config/s390/htmintrin.h: New file.
8562 * config/s390/s390intrin.h: New file.
8563 * doc/extend.texi: Document htm builtins.
8564 * config.gcc: Add the new header files to extra_headers.
8566 2013-06-26 Thomas Schwinge <thomas@codesourcery.com>
8568 * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
8569 (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
8571 2013-06-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8572 Pat Haugen <pthaugen@us.ibm.com>
8573 Peter Bergner <bergner@vnet.ibm.com>
8575 * config/rs6000/power8.md: New.
8576 * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
8577 setting for power8 entry.
8578 * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
8579 * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
8580 test for Power4/Power5 only.
8581 (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
8583 (force_new_group): Adjust comment.
8584 * config/rs6000/rs6000.md: Include power8.md.
8586 2013-06-26 Greta Yorsh <Greta.Yorsh@arm.com>
8588 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
8589 * config/arm/arm-protos.h (arm_max_conditional_execute): New
8591 (tune_params): Update comment.
8592 * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
8593 (arm_max_conditional_execute): New function.
8594 (thumb2_final_prescan_insn): Use max_insn_skipped and
8595 MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
8597 2013-06-25 Jakub Jelinek <jakub@redhat.com>
8599 PR tree-optimization/57705
8600 * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
8601 SSA_NAME step, provided that it is not defined inside the loop.
8602 (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
8603 (get_initial_def_for_induction): Handle SSA_NAME IV step.
8605 2013-06-25 Martin Jambor <mjambor@suse.cz>
8608 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
8609 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
8611 (ipa_note_param_call): Initialize member_ptr flag.
8612 (ipa_analyze_indirect_call_uses): Set member_ptr flag.
8613 (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
8614 (ipa_write_indirect_edge_info): Stream member_ptr flag.
8615 (ipa_read_indirect_edge_info): Likewise.
8617 2013-06-25 Richard Biener <rguenther@suse.de>
8620 * passes.c (init_optimization_passes): Move pass_fold_builtins
8621 and pass_dce earlier with -Og.
8623 2013-06-25 Eric Botcazou <ebotcazou@adacore.com>
8625 * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
8626 <BIT_FIELD_REF>: Remove trailing TAB.
8627 * varasm.c (output_constructor_bitfield): Fix formatting glitch and
8630 2013-06-24 Martin Jambor <mjambor@suse.cz>
8632 PR tree-optimization/57358
8633 * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
8634 (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
8635 (ipa_analyze_params_uses): Generate pessimistic info when true.
8637 2013-06-24 Martin Jambor <mjambor@suse.cz>
8639 PR tree-optimization/57539
8640 * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
8641 global.inlined_to of the new node to it. All callers changed.
8642 * ipa-inline-transform.c (clone_inlined_nodes): New variable
8643 inlining_into, pass it to cgraph_clone_node.
8644 * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
8645 ipa_free_edge_args_substructures.
8646 (ipa_edge_duplication_hook): Only add edges from inlined nodes to
8647 rdesc linked list. Do not assert rdesc edges have inlined caller.
8648 Assert we have found an rdesc in the rdesc list.
8650 2013-06-24 Richard Biener <rguenther@suse.de>
8652 * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
8653 (pointer_set_lookup): Declare.
8654 (class pointer_map): New template class implementing a
8655 generic pointer to T map.
8656 (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
8657 pointer_map<T>::contains, pointer_map<T>::insert,
8658 pointer_map<T>::traverse): New functions.
8659 * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
8660 (pointer_set_lookup): New function.
8661 (pointer_set_contains): Use pointer_set_lookup.
8662 (pointer_set_insert): Likewise.
8663 (insert_aux): Remove.
8664 (struct pointer_map_t): Embed a pointer_set_t.
8665 (pointer_map_create): Adjust.
8666 (pointer_map_destroy): Likewise.
8667 (pointer_map_contains): Likewise.
8668 (pointer_map_insert): Likewise.
8669 (pointer_map_traverse): Likewise.
8670 * tree-streamer.h (struct streamer_tree_cache_d): Use a
8671 pointer_map<unsigned> instead of a pointer_map_t.
8672 * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
8673 (streamer_tree_cache_lookup): Likewise.
8674 (streamer_tree_cache_create): Likewise.
8675 (streamer_tree_cache_delete): Likewise.
8676 * lto-streamer.h (struct lto_tree_ref_encoder): Use a
8677 pointer_map<unsigned> instead of a pointer_map_t.
8678 (lto_init_tree_ref_encoder): Adjust.
8679 (lto_destroy_tree_ref_encoder): Likewise.
8680 * lto-section-out.c (lto_output_decl_index): Likewise.
8681 (lto_record_function_out_decl_state): Likewise.
8682 * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
8684 2013-06-24 Richard Biener <rguenther@suse.de>
8686 PR tree-optimization/57488
8687 * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
8689 2013-06-24 Alan Modra <amodra@gmail.com>
8691 * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
8692 (gen_easy_altivec_constant): Likewise.
8693 * config/rs6000/predicates.md (easy_vector_constant_add_self,
8694 easy_vector_constant_msb): Likewise.
8696 2013-06-23 Jakub Jelinek <jakub@redhat.com>
8699 * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
8700 add missing return true.
8702 2013-06-23 Oleg Endo <olegendo@gcc.gnu.org>
8705 * config/sh/predicates.md (general_extend_operand): Invoke
8706 general_movsrc_operand for memory operands.
8707 (general_movsrc_operand): Allow reg+reg addressing, do not use
8708 general_operand for memory operands.
8710 2013-06-23 Sriraman Tallam <tmsriram@google.com>
8712 * config/i386/i386.c (ix86_pragma_target_parse): Restore target
8713 when current target options does not apply.
8714 * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
8715 * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
8716 * config/i386/bmiintrin.h: Pass appropriate target
8717 attributes to header.
8718 * config/i386/mmintrin.h: Ditto.
8719 * config/i386/nmmintrin.h: Ditto.
8720 * config/i386/avx2intrin.h: Ditto.
8721 * config/i386/fxsrintrin.h: Ditto.
8722 * config/i386/tbmintrin.h: Ditto.
8723 * config/i386/xsaveintrin.h: Ditto.
8724 * config/i386/f16cintrin.h: Ditto.
8725 * config/i386/xtestintrin.h: Ditto.
8726 * config/i386/xsaveoptintrin.h: Ditto.
8727 * config/i386/bmi2intrin.h: Ditto.
8728 * config/i386/lzcntintrin.h: Ditto.
8729 * config/i386/smmintrin.h: Ditto.
8730 * config/i386/wmmintrin.h: Ditto.
8731 * config/i386/x86intrin.h: Remove all header include guards.
8732 * config/i386/prfchwintrin.h: Ditto.
8733 * config/i386/pmmintrin.h: Ditto.
8734 * config/i386/tmmintrin.h: Ditto.
8735 * config/i386/xmmintrin.h: Ditto.
8736 * config/i386/popcntintrin.h: Ditto.
8737 * config/i386/rdseedintrin.h: Ditto.
8738 * config/i386/ammintrin.h: Ditto.
8739 * config/i386/emmintrin.h: Ditto.
8740 * config/i386/immintrin.h: Remove all header include guards.
8741 * config/i386/fma4intrin.h: Ditto.
8742 * config/i386/lwpintrin.h: Ditto.
8743 * config/i386/xopintrin.h: Ditto.
8744 * config/i386/ia32intrin.h: Ditto.
8745 * config/i386/avxintrin.h: Ditto.
8746 * config/i386/rtmintrin.h: Ditto.
8747 * config/i386/fmaintrin.h: Ditto.
8748 * config/i386/mm3dnow.h: Ditto.
8750 2013-06-22 Sriraman Tallam <tmsriram@google.com>
8752 * common/config/i386/i386-common.c: Handle LZCNT.
8754 2013-06-22 Andi Kleen <ak@linux.intel.com>
8756 * doc/extend.texi: Use __atomic_store_n instead of
8757 __atomic_store in HLE example.
8759 2013-06-22 Oleg Endo <olegendo@gcc.gnu.org>
8761 * config/sh/sh.c: Remove <cstdlib> workaround.
8763 2013-06-21 Andi Kleen <ak@linux.intel.com>
8765 * doc/extend.texi: Dont use __atomic_clear in HLE example. Fix typo.
8767 2013-06-21 Andi Kleen <ak@linux.intel.com>
8769 * doc/extend.texi: Document that __atomic_clear and
8770 __atomic_test_and_set should only be used with bool.
8772 2013-06-20 Jan Hubicka <jh@suse.cz>
8774 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
8776 * tree.c (decls_same_for_odr): New function.
8777 (same_for_edr): New function.
8778 (types_same_for_odr): New function.
8779 (get_binfo_at_offset): Use it.
8780 * tree.h (types_same_for_odr): Declare.
8782 2013-06-20 Oleg Endo <olegendo@gcc.gnu.org>
8783 Jason Merrill <jason@redhat.com>
8785 * system.h: Include <cstdlib> as well as <stdlib.h>.
8787 2013-06-20 Uros Bizjak <ubizjak@gmail.com>
8790 * config/i386/i386.c (construct_container): Report error if
8791 long double is used with disabled x87 float returns.
8793 2013-06-20 Jan Hubicka <jh@suse.cz>
8795 * lto-cgraph.c (input_symtab): Do not set cgraph state.
8797 2013-06-20 Joern Rennecke <joern.rennecke@embecosm.com>
8799 PR rtl-optimization/57425
8800 PR rtl-optimization/57569
8801 * alias.c (write_dependence_p): Remove parameters mem_mode and
8802 canon_mem_addr. Add parameters x_mode, x_addr and x_canonicalized.
8803 Changed all callers.
8804 (canon_anti_dependence): Get comments and semantics in sync.
8805 Add parameter mem_canonicalized. Changed all callers.
8806 * rtl.h (canon_anti_dependence): Update prototype.
8808 2013-06-20 Richard Biener <rguenther@suse.de>
8810 * data-streamer-in.c (streamer_read_uhwi): Optimize single
8811 byte case, inline streamer_read_uchar and defer section
8814 2013-06-20 Richard Biener <rguenther@suse.de>
8816 PR tree-optimization/57584
8817 * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
8818 SSA names into the expanded expression that take part in
8819 abnormal coalescing.
8821 2013-06-19 Sharad Singhai <singhai@google.com>
8823 * gcov.c (print_usage): Handle new option.
8824 (process_args): Ditto.
8825 (get_gcov_intermediate_filename): New function.
8826 (output_intermediate_file): New function.
8827 (output_gcov_file): New function
8828 (generate_results): Handle new option.
8829 (release_function): Relase demangled name.
8830 (read_graph_file): Handle demangled name.
8831 (output_lines): Ditto.
8832 * doc/gcov.texi: Document gcov intermediate format.
8834 2013-06-19 Vladimir Makarov <vmakarov@redhat.com>
8837 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
8838 (lra_emit_add): Use the functions. Add comment about Y as an
8841 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
8844 * collect2.c (collect_atexit): New.
8845 (collect_exit): Delete.
8846 (main): Register collect_atexit with atexit.
8847 (collect_wait): Change collect_exit to exit.
8849 * collect2.h (collect_exit): Delete.
8850 * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
8852 2013-06-19 Wei Mi <wmi@google.com>
8854 PR rtl-optimization/57518
8855 * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
8856 if regno is used in paradoxical subreg.
8857 (update_equiv_regs): Check pdx_subregs[regno] before
8858 set a reg to be equivalent with a mem.
8860 2013-06-19 Matthias Klose <doko@ubuntu.com>
8863 * file-find.h (find_a_file): Add a mode parameter.
8864 * file-find.c (find_a_file): Likewise.
8865 * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
8866 with X_OK for the executables.
8867 * collect2.c (main): Call find_a_file with X_OK.
8869 2013-06-19 Steve Ellcey <sellcey@mips.com>
8872 * config/mips/mips.md (casesi_internal_mips16_<mode>):
8873 Use NEXT_INSN instead of next_real_insn.
8875 2013-06-19 Jan Hubicka <jh@suse.cz>
8877 * cgraph.h (const_value_known_p): Replace by ...
8878 (ctor_for_folding): .. this one.
8879 * cgraphunit.c (process_function_and_variable_attributes): Use it.
8880 * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
8881 * expr.c (expand_expr_real_1): Likewise.
8882 (string_constant): Likewise.
8883 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
8884 * ipa.c (process_references): Likewise.
8885 (symtab_remove_unreachable_nodes): Likewise.
8886 * ipa-inline-analysis.c (param_change_prob): Likewise.
8887 * gimple-fold.c (canonicalize_constructor_val): Likewise.
8888 (get_base_constructor): Likwise.
8889 * varpool.c (varpool_remove_node): Likewise.
8890 (varpool_remove_initializer): LIkewise.
8891 (dump_varpool_node): LIkwise.
8892 (const_value_known_p): Rewrite to ...
8893 (ctor_for_folding): ... this one.
8895 2013-06-19 Jakub Jelinek <jakub@redhat.com>
8898 * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
8899 PERSONALITY in $PATH derived prefixes.
8901 2013-06-19 Jeff Law <law@redhat.com>
8903 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
8906 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
8907 (simplify_bitwise_binary): Use it to simpify certain binary ops on
8910 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
8912 * config/arm/vfp.md: Move VFP instruction classification documentation
8914 * config/arm/arm.md: ... here. Update instruction classification
8917 2013-06-19 Richard Earnshaw <rearnsha@arm.com>
8919 arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
8920 (peepholes for eq(reg, not-0)): Ensure condition register is dead after
8921 pattern. Use more efficient sequences on ARMv5 and Thumb2.
8923 2013-06-19 Steven Bosscher <steven@gcc.gnu.org>
8926 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8927 with NEXT_INSN. Use tablejump_p to check for jump table data
8930 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
8933 * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
8934 that now in C++ the value is correct per the C++ standards.
8936 2013-06-19 Richard Biener <rguenther@suse.de>
8938 * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
8941 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8944 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8947 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
8948 with next_active_insn.
8950 2013-06-18 Sriraman Tallam <tmsriram@google.com>
8952 * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
8953 functions are inlined during failures to flag an error.
8954 * tree-inline.c (expand_call_inline): Allow the error to be flagged
8955 in early inline pass.
8957 2013-06-18 H.J. Lu <hongjiu.lu@intel.com>
8959 * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
8962 2013-06-18 Julian Brown <julian@codesourcery.com>
8964 * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
8965 Permit virtual register pre-reload if !strict.
8966 (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
8968 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
8970 * config/arm/neon.md (movmisalign<mode>): Use
8971 neon_perm_struct_or_reg_operand instead of
8972 neon_struct_or_register_operand.
8973 (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
8974 neon_permissive_struct_operand instead of neon_struct_operand.
8975 * config/arm/constraints.md (Un, Um, Us): Adjust calls to
8976 neon_vector_mem_operand.
8977 * config/arm/predicates.md (neon_struct_operand): Adjust call to
8978 neon_vector_mem_operand.
8979 (neon_permissive_struct_operand): New.
8980 (neon_struct_or_register_operand): Rename to...
8981 (neon_perm_struct_or_reg_operand): This. Adjust call to
8982 neon_vector_mem_operand.
8984 2013-06-18 Richard Biener <rguenther@suse.de>
8986 * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
8987 * lto-streamer.h: Include pointer-set.h.
8988 (struct lto_decl_slot): Remove.
8989 (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
8990 Remove next_index entry.
8991 (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8992 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8993 (lto_init_tree_ref_encoder): Adjust.
8994 (lto_destroy_tree_ref_encoder): Likewise.
8995 * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
8996 lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
8997 (lto_output_decl_index): Adjust.
8998 (lto_new_out_decl_state): Likewise.
8999 (lto_record_function_out_decl_state): Likewise.
9000 * lto-streamer-out.c (copy_function): Likewise.
9002 2013-06-18 Richard Biener <rguenther@suse.de>
9004 * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
9005 * cgraphunit.c: Include cfgloop.h.
9006 (init_lowered_empty_function): Initialize the loop tree.
9007 (assemble_thunk): Insert new BBs into loops.
9009 2013-06-18 Richard Biener <rguenther@suse.de>
9011 * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
9012 * tree-streamer.c (streamer_tree_cache_create): Make maintaining
9013 the map from cache entry to cache index optional.
9014 (streamer_tree_cache_replace_tree): Adjust accordingly.
9015 (streamer_tree_cache_append): Likewise.
9016 (streamer_tree_cache_delete): Likewise.
9017 * lto-streamer-in.c (lto_data_in_create): Do not maintain the
9018 streamer cache map from cache entry to cache index.
9019 * lto-streamer-out.c (create_output_block): Adjust.
9021 2013-06-18 Sofiane Naci <sofiane.naci@arm.com>
9023 * config/arm/arm.md (attribute "insn"): Move multiplication and
9024 division attributes to...
9025 (attribute "type"): ... here. Remove mult.
9026 (attribute "mul32"): New attribute.
9027 (attribute "mul64"): Add umaal.
9028 (*arm_mulsi3): Update attributes.
9029 (*arm_mulsi3_v6): Likewise.
9030 (*thumb_mulsi3): Likewise.
9031 (*thumb_mulsi3_v6): Likewise.
9032 (*mulsi3_compare0): Likewise.
9033 (*mulsi3_compare0_v6): Likewise.
9034 (*mulsi_compare0_scratch): Likewise.
9035 (*mulsi_compare0_scratch_v6): Likewise.
9036 (*mulsi3addsi): Likewise.
9037 (*mulsi3addsi_v6): Likewise.
9038 (*mulsi3addsi_compare0): Likewise.
9039 (*mulsi3addsi_compare0_v6): Likewise.
9040 (*mulsi3addsi_compare0_scratch): Likewise.
9041 (*mulsi3addsi_compare0_scratch_v6): Likewise.
9042 (*mulsi3subsi): Likewise.
9043 (*mulsidi3adddi): Likewise.
9044 (*mulsi3addsi_v6): Likewise.
9045 (*mulsidi3adddi_v6): Likewise.
9046 (*mulsidi3_nov6): Likewise.
9047 (*mulsidi3_v6): Likewise.
9048 (*umulsidi3_nov6): Likewise.
9049 (*umulsidi3_v6): Likewise.
9050 (*umulsidi3adddi): Likewise.
9051 (*umulsidi3adddi_v6): Likewise.
9052 (*smulsi3_highpart_nov6): Likewise.
9053 (*smulsi3_highpart_v6): Likewise.
9054 (*umulsi3_highpart_nov6): Likewise.
9055 (*umulsi3_highpart_v6): Likewise.
9056 (mulhisi3): Likewise.
9057 (*mulhisi3tb): Likewise.
9058 (*mulhisi3bt): Likewise.
9059 (*mulhisi3tt): Likewise.
9060 (maddhisi4): Likewise.
9061 (*maddhisi4tb): Likewise.
9062 (*maddhisi4tt): Likewise.
9063 (maddhidi4): Likewise.
9064 (*maddhidi4tb): Likewise.
9065 (*maddhidi4tt): Likewise.
9067 (udivsi3): Likewise.
9068 * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
9069 (thumb2_mulsi_short_compare0): Likewise.
9070 (thumb2_mulsi_short_compare0_scratch): Likewise.
9071 * config/arm/arm1020e.md (1020mult1): Update attribute change.
9072 (1020mult2): Likewise.
9073 (1020mult3): Likewise.
9074 (1020mult4): Likewise.
9075 (1020mult5): Likewise.
9076 (1020mult6): Likewise.
9077 * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
9079 (cortex_a15_mult64): Likewise.
9080 (cortex_a15_sdiv): Likewise.
9081 (cortex_a15_udiv): Likewise.
9082 * config/arm/arm1026ejs.md (mult1): Update attribute change.
9088 * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
9089 (pj4_ir_div): Likewise.
9090 * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
9091 (11_mult2): Likewise.
9092 (11_mult3): Likewise.
9093 (11_mult4): Likewise.
9094 (11_mult5): Likewise.
9095 (11_mult6): Likewise.
9096 (11_mult7): Likewise.
9097 * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
9098 (cortex_a8_mla): Likewise.
9099 (cortex_a8_mull): Likewise.
9100 (cortex_a8_smulwy): Likewise.
9101 (cortex_a8_smlald): Likewise.
9102 * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
9103 * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
9104 (cortex_r4_mul_3): Likewise.
9105 (cortex_r4_mla_4): Likewise.
9106 (cortex_r4_mla_3): Likewise.
9107 (cortex_r4_smlald): Likewise.
9108 (cortex_r4_mull): Likewise.
9109 (cortex_r4_sdiv): Likewise.
9110 (cortex_r4_udiv): Likewise.
9111 * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
9112 (cortex_a7_idiv): Likewise.
9113 * config/arm/arm926ejs.md (9_mult1): Update attribute change.
9114 (9_mult2): Likewise.
9115 (9_mult3): Likewise.
9116 (9_mult4): Likewise.
9117 (9_mult5): Likewise.
9118 (9_mult6): Likewise.
9119 * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
9120 (cortex_a53_sdiv): Likewise.
9121 (cortex_a53_udiv): Likewise.
9122 * config/arm/fa726te.md (726te_mult_op): Update attribute change.
9123 * config/arm/fmp626.md (mp626_mult1): Update attribute change.
9124 (mp626_mult2): Likewise.
9125 (mp626_mult3): Likewise.
9126 (mp626_mult4): Likewise.
9127 * config/arm/fa526.md (526_mult1): Update attribute change.
9128 (526_mult2): Likewise.
9129 * config/arm/arm-generic.md (mult): Update attribute change.
9130 (mult_ldsched_strongarm): Likewise.
9131 (mult_ldsched): Likewise.
9132 (multi_cycle): Likewise.
9133 * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
9134 * config/arm/fa606te.md (606te_mult1): Update attribute change.
9135 (606te_mult2): Likewise.
9136 (606te_mult3): Likewise.
9137 (606te_mult4): Likewise.
9138 * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
9139 (cortex_a9_mac16): Likewise.
9140 (cortex_a9_multiply): Likewise.
9141 (cortex_a9_mac): Likewise.
9142 (cortex_a9_multiply_long): Likewise.
9143 * config/arm/fa626te.md (626te_mult1): Update attribute change.
9144 (626te_mult2): Likewise.
9145 (626te_mult3): Likewise.
9146 (626te_mult4): Likewise.
9148 2013-06-18 Richard Biener <rguenther@suse.de>
9151 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
9153 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9156 * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
9157 with next_active_insn.
9159 2013-06-18 Alan Modra <amodra@gmail.com>
9161 * config/rs6000/rs6000.h (enum data_align): New.
9162 (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
9163 (DATA_ABI_ALIGNMENT): Define.
9164 (CONSTANT_ALIGNMENT): Correct comment.
9165 * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
9166 * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
9168 2013-06-17 David Malcolm <dmalcolm@redhat.com>
9170 * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
9171 ATTRIBUTE_UNUSED marking.
9173 2013-06-17 Sofiane Naci <sofiane.naci@arm.com>
9175 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
9176 alternative and update.
9177 (aarch64_dup_lanedi): Delete.
9178 * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
9179 * config/aarch64/aarch64-simd-builtins.def: Update.
9181 2013-06-17 Richard Biener <rguenther@suse.de>
9183 * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
9184 (lto_input_scc): Declare.
9185 (lto_input_tree_1): Likewise.
9186 (struct lto_stats_d): Add num_tree_bodies_output and
9187 num_pickle_refs_output.
9188 * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
9189 (lto_read_tree_1): Split out from ...
9190 (lto_read_tree): ... this.
9191 (lto_input_scc): New function.
9192 (lto_input_tree_1): Split out from ...
9193 (lto_input_tree): ... this. Handle LTO_tree_scc.
9194 (lto_data_in_create): Create the streamer cache without hashes.
9195 * lto-streamer-out.c (create_output_block): Create the streamer
9196 cache with hashes when not doing WPA.
9197 (lto_write_tree_1): Split out from ...
9198 (lto_write_tree): ... this.
9199 (get_symbol_initial_value): New function.
9200 (lto_output_tree_1): Split out from ...
9201 (lto_output_tree): ... this. Write trees as series of SCCs
9202 using a DFS walk via DFS_write_tree.
9203 (struct sccs, struct scc_entry): New types.
9204 (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
9205 (DFS_write_tree_body): New function.
9206 (DFS_write_tree): Likewise.
9207 (hash_tree): Likewise.
9208 (scc_entry_compare): Likewise.
9209 (hash_scc): Likewise.
9210 (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
9211 * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
9212 TREE_CHAIN as regular reference.
9213 (streamer_read_integer_cst): Remove.
9214 (streamer_get_pickled_tree): Adjust.
9215 * tree-streamer-out.c (streamer_write_chain): Disable streaming
9216 of DECL_EXTERNALs in BLOCK_VARS for now.
9217 (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
9219 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
9220 Add hash value argument and record that if hashes are recorded
9222 (streamer_tree_cache_insert_1): Adjust.
9223 (streamer_tree_cache_insert): Likewise.
9224 (streamer_tree_cache_insert_at): Rename to ...
9225 (streamer_tree_cache_replace_tree): ... this and adjust.
9226 (streamer_tree_cache_append): Adjust.
9227 (record_common_node): Likewise.
9228 (streamer_tree_cache_create): Add argument whether to
9229 record hash values together with trees.
9230 (streamer_tree_cache_delete): Adjust.
9231 * tree-streamer.h (struct streamer_tree_cache_d): Add
9233 (streamer_read_integer_cst): Remove.
9234 (streamer_tree_cache_insert): Adjust.
9235 (streamer_tree_cache_append): Likewise.
9236 (streamer_tree_cache_insert_at): Rename to ...
9237 (streamer_tree_cache_replace_tree): ... this and adjust.
9238 (streamer_tree_cache_create): Add argument whether to record hashes.
9239 (streamer_tree_cache_get): Rename to ...
9240 (streamer_tree_cache_get_tree): ... this.
9241 (streamer_tree_cache_get_hash): New function.
9242 * tree.c (cache_integer_cst): New function.
9243 * tree.h (cache_integer_cst): Declare.
9244 (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
9245 * lto-symtab.c (lto_varpool_replace_node): Only release
9246 DECL_INITIAL of non-prevailing decls.
9247 * varpool.c (varpool_remove_initializer): Do not release
9248 DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
9250 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
9252 * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
9253 (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
9254 * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
9255 (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
9256 (mul<mode>3_mul3): Handle TARGET_MIPS5900.
9257 (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
9258 (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
9259 (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
9260 instead of TARGET_64BIT.
9261 (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
9262 Require ISA_HAS_<D>DIV.
9264 2013-06-16 Richard Sandiford <rdsandiford@googlemail.com>
9266 * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
9267 (mips*-*-linux*): Move default with_llsc setting to where other
9269 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
9271 (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
9272 Likewise. Remove default with_tune setting. Move default float
9273 setting to its own block. Handle with_llsc in the same block as above.
9275 2013-06-16 Joern Rennecke <joern.rennecke@embecosm.com>
9277 PR rtl-optimization/57425
9278 PR rtl-optimization/57569
9279 * alias.c (write_dependence_p): Add new parameters mem_mode,
9280 canon_mem_addr and mem_canonicalized. Change type of writep to bool.
9281 Changed all callers.
9282 (canon_anti_dependence): New function.
9283 * cse.c (check_dependence): Use canon_anti_dependence.
9284 * cselib.c (cselib_invalidate_mem): Likewise.
9285 * rtl.h (canon_anti_dependence): Declare.
9287 2013-06-16 Jürgen Urban <JuergenUrban@gmx.de>
9289 * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
9290 * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
9291 ".set mips3" for 64-bit targets.
9293 2013-06-15 Dehao Chen <dehao@google.com>
9295 * tree-flow.h (gimple_check_call_matching_types): Add new argument.
9296 * gimple-low.c (gimple_check_call_matching_types): Likewise.
9297 (gimple_check_call_args): Likewise.
9298 * value-prof.c (check_ic_target): Likewise.
9299 * ipa-inline.c (early_inliner): Likewise.
9300 * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
9301 * cgraph.c (cgraph_create_edge_1): Likewise.
9302 (cgraph_make_edge_direct): Likewise.
9304 2013-06-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9307 * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
9308 rs6000_output_move_128bit to handle emitting quad memory
9309 operations. Set attribute length to 8 bytes.
9311 2013-06-14 Vidya Praveen <vidyapraveen@arm.com>
9313 * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
9315 (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
9316 (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
9317 (aarch64_<su>mlsl<mode>): Likewise.
9319 2013-06-14 Mike Stump <mikestump@comcast.net>
9321 * Makefile.in (TARGET_H): Add insn-codes.h.
9323 2013-06-14 Alan Modra <amodra@gmail.com>
9327 * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
9328 EXPAND_MEMORY and EXPAND_WRITE to recursive call. Don't use
9329 bitfield expansion when EXPAND_MEMORY.
9330 (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
9332 2013-06-13 Michael Meissner <meissner@linux.vnet.ibm.com>
9334 * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
9335 test for clearing quad memory on 32-bit later.
9337 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
9339 * fold-const.c (negate_expr_p): Handle VECTOR_CST.
9340 (fold_negate_expr): Likewise.
9341 (fold_real_zero_addition_p): Handle vectors.
9342 (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
9344 2013-06-14 Alan Modra <amodra@gmail.com>
9346 * varasm.c (force_const_mem): Revert 2013-06-07 change.
9348 2013-06-13 Jan Hubicka <jh@suse.cz>
9350 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
9351 Local comdats are not externally visible.
9352 * symtab.c (dump_symtab_base): Dump externally visible.
9353 (verify_symtab_base): Verify back links in the symtab hash.
9355 2013-06-13 Bin Cheng <bin.cheng@arm.com>
9357 * fold-const.c (operand_equal_p): Consider NOP_EXPR and
9358 CONVERT_EXPR as equal nodes.
9360 2013-06-13 Bin Cheng <bin.cheng@arm.com>
9362 * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
9364 2013-06-13 Marc Glisse <marc.glisse@inria.fr>
9366 * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
9367 Generalize to complex and vector.
9368 * tree.c (build_all_ones_cst): New function.
9369 * tree.h (build_all_ones_cst): Declare it.
9371 2013-06-13 Alan Modra <amodra@gmail.com>
9373 * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
9374 * config/rs6000/rs6000.md (signbittf2): New insn.
9375 (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
9376 (abstf2_internal, cmptf_internal2): Likewise.
9377 * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
9379 2013-06-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9380 Pat Haugen <pthaugen@us.ibm.com>
9381 Peter Bergner <bergner@vnet.ibm.com>
9383 * config/rs6000/rs6000.c (emit_load_locked): Add support for
9384 power8 byte, half-word, and quad-word atomic instructions.
9385 (emit_store_conditional): Likewise.
9386 (rs6000_expand_atomic_compare_and_swap): Likewise.
9387 (rs6000_expand_atomic_op): Likewise.
9389 * config/rs6000/sync.md (larx): Add new modes for power8.
9391 (AINT): New mode iterator to include TImode as well as normal
9392 integer modes on power8.
9393 (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
9394 that VSX registers are not considered. Use AINT mode iterator
9395 instead of INT1 to allow inclusion of quad word atomic operations
9397 (load_locked<mode>): Likewise.
9398 (store_conditional<mode>): Likewise.
9399 (atomic_compare_and_swap<mode>): Likewise.
9400 (atomic_exchange<mode>): Likewise.
9401 (atomic_nand<mode>): Likewise.
9402 (atomic_fetch_<fetchop_name><mode>): Likewise.
9403 (atomic_nand_fetch<mode>): Likewise.
9404 (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
9406 (ATOMIC): On power8, add QImode, HImode modes.
9407 (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
9408 modes that promote to SImode.
9409 (load_lockedti): Convert TImode arguments to PTImode, so that we
9410 get a guaranteed even/odd register pair.
9411 (load_lockedpti): Likewise.
9412 (store_conditionalti): Likewise.
9413 (store_conditionalpti): Likewise.
9415 * config/rs6000/rs6000.md (QHI): New mode iterator for power8
9416 atomic load/store instructions.
9419 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
9421 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
9423 (insn_count): New attribute, with most cases extracted from...
9424 (length): ...here. Redefine most cases in terms of insn_count.
9425 (single_insn): Delete.
9426 (can_delay): Use insn_count to check for single instructions.
9427 (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
9428 (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
9429 (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
9430 (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
9431 (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
9432 (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
9433 (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
9434 (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
9435 (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
9436 (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
9437 rather than "length".
9438 (tls_get_tp_<mode>): Likewise. Remove redundant "no_delay" attribute.
9439 * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
9440 Use "insn_count" rather than "length".
9441 * config/mips/mips-dsp.md
9442 (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
9443 (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
9446 2013-06-12 Marc Glisse <marc.glisse@inria.fr>
9448 PR tree-optimization/57361
9449 * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
9451 2013-06-12 Sofiane Naci <sofiane.naci@arm.com>
9453 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
9455 (aarch64_simd_combine<mode>): New instruction expansion.
9456 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
9458 * config/aarch64/aarch64.c (aarch64_split_combine): New function.
9459 * config/aarch64/iterators.md (Vdbl): Add entry for DF.
9461 2013-06-12 Jan Hubicka <jh@suse.cz>
9463 * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
9464 decl has when in streaming stage.
9465 * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
9466 * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
9468 2013-06-12 Roland Stigge <stigge@antcom.de>
9471 * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
9473 2013-06-12 Jakub Jelinek <jakub@redhat.com>
9475 PR tree-optimization/57537
9476 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
9477 vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
9479 2013-06-12 Richard Biener <rguenther@suse.de>
9481 * data-streamer.h (streamer_write_char_stream): CSE
9482 obs->current_pointer.
9483 * data-streamer-out.c (streamer_write_uhwi_stream): Inline
9484 streamer_write_char_stream manually and optimize the resulting loop.
9485 (streamer_write_hwi_stream): Likewise.
9487 2013-06-12 Jan Hubicka <jh@suse.cz>
9489 * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
9490 * cgraph.h (varpool_create_empty_node): Declare.
9491 * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
9493 * symtab.c (symtab_unregister_node): Be lax about missin entries
9495 (symtab_get_node): Update comment.
9496 * varpool.c (varpool_create_empty_node): Break out from ...
9497 (varpool_node_for_decl): ... here.
9498 * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
9500 2013-06-12 Eric Botcazou <ebotcazou@adacore.com>
9502 * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
9503 <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
9504 part. Use straight-line flow at the end.
9505 <COMPONENT_REF>: Remove superfluous else.
9506 <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
9508 2013-06-12 Jakub Jelinek <jakub@redhat.com>
9511 * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
9512 target hook even for !TREE_PUBLIC decls. If no resolution info
9513 is available, return false for common and external decls.
9515 2013-06-12 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
9517 * config/rl78/constraints.md (U): New constraint.
9518 * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
9521 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9524 * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
9525 to allow returning address to AT_PLATFORM name.
9527 2013-06-11 Jan Hubicka <jh@suse.cz>
9529 * cgraph.c (cgraph_create_function_alias): Set weakref flag.
9530 * cgraph.h (symtab_node_base): Add weakref flag.
9531 * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
9532 (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
9533 (output_weakrefs): Use weakref flag.
9534 * fold-const.c (simple_operand_p): Handle WEAK.
9535 * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
9536 * ipa.c (varpool_externally_visible_p): Drop weakref.
9537 (function_and_variable_visibility): Update comment; fix weakref
9538 sanity checks; do not clear DECL_WEAK on them.
9539 * lto-cgraph.c (lto_output_node): update.
9540 (lto_output_varpool_node): Update.
9541 (input_overwrite_node): Update.
9542 (input_node): Update.
9543 (input_varpool_node): Update.
9544 * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
9545 (lto_symtab_merge_symbols): Add sanity check.
9546 (lto_symtab_prevailing_decl): Do not special case weakrefs.
9547 * passes.c (rest_of_decl_compilation): Set static flag, too.
9548 * symtab.c (dump_symtab_base): Dump weakref.
9549 (verify_symtab_base): Sanity check weakrefs.
9550 (symtab_make_decl_local): Remove duplicated code.
9551 (symtab_alias_ultimate_target): Simplify.
9552 * varpool.c (varpool_create_variable_alias): Set weakref flag.
9554 2013-06-11 Tom de Vries <tom@codesourcery.com>
9556 * genautomata.c (gen_regexp_sequence): Handle els_num == -1. Handle
9557 sequence_vect == NULL.
9559 2013-06-11 DJ Delorie <dj@redhat.com>
9561 * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
9562 (rl78_unwind_word_mode): New.
9564 2013-06-11 David Malcolm <dmalcolm@redhat.com>
9566 * final.c (debug_prefix_maps): Make static.
9568 2013-06-11 David Malcolm <dmalcolm@redhat.com>
9570 * function.c (initial_trampoline): Remove stray copy.
9572 2013-06-11 Sofiane Naci <sofiane.naci@arm.com>
9574 * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
9576 2013-06-11 Martin Jambor <mjambor@suse.cz>
9578 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
9579 within bounds at the beginning of the function.
9581 2013-06-11 Alan Modra <amodra@gmail.com>
9583 * varasm.c (get_section): Don't die on !DECL_P decl. Tidy error
9585 (get_named_section): Don't NULL !DECL_P decl.
9587 2013-06-11 Igor Zamyatin <igor.zamyatin@intel.com>
9589 * doc/invoke.texi (core-avx2): Document.
9591 (atom): Updated with MOVBE.
9593 2013-06-11 Richard Biener <rguenther@suse.de>
9595 * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
9597 2013-06-11 Anton Blanchard <anton@samba.org>
9599 * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
9600 correct shift value in little-endian mode.
9602 2013-06-11 Jakub Jelinek <jakub@redhat.com>
9605 * varasm.c (get_variable_align): Move #endif to the right place.
9607 2013-06-10 Cary Coutant <ccoutant@google.com>
9609 * dwarf2out.c (hash_external_ref): Use die_symbol or signature
9610 for hash so that hash table traversal order is deterministic.
9612 2013-06-10 Michael Meissner <meissner@linux.vnet.ibm.com>
9613 Pat Haugen <pthaugen@us.ibm.com>
9614 Peter Bergner <bergner@vnet.ibm.com>
9616 * config/rs6000/vector.md (GPR move splitter): Do not split moves
9617 of vectors in GPRS if they are direct moves or quad word load or
9620 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
9622 (direct_move_p): Likewise.
9623 (quad_load_store_p): Likewise.
9625 * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
9626 classes into bins based on the physical register type.
9627 (reg_class_to_reg_type): Likewise.
9628 (IS_STD_REG_TYPE): Likewise.
9629 (IS_FP_VECT_REG_TYPE): Likewise.
9630 (reload_fpr_gpr): Arrays to determine what insn to use if we can
9631 use direct move instructions.
9632 (reload_gpr_vsx): Likewise.
9633 (reload_vsx_gpr): Likewise.
9634 (rs6000_init_hard_regno_mode_ok): Precalculate the register type
9635 information that is a simplification of register classes. Also
9636 precalculate direct move reload helpers.
9637 (direct_move_p): New function to return true if the operation can
9638 be done as a direct move instruciton.
9639 (quad_load_store_p): New function to return true if the operation
9640 is a quad memory operation.
9641 (rs6000_legitimize_address): If quad memory, only allow register
9642 indirect for TImode addresses.
9643 (rs6000_legitimate_address_p): Likewise.
9644 (enum reload_reg_type): Delete, replace with rs6000_reg_type.
9645 (rs6000_reload_register_type): Likewise.
9646 (register_to_reg_type): Return register type.
9647 (rs6000_secondary_reload_simple_move): New helper function for
9648 secondary reload and secondary memory needed to identify anything
9649 that is a simple move, and does not need reloading.
9650 (rs6000_secondary_reload_direct_move): New helper function for
9651 secondary reload to identify cases that can be done with several
9652 instructions via the direct move instructions.
9653 (rs6000_secondary_reload_move): New helper function for secondary
9654 reload to identify moves between register types that can be done.
9655 (rs6000_secondary_reload): Add support for quad memory operations
9656 and for direct move.
9657 (rs6000_secondary_memory_needed): Likewise.
9658 (rs6000_debug_secondary_memory_needed): Change argument names.
9659 (rs6000_output_move_128bit): New function to return the move to
9660 use for 128-bit moves, including knowing about the various
9661 limitations of quad memory operations.
9663 * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
9664 memory operations. call rs6000_output_move_128bit for the actual
9665 instruciton(s) to generate.
9666 (vsx_movti_64bit): Likewise.
9668 * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
9669 (UNSPEC_P8V_MTVSRWZ): Likewise.
9670 (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
9671 (UNSPEC_P8V_MTVSRD): Likewise.
9672 (UNSPEC_P8V_XXPERMDI): Likewise.
9673 (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
9674 (UNSPEC_FUSION_GPR): Likewise.
9675 (FMOVE128_GPR): New iterator for direct move.
9676 (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
9679 (zero_extend<mode>di2_internal1): Add support for power8 32-bit
9680 loads and direct move instructions.
9681 (zero_extendsidi2_lfiwzx): Likewise.
9682 (extendsidi2_lfiwax): Likewise.
9683 (extendsidi2_nocell): Likewise.
9684 (floatsi<mode>2_lfiwax): Likewise.
9686 (floatunssi<mode>2_lfiwzx): Likewise.
9688 (fix_trunc<mode>_stfiwx): Likewise.
9689 (fixuns_trunc<mode>_stfiwx): Likewise.
9690 (mov<mode>_hardfloat, 32-bit floating point): Likewise.
9691 (mov<move>_hardfloat64, 64-bit floating point): Likewise.
9692 (parity<mode>2_cmpb): Set length/type attr.
9693 (unnamed shift right patterns, mov<mode>_internal2): Change type attr
9694 for 'mr.' to fast_compare.
9695 (bpermd_<mode>): Change type attr to popcnt.
9696 (p8_fmrgow_<mode>): New insns for power8 direct move support.
9697 (p8_mtvsrwz_1): Likewise.
9698 (p8_mtvsrwz_2): Likewise.
9699 (reload_fpr_from_gpr<mode>): Likewise.
9700 (p8_mtvsrd_1): Likewise.
9701 (p8_mtvsrd_2): Likewise.
9702 (p8_xxpermdi_<mode>): Likewise.
9703 (reload_vsx_from_gpr<mode>): Likewise.
9704 (reload_vsx_from_gprsf): Likewise.
9705 (p8_mfvsrd_3_<mode>): LIkewise.
9706 (reload_gpr_from_vsx<mode>): Likewise.
9707 (reload_gpr_from_vsxsf): Likewise.
9708 (p8_mfvsrd_4_disf): Likewise.
9709 (multi-word GPR splits): Do not split direct moves or quad memory
9712 2013-06-10 David Malcolm <dmalcolm@redhat.com>
9714 * tree-into-ssa.c (interesting_blocks): Make static.
9716 2013-06-10 Jakub Jelinek <jakub@redhat.com>
9719 * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
9720 CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
9721 Use DATA_ABI_ALIGNMENT for that case instead if defined.
9722 (get_variable_align): New function.
9723 (get_variable_section, emit_bss, emit_common,
9724 assemble_variable_contents, place_block_symbol): Use
9725 get_variable_align instead of DECL_ALIGN.
9726 (assemble_noswitch_variable): Add align argument, use it
9727 instead of DECL_ALIGN.
9728 (assemble_variable): Adjust caller. Use get_variable_align
9729 instead of DECL_ALIGN.
9730 * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
9732 (DATA_ABI_ALIGNMENT): Define.
9733 * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
9734 * config/i386/i386.c (x86_data_alignment): Add opt argument. If
9735 opt is false, only return the psABI mandated alignment increase.
9736 * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
9737 (DATA_ABI_ALIGNMENT): ... this.
9738 * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
9739 (DATA_ABI_ALIGNMENT): ... this.
9740 * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
9741 * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
9742 (DATA_ABI_ALIGNMENT): ... this.
9743 * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
9744 * doc/tm.texi: Regenerated.
9746 2013-06-10 Uros Bizjak <ubizjak@gmail.com>
9748 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
9749 cmp_code to construct REG_EQUAL note.
9751 2013-06-09 Jakub Jelinek <jakub@redhat.com>
9754 * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
9755 that operands[2] doesn't overlap with operands[0].
9757 2013-06-09 David Edelsohn <dje.gcc@gmail.com>
9758 Jan Hubicka <jh@suse.cz>
9760 * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
9761 hack to mark symbols as used.
9763 2013-06-08 Vladimir Makarov <vmakarov@redhat.com>
9765 PR rtl-optimization/57559
9766 * lra-constraints.c (process_alt_operands): Don't discourage
9767 memory with known offset for offsetable memory constraint.
9768 * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
9770 2013-06-08 Eric Botcazou <ebotcazou@adacore.com>
9772 * varasm.c (struct oc_local_state): Reorder fields.
9773 (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
9774 and adjust accordingly.
9775 (output_constructor): Reorder initialization code and adjust call to
9776 output_constructor_bitfield.
9778 2013-06-07 Jan Hubicka <jh@suse.cz>
9780 * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
9782 2013-06-07 David Malcolm <dmalcolm@redhat.com>
9784 * tree-object-size.c (unknown): Make const.
9786 2013-06-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9788 * config/s390/s390.md (cpu_facility): Add cpu_zarch.
9789 ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
9790 for last alternative in the cpu_facility attribute.
9792 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9795 * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
9796 (xordi3): Change operand 2 constraint to arm_xordi_operand.
9797 * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
9798 * config/arm/constraints.md (Dg): New constraint.
9799 * config/arm/neon.md (xordi3_neon): Remove.
9800 (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
9801 * config/arm/predicates.md (arm_xordi_operand): New predicate.
9803 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9805 * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
9806 Clean up alternatives.
9808 2013-06-07 Alan Modra <amodra@gmail.com>
9810 * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
9813 2013-06-07 Alan Modra <amodra@gmail.com>
9815 * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
9817 2013-06-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9819 * config/arm/constraints.md (Df): New constraint.
9820 * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
9821 Correct length attribute for last two alternatives.
9823 2013-06-07 Alan Modra <amodra@gmail.com>
9825 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
9826 override user -mfp-in-toc.
9827 (offsettable_ok_by_alignment): Consider just the current access
9828 rather than the whole object, unless BLKmode. Handle
9829 CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
9830 (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
9831 for -mcmodel=medium.
9832 * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
9833 override user -mfp-in-toc or -msum-in-toc. Default to
9834 -mno-fp-in-toc for -mcmodel=medium.
9836 2013-06-06 DJ Delorie <dj@redhat.com>
9838 * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
9839 TARGET_VALID_POINTER_MODE.
9841 2013-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
9842 Pat Haugen <pthaugen@us.ibm.com>
9843 Peter Bergner <bergner@vnet.ibm.com>
9845 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
9846 Document new power8 builtins.
9848 * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
9849 condition code register, to allow 128-bit logical operations to be
9850 done in the VSX or GPR registers.
9851 (nor<mode>3): Use the canonical form for nor.
9852 (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
9853 vclz*, and vpopcnt* vector instructions.
9854 (nand<mode>3): Likewise.
9855 (orc<mode>3): Likewise.
9856 (clz<mode>2): LIkewise.
9857 (popcount<mode>2): Likewise.
9859 * config/rs6000/predicates.md (int_reg_operand): Rework tests so
9860 that only the GPRs are recognized.
9862 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9863 support for new power8 builtins.
9865 * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
9867 (xscvdpspn): Likewise.
9873 (vpopcnt): Likewise.
9874 (vpopcntb): Likewise.
9875 (vpopcnth): Likewise.
9876 (vpopcntw): Likewise.
9877 (vpopcntd): Likewise.
9882 (eqv_v16qi3): Likewise.
9883 (eqv_v8hi3): Likewise.
9884 (eqv_v4si3): Likewise.
9885 (eqv_v2di3): Likewise.
9886 (eqv_v4sf3): Likewise.
9887 (eqv_v2df3): Likewise.
9889 (nand_v16qi3): Likewise.
9890 (nand_v8hi3): Likewise.
9891 (nand_v4si3): Likewise.
9892 (nand_v2di3): Likewise.
9893 (nand_v4sf3): Likewise.
9894 (nand_v2df3): Likewise.
9896 (orc_v16qi3): Likewise.
9897 (orc_v8hi3): Likewise.
9898 (orc_v4si3): Likewise.
9899 (orc_v2di3): Likewise.
9900 (orc_v4sf3): Likewise.
9901 (orc_v2df3): Likewise.
9903 * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
9904 allow power8 quad mode in 64-bit.
9905 (rs6000_builtin_vectorized_function): Add support to vectorize
9906 ISA 2.07 count leading zeros, population count builtins.
9907 (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
9908 V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
9909 (builtin_function_type): Add vgbbd builtin function which takes an
9911 (altivec_expand_vec_perm_const): Add support for new power8 merge
9914 * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
9915 that does not include TImdoe for use with 32-bit.
9916 (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
9918 (UNSPEC_VSX_CVDPSPN): Likewise.
9919 (vsx_xscvdpspn): Likewise.
9920 (vsx_xscvspdpn): Likewise.
9921 (vsx_xscvdpspn_scalar): Likewise.
9922 (vsx_xscvspdpn_directmove): Likewise.
9923 (vsx_and<mode>3): Split logical operations into 32-bit and
9924 64-bit. Add support to do logical operations on TImode as well as
9925 VSX vector types. Allow logical operations to be done in either
9926 VSX registers or in general purpose registers in 64-bit mode. Add
9927 splitters if GPRs were used. For AND, add clobber of CCmode to
9928 allow use of ANDI on GPRs. Rewrite nor to use the canonical RTL
9930 (vsx_and<mode>3_32bit): Likewise.
9931 (vsx_and<mode>3_64bit): Likewise.
9932 (vsx_ior<mode>3): Likewise.
9933 (vsx_ior<mode>3_32bit): Likewise.
9934 (vsx_ior<mode>3_64bit): Likewise.
9935 (vsx_xor<mode>3): Likewise.
9936 (vsx_xor<mode>3_32bit): Likewise.
9937 (vsx_xor<mode>3_64bit): Likewise.
9938 (vsx_one_cmpl<mode>2): Likewise.
9939 (vsx_one_cmpl<mode>2_32bit): Likewise.
9940 (vsx_one_cmpl<mode>2_64bit): Likewise.
9941 (vsx_nor<mode>3): Likewise.
9942 (vsx_nor<mode>3_32bit): Likewise.
9943 (vsx_nor<mode>3_64bit): Likewise.
9944 (vsx_andc<mode>3): Likewise.
9945 (vsx_andc<mode>3_32bit): Likewise.
9946 (vsx_andc<mode>3_64bit): Likewise.
9947 (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
9948 and xxlorc instructions.
9949 (vsx_eqv<mode>3_64bit): Likewise.
9950 (vsx_nand<mode>3_32bit): Likewise.
9951 (vsx_nand<mode>3_64bit): Likewise.
9952 (vsx_orc<mode>3_32bit): Likewise.
9953 (vsx_orc<mode>3_64bit): Likewise.
9955 * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
9957 * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
9959 (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
9960 (p8_vmrgow): Likewise.
9961 (altivec_and<mode>3): Add clobber of CCmode to allow AND using
9962 GPRs to be split under VSX.
9963 (p8v_clz<mode>2): Add power8 count leading zero support.
9964 (p8v_popcount<mode>2): Add power8 population count support.
9965 (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
9968 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
9971 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
9973 (vec_nand): Likewise.
9974 (vec_vclz): Likewise.
9975 (vec_vclzb): Likewise.
9976 (vec_vclzd): Likewise.
9977 (vec_vclzh): Likewise.
9978 (vec_vclzw): Likewise.
9979 (vec_vgbbd): Likewise.
9980 (vec_vmrgew): Likewise.
9981 (vec_vmrgow): Likewise.
9982 (vec_vpopcnt): Likewise.
9983 (vec_vpopcntb): Likewise.
9984 (vec_vpopcntd): Likewise.
9985 (vec_vpopcnth): Likewise.
9986 (vec_vpopcntw): Likewise.
9988 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
9990 PR rtl-optimization/57468
9991 * config/i386/i386.c (inline_secondary_memory_needed): Ignore
9994 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
9996 PR rtl-optimization/57459
9997 * lra-constraints.c (update_ebb_live_info): Fix typo for operand
9998 type when setting live regs.
10000 2013-06-06 Vladimir Makarov <vmakarov@redhat.com>
10002 * config/s390/s390.opt (mlra): New option.
10003 * config/s390/s390.c (s390_decompose_address): Check displacement
10004 for all registers for LRA.
10005 (s390_secondary_reload): Don't used secondary reloads for LRA.
10006 (s390_lra_p): New function.
10007 (TARGET_LRA_P): Define.
10008 * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
10009 of attribute cpu_facility to zarch for the last alternative.
10010 (*cmpmem_short): Ditto.
10012 2013-06-06 Eric Botcazou <ebotcazou@adacore.com>
10014 * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
10015 (arm_compute_static_chain_stack_bytes): Use it. Tidy up.
10016 (arm_expand_prologue): Likewise.
10018 2013-06-06 Teresa Johnson <tejohnson@google.com>
10021 * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
10022 as this is now done by redirect_edge_and_branch_force.
10023 * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
10024 barriers, and fix interaction with splitting.
10025 * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
10026 * cfgcleanup.c (try_forward_edges): Fix early return value to properly
10027 reflect changes made in the routine.
10028 * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
10029 (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
10030 since this is called in cfglayout mode, and replace partition fixup
10031 with assert as that is now done by force_nonfallthru_and_redirect.
10032 (add_reg_crossing_jump_notes): Handle the fact that some jumps may
10033 already be marked with region crossing note.
10034 (insert_section_boundary_note): Make non-static, gate on flag
10035 has_bb_partition, rewrite to also check for multiple partitions.
10036 (rest_of_handle_reorder_blocks): Remove call to
10037 insert_section_boundary_note, now done later during free_cfg.
10038 (duplicate_computed_gotos): Don't duplicate partition crossing edge.
10039 * bb-reorder.h (insert_section_boundary_note): Declare.
10040 * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
10041 * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
10042 invoke insert_section_boundary_note.
10043 (try_redirect_by_replacing_jump): Remove unnecessary
10044 check for region crossing note.
10045 (fixup_partition_crossing): New function.
10046 (rtl_redirect_edge_and_branch): Fixup partition boundaries.
10047 (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
10048 in non-cfglayout mode.
10049 (force_nonfallthru_and_redirect): Fixup partition boundaries,
10050 remove old code that tried to do this. Emit barrier correctly
10051 when we are in cfglayout mode.
10052 (last_bb_in_partition): New function.
10053 (rtl_split_edge): Correctly fixup partition boundaries.
10054 (commit_one_edge_insertion): Remove old code that tried to
10055 fixup region crossing edge since this is now handled in
10056 split_block, and set up insertion point correctly since
10057 block may now end in a jump.
10058 (verify_hot_cold_block_grouping): Guard against checking when not in
10059 linearized RTL mode.
10060 (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
10062 (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
10063 rtl_verify_flow_info, so not called in cfglayout mode.
10064 (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
10065 (fixup_reorder_chain): Remove old code that attempted to fixup region
10066 crossing note as this is now handled in force_nonfallthru_and_redirect.
10067 (duplicate_insn_chain): Don't duplicate switch section notes.
10068 (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
10070 * basic-block.h (emit_barrier_after_bb): Declare.
10072 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10074 * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
10075 sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
10076 arm_usatsihi): Adjust alternatives for arm_restrict_it.
10078 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10080 * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
10082 * config/arm/ldmstm.md: Regenerate.
10084 2013-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10086 * config/arm/sync.md (atomic_loaddi_1):
10087 Disable predication for arm_restrict_it.
10088 (arm_load_exclusive<mode>): Likewise.
10089 (arm_load_exclusivesi): Likewise.
10090 (arm_load_exclusivedi): Likewise.
10091 (arm_load_acquire_exclusive<mode>): Likewise.
10092 (arm_load_acquire_exclusivesi): Likewise.
10093 (arm_load_acquire_exclusivedi): Likewise.
10094 (arm_store_exclusive<mode>): Likewise.
10095 (arm_store_exclusive<mode>): Likewise.
10096 (arm_store_release_exclusivedi): Likewise.
10097 (arm_store_release_exclusive<mode>): Likewise.
10099 2013-06-06 Richard Biener <rguenther@suse.de>
10101 * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
10103 (lto_tag_is_tree_code_p): Adjust.
10104 (lto_tag_is_gimple_code_p): Likewise.
10105 (lto_gimple_code_to_tag): Likewise.
10106 (lto_tag_to_gimple_code): Likewise.
10107 (lto_tree_code_to_tag): Likewise.
10108 (lto_tag_to_tree_code): Likewise.
10109 * data-streamer.h (streamer_write_hwi_in_range): Use
10110 uhwi streaming to stream the normalized range.
10111 (streamer_read_hwi_in_range): Likewise.
10113 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10115 * config/arm/arm.md (enabled_for_depr_it): New attribute.
10116 (predicable_short_it): Likewise.
10117 (predicated): Likewise.
10118 (enabled): Handle above.
10119 (define_cond_exec): Set predicated attribute to yes.
10121 2013-06-05 Mike Stump <mikestump@comcast.net>
10123 * gdbinit.in (__FUNCTION__): Add.
10125 2013-06-05 Uros Bizjak <ubizjak@gmail.com>
10127 * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
10128 GE, GT, GEU and GTU compares, modulo DImode compares with zero.
10130 2013-06-05 Jan Hubicka <jh@suse.cz>
10132 * varasm.c (mark_decl_referenced): Revert the removal until targets
10135 2013-06-05 David Edelsohn <dje.gcc@gmail.com>
10137 * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
10138 instead of mark_decl_referenced.
10140 2013-06-05 Jan Hubicka <jh@suse.cz>
10142 * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
10143 (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
10144 and symtab_used_from_object_file_p.
10145 (cgraph_make_node_local_1): Clear forced_by_abi.
10146 (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
10147 * cgraph.h (symtab_node_base): Add forced_by_abi.
10148 (decide_is_variable_needed): Remove.
10149 (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
10150 * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
10151 (decide_is_symbol_needed): ... this one; handle symbols in general;
10152 always analyze virtuals; honnor forced_by_abi.
10153 (cgraph_finalize_function): Update.
10154 (varpool_finalize_decl): Update.
10155 (symbol_defined_and_needed): Remove.
10156 (analyze_functions): Update.
10157 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
10158 output_refs, input_overwrite_node): Handle forced_by_abi.
10159 * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
10160 (address_taken_from_non_vtable_p): ... this one.
10161 (comdat_can_be_unshared_p_1): New function.
10162 (cgraph_comdat_can_be_unshared_p): Rename to ...
10163 (comdat_can_be_unshared_p): ... this one; handle symbols in general.
10164 (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
10165 (function_and_variable_visibility): Clear forced_by_abi as needed.
10166 * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
10167 (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
10168 * symtab.c (dump_symtab_base): Dump forced_by_abi.
10169 * varpool.c (decide_is_variable_needed): Remove.
10171 2013-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10173 * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
10174 (arm_option_override): Override arm_restrict_it where appropriate.
10175 (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
10176 * config/arm/arm.opt (mrestrict-it): New command-line option.
10177 * doc/invoke.texi: Document -mrestrict-it.
10179 2013-06-05 David Malcolm <dmalcolm@redhat.com>
10181 * tsan.c (tsan_atomic_table): Make const.
10183 2013-06-05 Richard Biener <rguenther@suse.de>
10185 * tree-streamer.c (streamer_tree_cache_insert_1): Update the
10186 index associated with the tree we are supposed to replace.
10187 * tree-streamer-out.c (pack_ts_base_value_fields): Output
10188 TREE_ASM_WRITTEN as zero for everything but SSA names.
10190 2013-06-05 David Malcolm <dmalcolm@redhat.com>
10192 * tree-ssa-structalias.c (call_stmt_vars): Make static.
10194 2013-06-04 Jan Hubicka <jh@suse.cz>
10196 * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
10197 (input_node, input_varpool_node): Handle correctly external same
10199 * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
10200 nodes at ltrans stage.
10202 2013-06-04 Jan Hubicka <jh@suse.cz>
10204 * ipa-inline.c (update_caller_keys): Fix availability test.
10205 (update_callee_keys): Likewise.
10206 * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
10207 to follow ELF standard.
10209 2013-06-04 Jürgen Urban <JuergenUrban@gmx.de>
10211 * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
10212 (mips64r5900el-*-elf*): New configurations.
10213 * config/mips/mips-cpus.def (r5900): New processor.
10214 * config/mips/mips-tables.opt: Regenerate.
10215 * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
10216 (mips_issue_rate): Handle PROCESSOR_R5900.
10217 (mips_reorg_process_insns): Force reorder mode for the R5900.
10218 * config/mips/mips.h (TARGET_MIPS5900): Define.
10219 (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
10221 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
10223 * config/mips/mips.md (processor): Add r5900.
10224 (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
10226 2013-06-04 Ian Bolton <ian.bolton@arm.com>
10228 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
10229 into function to generate MOVI instruction.
10230 * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
10231 (aarch64_preferred_simd_mode): Turn into wrapper.
10232 (aarch64_output_scalar_simd_mov_immediate): New function.
10233 * config/aarch64/aarch64-protos.h: Add prototype for above.
10235 2013-06-04 Ian Bolton <ian.bolton@arm.com>
10237 * config/aarch64/aarch64.c (simd_immediate_info): Remove
10238 element_char member.
10239 (sizetochar): Return signed char.
10240 (aarch64_simd_valid_immediate): Remove elchar and other
10241 unnecessary variables.
10242 (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
10243 Calculate element_char as required.
10244 * config/aarch64/aarch64-protos.h: Update and move prototype
10245 for aarch64_output_simd_mov_immediate.
10246 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
10249 2013-06-04 Ian Bolton <ian.bolton@arm.com>
10251 * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
10252 information completed by aarch64_simd_valid_immediate.
10253 (aarch64_legitimate_constant_p): Update arguments.
10254 (aarch64_simd_valid_immediate): Work with struct rather than many
10256 (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
10257 (aarch64_simd_make_constant): Update arguments.
10258 (aarch64_output_simd_mov_immediate): Work with struct rather than
10259 many pointers. Output immediate directly rather than as operand.
10260 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
10262 * config/aarch64/constraints.md (Dn): Update arguments.
10264 2013-06-04 Ian Bolton <ian.bolton@arm.com>
10266 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
10268 (aarch64_simd_immediate_valid_for_move): Remove.
10269 (aarch64_simd_scalar_immediate_valid_for_move): Update call.
10270 (aarch64_simd_make_constant): Update call.
10271 (aarch64_output_simd_mov_immediate): Update call.
10272 * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
10274 * config/aarch64/constraints.md (Dn): Update call.
10276 2013-06-04 Ian Bolton <ian.bolton@arm.com>
10278 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
10279 return type to bool for prototype.
10280 (aarch64_legitimate_constant_p): Check for true instead of not -1.
10281 (aarch64_simd_valid_immediate): Fix up each return to return a bool.
10282 (aarch64_simd_immediate_valid_for_move): Update retval for bool.
10284 2013-06-04 Catherine Moore <clm@codesourcery.com>
10286 * config/mips/mips.opt (meva): New.
10287 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
10288 (ASM_SPEC): Handle -meva.
10289 * doc/invoke.texi (meva): Document.
10291 2013-06-04 Alan Modra <amodra@gmail.com>
10293 * config/rs6000/rs6000.c (output_toc): Correct little-endian float
10296 2013-06-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10298 * rtl.def: Add extra fourth optional field to define_cond_exec.
10299 * gensupport.c (process_one_cond_exec): Process attributes from
10301 * doc/md.texi: Document fourth field in define_cond_exec.
10303 2013-06-04 Eric Botcazou <ebotcazou@adacore.com>
10305 * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
10306 out the processing order as in store_bit_field_1.
10308 2013-06-04 Jan Hubicka <jh@suse.cz>
10310 PR middle-end/57500
10311 * cgraphunit.c (cgraph_process_same_body_aliases): Create
10312 non-VAR_DECL node if it does not exist yet.
10314 2013-06-03 Richard Sandiford <rdsandiford@googlemail.com>
10316 * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
10317 (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
10318 (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
10319 target_cpu_default setting.
10321 2013-06-03 Teresa Johnson <tejohnson@google.com>
10323 * dumpfile.c (opt_info_switch_p): Change -fopt-info
10324 default to -fopt-info=optimized instead of all.
10325 * doc/invoke.texi: Ditto.
10326 * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
10327 success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
10328 (execute_vect_slp): Emit BB vectorization success under
10329 MSG_OPTIMIZED_LOCATIONS.
10330 * tree-vect-slp.c (vect_slp_transform_bb): Change
10331 MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
10332 * tree-vect-loop.c (vect_transform_loop): Ditto.
10334 2013-06-03 Jason Merrill <jason@redhat.com>
10337 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10338 Use TARGET_EXPR for C++.
10340 2013-06-03 Jakub Jelinek <jakub@redhat.com>
10342 PR rtl-optimization/57268
10343 * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
10344 if DEBUG_INSN_P (insn).
10347 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
10349 PR rtl-optimization/57268
10350 * sched-deps.c (sched_analyze_2): Flush dependence lists if
10351 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10353 2013-06-03 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
10355 * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
10356 (ix86_avoid_lea_for_addr): Likewise.
10357 (exact_dependency_1): Likewise.
10358 (ix86_adjust_cost): Likewise.
10359 (swap_top_of_ready_list): Fix formatting and !reload_completed check
10361 (do_reorder_for_imul): Fix typo, formatting and
10362 !reload_completed check removed.
10363 (ix86_sched_reorder): Fix typo and formatting.
10364 (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
10367 2013-06-03 Sofiane Naci <sofiane.naci@arm.com>
10369 * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
10371 2013-06-03 Eric Botcazou <ebotcazou@adacore.com>
10373 * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
10374 <STRING_CST>: Likewise.
10375 <VECTOR_CST>: Likewise.
10377 2013-06-01 Janus Weil <janus@gcc.gnu.org>
10378 Mikael Morin <mikael@gcc.gnu.org>
10380 * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
10381 * config.in: Regenerated.
10382 * configure: Regenerated.
10384 2013-06-01 Jan Hubicka <jh@suse.cz>
10386 PR middle-end/57366
10387 * cgraphunit.c (compile): When weakref is not supported,
10388 set up transparent aliases before final output pass.
10389 * varasm.c (assemble_alias): Do not try to do it here.
10391 2013-06-01 Jan Hubicka <jh@suse.cz>
10393 PR middle-end/57467
10394 * passes.c (for_per_function): Skip unanalyzed functions.
10396 2013-06-01 Jan Hubicka <jh@suse.cz>
10398 * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
10399 (lto_symtab_merge_symbols_1): ... this one.
10400 (lto_symtab_merge_cgraph_nodes): Rename to ...
10401 (lto_symtab_merge_symbols): ... this one; simplify.
10402 * cgraph.c (same_body_aliases_done): Rename to ...
10403 (cpp_implicit_aliases_done): ... this one.
10404 (cgraph_create_function_alias): Update.
10405 (cgraph_same_body_alias): Update.
10406 (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
10407 (verify_edge_corresponds_to_fndecl): Simplify.
10408 * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
10409 (cgraph_node): Remove same_body_alias.
10410 (varpool_node): Remove alias_of and extra_name_alias.
10411 (same_body_aliases_done): Rename to ..
10412 (cpp_implicit_aliases_done): ... this one.
10413 (symtab_alias_ultimate_target): Add default parameter.
10414 (symtab_resolve_alias): New function.
10415 (fixup_same_cpp_alias_visibility): Declare.
10416 (cgraph_function_node): Add default parameter.
10417 (cgraph_node_asm_name): Likewise.
10418 (cgraph_function_or_thunk_node): Add default parameter; do
10419 not ICE when it is NULL.
10420 (varpool_variable_node): Likewise.
10421 * tree-emutls.c (create_emultls_var): Update.
10422 (ipa_lower_emutls): Update.
10423 * cgraphunit.c (cgraph_decide_is_function_needed): Update.
10424 (cgraph_reset_node): Reset alias info.
10425 (cgraph_finalize_function): Update.
10426 (fixup_same_cpp_alias_visibility): Move to symtab.c.
10427 (analyze_function): Simplify.
10428 (cgraph_process_same_body_aliases): Simplify.
10429 (analyze_functions): Fixup same body aliases.
10430 (handle_alias_pairs): Simplify.
10431 (assemble_thunk): Update.
10432 (assemble_thunks_and_aliases): Update.
10433 (output_weakrefs): Rewrite.
10434 * lto-cgraph.c (lto_output_node): Rewrite alias handling.
10435 (lto_output_varpool_node): Likewise.
10436 (compute_ltrans_boundary): Remve assert.
10437 (get_alias_symbol): New functoin.
10438 (input_node): Rewrite alias handling.
10439 (input_varpool_node): Likewise.
10440 * ipa-pure-const.c (propagate_pure_const): Fix formating.
10441 * ipa.c (process_references): Handle weakrefs correctly.
10442 (symtab_remove_unreachable_nodes): Likewise.
10443 * trans-mem.c (get_cg_data): Update.
10444 (ipa_tm_create_version_alias): Update.
10445 (ipa_tm_execute): Update.
10446 * symtab.c (dump_symtab_base): Dump aliases.
10447 (verify_symtab_base): Verify aliases.
10448 (symtab_node_availability): New function.
10449 (symtab_alias_ultimate_target): Simplify.
10450 (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
10451 handle all the fixup cases.
10452 (symtab_resolve_alias): New function.
10453 * passes.c (ipa_write_summaries): Handle weakrefs.
10454 * varpool.c (varpool_analyze_node): Simplify.
10455 (assemble_aliases): Update.
10456 (varpool_create_variable_alias): Simplify.
10457 (varpool_extra_name_alias): Simplify.
10458 * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
10459 (lto_symtab_merge_symbols): ... this one.
10461 2013-06-01 Dinar Temirbulatov <dinar@kugelworks.com>
10464 PR rtl-optimization/57268
10465 * sched-deps.c (sched_analyze_2): Flush dependence lists if
10466 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10468 2013-06-01 Tobias Burnus <burnus@net-b.de>
10470 Partially reverted:
10471 2013-05-31 Tobias Burnus <burnus@net-b.de>
10473 PR middle-end/57073
10474 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
10477 2013-05-31 Dinar Temirbulatov <dinar@kugelworks.com>
10479 PR rtl-optimization/57268
10480 * sched-deps.c (sched_analyze_2): Flush dependence lists if
10481 the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
10483 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
10485 * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
10486 unordered comparison operators when -fno-trapping-math is in effect
10488 * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
10489 and implement unordered comparison operators properly on the e500.
10491 2013-05-31 Eric Botcazou <ebotcazou@adacore.com>
10493 * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
10494 for constant scalar integers.
10495 (simplify_relational_operation_1): Likewise.
10497 2013-05-31 Segher Boessenkool <segher@kernel.crashing.org>
10499 * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
10500 * config/rs6000/rs6000.md (cpu): Reorder. Split long line.
10503 2013-05-31 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
10504 Igor Zamyatin <igor.zamyatin@intel.com>
10506 Silvermont (SLM) architecture performance tuning.
10507 * config/i386/i386.h (enum ix86_tune_indices): Add
10508 X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
10509 (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
10511 * config/i386/i386.c (initial_ix86_tune_features)
10512 <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
10513 (ix86_lea_outperforms): Handle Silvermont tuning.
10514 (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
10516 (ix86_use_lea_for_mov): Likewise.
10517 (ix86_avoid_lea_for_addr): Likewise.
10518 (ix86_lea_for_add_ok): Likewise.
10519 (exact_dependency_1): New function.
10520 (exact_store_load_dependency): Likewise.
10521 (ix86_adjust_cost): Handle Silvermont tuning.
10522 (do_reoder_for_imul): Likewise.
10523 (swap_top_of_ready_list): New function.
10524 (ix86_sched_reorder): Changed to handle Silvermont tuning.
10526 * config/i386/i386.md (peepholes that split memory operand in fp
10529 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
10531 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10532 Remove un-necessary braces.
10534 2013-05-31 Marcus Shawcroft <marcus.shawcroft@arm.com>
10536 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10537 Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
10539 2013-05-31 Tobias Burnus <burnus@net-b.de>
10541 PR middle-end/57073
10542 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
10544 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10547 * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
10548 * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
10549 * config/arm/neon.md (iordi3_neon): Remove.
10550 (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
10551 * config/arm/predicates.md (imm_for_neon_logic_operand):
10552 Move to earlier in the file.
10553 (neon_logic_op2): Likewise.
10554 (arm_iordi_operand_neon): New predicate.
10556 2013-05-31 Richard Biener <rguenther@suse.de>
10558 PR tree-optimization/57478
10559 PR tree-optimization/57453
10560 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
10563 2013-05-31 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
10565 * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
10566 (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
10568 2013-05-30 Tobias Burnus <burnus@net-b.de>
10569 Thomas Koenig <tkoenig@gcc.gnu.org>
10571 PR middle-end/57073
10572 * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
10573 powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
10575 2013-05-30 Steven Bosscher <steven@gcc.gnu.org>
10577 * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
10579 2013-05-30 Vladimir Makarov <vmakarov@redhat.com>
10581 * target.def (register_usage_leveling_p): New hook.
10582 * targhooks.c (default_register_usage_leveling_p): New.
10583 * targhooks.h (default_register_usage_leveling_p): New prototype.
10584 * lra-assigns.c (register_usage_leveling_p): Use the hook.
10585 * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
10586 * doc/tm.texi: Update.
10587 * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
10589 2013-05-30 Ian Bolton <ian.bolton@arm.com>
10591 * config/aarch64/aarch64.md (insv<mode>): New define_expand.
10592 (*insv_reg<mode>): New define_insn.
10594 2013-05-30 Joern Rennecke <joern.rennecke@embecosm.com>
10596 PR rtl-optimization/57439
10597 * postreload.c (move2add_valid_value_p): Check that we have
10598 a zero subreg_regno_offset when accessing the register in
10599 the requested mode.
10601 2013-05-30 Yuri Rumyantsev <yuri.s.rumyantsev@intel.com>
10602 Igor Zamyatin <igor.zamyatin@intel.com>
10604 Silvermont (SLM) architecture pipeline model, tuning and
10606 * config.gcc: Add slm config options and target.
10608 * config/i386/slm.md: New.
10610 * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
10612 * config/i386/i386-c.c (ix86_target_macros_internal): New case
10614 (ix86_target_macros_internal): Likewise.
10616 * config/i386/i386.c (slm_cost): New cost.
10617 (m_SLM): New macro flag.
10618 (initial_ix86_tune_features): Set m_SLM.
10619 (x86_accumulate_outgoing_args): Likewise.
10620 (x86_arch_always_fancy_math_387): Likewise.
10621 (processor_target_table): Add slm cost.
10622 (cpu_names): Add slm cpu name.
10623 (x86_option_override_internal): Set SLM ISA.
10624 (ix86_issue_rate): New case PROCESSOR_SLM.
10625 (ia32_multipass_dfa_lookahead): Likewise.
10626 (fold_builtin_cpu): Add slm.
10628 * config/i386/i386.h (TARGET_SLM): New target macro.
10629 (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
10630 (processor_type): Add PROCESSOR_SLM.
10632 * config/i386/i386.md (cpu): Add new value "slm".
10633 (slm.md): Include slm.md.
10635 2013-05-30 Bernd Schmidt <bernds@codesourcery.com>
10636 Zhenqiang Chen <zhenqiang.chen@linaro.org>
10638 * config/arm/arm-protos.h: Add and update function protos.
10639 * config/arm/arm.c (use_simple_return_p): New added.
10640 (thumb2_expand_return): Check simple_return flag.
10641 * config/arm/arm.md: Add simple_return and conditional simple_return.
10642 * config/arm/iterators.md: Add iterator for return and simple_return.
10644 2013-05-30 Zhenqiang Chen <zhenqiang.chen@linaro.org>
10646 * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
10647 (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
10648 (arm_emit_vfp_multi_reg_pop): Likewise.
10649 (thumb2_emit_ldrd_pop): Likewise.
10650 (arm_expand_epilogue): Add misc REG_CFA notes.
10651 (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
10653 2013-05-29 Lawrence Crowl <crowl@google.com>
10655 * config/arm/t-arm: Update for below.
10657 * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
10658 Change type to hash_table. Update dependent calls and types.
10660 * config/i386/t-cygming: Update for below.
10662 * config/i386/t-interix: Update for below.
10664 * config/i386/winnt.c (i386_pe_section_type_flags::htab):
10665 Change type to hash_table. Update dependent calls and types.
10666 (i386_find_on_wrapper_list::wrappers): Likewise.
10668 * config/ia64/t-ia64: Update for below.
10670 * config/ia64/ia64.c (bundle_state_table):
10671 Change type to hash_table. Update dependent calls and types.
10673 * config/mips/mips.c (mips_reorg_process_insns::htab):
10674 Change type to hash_table. Update dependent calls and types.
10676 * config/sol2.c (solaris_comdat_htab):
10677 Change type to hash_table. Update dependent calls and types.
10679 * config/t-sol2: Update for above.
10681 2013-05-29 Teresa Johnson <tejohnson@google.com>
10683 * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
10684 functions are not yet marked as defined.
10686 2013-05-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10687 Pat Haugen <pthaugen@us.ibm.com>
10688 Peter Bergner <bergner@vnet.ibm.com>
10690 * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
10694 (vrotl<mode>3): Likewise.
10695 (vashl<mode>3): Likewise.
10696 (vlshr<mode>3): Likewise.
10697 (vashr<mode>3): Likewise.
10699 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10700 support for power8 V2DI builtins.
10702 * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
10703 power8 V2DI builtins.
10704 (vupkhsw): Likewise.
10705 (vupklsw): Likewise.
10706 (vaddudm): Likewise.
10707 (vminsd): Likewise.
10708 (vmaxsd): Likewise.
10709 (vminud): Likewise.
10710 (vmaxud): Likewise.
10711 (vpkudum): Likewise.
10712 (vpksdss): Likewise.
10713 (vpkudus): Likewise.
10714 (vpksdus): Likewise.
10719 (vsubudm): Likewise.
10720 (vcmpequd): Likewise.
10721 (vcmpgtsd): Likewise.
10722 (vcmpgtud): Likewise.
10723 (vcmpequd_p): Likewise.
10724 (vcmpgtsd_p): Likewise.
10725 (vcmpgtud_p): Likewise.
10726 (vupkhsw): Likewise.
10727 (vupklsw): Likewise.
10728 (vaddudm): Likewise.
10729 (vmaxsd): Likewise.
10730 (vmaxud): Likewise.
10731 (vminsd): Likewise.
10732 (vminud): Likewise.
10733 (vpksdss): Likewise.
10734 (vpksdus): Likewise.
10735 (vpkudum): Likewise.
10736 (vpkudus): Likewise.
10741 (vsubudm): Likewise.
10743 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
10744 support for power8 V2DI instructions.
10746 * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
10747 power8 V2DI instructions. Combine pack and unpack insns to use an
10748 iterator for each mode. Check whether a particular mode supports
10749 Altivec instructions instead of just checking TARGET_ALTIVEC.
10750 (UNSPEC_VPKUWUM): Likewise.
10751 (UNSPEC_VPKSHSS): Likewise.
10752 (UNSPEC_VPKSWSS): Likewise.
10753 (UNSPEC_VPKUHUS): Likewise.
10754 (UNSPEC_VPKSHUS): Likewise.
10755 (UNSPEC_VPKUWUS): Likewise.
10756 (UNSPEC_VPKSWUS): Likewise.
10757 (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
10758 (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
10759 (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
10760 (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
10761 (UNSPEC_VUPKHSB): Likewise.
10762 (UNSPEC_VUNPACK_HI_SIGN): Likewise.
10763 (UNSPEC_VUNPACK_LO_SIGN): Likewise.
10764 (UNSPEC_VUPKHSH): Likewise.
10765 (UNSPEC_VUPKLSB): Likewise.
10766 (UNSPEC_VUPKLSH): Likewise.
10768 (VI_char): Likewise.
10769 (VI_scalar): Likewise.
10770 (VI_unit): Likewise.
10772 (VP_small): Likewise.
10773 (VP_small_lc): Likewise.
10774 (VU_char): Likewise.
10775 (add<mode>3): Likewise.
10776 (altivec_vaddcuw): Likewise.
10777 (altivec_vaddu<VI_char>s): Likewise.
10778 (altivec_vadds<VI_char>s): Likewise.
10779 (sub<mode>3): Likewise.
10780 (altivec_vsubcuw): Likewise.
10781 (altivec_vsubu<VI_char>s): Likewise.
10782 (altivec_vsubs<VI_char>s): Likewise.
10783 (altivec_vavgs<VI_char>): Likewise.
10784 (altivec_vcmpbfp): Likewise.
10785 (altivec_eq<mode>): Likewise.
10786 (altivec_gt<mode>): Likewise.
10787 (altivec_gtu<mode>): Likewise.
10788 (umax<mode>3): Likewise.
10789 (smax<mode>3): Likewise.
10790 (umin<mode>3): Likewise.
10791 (smin<mode>3): Likewise.
10792 (altivec_vpkuhum): Likewise.
10793 (altivec_vpkuwum): Likewise.
10794 (altivec_vpkshss): Likewise.
10795 (altivec_vpkswss): Likewise.
10796 (altivec_vpkuhus): Likewise.
10797 (altivec_vpkshus): Likewise.
10798 (altivec_vpkuwus): Likewise.
10799 (altivec_vpkswus): Likewise.
10800 (altivec_vpks<VI_char>ss): Likewise.
10801 (altivec_vpks<VI_char>us): Likewise.
10802 (altivec_vpku<VI_char>us): Likewise.
10803 (altivec_vpku<VI_char>um): Likewise.
10804 (altivec_vrl<VI_char>): Likewise.
10805 (altivec_vsl<VI_char>): Likewise.
10806 (altivec_vsr<VI_char>): Likewise.
10807 (altivec_vsra<VI_char>): Likewise.
10808 (altivec_vsldoi_<mode>): Likewise.
10809 (altivec_vupkhsb): Likewise.
10810 (altivec_vupkhs<VU_char>): Likewise.
10811 (altivec_vupkls<VU_char>): Likewise.
10812 (altivec_vupkhsh): Likewise.
10813 (altivec_vupklsb): Likewise.
10814 (altivec_vupklsh): Likewise.
10815 (altivec_vcmpequ<VI_char>_p): Likewise.
10816 (altivec_vcmpgts<VI_char>_p): Likewise.
10817 (altivec_vcmpgtu<VI_char>_p): Likewise.
10818 (abs<mode>2): Likewise.
10819 (vec_unpacks_hi_v16qi): Likewise.
10820 (vec_unpacks_hi_v8hi): Likewise.
10821 (vec_unpacks_lo_v16qi): Likewise.
10822 (vec_unpacks_hi_<VP_small_lc>): Likewise.
10823 (vec_unpacks_lo_v8hi): Likewise.
10824 (vec_unpacks_lo_<VP_small_lc>): Likewise.
10825 (vec_pack_trunc_v8h): Likewise.
10826 (vec_pack_trunc_v4si): Likewise.
10827 (vec_pack_trunc_<mode>): Likewise.
10829 * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
10831 (vec_vmaxsd): Likewise.
10832 (vec_vmaxud): Likewise.
10833 (vec_vminsd): Likewise.
10834 (vec_vminud): Likewise.
10835 (vec_vpksdss): Likewise.
10836 (vec_vpksdus): Likewise.
10837 (vec_vpkudum): Likewise.
10838 (vec_vpkudus): Likewise.
10839 (vec_vrld): Likewise.
10840 (vec_vsld): Likewise.
10841 (vec_vsrad): Likewise.
10842 (vec_vsrd): Likewise.
10843 (vec_vsubudm): Likewise.
10844 (vec_vupkhsw): Likewise.
10845 (vec_vupklsw): Likewise.
10847 2013-05-29 Jan Hubicka <jh@suse.cz>
10849 * cgraph.h (symtab_node_base): Add definition, alias and analyzed
10850 flags; reorder rest of fields in more consistent way.
10851 (varpool_node): Remove analyzed, finalized and alias.
10852 (cgraph_ndoe): Likewise.
10853 (symtab_alias_ultimate_target): New function.
10854 (cgraph_function_node): Move offline.
10855 (cgraph_reset_node): Declare.
10856 (cgraph_comdat_can_be_unshared_p): Remove.
10857 (varpool_remove_initializer): Declare.
10858 (varpool_first_defined_variable, varpool_next_defined_variable
10859 cgraph_first_defined_function, cgraph_next_defined_function): Update.
10860 (cgraph_function_with_gimple_body_p): Update.
10861 (varpool_all_refs_explicit_p): Update.
10862 (symtab_alias_target): New function.
10863 (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
10864 (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
10865 (cgraph_function_or_thunk_node): Simplify using
10866 symtab_alias_ultimate_target.
10867 (varpool_variable_node): Likewise.
10868 * cgraph.c (cgraph_create_function_alias): Update.
10869 (cgraph_add_thunk): Update.
10870 (cgraph_remove_node): Update.
10871 (dump_cgraph_node): Do not dump removed flags.
10872 (cgraph_function_body_availability): Update.
10873 (cgraph_propagate_frequency): Update.
10874 (verify_cgraph_node): Check sanity of local flag.
10875 (cgraph_function_node): Move here from cgraph.h; revamp for
10876 cgraph_function_or_thunk_node.
10877 * lto-symtab.c (lto_varpool_replace_node): Update.
10878 (lto_symtab_resolve_can_prevail_p): Update.
10879 (lto_symtab_merge_cgraph_nodes): Update.
10880 * ipa-cp.c (determine_versionability, initialize_node_lattices,
10881 propagate_constants_accross_call, devirtualization_time_bonus,
10882 ipcp_propagate_stage): Update.
10883 * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
10884 * ipa-inline-transform.c (clone_inlined_nodes,
10885 preserve_function_body_p): Update.
10886 * ipa-reference.c (propagate): Update.
10887 (write_node_summary_p): Update.
10888 * toplev.c (wrapup_global_declaration_2): Update.
10889 * cgraphunit.c (cgraph_analyze_function): Rename to ...
10890 (analyze_function) ... this one.
10891 (cgraph_process_new_functions): Update.
10892 (cgraph_reset_node): Export.
10893 (cgraph_finalize_function): Update.
10894 (cgraph_add_new_function): Update.
10895 (process_function_and_variable_attributes): Update.
10896 (varpool_finalize_decl): Update.
10897 (symbol_finalized): Remove.
10898 (symbol_finalized_and_needed): Rename to ...
10899 (symbol_defined_and_needed): ... update.
10900 (cgraph_analyze_functions): Update.
10901 (handle_alias_pairs): Update.
10902 (mark_functions_to_output): Update.
10903 (assemble_thunk): Update.
10904 (output_in_order): Update.
10905 (output_weakrefs): Update.
10906 (finalize_compilation_unit): Update.
10907 * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
10908 lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
10909 input_node, input_varpool_node): Update.
10910 * dbxout.c (dbxout_expand_expr): Update.
10911 * cgraphclones.c (cgraph_clone_node): Update.
10912 (cgraph_copy_node_for_versioning): Update.
10913 (cgraph_materialize_clone): Update.
10914 (cgraph_materialize_all_clones): Update.
10915 * ipa-pure-const.c (analyze_function, pure_const_write_summary,
10916 propagate_pure_const, propagate_nothrow): Update.
10917 * lto-streamer-out.c (lto_output, write_symbol): Update.
10918 * ipa-utils.c (ipa_reverse_postorder): Update.
10919 * ipa-inline.c (can_inline_edge_p): Update.
10920 (update_caller_keys, ipa_inline): Update.
10921 * dwarf2out.c (reference_to_unused,
10922 premark_types_used_by_global_vars_helper): Update.
10923 * tree-eh.c (tree_could_trap_p): Update.
10924 * ipa-split.c (consider_split, execute_split_functions): Update.
10925 * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
10926 has_addr_references_p): Update; move ahead in file for better
10928 (process_references): Simplify.
10929 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
10930 bodies are removed.
10931 (cgraph_comdat_can_be_unshared_p): Make static.
10932 (cgraph_externally_visible_p): Update.
10933 (varpool_externally_visible_p): Update.
10934 (function_and_variable_visibility): Update.
10935 * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
10936 ipa_tm_mark_force_output_node): Update.
10937 * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
10938 estimate_edge_devirt_benefit, inline_generate_summary,
10939 inline_write_summary): Update.
10940 * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
10941 * ipa-prop.c (ipa_compute_jump_functions): Update.
10942 (ipa_print_node_params, ipa_prop_read_section,
10943 ipa_update_after_lto_read, read_replacements_section): Update.
10944 * varasm.c (mark_decl_referenced): Update.
10945 (assemble_alias, dump_tm_clone_pairs): Update.
10946 * tree-inline.c (copy_bb): Update.
10947 (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
10949 * symtab.c (dump_symtab_base): Print new flags.
10950 (verify_symtab_base): Verify new flags.
10951 (symtab_alias_ultimate_target): New function.
10952 * tree-ssa-structalias.c (get_constraint_for_ssa_var,
10953 create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
10955 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
10957 * i386.c (ix86_get_function_versions_dispatcher,
10958 ix86_generate_version_dispatcher_body): Update.
10959 (fold_builtin_cpu): Use varpool_add_new_variable.
10960 * varpool.c (varpool_remove_initializer): Break out from ...
10961 (varpool_remove_node): ... this one.
10962 (dump_varpool_node, varpool_node_for_asm,
10963 cgraph_variable_initializer_availability, varpool_analyze_node,
10964 varpool_assemble_decl, varpool_remove_unreferenced_decls,
10965 varpool_finalize_named_section_flags, varpool_create_variable_alias):
10968 2013-05-29 Jan Hubicka <jh@suse.cz>
10970 * passes.c (init_optimization_passes): Move OMP expansion into lowering.
10972 2013-05-29 Easwaran Raman <eraman@google.com>
10974 PR tree-optimization/57442
10975 * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
10976 when control exits the main loop.
10978 2013-05-29 Sandeep Kumar Singh <Sandeep.Singh2@kpitcummins.com>
10980 * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
10982 * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
10983 * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
10984 * rx/t-rx: Add rx100 under multi library matches option for nofpu
10987 2013-05-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10989 PR tree-optimization/57441
10990 * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
10991 Don't limit size of incr_vec to number of candidates.
10993 2013-05-29 Steve Ellcey <sellcey@imgtec.com>
10995 * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
10996 and mips16 directories.
10997 * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
10998 (MULTILIB_DIRNAMES): Ditto.
10999 (MULTILIB_EXCEPTIONS): Add new exceptions.
11000 * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
11001 (MULTILIB_DIRNAMES): Ditto.
11002 (MULTILIB_EXCEPTIONS): Add new exceptions.
11004 2012-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
11005 Marcus Shawcroft <marcus.shawcroft@arm.com>
11007 * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
11008 SYMBOL_TINY_ABSOLUTE.
11009 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
11010 SYMBOL_TINY_ABSOLUTE.
11011 (aarch64_expand_mov_immediate): Likewise.
11012 (aarch64_classify_symbol): Likewise.
11013 (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
11014 Permit SYMBOL_TINY_ABSOLUTE.
11015 * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
11017 2013-05-29 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
11018 Marcus Shawcroft <marcus.shawcroft@arm.com>
11020 * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
11021 Refactor if/switch. Replace gcc_assert with if.
11023 2013-05-29 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
11025 * config/i386/i386.c (initial_ix86_tune_features): Enable
11026 FP Reassociation for AMD bdver1 and bdver2.
11028 2013-05-29 Martin Jambor <mjambor@suse.cz>
11030 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
11031 and IMAGPART_EXPR do not occur within other handled_components.
11033 2013-05-29 Richard Biener <rguenther@suse.de>
11035 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
11036 access on whether the use is in the BB we currently try to
11038 (vect_bb_vectorization_profitable_p): Pass the BB we currently
11039 vectorize to vect_bb_slp_scalar_cost.
11041 2013-05-29 Richard Biener <rguenther@suse.de>
11043 * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
11044 computing scalar cost offsetted by stmts that are kept live
11046 (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
11047 for computation of scalar cost.
11049 2013-05-28 Steve Ellcey <sellcey@mips.com>
11051 * config/mips/mips-cpus.def (mips32r2): Change processor type.
11053 2013-05-28 Balaji V. Iyer <balaji.v.iyer@intel.com>
11055 * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
11056 array notation built-in reduction functions.
11057 * doc/passes.texi (Passes): Added documentation about changes done
11059 * doc/invoke.texi (C Dialect Options): Added documentation about
11060 the -fcilkplus flag.
11061 * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
11062 (BUILTINS_DEF): Depend on cilkplus.def.
11063 * builtins.def: Include cilkplus.def. Define DEF_CILKPLUS_BUILTIN.
11064 * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
11065 * cilkplus.def: New file.
11067 2013-05-28 Joern Rennecke <joern.rennecke@embecosm.com>
11069 PR rtl-optimization/57439
11070 * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
11072 2013-05-28 Easwaran Raman <eraman@google.com>
11074 PR tree-optimization/57337
11075 * tree-ssa-reassoc.c (appears_later_in_bb): New function.
11076 (find_insert_point): Correctly identify the insertion point
11077 when two statements with the same UID is compared.
11079 2013-05-28 Richard Biener <rguenther@suse.de>
11081 PR tree-optimization/56787
11082 * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
11083 from the list of data references.
11084 * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
11086 (vect_analyze_loop_operations): Likewise.
11087 (vect_transform_loop): Remove clobbers.
11089 2013-05-28 Martin Jambor <mjambor@suse.cz>
11091 * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
11092 and REALPART_EXPRs have scalar type.
11094 2013-05-28 Richard Biener <rguenther@suse.de>
11096 PR tree-optimization/57411
11097 * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
11099 * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
11100 virtual operand propagation.
11102 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
11104 * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
11105 destination register for bmasksi_vis.
11106 (vector_init_bshuffle): Likewise.
11107 * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
11109 2013-05-28 Eric Botcazou <ebotcazou@adacore.com>
11111 * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
11112 * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
11113 mode if the instruction isn't available in the original mode.
11114 * config/sparc/sparc.opt (mfix-ut699): New option.
11115 * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
11116 (divdf3): Turn into expander.
11117 (divdf3_nofix): New insn.
11118 (divdf3_fix): Likewise.
11119 (divsf3): Disable if -mfix-ut699.
11120 (sqrtdf2): Turn into expander.
11121 (sqrtdf2_nofix): New insn.
11122 (sqrtdf2_fix): Likewise.
11123 (sqrtsf2): Disable if -mfix-ut699.
11125 2013-05-27 Richard Biener <rguenther@suse.de>
11127 PR middle-end/57412
11128 * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
11129 block for the new loop.
11131 2013-05-27 Richard Biener <rguenther@suse.de>
11133 PR tree-optimization/57343
11134 * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
11135 use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
11136 (number_of_iterations_cond): Do not build the folded tree.
11138 2013-05-27 Richard Biener <rguenther@suse.de>
11141 PR middle-end/57381
11142 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
11143 OEP_CONSTANT_ADDRESS_OF retained.
11145 PR tree-optimization/57417
11146 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
11147 for unchanged base.
11148 (set_ssa_val_to): Compare addresses using
11149 get_addr_base_and_unit_offset.
11151 2013-05-27 Joern Rennecke <joern.rennecke@embecosm.com>
11153 PR rtl-optimization/56833
11154 * postreload.c (move2add_record_mode): New function.
11155 (move2add_record_sym_value, move2add_valid_value_p): Likewise.
11156 (move2add_use_add2_insn): Use move2add_record_sym_value.
11157 (move2add_use_add3_insn): Likewise.
11158 (reload_cse_move2add): Use move2add_valid_value_p and
11159 move2add_record_mode. Invalidate call-clobbered and REG_INC
11160 affected regs by setting reg_mode to VOIDmode.
11161 (move2add_note_store): Don't pretend the inside of a SUBREG is
11162 the actual destination. Invalidate single/leading registers by
11163 setting reg_mode to VOIDmode.
11164 Use move2add_record_sym_value, move2add_valid_value_p and
11165 move2add_record_mode.
11167 2013-05-27 Richard Biener <rguenther@suse.de>
11169 PR tree-optimization/57396
11170 * tree-affine.c (double_int_constant_multiple_p): Properly
11171 return false for val == 0 and div != 0.
11173 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
11175 * config/mips/mips.h: Use #elif in preprocessor conditions.
11177 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
11180 * config/mips/constraints.md (kl): New constraint.
11181 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
11182 (divmod<mode>4_internal): Rename to divmod<mode>4. Use "kl" as the
11183 constraint for operand 0. Split after CSE for MIPS16. Emit a move
11184 from LO for MIPS16.
11185 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
11187 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
11190 * config/mips/mips.c (mips_can_inline_p): New function.
11191 (TARGET_CAN_INLINE_P): Define.
11193 2013-05-25 Steven Bosscher <steven@gcc.gnu.org>
11195 * sched-int.h (ds_t, dw_t): Make unsigned int.
11196 Fix documentation that describes how all the ds_t bits are used.
11197 Reserve the last bit for delayed-branch scheduling.
11198 (BITS_PER_DEP_STATUS): Move to ds_t typedef.
11199 (BITS_PER_DEP_WEAK): Fix definition and documentation.
11200 (gen_dep_weak_1): Remove prototype.
11201 * sched-deps.c (get_dep_weak_1): Make static.
11202 * target.def (speculate_insn, needs_block_p, gen_spec_check,
11203 get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
11204 * doc/tm.texi: Regenerate.
11205 * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
11207 2013-05-24 Steven Bosscher <steven@gcc.gnu.org>
11210 * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
11212 2013-05-24 Nathan Sidwell <nathan@codesourcery.com>
11213 Sandra Loosemore <sandra@codesourcery.com>
11215 * config.gcc (powerpc-*): Allow native for with-cpu.
11217 2013-05-24 Jeff Law <law@redhat.com>
11219 PR tree-optimization/57124
11220 * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
11221 conversion feeding a condition if the range has an overflow
11222 if -fstrict-overflow. Add warnings for when we do make the
11225 2013-05-24 Dehao Chen <dehao@google.com>
11227 * tree-cfg.c (locus_discrim_map): Fix the typo.
11228 (locus_discrim_hasher): Likewise.
11229 (locus_discrim_hasher::hash): Likewise.
11230 (locus_discrim_hasher::equal): Likewise.
11232 2013-05-24 Martin Jambor <mjambor@suse.cz>
11234 PR tree-optimization/57294
11235 * cgraph.h (ipa_record_stmt_references): Declare.
11236 * cgraphbuild.c (ipa_record_stmt_references): New function.
11237 (build_cgraph_edges): Use ipa_record_stmt_references.
11238 (rebuild_cgraph_edges): Likewise.
11239 (cgraph_rebuild_references): Likewise.
11240 * ipa-prop.c (ipa_modify_call_arguments): Discard references
11241 associated with the old statement and build references from the
11242 newly built statements.
11243 * ipa-ref.c (ipa_remove_stmt_references): New function.
11244 * ipa-ref.h (ipa_remove_stmt_references): Declare.
11246 2013-05-24 Vladimir Makarov <vmakarov@redhat.com>
11248 * lra-constraints.c (emit_spill_move): Use smaller mode for
11250 (check_and_process_move): Consider mem-reg moves for secondary
11252 (curr_insn_transform): Don't lose insns emitted before for
11253 secondary memory moves.
11254 (inherit_in_ebb): Mark defined reg. Add usage only if it is not a
11255 reg set up in the current insn.
11257 2013-05-24 Dehao Chen <dehao@google.com>
11259 * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
11261 (locus_descrim_hasher::equal): Likewise.
11262 (build_gimple_cfg): New discriminator assignment algorithm.
11263 (make_edges): Likewise.
11264 (next_discriminator_for_locus): Likewise.
11265 (same_line_p): Likewise.
11266 (assign_discriminators): Likewise.
11267 (make_cond_expr_edges): Likewise.
11268 (make_gimple_switch_edges): Likewise.
11269 (make_goto_expr_edges): Likewise.
11270 (make_gimple_asm_edges): Likewise.
11272 2013-05-24 Ian Bolton <ian.bolton@arm.com>
11274 * config/aarch64/aarch64.c (aarch64_print_operand): Change the
11275 X format specifier to only display bottom 16 bits.
11276 * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
11277 immediate to match for operand 2, since it will be masked.
11279 2013-05-24 Richard Biener <rguenther@suse.de>
11281 PR tree-optimization/57287
11282 * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
11283 all SSA names that occur in abnormal PHIs.
11285 2013-05-24 Alexander Ivchenko <alexander.ivchenko@intel.com>
11288 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
11289 that index is not negative.
11291 2013-05-24 Eric Botcazou <ebotcazou@adacore.com>
11293 PR rtl-optimization/55177
11294 * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
11295 (simplify_byte_swapping_operation): New.
11296 (simplify_binary_operation_1): Call it for AND, IOR and XOR.
11297 (simplify_relational_operation_1): Deal with BSWAP.
11299 2013-05-23 Richard Henderson <rth@redhat.com>
11302 * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
11303 (ix86_reorg): Call it.
11305 2013-05-23 Uros Bizjak <ubizjak@gmail.com>
11308 * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
11309 * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
11310 REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
11312 2013-05-23 Christian Bruel <christian.bruel@st.com>
11315 * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
11317 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
11318 Marcus Shawcroft <marcus.shawcroft@arm.com>
11320 * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
11321 * config/aarch64/constraints.md (Usa): Remove.
11322 * doc/md.texi (AArch64 Usa): Remove.
11324 2013-05-23 Chris Schlumberger-Socha <chris.schlumberger-socha@arm.com>
11325 Marcus Shawcroft <marcus.shawcroft@arm.com>
11327 * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
11328 * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
11329 * config/aarch64/predicates.md (aarch64_const_address): Remove.
11330 (aarch64_mov_operand): Use aarch64_mov_operand_p.
11332 2013-05-23 Vidya Praveen <vidyapraveen@arm.com>
11334 * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
11335 instruction (AdvSIMD).
11336 * config/aarch64/aarch64-builtins.c
11337 (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
11338 * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
11340 2013-05-23 Martin Jambor <mjambor@suse.cz>
11342 PR middle-end/57347
11343 * tree.h (contains_bitfld_component_ref_p): Declare.
11344 * tree-sra.c (contains_bitfld_comp_ref_p): Move...
11345 * tree.c (contains_bitfld_component_ref_p): ...here. Adjust its
11347 * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
11348 not access a bit-field. Assert all final offsets are byte-aligned.
11350 2013-05-23 Richard Biener <rguenther@suse.de>
11352 PR tree-optimization/57380
11353 * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
11354 least one invariant or re-used load.
11355 * passes.c (init_optimization_passes): Move pass_phiprop before
11358 2013-05-23 James Greenhalgh <james.greenhalgh@arm.com>
11360 * config/aarch64/aarch64-simd.md
11361 (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
11363 2013-05-23 Richard Biener <rguenther@suse.de>
11365 PR middle-end/57381
11366 * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
11367 OEP_CONSTANT_ADDRESS_OF retained.
11369 2013-05-23 Jakub Jelinek <jakub@redhat.com>
11371 PR middle-end/57344
11372 * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
11373 don't lower unit. Handle unit not being always BITS_PER_WORD.
11375 2013-05-23 Richard Biener <rguenther@suse.de>
11377 PR rtl-optimization/57341
11378 * ira.c (validate_equiv_mem_from_store): Use anti_dependence
11379 instead of true_dependence.
11381 2013-05-22 David Malcolm <dmalcolm@redhat.com>
11383 * bb-reorder.c (branch_threshold): Make const.
11384 (exec_threshold): Ditto.
11386 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11387 Pat Haugen <pthaugen@us.ibm.com>
11388 Peter Bergner <bergner@vnet.ibm.com>
11390 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
11391 documentation for the power8 crypto builtins.
11393 * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
11395 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
11396 macros for defining power8 builtin functions.
11397 (BU_P8V_AV_2): Likewise.
11398 (BU_P8V_AV_P): Likewise.
11399 (BU_P8V_VSX_1): Likewise.
11400 (BU_P8V_OVERLOAD_1): Likewise.
11401 (BU_P8V_OVERLOAD_2): Likewise.
11402 (BU_CRYPTO_1): Likewise.
11403 (BU_CRYPTO_2): Likewise.
11404 (BU_CRYPTO_3): Likewise.
11405 (BU_CRYPTO_OVERLOAD_1): Likewise.
11406 (BU_CRYPTO_OVERLOAD_2): Likewise.
11407 (XSCVSPDP): Fix typo, point to the correct instruction.
11408 (VCIPHER): Add power8 crypto builtins.
11409 (VCIPHERLAST): Likewise.
11410 (VNCIPHER): Likewise.
11411 (VNCIPHERLAST): Likewise.
11412 (VPMSUMB): Likewise.
11413 (VPMSUMH): Likewise.
11414 (VPMSUMW): Likewise.
11415 (VPERMXOR_V2DI): Likewise.
11416 (VPERMXOR_V4SI: Likewise.
11417 (VPERMXOR_V8HI: Likewise.
11418 (VPERMXOR_V16QI: Likewise.
11419 (VSHASIGMAW): Likewise.
11420 (VSHASIGMAD): Likewise.
11421 (VPMSUM): Likewise.
11422 (VPERMXOR): Likewise.
11423 (VSHASIGMA): Likewise.
11425 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
11426 __CRYPTO__ if the crypto instructions are available.
11427 (altivec_overloaded_builtins): Add support for overloaded power8
11430 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
11431 support for power8 crypto builtins.
11432 (builtin_function_type): Likewise.
11433 (altivec_init_builtins): Add support for builtins that take vector
11434 long long (V2DI) arguments.
11436 * config/rs6000/crypto.md: New file, define power8 crypto
11439 2013-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11440 Pat Haugen <pthaugen@us.ibm.com>
11441 Peter Bergner <bergner@vnet.ibm.com>
11443 * doc/invoke.texi (Option Summary): Add power8 options.
11444 (RS/6000 and PowerPC Options): Likewise.
11446 * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
11447 constraints.md instead of rs6000.h. Reorder w* constraints. Add
11448 wm, wn, wr documentation.
11450 * config/rs6000/constraints.md (wm): New constraint for VSX
11451 registers if direct move instructions are enabled.
11452 (wn): New constraint for no registers.
11453 (wq): New constraint for quad word even GPR registers.
11454 (wr): New constraint if 64-bit instructions are enabled.
11455 (wv): New constraint if power8 vector instructions are enabled.
11456 (wQ): New constraint for quad word memory locations.
11458 * config/rs6000/predicates.md (const_0_to_15_operand): New
11459 constraint for 0..15 for crypto instructions.
11460 (gpc_reg_operand): If VSX allow registers in VSX registers as well
11461 as GPR and floating point registers.
11462 (int_reg_operand): New predicate to match only GPR registers.
11463 (base_reg_operand): New predicate to match base registers.
11464 (quad_int_reg_operand): New predicate to match even GPR registers
11465 for quad memory operations.
11466 (vsx_reg_or_cint_operand): New predicate to allow vector logical
11467 operations in both GPR and VSX registers.
11468 (quad_memory_operand): New predicate for quad memory operations.
11469 (reg_or_indexed_operand): New predicate for direct move support.
11471 * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
11472 Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
11473 (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
11474 (POWERPC_MASKS): Add power8 options.
11475 (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
11478 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
11479 Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
11481 * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
11482 (-mpower8-fusion): New power8 options.
11483 (-mpower8-fusion-sign): Likewise.
11484 (-mpower8-vector): Likewise.
11485 (-mcrypto): Likewise.
11486 (-mdirect-move): Likewise.
11487 (-mquad-memory): Likewise.
11489 * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
11490 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
11492 (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
11493 (rs6000_debug_vector_unit): Add p8_vector.
11494 (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
11495 definitions. Also print fusion state.
11496 (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
11497 (rs6000_builtin_mask_calculate): Add power8 builtin support.
11498 (rs6000_option_override_internal): Add support for power8.
11499 (rs6000_common_init_builtins): Add debugging for skipped builtins
11500 if -mdebug=builtin.
11501 (rs6000_adjust_cost): Add power8 support.
11502 (rs6000_issue_rate): Likewise.
11503 (insn_must_be_first_in_group): Likewise.
11504 (insn_must_be_last_in_group): Likewise.
11505 (force_new_group): Likewise.
11506 (rs6000_register_move_cost): Likewise.
11507 (rs6000_opt_masks): Likewise.
11509 * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
11510 power8 capable assembler, default to power7 options.
11511 (TARGET_DIRECT_MOVE): Likewise.
11512 (TARGET_CRYPTO): Likewise.
11513 (TARGET_P8_VECTOR): Likewise.
11514 (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
11515 (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
11516 (VECTOR_MEM_P8_VECTOR_P): Likewise.
11517 (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
11518 (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
11519 (TARGET_XSCVDPSPN): Likewise.
11520 (TARGET_XSCVSPDPN): Likewsie.
11521 (TARGET_SYNC_HI_QI): Likewise.
11522 (TARGET_SYNC_TI): Likewise.
11523 (MASK_CRYPTO): Likewise.
11524 (MASK_DIRECT_MOVE): Likewise.
11525 (MASK_P8_FUSION): Likewise.
11526 (MASK_P8_VECTOR): Likewise.
11527 (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
11528 temporary used by some of the direct move instructions to get two FP
11529 temporary registers does not force creation of a stack frame.
11530 (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
11531 (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
11532 that any VSX registers are tieable, even if they are also an
11533 Altivec vector mode.
11534 (r6000_reg_class_enum): Add wm, wr, wv constraints.
11535 (RS6000_BTM_P8_VECTOR): Power8 builtin support.
11536 (RS6000_BTM_CRYPTO): Likewise.
11537 (RS6000_BTM_COMMON): Likewise.
11539 * config/rs6000/rs6000.md (cpu attribute): Add power8.
11540 * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
11541 (enum rs6000_vector): Add power8 vector support.
11543 2013-05-22 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11547 * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
11548 (any_sibcall_could_use_r3): this and handle indirect calls.
11549 (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
11551 2013-05-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11553 * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
11555 2013-05-22 Richard Biener <rguenther@suse.de>
11557 PR middle-end/57349
11558 * profile.c (branch_prob): Do not split blocks that are
11559 abnormally receiving from ECF_RETURNS_TWICE functions.
11561 2013-05-22 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
11563 * recog.c (offsettable_address_addr_space_p): Fix calculation of
11564 address mode. Move pointer mode initialization to the same place.
11566 2013-05-22 Michael Zolotukhin <michael.v.zolotukhin@intel.com>
11568 * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
11569 while it has any effect.
11571 2013-05-21 Easwaran Raman <eraman@google.com>
11573 PR tree-optimization/57322
11574 * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
11575 UID of the statement added to the BB to be 1.
11577 2013-05-21 Jakub Jelinek <jakub@redhat.com>
11579 PR tree-optimization/57331
11580 * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
11581 of conversion from pointer type to integral type with integer.
11583 2013-05-21 Martin Jambor <mjambor@suse.cz>
11586 * ipa-prop.c (ipa_read_node_info): Process param_used and
11587 controlled_uses in the same order as when writing.
11589 2013-05-21 Magnus Granberg <baldrick@free.fr>
11592 * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
11594 2013-05-21 Richard Biener <rguenther@suse.de>
11596 PR tree-optimization/57318
11597 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
11598 estimate stmts with side-effects as likely eliminated.
11600 2013-05-21 Richard Biener <rguenther@suse.de>
11602 PR tree-optimization/57330
11603 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
11604 preserve the call stmts fntype.
11606 2013-05-21 Richard Biener <rguenther@suse.de>
11608 PR tree-optimization/57303
11609 * tree-ssa-sink.c (statement_sink_location): Improve killing
11610 stmt detection and properly handle self-assignments.
11612 2013-05-21 Christian Bruel <christian.bruel@st.com>
11614 * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
11615 spanning registers. LEAF_REG_REMAP is supported only for contiguous
11616 registers. Set register size out of the PARALLEL loop.
11618 2013-05-20 Oleg Endo <olegendo@gcc.gnu.org>
11621 * config/sh/sh.md (fmasf4): Remove empty constraints strings.
11622 (*fmasf4, *fmasf4_media): New insns.
11624 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
11626 * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
11627 * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
11628 (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
11629 (mips_idiv_insns): Update the comments to say that the returned
11630 instruction counts are in units of BASE_INSN_LENGTH.
11631 (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
11632 by BASE_INSN_LENGTH rather than 4. Add the jump separately,
11633 using 2 rather than 4 as the length of indirect MIPS16 and
11634 microMIPS jumps. Use NOP_INSN_LENGTH rather than 4 as the
11635 length of a NOP. Don't divide MIPS16 lengths by 2.
11636 (mips16_split_long_branches): Assume a branch is long if the
11637 length is greater than 4 rather than 8.
11638 * config/mips/mips.md (length): Give MIPS16 lengths directly,
11639 rather than multiplying them by 2. Multiply instruction counts
11640 by BASE_INSN_LENGTH rather than 4.
11641 (*jump_mips16, tls_get_tp_mips16_<mode>)
11642 (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
11644 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
11646 * config/mips/mips.md (extended_mips16): Remove branch case.
11647 (length): Remove duplicated extended_mips16 test.
11649 2013-05-19 Richard Sandiford <rdsandiford@googlemail.com>
11651 * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
11653 2013-05-18 Richard Sandiford <rdsandiford@googlemail.com>
11655 * recog.h (Recog_data): Rename to...
11656 (recog_data_d): ...this.
11657 (recog_data): Update accordingly.
11658 * recog.c (recog_data): Likewise.
11659 * reload.c (save_recog_data): Likewise.
11660 * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
11661 (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
11663 2013-05-17 Julian Brown <julian@codesourcery.com>
11665 * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
11666 found in a REG_EQUAL note, invalidate it.
11668 2013-05-17 Easwaran Raman <eraman@google.com>
11670 * tree-ssa-reassoc.c (find_insert_point): New function.
11671 (insert_stmt_after): Likewise.
11672 (get_def_stmt): Likewise.
11673 (ensure_ops_are_available): Likewise.
11674 (not_dominated_by): Likewise.
11675 (rewrite_expr_tree): Do not move statements beyond what is
11676 necessary. Remove call to swap_ops_for_binary_stmt...
11677 (reassociate_bb): ... and move it here.
11678 (build_and_add_sum): Assign UIDs for new statements.
11679 (linearize_expr): Likewise.
11680 (do_reassoc): Renumber gimple statement UIDs.
11682 2013-05-17 Jan Hubicka <jh@suse.cz>
11684 * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
11686 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
11687 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
11689 (output_weakrefs): Update.
11691 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
11692 Martin Jambor <mjambor@suse.cz>
11694 PR middle-end/57276
11695 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
11696 value that corresponds to the given aggval is found in values vector.
11698 2013-05-17 Uros Bizjak <ubizjak@gmail.com>
11700 * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
11701 sse, sse2, sse3, ssse3 and sse4a flags to options.
11703 2013-05-17 David Malcolm <dmalcolm@redhat.com>
11705 * gengtype-state.c: (s_expr_writer): New class, to handle
11706 prettifying of output layout of s-expressions.
11707 (state_writer): New class, to write out gtype.state.
11708 (state_written_type_count): Move this variable into member data of
11710 (s_expr_writer::s_expr_writer): New code: constructor for new class
11711 (state_writer::state_writer(): ditto
11712 (s_expr_writer::write_new_line): New function
11713 (s_expr_writer::write_any_indent): ditto
11714 (s_expr_writer::begin_s_expr): ditto
11715 (s_expr_writer::end_s_expr): ditto
11716 (write_state_fileloc): convert to method of state_writer...
11717 (state_writer:: write_state_fileloc): ...and use methods of
11718 s_expr_writer to write indentation into the gtype.state output file
11719 to visually represent the hierarchical structure of the list
11721 (write_state_fields): ditto, renaming to...
11722 (state_writer::write_state_fields)
11723 (write_state_a_string): ditto, renaming to...
11724 (state_writer::write_state_a_string)
11725 (write_state_string_option): ditto, renaming to...
11726 (state_writer::write_state_string_option)
11727 (write_state_type_option): ditto, renaming to...
11728 (state_writer::write_state_type_option)
11729 (write_state_nested_option): ditto, renaming to...
11730 (state_writer::write_state_nested_option)
11731 (write_state_option): ditto, renaming to...
11732 (state_writer::write_state_option)
11733 (write_state_options): ditto, renaming to...
11734 (state_writer::write_state_options)
11735 (write_state_lang_bitmap): ditto, renaming to...
11736 (state_writer::write_state_lang_bitmap)
11737 (write_state_version): ditto, renaming to...
11738 (state_writer::write_state_version)
11739 (write_state_scalar_type): ditto, renaming to...
11740 (state_writer::write_state_scalar_type)
11741 (write_state_string_type): ditto, renaming to...
11742 (state_writer::write_state_string_type)
11743 (write_state_undefined_type): ditto, renaming to...
11744 (state_writer::write_state_undefined_type)
11745 (write_state_struct_union_type): ditto, renaming to...
11746 (state_writer::write_state_struct_union_type)
11747 (write_state_struct_type): ditto, renaming to...
11748 (state_writer::write_state_struct_type)
11749 (write_state_user_struct_type): ditto, renaming to...
11750 (state_writer::write_state_user_struct_type)
11751 (write_state_lang_struct_type): ditto, renaming to...
11752 (state_writer::write_state_lang_struct_type)
11753 (write_state_param_struct_type): ditto, renaming to...
11754 (state_writer::write_state_param_struct_type)
11755 (write_state_pointer_type): ditto, renaming to...
11756 (state_writer::write_state_pointer_type)
11757 (write_state_array_type): ditto, renaming to...
11758 (state_writer::write_state_array_type)
11759 (write_state_gc_used): ditto, renaming to...
11760 (state_writer::write_state_gc_used)
11761 (write_state_common_type_content): ditto, renaming to...
11762 (state_writer::write_state_common_type_content)
11763 (write_state_type): ditto, renaming to...
11764 (state_writer::write_state_type)
11765 (write_state_pair_list): ditto, renaming to...
11766 (state_writer::write_state_pair_list)
11767 (write_state_pair): ditto, renaming to...
11768 (state_writer::write_state_pair)
11769 (write_state_typedefs): ditto, renaming to...
11770 (state_writer::write_state_typedefs)
11771 (write_state_structures): ditto, renaming to...
11772 (state_writer::write_state_structures)
11773 (write_state_param_structs): ditto, renaming to...
11774 (state_writer::write_state_param_structs)
11775 (write_state_variables): ditto, renaming to...
11776 (state_writer::write_state_variables)
11777 (write_state_srcdir): ditto, renaming to...
11778 (state_writer::write_state_srcdir)
11779 (write_state_files_list): ditto, renaming to...
11780 (state_writer::write_state_files_list)
11781 (write_state_languages): ditto, renaming to...
11782 (state_writer::write_state_languages)
11783 (write_state): create a state_writer instance and use it when
11784 writing out the state file
11786 2013-05-17 Mike Stump <mikestump@comcast.net>
11788 PR rtl-optimization/57304
11789 * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
11790 accessing DF_REF_REAL_LOC.
11792 2013-05-17 Jakub Jelinek <jakub@redhat.com>
11794 PR rtl-optimization/57281
11795 PR rtl-optimization/57300
11796 * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
11797 (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
11798 what the other splitter did if the registers are dead.
11800 2013-05-17 Richard Biener <rguenther@suse.de>
11802 * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
11805 2013-05-17 Jakub Jelinek <jakub@redhat.com>
11807 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
11810 2013-05-17 Marek Polacek <polacek@redhat.com>
11812 * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
11813 length when doing non-zero store of storing '\0' to '\0'.
11815 2013-05-17 Jakub Jelinek <jakub@redhat.com>
11817 * tree-vect-patterns.c (vect_recog_rotate_pattern): For
11818 vect_external_def oprnd1 with loop_vinfo, try to emit
11819 optional cast, negation and and stmts on the loop preheader
11820 edge instead of into the pattern def seq.
11822 PR tree-optimization/57051
11823 * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
11824 case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
11826 2013-05-16 Nick Clifton <nickc@redhat.com>
11828 * config/rl78/rl78.c (rl78_attribute_table): Add naked.
11829 (rl78_is_naked_func): New function.
11830 (rl78_expand_prologue): Skip prologue generation for naked functions.
11831 (rl78_expand_epilogue): Skip epilogue generation for naked functions.
11832 * doc/extend.texi (naked): Add RL78 to the list of processors
11833 that supports this attribute.
11835 2013-05-16 Jeff Law <law@redhat.com>
11837 * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
11839 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
11841 * config/i386/driver-i386.c (host_detect_local_cpu): Determine
11842 cache parameters using detect_caches_amd also for CYRIX,
11843 NSC and TM2 signatures.
11845 2013-05-16 Uros Bizjak <ubizjak@gmail.com>
11846 Dzianis Kahanovich <mahatma@eu.by>
11850 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11851 VIA/Centaur processors and determine their cache parameters
11852 using detect_caches_amd.
11854 2013-05-16 Teresa Johnson <tejohnson@google.com>
11856 * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
11857 (rtl_verify_edges): New function.
11858 (rtl_verify_bb_insns): Ditto.
11859 (rtl_verify_bb_pointers): Ditto.
11860 (rtl_verify_bb_insn_chain): Ditto.
11861 (rtl_verify_fallthru): Ditto.
11862 (rtl_verify_bb_layout): Ditto.
11863 (rtl_verify_flow_info_1): Outline checks into new functions.
11864 (rtl_verify_flow_info): Ditto.
11866 2013-05-16 Steve Ellcey <sellcey@imgtec.com>
11868 * cfghooks.c (copy_bbs): Add update_dominance argument.
11869 * cfghooks.h (copy_bbs): Update prototype.
11870 * tree-cfg.c (gimple_duplicate_sese_region):
11871 Add update_dominance argument.
11872 * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
11873 * tree-ssa-loop-ch.c (copy_loop_headers): Update
11874 gimple_duplicate_sese_region call.
11875 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
11876 Update copy_bbs call.
11877 * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
11878 * trans-mem.c (ipa_uninstrument_transaction): Ditto.
11880 2013-05-16 Jakub Jelinek <jakub@redhat.com>
11882 * tree-vectorizer.h (NUM_PATTERNS): Increment.
11883 * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
11884 vect_recog_rotate_pattern.
11885 (vect_recog_rotate_pattern): New function.
11887 2013-05-16 Jason Merrill <jason@redhat.com>
11889 * Makefile.in (LLINKER): New variable.
11890 (mostlyclean): Remove link mutex.
11891 * configure.ac: Handle --enable-link-mutex.
11892 * lock-and-run.sh: New script.
11894 2013-05-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11897 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
11900 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11902 * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
11904 2013-05-16 Greta Yorsh <Greta.Yorsh@arm.com>
11906 * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
11907 * config/arm/arm.c (next_consecutive_mem): New function.
11908 (gen_movmem_ldrd_strd): Likewise.
11909 * config/arm/arm.md (movmemqi): Update condition and code.
11910 (unaligned_loaddi, unaligned_storedi): New patterns.
11912 2013-05-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11914 * config.gcc: Obsolete *-*-solaris2.9*.
11915 * doc/install.texi (Specific, *-*-solaris2*): Document it.
11917 2013-05-16 Richard Biener <rguenther@suse.de>
11919 * passes.c (init_optimization_passes): Move pass_parallelize_loops
11920 earlier, after GRAPHITE transforms and IV canonicalization.
11922 2013-05-16 Jakub Jelinek <jakub@redhat.com>
11924 * omp-low.c (extract_omp_for_data): For collapsed loops,
11925 if at least one of the loops is known at compile time to
11926 iterate zero times, set count to 0.
11927 (expand_omp_regimplify_p): New function.
11928 (expand_omp_for_generic): For collapsed loops, if at least
11929 one of the loops isn't known to iterate at least once,
11930 add runtime check with setting count to 0.
11931 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
11932 For unsigned types if it isn't known at compile time that
11933 the loop will iterate at least once, add runtime check to bypass
11934 the whole loop if initial condition isn't true.
11936 2013-05-16 Nathan Sidwell <nathan@codesourcery.com>
11938 * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
11940 2013-05-16 Marc Glisse <marc.glisse@inria.fr>
11942 PR middle-end/57286
11943 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
11944 transformations to avoid an infinite loop.
11946 2013-05-16 Marek Polacek <polacek@redhat.com>
11948 * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
11950 2013-05-15 Leif Ekblad <leif@rdos.net>
11952 * config/i386/i386.c (ix86_decompose_address): Use
11953 DEFAULT_TLS_SEG_REG to access TLS segment register.
11954 * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
11955 * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
11956 (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
11958 2013-05-15 Richard Sandiford <rdsandiford@googlemail.com>
11961 * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
11962 sibling calls to functions that would normally be lazily bound,
11963 unless $gp is call-clobbered.
11965 2013-05-15 Uros Bizjak <ubizjak@gmail.com>
11967 * config/i386/i386.c (ix86_option_override_internal): Update
11968 processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags. Add
11969 PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
11970 non-SSE 3dNow! targets. Enable TARGET_PRFCHW for TARGET_3DNOW targets.
11971 * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
11973 (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
11975 2013-05-15 Andreas Schwab <schwab@suse.de>
11977 * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
11978 for rotlhi3+1 and rotlqi3+1, resp. Fix reference to non-existing
11981 2013-05-15 Teresa Johnson <tejohnson@google.com>
11983 * loop-unroll.c (report_unroll_peel): Check decision before
11984 emitting unroll/peel message.
11986 2013-05-15 Teresa Johnson <tejohnson@google.com>
11988 * function.h (has_bb_partition): New rtl_data flag.
11989 (bb_reorder_complete): Ditto.
11990 * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
11991 instead of flag_reorder_blocks_and_partition.
11992 * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
11993 with some enhancements.
11994 (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
11995 * bb-reorder.c (connect_traces): Check for has_bb_partition
11996 instead of flag_reorder_blocks_and_partition.
11997 (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
11998 (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
11999 verify_hot_cold_block_grouping.
12000 (partition_hot_cold_basic_blocks): Set has_bb_partition.
12002 2013-05-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12005 * config/arm/predicates.md (call_insn_operand): New predicate.
12006 * config/arm/constraints.md ("Cs", "Ss"): New constraints.
12007 * config/arm/arm.md (*call_insn, *call_value_insn): Match only
12008 if insn is not a tail call.
12009 (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
12011 * config/arm/arm.h (enum reg_class): New caller save register class.
12012 (REG_CLASS_NAMES): Likewise.
12013 (REG_CLASS_CONTENTS): Likewise.
12014 * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
12017 2013-05-15 Richard Biener <rguenther@suse.de>
12019 * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
12020 of MSG_OPTIMIZED_LOCATIONS.
12021 * tree-vect-slp.c (vect_make_slp_decision): Likewise.
12022 (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
12024 * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
12025 of MSG_OPTIMIZED_LOCATIONS.
12026 (execute_vect_slp): Likewise.
12027 * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
12028 (vect_create_cond_for_alias_checks): Likewise.
12029 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12030 (vect_recog_widen_mult_pattern): Likewise.
12031 (vect_recog_widen_sum_pattern): Likewise.
12032 (vect_recog_over_widening_pattern): Likewise.
12033 (vect_recog_widen_shift_pattern): Likewise.
12034 (vect_recog_vector_vector_shift_pattern): Likewise.
12035 (vect_recog_divmod_pattern): Likewise.
12036 (vect_recog_mixed_size_cond_pattern): Likewise.
12037 (vect_recog_bool_pattern): Likewise.
12038 (vect_pattern_recog_1): Likewise.
12040 2013-05-15 Martin Jambor <mjambor@suse.cz>
12042 * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
12043 non-functions to builtin_unreachable.
12044 * ipa-inline-transform.c (inline_call): Do not assert estimates were
12045 correct when new direct edges were discovered.
12047 2013-05-15 Martin Jambor <mjambor@suse.cz>
12049 * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
12050 header, print symbol order instead of node uid, print more information
12051 about indirect edge targets.
12052 (ipa_make_edge_direct_to_target): Print symbol order instead of node
12054 (ipa_make_edge_direct_to_target): Likewise.
12055 (remove_described_reference): Likewise.
12056 (propagate_controlled_uses): Likewise.
12057 (ipa_print_node_params): Also print symbol order.
12058 (ipcp_transform_function): Print symbol order instead of node uids.
12059 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
12060 (cgraph_get_create_real_symbol_node): Likewise.
12061 * ipa-cp.c (print_lattice): Likewise.
12062 (print_all_lattices): Likewise.
12063 (determine_versionability): Likewise.
12064 (initialize_node_lattices): Likewise.
12065 (estimate_local_effects): Likewise.
12066 (update_profiling_info): Likewise.
12067 (create_specialized_node): Likewise.
12068 (perhaps_add_new_callers): Likewise.
12069 (decide_about_value): Likewise.
12070 (decide_whether_version_node): Likewise.
12071 (identify_dead_nodes): Likewise.
12072 * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
12073 (dump_inline_summary): Likewise.
12074 (estimate_node_size_and_time): Likewise.
12075 (inline_analyze_function): Likewise.
12076 * ipa-inline.c (report_inline_failed_reason): Likewise.
12077 (want_early_inline_function_p): Likewise.
12078 (edge_badness): Likewise.
12079 (update_edge_key): Likewise.
12080 (inline_small_functions): Likewise. Add dumping of order to two other
12082 * ipa-pure-const.c (pure_const_read_summary): Print symbol order
12083 instead of node uids.
12084 (propagate_pure_const): Likewise.
12085 (propagate_pure_const): Likewise.
12086 * ipa-utils.c (dump_cgraph_node_set): Likewise.
12087 * lto-cgraph.c (input_node): Explicitly specify we dump uid.
12088 * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
12090 * tree-pretty-print.c (dump_function_header): Likewise.
12091 * tree-sra.c (convert_callers_for_node): Dump in traditional format.
12092 Print symbol order instead of node uids.
12094 2013-05-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12096 * config/s390/s390.c (s390_register_move_cost): Don't impose the
12097 FPR<->GPR move cost penalty if ldgr/lgdr can be used.
12099 2013-05-15 Richard Biener <rguenther@suse.de>
12101 PR tree-optimization/57275
12102 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
12103 return value for fail to do runtime alias checks for gather loads.
12105 2013-05-15 Jan Hubicka <jh@suse.cz>
12109 * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
12110 weakrefs are not external.
12111 (lto_symtab_merge_decls): Fix thinko when dealing with
12112 non-lto_symtab decls.
12113 (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
12114 (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
12115 * varpool.c (dump_varpool_node): Dump more flags.
12117 2013-05-15 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
12119 * config/i386/i386.c (processor_alias_table): Add instruction
12120 FSGSBASE for AMD bdver3 architecture.
12122 2013-05-14 Jakub Jelinek <jakub@redhat.com>
12124 * tree.c (warn_deprecated_use): Print file:line using locus color.
12125 * diagnostic.c (diagnostic_report_current_module): Print file:line
12126 and file:line:column using locus color.
12128 2013-05-14 Mike Stump <mikestump@comcast.net>
12130 * gdbinit.in: Add __null.
12132 2013-05-14 Mike Stump <mikestump@comcast.net>
12134 * recog.h: Rename struct recog_data to Recog_data.
12135 * recog.c: Likewise.
12136 * reload.c (can_reload_into): Likewise.
12137 * config/picochip/picochip.c: Likewise.
12139 2013-05-14 Mike Stump <mikestump@comcast.net>
12141 * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
12143 2013-05-14 Steven Bosscher <steven@gcc.gnu.org>
12145 * resource.h (struct resources): Remove unch_memory member.
12146 (CLEAR_RESOURCE): Don't clear unch_memory.
12147 * resource.c (mark_referenced_resources): Don't set it.
12148 (mark_set_resources): Likewise.
12149 (mark_target_live_regs): Don't clear it.
12150 (init_resource_info): Likewise.
12151 * reorg.c (resource_conflicts_p): Don't compare it.
12152 (redundant_insn): Don't set it.
12154 * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
12156 * emit-rtl.c (next_label): Remove unused function.
12157 (skip_consecutive_labels, link_cc0_insns): Move to ...
12158 * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
12159 only place where these functions are used, and make them static.
12161 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
12163 * fold-const.c (fold_negate_expr): Handle vectors.
12164 (fold_truth_not_expr): Make it static.
12165 (fold_invert_truthvalue): New static function.
12166 (invert_truthvalue_loc): Handle vectors. Do not call
12167 fold_truth_not_expr directly.
12168 (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
12169 <TRUTH_NOT_EXPR>: Do not cast to boolean.
12170 (fold_comparison): Handle vector constants.
12171 (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
12172 (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
12173 * tree.h (fold_truth_not_expr): Remove declaration.
12175 2013-05-14 James Greenhalgh <james.greenhalgh@arm.com>
12177 * config/aarch64/aarch64-simd.md
12178 (aarch64_vcond_internal<mode>): Rename to...
12179 (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
12180 (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
12181 float modes. Clarify all iterator modes.
12182 (vcond<mode><mode>): Use new name for vcond expanders.
12183 (vcond<v_cmp_result><mode>): Likewise.
12184 (vcondu<mode><mode>: Likewise.
12185 * config/aarch64/iterators.md (VDQF_COND): New.
12187 2013-05-14 Marc Glisse <marc.glisse@inria.fr>
12190 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
12191 variable for the shift amount. Check that we shift by non-negative
12194 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
12197 * config/arm/arm.h (EPILOGUE_USES): Only return true
12198 for LR_REGNUM after epilogue_completed.
12200 2013-05-14 Joern Rennecke <joern.rennecke@embecosm.com>
12202 * config/avr/avr.c (avr_encode_section_info): Bail out if the type
12203 is error_mark_node.
12205 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12208 * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
12209 and Solaris 11+/x86 with gld.
12210 * configure: Regenerate.
12212 2013-05-14 Jakub Jelinek <jakub@redhat.com>
12214 * expmed.c (expand_shift_1): Canonicalize rotates by
12215 constant bitsize / 2 to bitsize - 1.
12216 * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
12217 case ROTATERT>: Likewise.
12220 2013-05-10 Jakub Jelinek <jakub@redhat.com>
12222 * config/i386/i386.md (rotateinv): New code attr.
12223 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12224 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12225 roll $31, %eax, etc.
12227 2013-05-14 Richard Biener <rguenther@suse.de>
12229 PR middle-end/57235
12230 * tree-eh.c (sink_clobbers): Give up for successors with
12231 multiple predecessors and no virtual uses.
12233 2013-05-14 Eric Botcazou <ebotcazou@adacore.com>
12235 * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
12236 * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
12238 2013-05-14 Jakub Jelinek <jakub@redhat.com>
12240 PR middle-end/57251
12241 * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
12242 the case when both op0 and op1 have VOIDmode.
12244 2013-05-14 Kaushik Phatak <kaushik.phatak@kpitcummins.com>
12246 * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
12247 in multiply-accumulate mode.
12249 2013-05-13 Guozhi Wei <carrot@google.com>
12251 * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
12253 2013-05-13 Kai Tietz <ktietz@redhat.com>
12256 * config/i386/cygming.h (TARGET_PECOFF): Define as true.
12257 * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
12258 (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
12259 * config/i386/i386.c (ix86_option_override_internal): Likewise.
12260 (ix86_expand_prologue): Likewise.
12261 (ix86_expand_split_stack_prologue): Likewise.
12262 (legitimate_pic_address_disp_p): Likewise.
12263 (legitimize_pic_address): Likewise.
12264 (legitimize_tls_address): Likewise.
12265 (legitimize_pe_coff_symbol): Likewise.
12266 (output_pic_addr_const): Likewise.
12267 (construct_plt_address): Likewise.
12268 (ix86_expand_call): Likewise.
12269 (x86_output_mi_thunk): Likewise.
12270 (x86_function_profiler): Likewise.
12272 2013-05-13 Sofiane Naci <sofiane.naci@arm.com>
12274 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
12275 similar switch cases.
12276 (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
12277 (aarch64_simd_mov_to_<mode>low): Delete.
12278 (aarch64_simd_mov_to_<mode>high): Delete.
12279 (move_lo_quad_<mode>): Add w<-r alternative.
12280 (aarch64_simd_move_hi_quad_<mode>): Likewise.
12281 (aarch64_simd_mov_from_*): Update type attribute.
12282 * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
12285 2013-05-13 Jan Hubicka <jh@suse.cz>
12287 * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
12288 * config/i386/i386.c (ix86_compute_frame_layout,
12289 ix86_expand_epilogue, emit_i387_cw_initialization,
12290 ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
12291 ix86_local_alignment): Fix use of size/speed predicates.
12293 2013-05-13 Jakub Jelinek <jakub@redhat.com>
12295 PR tree-optimization/45216
12296 PR tree-optimization/57157
12297 * tree-ssa-forwprop.c (simplify_rotate): Only recognize
12298 the (-Y) & (B - 1) variant if OP is |.
12299 * expmed.c (expand_shift_1): For rotations by const0_rtx just
12300 return shifted. Use (-op1) & (prec - 1) as other_amount
12301 instead of prec - op1.
12303 2013-05-13 Martin Jambor <mjambor@suse.cz>
12305 PR middle-end/42371
12306 * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
12307 (ipa_constant_data): New type.
12308 (ipa_jump_func): Use ipa_constant_data to hold information about
12309 constant jump functions.
12310 (ipa_get_jf_constant): Adjust to jump function type changes.
12311 (ipa_get_jf_constant_rdesc): New function.
12312 (ipa_param_descriptor): New field controlled_uses.
12313 (ipa_get_controlled_uses): New function.
12314 (ipa_set_controlled_uses): Likewise.
12315 * ipa-ref.h (ipa_find_reference): Declare.
12316 * ipa-prop.c (ipa_cst_ref_desc): New type.
12317 (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
12319 (ipa_set_jf_constant): Likewise. Also create reference descriptions.
12320 New parameter cs. Adjust all callers.
12321 (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
12322 (remove_described_reference): New function.
12323 (jfunc_rdesc_usable): Likewise.
12324 (try_make_edge_direct_simple_call): Decrement controlled use count,
12325 attempt to remove reference if it hits zero.
12326 (combine_controlled_uses_counters): New function.
12327 (propagate_controlled_uses): Likewise.
12328 (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
12329 (ipa_edge_duplication_hook): Duplicate reference descriptions.
12330 (ipa_print_node_params): Print described use counter.
12331 (ipa_write_jump_function): Adjust to jump function type changes.
12332 (ipa_read_jump_function): New parameter CS, pass it to
12333 ipa_set_jf_constant. Adjust caller.
12334 (ipa_write_node_info): Stream controlled use count
12335 (ipa_read_node_info): Likewise.
12336 * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
12338 * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
12339 count. Remove cloning-added reference if it reaches zero.
12340 * ipa-ref.c (ipa_find_reference): New function.
12342 2013-05-13 Ganesh Gopalasubramanian <Ganesh.Gopalasubramanian@amd.com>
12344 * config/i386/i386.c (processor_target_table): Modified default
12345 alignment values for AMD BD and BT architectures.
12347 2013-05-13 Marc Glisse <marc.glisse@inria.fr>
12349 * tree-vect-generic.c (uniform_vector_p): Move ...
12350 * tree.c (uniform_vector_p): ... here.
12351 * tree.h (uniform_vector_p): Declare it.
12352 * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
12355 2013-05-13 Jakub Jelinek <jakub@redhat.com>
12357 PR tree-optimization/57230
12358 * tree-ssa-strlen.c (handle_char_store): Record length for
12359 array store from STRING_CST.
12361 PR tree-optimization/57230
12362 * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
12365 2013-05-12 Joern Rennecke <joern.rennecke@embecosm.com>
12367 * config/epiphany/epiphany.c (epiphany_init): Check size of
12368 NUM_MODES_FOR_MODE_SWITCHING.
12369 (epiphany_expand_prologue):
12370 Remove CONFIG_REGNUM initial value handling code.
12371 (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
12372 (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
12373 (emit_set_fp_mode, epiphany_mode_after): Likewise.
12374 (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
12375 Don't return 1 for FP_MODE_NONE.
12376 * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
12377 Add value for EPIPHANY_MSW_ENTITY_CONFIG.
12378 (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
12379 * config/epiphany/epiphany.md (save_config): New pattern.
12381 2013-05-12 Uros Bizjak <ubizjak@gmail.com>
12383 * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
12385 2013-05-10 Uros Bizjak <ubizjak@gmail.com>
12387 * config/i386/i386.md (memory): Handle sseishft1.
12388 * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
12389 (*vec_extractv2di_1): Ditto.
12391 2013-05-10 Vladimir Makarov <vmakarov@redhat.com>
12393 * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
12396 2013-05-10 Sebastian Huber <sebastian.huber@embedded-brains.de>
12398 * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
12399 Add mthumb/march=armv7-a multilib.
12400 Add mthumb/march=armv7-r multilib.
12401 Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
12403 2013-05-10 Ralf Corsépius <ralf.corsepius@rtems.org>
12405 * config/v850/t-rtems: Add more multilibs.
12407 2013-05-10 Richard Biener <rguenther@suse.de>
12409 PR tree-optimization/57214
12410 * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
12411 not propagate from SSA names that occur in abnormal PHI nodes.
12413 2013-05-10 Marc Glisse <marc.glisse@inria.fr>
12415 * stor-layout.c (element_precision): New function.
12416 * machmode.h (element_precision): Declare it.
12417 * tree.c (build_minus_one_cst): New function.
12418 (element_precision): Likewise.
12419 * tree.h (build_minus_one_cst): Declare new function.
12420 (element_precision): Likewise.
12421 * fold-const.c (operand_equal_p): Use element_precision.
12422 (fold_binary_loc): Handle vector types.
12423 * convert.c (convert_to_integer): Use element_precision.
12424 * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
12427 2013-05-10 Richard Sandiford <rdsandiford@googlemail.com>
12429 * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
12430 (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
12431 (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
12432 (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12433 * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
12434 (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
12435 (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
12436 (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
12437 * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
12438 (Uuw8): New constraints.
12439 (Usb4): Move into alphabetical order.
12440 * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
12441 (sd8_operand, ub8_operand, uw8_operand): New predicates.
12442 * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
12443 previously unnamed patterns.
12444 (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
12445 (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
12446 (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
12447 (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
12448 of set_attr_alternative/if_then_else. Use extended_mips16 instead
12449 of specific lengths.
12451 2013-05-10 Jakub Jelinek <jakub@redhat.com>
12453 * config/i386/i386.md (rotateinv): New code attr.
12454 (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
12455 *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
12456 roll $31, %eax, etc.
12458 PR tree-optimization/45216
12459 PR tree-optimization/57157
12460 * tree-ssa-forwprop.c (simplify_rotate): New function.
12461 (ssa_forward_propagate_and_combine): Call it.
12463 2013-05-10 Richard Biener <rguenther@suse.de>
12465 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
12466 disable peeling when we version for aliasing.
12467 (vector_alignment_reachable_p): Honor explicit user alignment.
12468 (vect_supportable_dr_alignment): Likewise.
12469 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
12470 STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
12471 * tree-vect-loop.c (vect_transform_loop): First apply versioning,
12472 then peeling to arrange for the cost-model check to come first.
12474 2013-05-10 Alan Modra <amodra@gmail.com>
12476 * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
12477 (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
12478 * configure: Regenerate.
12480 2013-05-10 Alan Modra <amodra@gmail.com>
12483 * varasm.c (default_elf_select_section): Move !DECL_P check..
12484 (get_named_section): ..to here before calling get_section_name.
12486 (default_section_type_flags): Add DECL_P check.
12487 * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
12488 * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
12490 2013-05-09 Joern Rennecke <joern.rennecke@embecosm.com>
12492 * config/epiphany/epiphany.c (epiphany_expand_prologue):
12493 When using gen_stack_adjust_str with a register offset, add a
12494 REG_FRAME_RELATED_EXPR note.
12496 2013-05-09 Uros Bizjak <ubizjak@gmail.com>
12498 * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
12499 (*vec_extractv4si_zext_mem): Ditto.
12500 (*vec_extractv2di): Add 0->x and x->x alternatives.
12501 * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
12502 * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
12504 2013-05-09 Jason Merrill <jason@redhat.com>
12506 N3639 C++1y VLA support
12507 * gimplify.c (gimplify_vla_decl): Don't touch an existing
12510 * tree.c (build_constructor_va): New.
12511 * tree.h: Declare it.
12513 2013-05-09 Martin Jambor <mjambor@suse.cz>
12516 * gimple-fold.c (canonicalize_constructor_val): Call
12517 cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
12519 2013-05-09 Jan Hubicka <jh@suse.cz>
12520 Richard Biener <rguenther@suse.de>
12523 * symtab.c (symtab_make_decl_local): Do not add private names.
12525 2013-05-09 Jan Hubicka <jh@suse.cz>
12528 * symtab.c (insert_to_assembler_name_hash): Handle clones.
12529 (unlink_from_assembler_name_hash): Likewise.
12530 (symtab_prevail_in_asm_name_hash, symtab_register_node,
12531 symtab_unregister_node, symtab_initialize_asm_name_hash,
12532 change_decl_assembler_name): Update.
12534 2013-05-09 Sofiane Naci <sofiane.naci@arm.com>
12536 * config/aarch64/aarch64.md: New movtf split.
12537 (*movtf_aarch64): Update.
12538 (aarch64_movdi_tilow): Handle TF modes and rename to
12539 aarch64_movdi_<mode>low.
12540 (aarch64_movdi_tihigh): Handle TF modes and rename to
12541 aarch64_movdi_<mode>high
12542 (aarch64_movtihigh_di): Handle TF modes and rename to
12543 aarch64_mov<mode>high_di
12544 (aarch64_movtilow_di): Handle TF modes and rename to
12545 aarch64_mov<mode>low_di
12546 (aarch64_movtilow_tilow): Remove spurious whitespace.
12547 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
12549 (aarch64_print_operand): Update.
12551 2013-05-09 Alan Modra <amodra@gmail.com>
12553 * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
12555 * configure: Regenerate.
12557 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
12559 * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
12560 splitter preparation statements.
12561 * config/i386/sse.md (*vec_extract* splitters): Ditto.
12562 (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
12565 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12567 * gimple-ssa-strength-reduction.c (count_candidates): Change
12568 return value to int.
12569 (analyze_candidates_and_replace): Change type of length to int.
12571 2013-05-08 Uros Bizjak <ubizjak@gmail.com>
12573 * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
12574 (*vec_extract<mode>): Use VI12_128 mode iterator.
12575 (*vec_extract<mode>_mem): Ditto.
12576 (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
12579 2013-05-08 Diego Novillo <dnovillo@google.com>
12584 2012-08-17 Diego Novillo <dnovillo@google.com>
12587 * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
12588 * config.in: Regenerate.
12589 * configure: Regenerate.
12590 * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
12592 2013-05-08 Jan Hubicka <jh@suse.cz>
12595 * cgraph.c (cgraph_make_node_local_1): Se unique_name.
12596 * cgraph.h (symtab_node_base): Add unique_name.
12597 * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
12598 input_overwrite_node, input_varpool_node): Stream unique_name.
12599 * cgraphclones.c (cgraph_create_virtual_clone,
12600 cgraph_function_versioning): Set unique_name.
12601 * ipa.c (function_and_variable_visibility): Set unique_name.
12603 2013-05-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12605 * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
12606 (alloc_cand_and_find_basis): Restrict conditional candidate
12607 processing to CAND_MULTs.
12609 2013-05-08 Jan Hubicka <jh@suse.cz>
12612 lto-symtab.c (lto_symtab_symbol_p): New function.
12613 (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
12614 lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
12615 lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
12616 Skip static symbols.
12618 2013-05-08 Paolo Carlini <paolo.carlini@oracle.com>
12620 PR tree-optimization/57200
12621 * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
12622 Only call inform if the preceding warning_at returns true.
12624 2013-05-07 Han Shen <shenhan@google.com>
12626 * cfgexpand.c (record_or_union_type_has_array_p): New function.
12627 (expand_used_vars): Add logic handling '-fstack-protector-strong'.
12628 * common.opt (fstack-protector-strong): New option.
12629 * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
12630 * doc/invoke.texi (Optimization Options): Document
12631 "-fstack-protector-strong".
12632 * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
12634 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
12636 * config/mips/mips.c (mips_machine_reorg2): Return 0.
12638 2013-05-07 Vladimir Makarov <vmakarov@redhat.com>
12640 * ira.c (update_equiv_regs): Add insn having equiv memory even if
12641 it is not lhs of the insn.
12642 (setup_reg_equiv): Remove insn having equiv memory which it is not
12644 * lra-constraints.c (process_address): Try to improve generation
12645 code for address base + disp.
12646 (lra_constraints): Make correct the code for checking insn setting
12647 up backward equivalence. Remove insn only if it is in the init
12649 * lra-eliminations.c (update_reg_eliminate): Change return value.
12650 (lra_eliminate): Use the result.
12652 2013-05-07 Uros Bizjak <ubizjak@gmail.com>
12654 * config/i386/sse.md (ssescalarnummask): New mode attribute.
12655 (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
12656 (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
12657 *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
12658 register target operands.
12659 (*vec_extractv8hi_sse2): New pattern.
12660 (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
12661 (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
12662 (*vec_extract<mode>_mem): New insn and split pattern.
12664 2013-05-07 Christophe Lyon <christophe.lyon@linaro.org>
12666 * config/arm/arm.c (arm_asan_shadow_offset): New function.
12667 (TARGET_ASAN_SHADOW_OFFSET): Define.
12668 * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
12669 (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
12671 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12673 * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
12674 (incr_vec_index): Return -1 if increment not found.
12675 (create_add_on_incoming_edge): Assert if increment not found.
12676 (record_increment): Limit number of increments recorded.
12677 (all_phi_incrs_profitable): Return false if an increment not found.
12678 (replace_profitable_candidates): Don't process increments that were
12680 (analyze_candidates_and_replace): Limit size of incr_vec.
12682 2013-05-07 Richard Biener <rguenther@suse.de>
12684 * calls.c (special_function_p): setjmp-like functions are leaf.
12685 * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
12686 * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
12688 2013-05-07 Sofiane Naci <sofiane.naci@arm.com>
12690 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
12691 (aarch64_simd_mov<mode>): New expander.
12692 (aarch64_simd_mov_to_<mode>low): New instruction pattern.
12693 (aarch64_simd_mov_to_<mode>high): Likewise.
12694 (aarch64_simd_mov_from_<mode>low): Likewise.
12695 (aarch64_simd_mov_from_<mode>high): Likewise.
12696 (aarch64_dup_lane<mode>): Update.
12697 (aarch64_dup_lanedi): New instruction pattern.
12698 * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
12699 * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
12701 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12703 * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
12704 (replace_mult_candidate): Remove unnecessary argument; remove
12705 unnecessary parameter from call to introduce_cast_before_cand.
12706 (replace_unconditional_candidate): Remove unnecessary parameter
12707 from call to replace_mult_candidate.
12708 (replace_conditional_candidate): Likewise.
12709 (insert_initializers): Use make_temp_ssa_name.
12710 (introduce_cast_before_cand): Remove unnecessary argument; use
12711 make_temp_ssa_name.
12712 (replace_one_candidate): Remove unnecessary argument; remove
12713 unnecessary parameter from calls to introduce_cast_before_cand.
12714 (replace_profitable_candidates): Remove unnecessary parameters
12715 from calls to replace_one_candidate.
12717 2013-05-07 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12719 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12720 phi def as possibly hiding a basis for a CAND_ADD whose operands
12721 have been commuted in the analysis.
12722 (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
12724 2013-05-07 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
12726 * config/aarch64/aarch64.md
12727 (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
12728 shift value between 0-4.
12730 2013-05-07 Richard Biener <rguenther@suse.de>
12732 * double-int.h (rshift): New overload.
12733 * double-int.c (rshift): New function.
12734 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
12735 (create_reference_ops_from_ref): Remove.
12736 (vn_reference_insert): Use shared ops for constructing the
12737 reference and copy it.
12739 2013-05-07 Richard Biener <rguenther@suse.de>
12741 PR middle-end/57190
12742 * tree-eh.c (sink_clobbers): Properly propagate
12743 SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
12745 2013-05-07 Jakub Jelinek <jakub@redhat.com>
12747 PR tree-optimization/57149
12748 * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
12749 (can_skip_redundant_opnd, compute_uninit_opnds_pos,
12750 collect_phi_def_edges, execute_late_warn_uninitialized): Use
12751 uninit_undefined_value_p instead of ssa_undefined_value_p.
12754 * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
12755 for modifier == EXPAND_INITIALIZER.
12757 2013-05-07 Anton Blanchard <anton@samba.org>
12759 * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
12760 for powerpc64 little endian.
12761 * configure: Regenerate.
12763 2013-05-06 Graham Stott <grahams@btinternet.com>
12765 * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
12766 mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
12767 wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
12768 shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
12770 2013-05-06 Graham Stott <grahams@btinternet.com>
12772 * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
12773 codes which allow non-lvalues.
12775 2013-05-06 Marc Glisse <marc.glisse@inria.fr>
12777 * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
12778 components are all 1s.
12779 (integer_minus_onep): New function.
12780 * tree.h (integer_minus_onep): Declare it.
12781 * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
12782 integer_minus_onep instead of integer_all_onesp.
12784 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
12787 * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
12788 variations of these patterns.
12790 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
12792 * config/i386/i386.md (isa): Add x64_sse4 member.
12793 (enabled): Handle x64_sse4.
12794 (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
12795 instruction for 64bit SSE4_1 targets. Update insn attributes.
12796 (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
12797 instruction for SSE4_1 targets. Update insn attributes.
12798 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12799 with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
12800 (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
12802 (*vec_extractv4si): Rename from *sse4_1_pextrd.
12803 (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
12804 (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
12806 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
12809 * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
12811 2013-05-06 Maxim Kuznetsov <maks.kuznetsov@gmail.com>
12813 * final.c (do_assembler_dialects): Don't handle curly braces and
12814 vertical bar escaped by % as dialect delimiters.
12815 (output_asm_insn): Print curly braces and vertical bar if escaped
12816 by % and ASSEMBLER_DIALECT defined.
12817 * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
12818 * doc/tm.texi: Regenerated.
12820 2013-05-06 Steven Bosscher <steven@gcc.gnu.org>
12822 * config/mips/mips.c: Include tree-pass.h.
12823 (mips_reorg): Split in pre- and post-dbr_schedule parts.
12824 (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
12825 (pass_mips_machine_reorg2): New machine specific pass.
12826 (insert_pass_mips_machine_reorg2): New pass plugin definition.
12827 (mips_option_override): Register the new pass.
12828 * rtl.h (cleanup_barriers): Remove prototype.
12829 (dbr_schedule): Likewise.
12830 * jump.c (cleanup_barriers): Make static.
12831 * reorg.c (dbr_schedule): Likewise.
12833 2013-05-06 Richard Biener <rguenther@suse.de>
12835 PR tree-optimization/57185
12836 * tree-parloops.c (add_field_for_reduction): Handle anonymous
12837 SSA names properly.
12839 2013-05-06 Uros Bizjak <ubizjak@gmail.com>
12842 * config/i386/i386.c (add_parameter_dependencies): Add dependence
12843 between "first_arg" and "insn", not "last" and "insn".
12845 2013-05-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12847 * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
12848 (find_candidates_in_block): Re-enable slsr_process_phi.
12849 (create_phi_basis): Fix double counting of candidate adjustment.
12851 2013-05-06 Richard Biener <rguenther@suse.de>
12853 PR middle-end/57147
12854 * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
12855 the edge is also fallthru, preserve it and just clear the
12857 * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
12858 also complex, preserve that and just clear the fallthru flag.
12859 * tree-inline.c (update_ssa_across_abnormal_edges): Also
12860 update virtual operands.
12862 2013-05-06 Alan Modra <amodra@gmail.com>
12864 * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
12865 (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
12866 * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
12867 * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
12868 (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12869 * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
12871 2013-05-06 Alan Modra <amodra@gmail.com>
12873 * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
12874 (ASM_SPEC): ..here. Emit DEFAULT_ASM_ENDIAN too.
12875 (DEFAULT_ASM_ENDIAN): Define.
12876 (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
12877 * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
12878 Update -K PIC clause from sysv4.h.
12879 (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
12880 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
12882 2013-05-06 Alan Modra <amodra@gmail.com>
12884 * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
12885 twice for little-endian.
12886 (ashrdi3_no_power, ashrdi3): Support little-endian.
12888 2013-05-06 Oleg Endo <olegendo@gcc.gnu.org>
12891 * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
12892 * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
12894 * config/sh/iterators.md (SMIN_SMAX): New code iterator.
12895 * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
12896 clips_min_const_int, clips_max_const_int, clipu_max_const_int):
12899 2013-05-05 Steven Bosscher <steven@gcc.gnu.org>
12900 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
12902 * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
12903 * config/pa/pa.opt: Make mbig-switch a no-op.
12904 * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
12905 (CASE_VECTOR_MODE): Always return SImode.
12906 (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
12907 for the !TARGET_BIG_SWITCH case.
12908 * config/pa/pa-linux.h: Likewise.
12909 * config/pa/pa-openbsd.h: Likewise.
12910 * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
12911 * config/pa/pa.md (short_jump): Remove define_insn.
12912 (casesi): Remove code for the !TARGET_BIG_SWITCH case.
12913 (casesi0): Remove define_insn.
12914 (type): Remove btable_branch.
12915 (pa_combine_type): Likewise.
12916 (in_nullified_branch_delay): Likewise.
12917 (in_call_delay): Likewise.
12918 (define_delay): Likewise.
12919 (define_insn_reservation "Z3"): Likewise.
12920 (define_insn_reservation "Z4"): Likewise.
12921 * config/pa/pa.c (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
12922 (pa_adjust_insn_length): Remove adjustment for btable branches.
12923 * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
12926 2013-05-05 Uros Bizjak <ubizjak@gmail.com>
12928 * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
12929 from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
12930 Add m->r,x alternatives.
12931 (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
12932 splitters using SWI48x mode iterator.
12933 (*vec_extract_v2di_0_sse): Rename from *sse2_storeq. Disable for
12934 TARGET_64BIT. Add m->x alternative.
12935 (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
12936 Add o->x alternative. Enable for TARGET_SSE.
12937 (sse_storeq): Remove expander.
12938 (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
12939 with memory input operand.
12940 (*vec_extractv2di_1 splitter): New.
12941 (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
12942 * config/i386/i386.md (ssevecmodelower): New mode attribute.
12944 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
12946 * config/rs6000/rs6000.c (INT_P): Reformat. Delete obsolete comment.
12947 (INT_LOWPART): Delete.
12948 (extract_MB): Adjust.
12949 (extract_ME): Adjust.
12950 (print_operand): Adjust.
12952 2013-05-04 Segher Boessenkool <segher@kernel.crashing.org>
12954 * config/rs6000/predicates.md (reg_or_add_cint_operand,
12955 reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12956 (reg_or_logical_cint_operand, easy_fp_constant,
12957 logical_const_operand): Delete "CONST_DOUBLE" case.
12958 * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
12959 "HOST_BITS_PER_WIDE_INT == 64" test.
12960 (num_insns_constant): Ditto. Delete CONST_DOUBLE DImode/VOIDmode case.
12961 (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
12962 (rs6000_emit_set_const): Delete CONST_DOUBLE case.
12963 (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
12965 (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
12966 CONST_DOUBLE DImode/VOIDmode case.
12967 (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
12968 (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case. Delete
12969 CONST_DOUBLE VOIDmode case.
12970 (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
12971 (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
12972 * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
12973 Delete CONST_DOUBLE case.
12974 (splitters for mov FMOVE64 const_double): Delete
12975 "HOST_BITS_PER_WIDE_INT == 32" case. Delete
12976 "HOST_BITS_PER_WIDE_INT >= 64" test.
12977 (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
12979 (mov DI const_double): Delete.
12981 2013-05-04 Jakub Jelinek <jakub@redhat.com>
12983 * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
12984 on op shows all bits zero in mode of a lowpart subreg, return zero.
12986 2013-05-03 Michael Meissner <meissner@linux.vnet.ibm.com>
12989 * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
12990 to save TFmode registers and DImode to save TImode registers for
12991 caller save operations.
12992 (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
12993 mark being partially clobbered since they only use the first
12996 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
12997 and TDmode only use the upper 64-bits of each VSX register.
12999 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13001 * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
13002 (find_candidates_in_block): Disable slsr_process_phi.
13004 2013-05-03 Guozhi Wei <carrot@google.com>
13006 * coverage.c (coverage_obj_init): Move the construction of gcov
13008 (build_init_ctor): ... here.
13010 2013-05-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13012 * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
13013 (slsr_cand_d): Redefine def_phi.
13014 (stride_status, phi_adjust_status, count_phis_status): New enums.
13015 (find_phi_def): New.
13016 (find_basis_for_base_expr): New.
13017 (find_basis_for_candidate): Handle hidden bases.
13018 (alloc_cand_and_find_basis): Handle phi candidates.
13019 (slsr_process_phi): New.
13020 (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
13021 (create_mul_imm_cand): Likewise.
13022 (create_add_ssa_cand): Exclude phi base candidates.
13023 (create_add_imm_cand): Likewise.
13024 (slsr_process_cast): Likewise.
13025 (slsr_process_copy): Likewise.
13026 (find_candidates_in_block): Handle phi candidates.
13027 (dump_candidate): Likewise.
13028 (unconditional_cands): Delete.
13029 (unconditional_cands_with_known_stride_p): Delete.
13030 (phi_dependent_cand_p): New.
13031 (cand_increment): Handle phi-dependent candidates.
13032 (replace_dependent): Delete.
13033 (replace_mult_candidate): New.
13034 (replace_unconditional_candidate): New.
13035 (incr_vec_index): Move to avoid forward reference.
13036 (create_add_on_incoming_edge): New.
13037 (create_phi_basis): New.
13038 (replace_dependents): Delete.
13039 (replace_conditional_candidate): New.
13040 (phi_add_costs): New.
13041 (replace_uncond_cands_and_profitable_phis): New.
13042 (record_increment): Handle phi adjustments.
13043 (record_phi_increments): New.
13044 (record_increments): Handle phi adjustments.
13045 (phi_incr_cost): New.
13046 (lowest_cost_path): Handle phis.
13047 (total_savings): Likewise.
13048 (analyze_increments): Likewise.
13049 (ncd_with_phi): New.
13050 (ncd_of_cand_and_phis): New.
13051 (nearest_common_dominator_for_cands): Handle phi increments.
13052 (all_phi_incrs_profitable): New.
13053 (replace_profitable_candidates): Handle phi-dependent candidates.
13054 (analyze_candidates_and_replace): Likewise.
13056 2013-05-03 Teresa Johnson <tejohnson@google.com>
13059 * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
13060 do not exceed REG_BR_PROB_BASE.
13062 2013-05-03 Jeff Law <law@redhat.com>
13064 PR tree-optimization/57144
13065 * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
13066 operand of the condition will bit into the new type when eliminating
13067 a cast feeding a condition.
13069 2013-05-03 Jakub Jelinek <jakub@redhat.com>
13071 PR rtl-optimization/57130
13072 * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
13073 of COMPARE as in_code to the recursive call if needed.
13075 2013-05-03 Uros Bizjak <ubizjak@gmail.com>
13077 * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
13078 (enabled): Handle new members.
13079 * config/i386/sse.md (*vec_concatv2si): Merge from
13080 *vec_concatv2si_sse2 and vec_concatv2si_sse.
13081 (vec_concatv2di): Merge with *vec_concatv2di_rex64.
13083 2013-05-03 Joern Rennecke <joern.rennecke@embecosm.com>
13085 PR tree-optimization/57027
13086 * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
13087 for fnms opportunity, check we got the prerequisite kind
13088 of tree / gimple before using accessor functions.
13090 2013-05-03 Richard Biener <rguenther@suse.de>
13092 * double-int.h (lshift): New overload without precision
13093 and arith argument.
13094 (operator *=, operator +=, operator -=): Move ...
13095 * double-int.c (operator *=, operator +=, operator -=): ... here
13096 and implement more efficiently.
13097 (mul_double_with_sign): Remove.
13098 (lshift_double): Adjust to take unsinged shift argument, push
13099 dispatching code to callers.
13100 (mul_double_wide_with_sign): Add early out for callers that
13101 are not interested in high parts or overflow.
13102 (lshift): New function.
13103 (lshift, rshift, alshift, arshift, llshift, lrshift): Add
13104 dispatch code here.
13105 (lrotate, rrotate): Use logical shifts.
13106 * expr.c (get_inner_reference): Use lshift.
13107 * fixed-value.c (do_fixed_divide): Likewise.
13108 * tree-dfa.c (get_ref_base_and_extent): Likewise.
13109 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
13110 (indirect_refs_may_alias_p): Likewise.
13111 (stmt_kills_ref_p_1): Likewise.
13113 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
13115 * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
13117 2013-05-03 Vidya Praveen <vidyapraveen@arm.com>
13119 * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
13120 scalar form of FABD instruction.
13122 2013-05-02 Vladimir Makarov <vmakarov@redhat.com>
13124 * lra-constraints.c (process_alt_operands): Add checking alt
13125 number to choose the best alternative.
13127 2013-05-02 Richard Biener <rguenther@suse.de>
13129 * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
13130 bitmap and its handling.
13131 (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
13133 2013-05-02 Richard Biener <rguenther@suse.de>
13135 PR middle-end/57140
13136 * tree-inline.c (copy_loops): Properly handle removed loops.
13137 (copy_cfg_body): Mark destination loops for fixup if source
13138 loops needed fixup.
13140 2013-05-02 Greta Yorsh <Greta.Yorsh@arm.com>
13143 * config/arm/arm.c (arm_expand_epilogue): Check really_return before
13144 generating simple_return for naked functions.
13146 2013-05-02 Martin Jambor <mjambor@suse.cz>
13148 PR middle-end/56988
13149 * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
13150 * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
13152 (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
13153 ipa_agg_replacement_value structures.
13154 (known_aggs_to_agg_replacement_list): Likewise.
13155 * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
13156 (read_agg_replacement_chain): Likewise.
13157 (ipcp_transform_function): Also check that by_ref flags match.
13159 2013-05-02 Richard Biener <rguenther@suse.de>
13161 * graphds.h (struct graph): Add obstack member.
13162 * graphds.c (new_graph): Initialize obstack and allocate
13164 (add_edge): Allocate edge from the obstack.
13165 (free_graph): Free the obstack instead of all edges and vertices.
13167 2013-05-02 Teresa Johnson <tejohnson@google.com>
13169 * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
13171 * cfg.c (update_bb_profile_for_threading): Ditto.
13172 * tree-inline.c (copy_bb): Ditto.
13173 (copy_edges_for_bb): Ditto.
13174 (initialize_cfun): Ditto.
13175 (copy_cfg_body): Ditto.
13176 (expand_call_inline): Ditto.
13177 * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
13178 (estimate_node_size_and_time): Ditto.
13179 (inline_merge_summary): Ditto.
13180 * cgraphclones.c (cgraph_clone_edge): Ditto.
13181 (cgraph_clone_node): Ditto.
13182 * sched-rgn.c (compute_dom_prob_ps): Ditto.
13183 (compute_trg_info): Ditto.
13185 2013-05-02 Ian Bolton <ian.bolton@arm.com>
13187 * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
13188 S reg when fp attribute set.
13189 (movdi_aarch64): Only allow to/from D reg when fp attribute set.
13191 2013-05-02 Ian Bolton <ian.bolton@arm.com>
13193 * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
13195 (*and_one_cmplsi3_compare0_uxtw): Likewise.
13196 (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
13197 (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
13199 2013-05-02 Richard Biener <rguenther@suse.de>
13201 * tree-scalar-evolution.c (scev_info_hasher): Remove.
13202 (struct instantiate_cache_entry): New type.
13203 (struct instantiate_cache_entry_hasher): New hashtable descriptor.
13204 (struct instantiate_cache_type): New type.
13205 (set_instantiated_value, get_instantiated_value): Remove.
13206 (get_instantiated_value_entry): New function.
13207 (instantiate_scev_name): Use the new cache and adjust.
13208 (instantiate_scev_poly): Adjust.
13209 (instantiate_scev_binary): Likewise.
13210 (instantiate_array_ref): Likewise.
13211 (instantiate_scev_convert): Likewise.
13212 (instantiate_scev_not): Likewise.
13213 (instantiate_scev_3): Likewise.
13214 (instantiate_scev_2): Likewise.
13215 (instantiate_scev_r): Likewise.
13216 (instantiate_scev): Likewise.
13217 (resolve_mixers): Likewise.
13219 2013-05-01 Vladimir Makarov <vmakarov@redhat.com>
13222 * lra-constraints.c (best_small_class_operands_num): Remove.
13223 (process_alt_operands): Remove small_class_operands_num. Take
13224 small classes operands into losers and only if the operand is not
13225 matched. Modify debugging output.
13226 (curr_insn_transform): Remove best_small_class_operands_num.
13229 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13231 * config/aarch64/aarch64-builtins.c
13232 (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
13233 * config/aarch64/aarch64-simd-builtins.def
13234 (reduc_splus_): Add new modes.
13235 (reduc_uplus_): New.
13236 * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
13237 (reduc_uplus_v4sf): Likewise.
13238 (reduc_splus_v4sf): Likewise.
13239 (aarch64_addv<mode>): Likewise.
13240 (reduc_uplus_<mode>): Likewise.
13241 (reduc_splus_<mode>): Likewise.
13242 (aarch64_addvv2di): Likewise.
13243 (reduc_uplus_v2di): Likewise.
13244 (reduc_splus_v2di): Likewise.
13245 (aarch64_addvv2si): Likewise.
13246 (reduc_uplus_v2si): Likewise.
13247 (reduc_splus_v2si): Likewise.
13248 (reduc_<sur>plus_<mode>): New.
13249 (reduc_<sur>plus_v2di): Likewise.
13250 (reduc_<sur>plus_v2si): Likewise.
13251 (reduc_<sur>plus_v4sf): Likewise.
13252 (aarch64_addpv4sf): Likewise.
13253 * config/aarch64/arm_neon.h
13254 (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
13255 * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
13256 add UNSPEC_SADDV, UNSPEC_UADDV.
13258 (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
13260 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13262 * config/aarch64/arm_neon.h
13263 (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
13265 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13267 * config/aarch64/aarch64-builtins
13268 (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
13270 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13272 * config/aarch64/aarch64-simd-builtins.def
13273 (reduc_smax_): New.
13274 (reduc_smin_): Likewise.
13275 (reduc_umax_): Likewise.
13276 (reduc_umin_): Likewise.
13277 (reduc_smax_nan_): Likewise.
13278 (reduc_smin_nan_): Likewise.
13281 (smax): Update for V2SF, V4SF and V2DF modes.
13284 (smin_nan): Likewise.
13285 * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
13286 (<su><maxmin><mode>3): ...This, refactor.
13287 (s<maxmin><mode>3): New.
13288 (<maxmin_uns><mode>3): Likewise.
13289 (reduc_<maxmin_uns>_<mode>): Refactor.
13290 (reduc_<maxmin_uns>_v4sf): Likewise.
13291 (reduc_<maxmin_uns>_v2si): Likewise.
13292 (aarch64_<fmaxmin><mode>: Remove.
13293 * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
13295 (vmin<q>_f<32,64>): Likewise.
13296 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
13298 (su): Add mappings for smax, smin, umax, umin.
13300 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
13301 (FMAXMIN): Rename as...
13302 (FMAXMIN_UNS): ...This.
13304 (fmaxminv): Likewise.
13305 (fmaxmin): Likewise.
13307 (maxmin_uns_op): Likewise.
13309 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13311 * config/aarch64/arm_neon.h
13312 (vac<ge, gt><sd>_f<32, 64>): Rename to...
13313 (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
13314 (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
13316 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13318 * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
13319 * config/aarch64/iterators.md (FAC_COMPARISONS): New.
13321 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13323 * config/aarch64/aarch64-simd.md
13324 (vcond<mode>_internal): Handle special cases for constant masks.
13325 (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
13326 (vcondu<mode><mode>): Likewise.
13327 (vcond<v_cmp_result><mode>): New.
13329 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13331 * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
13332 (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
13333 * config/aarch64/aarch64-simd-builtins.def
13334 (cmeq): Update to BUILTIN_VALLDI.
13339 * config/aarch64/arm_neon.h
13340 (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
13341 to builtins or C as appropriate.
13343 2013-05-01 James Greenhalgh <james.greenhalgh@arm.com>
13345 * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
13347 (cmhi): Rename to...
13349 * config/aarch64/aarch64-simd.md
13350 (simd_mode): Add SF.
13351 (aarch64_vcond_internal): Use new names for unsigned comparison insns.
13352 (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
13353 * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
13354 (cstore<mode>_neg): ...This.
13355 * config/aarch64/iterators.md
13357 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
13358 (COMPARISONS): New.
13359 (UCOMPARISONS): Likewise.
13360 (optab): Add missing comparisons.
13366 (VCMP_S): Likewise.
13367 (VCMP_U): Likewise.
13368 (V_cmp_result): Add DF, SF modes.
13369 (v_cmp_result): Likewise.
13371 (vmtype): Likewise.
13372 * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
13374 2013-05-01 Greta Yorsh <Greta.Yorsh@arm.com>
13376 * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
13377 define_insn to define_insn_and_split.
13378 (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
13379 (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
13380 (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
13381 (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
13382 (thumb2_negscc): Likewise.
13384 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
13386 * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
13388 2013-04-30 Greta Yorsh <Greta.Yorsh@arm.com>
13390 * config/arm/thumb2.md: Remove trailing whitespaces.
13392 2013-04-30 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
13394 * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
13395 Use gen_int_mode rather than GEN_INT.
13397 2013-04-30 H.J. Lu <hongjiu.lu@intel.com>
13399 * value-prof.c (stream_in_histogram_value): Remove the strayed
13402 2013-04-30 Richard Biener <rguenther@suse.de>
13404 PR middle-end/57122
13405 * cfghooks.c (split_edge): Properly check for the loop latch edge.
13407 2013-04-30 Richard Biener <rguenther@suse.de>
13409 PR middle-end/57107
13410 * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
13412 2013-04-30 Andrey Belevantsev <abel@ispras.ru>
13414 PR rtl-optimization/56957
13415 PR rtl-optimization/57105
13416 * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
13417 variable. Use just INSN_UID for determining whether an insn
13418 should be only disconnected from the insn stream.
13419 * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
13421 2013-04-30 Jakub Jelinek <jakub@redhat.com>
13423 PR tree-optimization/57104
13424 * tsan.c (instrument_expr): Don't instrument accesses to
13425 DECL_HARD_REGISTER VAR_DECLs.
13427 2013-04-30 Richard Biener <rguenther@suse.de>
13429 * function.h (loops_for_fn): New inline function.
13430 (set_loops_for_fn): Likewise.
13431 * cfgloop.h (place_new_loop): Add struct function parameter.
13432 (get_loop): Likewise.
13433 (get_loops): Likewise.
13434 (number_of_loops): Likewise.
13435 (fel_next): Adjust.
13436 (fel_init): Likewise.
13437 * cfg.c (get_loop_copy): Adjust.
13438 * cfgloop.c (flow_loops_dump): Likewise.
13439 (record_loop_exits): Likewise.
13440 (verify_loop_structure): Likewise.
13441 * cfgloopanal.c (mark_irreducible_loops): Likewise.
13442 (estimate_reg_pressure_cost): Likewise.
13443 (mark_loop_exit_edges): Likewise.
13444 * cfgloopmanip.c (place_new_loop): Likewise.
13445 (add_loop): Likewise.
13446 (duplicate_loop): Likewise.
13447 * graph.c (draw_cfg_nodes): Likewise.
13448 * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
13449 * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
13450 (extract_affine_chrec): Likewise.
13451 (build_scop_iteration_domain): Likewise.
13452 * graphite.c (graphite_initialize): Likewise.
13453 * ira-build.c (create_loop_tree_nodes): Likewise.
13454 (more_one_region_p): Likewise.
13455 (rebuild_regno_allocno_maps): Likewise.
13456 (mark_loops_for_removal): Likewise.
13457 (mark_all_loops_for_removal): Likewise.
13458 (remove_unnecessary_regions): Likewise.
13459 (ira_build): Likewise.
13460 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
13461 * loop-init.c (fix_loop_structure): Likewise.
13462 (gate_rtl_move_loop_invariants): Likewise.
13463 (gate_rtl_unswitch): Likewise.
13464 (gate_rtl_unroll_and_peel_loops): Likewise.
13465 (rtl_doloop): Likewise.
13466 * lto-streamer-in.c (input_cfg): Likewise.
13467 * lto-streamer-out.c (output_cfg): Likewise.
13468 * modulo-sched.c (sms_schedule): Likewise.
13469 * predict.c (tree_estimate_probability): Likewise.
13470 (tree_estimate_probability_driver): Likewise.
13471 (estimate_loops): Likewise.
13472 * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
13473 (move_sese_region_to_fn): Likewise.
13474 (debug_loop_num): Likewise.
13475 * tree-chrec.c (chrec_evaluate): Likewise.
13476 (hide_evolution_in_other_loops_than_loop): Likewise.
13477 (chrec_component_in_loop_num): Likewise.
13478 (reset_evolution_in_loop): Likewise.
13479 (evolution_function_is_invariant_rec_p): Likewise.
13480 * tree-if-conv.c (main_tree_if_conversion): Likewise.
13481 * tree-inline.c (copy_loops): Likewise.
13482 (copy_cfg_body): Likewise.
13483 (tree_function_versioning): Likewise.
13484 * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
13485 * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
13487 (add_to_evolution_1): Likewise.
13488 (scev_const_prop): Likewise.
13489 * tree-scalar-evolution.h (get_chrec_loop): Likewise.
13490 * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
13491 * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
13492 (tree_ssa_lim_initialize): Likewise.
13493 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
13494 (verify_loop_closed_ssa): Likewise.
13495 * tree-ssa-loop.c (tree_ssa_loop_init): Likewise.
13496 (tree_ssa_loop_im): Likewise.
13497 (tree_ssa_loop_unswitch): Likewise.
13498 (tree_vectorize): Likewise.
13499 (check_data_deps): Likewise.
13500 (tree_ssa_loop_ivcanon): Likewise.
13501 (tree_ssa_loop_bounds): Likewise.
13502 (tree_complete_unroll): Likewise.
13503 (tree_complete_unroll_inner): Likewise.
13504 (tree_parallelize_loops): Likewise.
13505 (tree_ssa_loop_prefetch): Likewise.
13506 (tree_ssa_loop_ivopts): Likewise.
13507 * tree-ssa.c (execute_update_addresses_taken): Liekwise.
13508 * tree-vectorizer.c (vectorize_loops): Likewise.
13510 2013-04-29 Mike Frysinger <vapier@gentoo.org>
13512 * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
13513 (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
13514 * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
13515 with EABI_LINK_SPEC.
13517 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
13520 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
13523 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
13526 * lra-constraints.c (process_alt_operands): Discourage a bit more
13527 using memory for pseudos. Print cost dump for alternatives.
13528 Modify cost values for conflicts with early clobbers.
13529 (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
13531 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
13534 * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
13536 2013-04-29 Ian Bolton <ian.bolton@arm.com>
13538 * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
13539 from/to S register.
13540 (movdi_aarch64): Support LDR/STR from/to D register.
13542 2013-04-29 Ian Bolton <ian.bolton@arm.com>
13544 * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
13545 or higher by default.
13547 2013-04-29 Richard Biener <rguenther@suse.de>
13549 PR middle-end/57075
13550 * tree-inline.c (copy_edges_for_bb): Still split the bbs,
13551 even if not adding abnormal edges for calls that can make
13554 2013-04-29 Richard Biener <rguenther@suse.de>
13556 PR middle-end/57103
13557 * tree-cfg.c (move_stmt_op): Fix condition under which to update
13559 (move_stmt_r): Remove redundant checking.
13561 2013-04-29 Teresa Johnson <tejohnson@google.com>
13564 * basic-block.h (apply_scale): New function.
13565 (apply_probability): Use apply_scale.
13566 * gimple-streamer-in.c (input_bb): Ditto.
13567 * lto-streamer-in.c (input_cfg): Ditto.
13568 * lto-cgraph.c (merge_profile_summaries): Ditto.
13569 * tree-optimize.c (execute_fixup_cfg): Ditto.
13570 * tree-inline.c (copy_bb): Update comment to use apply_scale.
13571 (copy_edges_for_bb): Ditto.
13572 (copy_cfg_body): Ditto.
13574 2013-04-29 Tom de Vries <tom@codesourcery.com>
13576 * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
13577 (replace_block_by): Don't set LOOPS_NEED_FIXUP.
13578 (tail_merge_optimize): Handle current_loops == NULL.
13580 2013-04-26 Jeff Law <law@redhat.com>
13582 * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
13583 (simplify_cond_using_ranges): Generalize code to simplify
13584 COND_EXPRs where one argument is a constant and the other
13585 is an SSA_NAME created by an integral type conversion.
13587 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13589 * config/arm/arm.md (store_minmaxsi): Use only when
13590 optimize_insn_for_size_p.
13592 2013-04-29 Christian Bruel <christian.bruel@st.com>
13595 * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
13597 2013-04-29 Richard Biener <rguenther@suse.de>
13599 PR middle-end/57089
13600 * omp-low.c (expand_omp_taskreg): If the parent function had a broken
13601 loop tree make sure to schedule a fixup for the child as well.
13602 (expand_omp_for_generic): Properly add loops.
13603 (expand_omp_for_static_nochunk): Likewise.
13604 (expand_omp_for_static_chunk): Likewise.
13605 (expand_omp_for): For the degenerate case fixup loops.
13606 (expand_omp_sections): Fix default bb placement in loops.
13607 (expand_omp_atomic_pipeline): Properly add loops.
13609 2013-04-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13611 * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
13613 2013-04-29 Tom de Vries <tom@codesourcery.com>
13615 * tree-ssa-tail-merge.c: Update header comment.
13617 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13619 * config/aarch64/arm_neon.h
13620 (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
13621 (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
13622 (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
13623 (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
13624 (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
13625 (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
13626 (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
13627 (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
13629 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13631 * config/aarch64/aarch64-simd.md
13632 (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
13633 (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
13634 fix_trunc, fixuns_trunc.
13635 (ftrunc<VDQF:mode>2): New.
13636 * config/aarch64/iterators.md (optab): Add fix, fixuns.
13637 (fix_trunc_optab): New.
13639 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13641 * config/aarch64/aarch64-builtins.c
13642 (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
13643 iceilf, lround, iroundf.
13645 2013-04-29 Uros Bizjak <ubizjak@gmail.com>
13648 * config/i386/i386.h (enum ix86_tune_indices)
13649 <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
13650 New, split from X86_TUNE_INTER_UNIT_MOVES.
13651 <X86_TUNE_INTER_UNIT_MOVES>: Remove.
13652 (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
13653 (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
13654 (TARGET_INTER_UNIT_MOVES): Remove.
13655 * config/i386/i386.c (initial_ix86_tune_features): Update.
13656 Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
13657 (ix86_expand_convert_uns_didf_sse): Use
13658 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13659 (ix86_expand_vector_init_one_nonzero): Ditto.
13660 (ix86_expand_vector_init_interleave): Ditto.
13661 (inline_secondary_memory_needed): Return true for moves from SSE class
13662 registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
13663 to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13664 * config/i386/constraints.md (Yi, Ym): Depend on
13665 TARGET_INTER_UNIT_MOVES_TO_VEC.
13666 (Yj, Yn): New constraints.
13667 * config/i386/i386.md (*movdi_internal): Change constraints of
13668 operand 1 from Yi to Yj and from Ym to Yn.
13669 (*movsi_internal): Ditto.
13670 (*movdf_internal): Ditto.
13671 (*movsf_internal): Ditto.
13672 (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
13673 TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
13674 (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
13675 (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
13676 (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
13677 * config/i386/sse.md (movdi_to_sse): Ditto.
13678 (sse2_stored): Change constraint of operand 1 from Yi to Yj.
13679 Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
13680 TARGET_INTER_UNIT_MOVES.
13681 (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
13682 (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
13683 instead of TARGET_INTER_UNIT_MOVES.
13684 * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
13685 operand 1 from Yi to Yj and from Ym to Yn.
13687 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13689 * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
13690 (float_truncate_hi_): Likewise.
13691 (float_extend_lo_): Likewise.
13692 (float_truncate_lo_): Likewise.
13693 * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
13694 (aarch64_float_extend_lo_v2df): Likewise.
13695 (vec_unpacks_hi_v4sf): Likewise.
13696 (aarch64_float_truncate_lo_v2sf): Likewise.
13697 (aarch64_float_truncate_hi_v4sf): Likewise.
13698 (vec_pack_trunc_v2df): Likewise.
13699 (vec_pack_trunc_df): Likewise.
13701 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13703 * config/aarch64/aarch64-builtins.c
13704 (aarch64_fold_builtin): Fold float conversions.
13705 * config/aarch64/aarch64-simd-builtins.def
13706 (floatv2si, floatv4si, floatv2di): New.
13707 (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
13708 * config/aarch64/aarch64-simd.md
13709 (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
13710 * config/aarch64/iterators.md (FLOATUORS): New.
13711 (optab): Add float, floatuns.
13712 (su_optab): Likewise.
13714 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13716 * config/aarch64/aarch64-builtins.c
13717 (aarch64_builtin_vectorized_function): Use new names for
13719 * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
13720 (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
13721 (fcvtzu): Split as...
13722 (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
13723 (fcvtas): Split as...
13724 (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
13725 (fcvtau): Split as...
13726 (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
13727 (fcvtps): Split as...
13728 (lceilv2sf, lceilv4sf, lceilv2df): ...This.
13729 (fcvtpu): Split as...
13730 (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
13731 (fcvtms): Split as...
13732 (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
13733 (fcvtmu): Split as...
13734 (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
13735 (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
13736 (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
13737 (lfrintnusf, lfrintnudf): Likewise.
13738 * config/aarch64/aarch64-simd.md
13739 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
13741 (aarch64_fcvt<frint_suffix><su><mode>): Remove.
13742 * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
13743 (fcvt_pattern): Likewise.
13745 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13747 * config/aarch64/aarch64-simd.md
13748 (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
13749 (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
13751 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13753 * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
13754 (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
13755 (vrnd<a,m,n,p>_f32): Implement using builtins.
13756 (vrnd<i,x><q>_f<32, 64>): New.
13758 2013-04-29 James Greenhalgh <james.greenhalgh@arm.com>
13760 * config/aarch64/aarch64-builtins.c
13761 (aarch64_builtin_vectorized_function): Fold to standard pattern names.
13762 * config/aarch64/aarch64-simd-builtins.def (frintn): New.
13763 (frintz): Rename to...
13765 (frintp): Rename to...
13767 (frintm): Rename to...
13769 (frinti): Rename to...
13770 (nearbyint): ...this.
13771 (frintx): Rename to...
13773 (frinta): Rename to...
13775 * config/aarch64/aarch64-simd.md
13776 (aarch64_frint<frint_suffix><mode>): Delete.
13777 (<frint_pattern><mode>2): Convert to insn.
13778 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
13779 * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
13780 (frint_pattern): Likewise.
13781 (frint_suffix): Likewise.
13783 2013-04-29 Richard Biener <rguenther@suse.de>
13785 PR tree-optimization/57081
13786 * loop-init.c: Include tree-flow.h.
13787 (loop_optimizer_finalize): Free number of iteration estimates.
13788 * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
13790 2013-04-29 Jakub Jelinek <jakub@redhat.com>
13792 PR tree-optimization/57083
13793 * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
13794 non-singleton shift count range, zero extend low_bound for uns case.
13796 * config/i386/predicates.md (general_vector_operand): New predicate.
13797 * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
13798 (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
13799 if they aren't nonimmediate operands. If their original values
13800 satisfy const_vector_equal_evenodd_p, don't shift them.
13801 * config/i386/sse.md (mul<mode>3): Use general_vector_operand
13802 predicates. For the SSE4.1 case force operands[{1,2}] into registers
13803 if not nonimmediate_operand.
13804 (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
13805 instead of register_operand.
13806 (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
13808 2013-04-28 Eric Botcazou <ebotcazou@adacore.com>
13810 * stor-layout.c (finalize_size_functions): Allocate a structure and
13811 reset cfun before dumping the functions.
13813 2013-04-27 Jakub Jelinek <jakub@redhat.com>
13815 * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
13818 * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
13819 use xop_pmacsdqh if uns_p.
13820 * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
13821 the immediate rotate count.
13823 2013-04-26 Vladimir Makarov <vmakarov@redhat.com>
13825 * rtl.h (struct rtx_def): Add comment for field jump.
13826 (LRA_SUBREG_P): New macro.
13827 * recog.c (register_operand): Check LRA_SUBREG_P.
13828 * lra.c (lra): Add note at the end of RTL code. Align non-empty
13830 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
13831 (lra_final_code_change): Skip subreg change for operators.
13832 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
13833 if there are no operand changes.
13834 * lra-constraints.c (curr_insn_set): New.
13835 (match_reload): Set LRA_SUBREG_P.
13836 (emit_spill_move): Ditto.
13837 (check_and_process_move): Use curr_insn_set. Process only single
13838 set insns. Don't initialize sec_mem_p and change_p.
13839 (simplify_operand_subreg): Use LRA_SUBREG_P.
13840 (reg_in_class_p): New function.
13841 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
13842 of #ifdef. Add code to remove cycling.
13843 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
13844 non-null disp. Reload inner instead of disp when base and index
13845 are null. Try to put lo_sum into register.
13846 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
13847 (check_and_process_move): Move code for move cost check to
13848 simple_move_p. Remove equiv_substitution.
13849 (simple_move_p): New function.
13850 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
13851 curr_insn_set. Call check_and_process_move only for single set
13852 insns. Use the new function. Move call of check_and_process_move
13853 after operand equiv substitution and address process.
13855 2013-04-26 Jakub Jelinek <jakub@redhat.com>
13858 * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
13859 with nonlocal goto receivers or returns twice calls, ignore
13860 unininitialized values from abnormal edges to nl goto receiver
13861 or returns twice call.
13863 2013-04-26 Jakub Jelinek <jakub@redhat.com>
13865 PR tree-optimization/57051
13866 * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
13867 and VEC_RSHIFT_EXPR if shift count is a multiple of element
13870 2013-04-26 Richard Biener <rguenther@suse.de>
13872 * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
13873 (expand_omp_taskreg): Likewise. Mark loops for fixup.
13874 * tree-cfg.c (move_block_to_fn): Remap loop fathers.
13875 (fixup_loop_arrays_after_move): New function.
13876 (move_sese_region_to_fn): Properly outline the loop tree parts
13877 of the SESE region.
13879 2013-04-26 Uros Bizjak <ubizjak@gmail.com>
13881 * config/i386/i386.md (type, unit): Fix long lines.
13883 2013-04-26 Richard Biener <rguenther@suse.de>
13885 * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
13886 (lto-streamer-out.o): Likewise.
13887 * cfgloop.c (init_loops_structure): Export, add struct function
13888 argument and adjust.
13889 (flow_loops_find): Adjust.
13890 * cfgloop.h (enum loop_estimation): Add EST_LAST.
13891 (init_loops_structure): Declare.
13892 * lto-streamer-in.c: Include cfgloop.h.
13893 (input_cfg): Input the loop tree.
13894 * lto-streamer-out.c: Include cfgloop.h.
13895 (output_cfg): Output the loop tree.
13896 (output_struct_function_base): Do not drop PROP_loops.
13898 2013-03-26 Richard Biener <rguenther@suse.de>
13900 * tree-cfg.c (execute_build_cfg): Build the loop tree.
13901 (pass_build_cfg): Provide PROP_loops.
13902 (move_sese_region_to_fn): Remove loops that are outlined into fn
13904 * tree-inline.c: Include cfgloop.h.
13905 (initialize_cfun): Do not drop PROP_loops.
13906 (copy_loops): New function.
13907 (copy_cfg_body): Copy loop structure.
13908 (tree_function_versioning): Initialize destination loop tree.
13909 * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
13910 (pass_parallelize_loops): Do IL verification.
13911 * loop-init.c (loop_optimizer_init): Fixup loops if required.
13912 * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
13913 the CFG make sure we fixup loops as well.
13914 * tree-ssa-tail-merge.c: Include cfgloop.h.
13915 (replace_block_by): When merging loop latches mark loops for fixup.
13916 * lto-streamer-out.c (output_struct_function_base): Drop
13917 PROP_loops for now.
13918 * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
13919 (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
13920 * ipa-split.c: Include cfgloop.h.
13921 (split_function): Add the new return block to the loop tree root.
13922 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
13923 whether we have removed the forwarder block.
13924 (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
13925 * cfgloop.h (place_new_loop): Declare.
13926 * cfgloopmanip.c (place_new_loop): Export.
13927 * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
13928 (tree-switch-conversion.o): Likewise.
13929 (tree-complex.o): Likewise.
13930 (tree-inline.o): Likewise.
13931 (tree-ssa-tailmerge.o): Likewise.
13932 (ipa-split.o): Likewise.
13933 (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
13934 (tree-ssa-copy.o): Likewise.
13935 * tree-switch-conversion.c: Include cfgloop.h
13936 (process_switch): If we emit a bit-test cascade, schedule loops
13938 * tree-complex.c: Include cfgloop.h.
13939 (expand_complex_div_wide): Properly add new basic-blocks to loops.
13940 * asan.c: Include cfgloop.h.
13941 (create_cond_insert_point): Properly add new basic-blocks to
13942 loops, schedule loop fixup.
13943 * cfgloop.c (verify_loop_structure): Check that looks are not
13945 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
13947 (expand_omp_for_generic): Likewise.
13948 (expand_omp_sections): Likewise.
13949 (expand_omp_atomic_pipeline): Schedule loops for fixup.
13950 * tree-ssa-copy.c: Include tree-scalar-evolution.h.
13951 (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
13952 is initialized, not when loops are present.
13953 * tree-parloops.c (parallelize_loops): Remove checking here.
13954 * passes.c (init_optimization_passes): Schedule a copy-propagation
13955 pass before complete unrolling of inner loops.
13957 2013-04-26 Jakub Jelinek <jakub@redhat.com>
13959 * Makefile.in (toplev.o): Depend on diagnostic-color.h.
13960 * diagnostic-color.c (should_colorize): Remove _WIN32 version.
13961 (colorize_init): Add argument to _WIN32 version.
13962 * toplev.c: Include diagnostic-color.h.
13963 (process_options): Default to -fdiagnostics-color=auto if
13964 GCC_COLORS env var is in the environment.
13965 * common.opt (fdiagnostics-color=): Add Var and Init.
13966 * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
13967 env var is in the environment, the default is auto rather than never.
13969 * diagnostic.h (file_name_as_prefix): Add context argument.
13970 * diagnostic.c (file_name_as_prefix): Likewise. Colorize
13971 the string as locus.
13972 * langhooks.c (lhd_print_error_function): Adjust caller.
13974 2013-04-25 Lawrence Crowl <crowl@google.com>
13976 * var-tracking.c (shared_hash_def::htab):
13977 Change type to hash_table. Update dependent calls and types.
13979 2013-04-25 Lawrence Crowl <crowl@google.com>
13981 * Makefile.in: Update as needed below.
13983 * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
13984 Move declaration to after the type's method definitons.
13986 * attribs.c (htab_t scoped_attributes::attribute_hash):
13987 Change type to hash_table. Update dependent calls and types.
13989 * bitmap.c (htab_t bitmap_desc_hash):
13990 Change type to hash_table. Update dependent calls and types.
13992 * cselib.c (htab_t cselib_hash_table):
13993 Change type to hash_table. Update dependent calls and types.
13995 * data-streamer.h (struct string_slot): Move to lto-streamer.h.
13996 (hash_string_slot_node): Move implementation into lto-streamer.h
13997 struct string_slot_hasher.
13998 (eq_string_slot_node): Likewise.
14000 * data-streamer-out.c: Update output_block::string_hash_table
14001 dependent calls and types.
14003 * dwarf2cfi.c (htab_t trace_index):
14004 Change type to hash_table. Update dependent calls and types.
14006 * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
14007 Change type to hash_table. Update dependent calls and types.
14008 (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
14009 (htab_t optimize_external_refs::map): Likewise.
14010 (htab_t output_comp_unit::extern_map): Likewise.
14011 (htab_t output_comdat_type_unit::extern_map): Likewise.
14012 (htab_t output_macinfo::macinfo_htab): Likewise.
14013 (htab_t optimize_location_lists::htab): Likewise.
14014 (htab_t dwarf2out_finish::comdat_type_table): Likewise.
14016 * except.c (htab_t ehspec_hash_type):
14017 Change type to hash_table. Update dependent calls and types.
14018 (assign_filter_values::ttypes): Likewise.
14019 (assign_filter_values::ehspec): Likewise.
14020 (sjlj_assign_call_site_values::ar_hash): Likewise.
14021 (convert_to_eh_region_ranges::ar_hash): Likewise.
14023 * gcse.c (htab_t pre_ldst_table):
14024 Change type to hash_table. Update dependent calls and types.
14026 * ggc-common.c (htab_t saving_htab):
14027 Change type to hash_table. Update dependent calls and types.
14028 (htab_t loc_hash): Likewise.
14029 (htab_t ptr_hash): Likewise.
14030 (call_count): Rename ggc_call_count.
14031 (call_alloc): Rename ggc_call_alloc.
14032 (loc_descriptor): Rename make_loc_descriptor.
14033 (add_statistics): Rename ggc_add_statistics.
14035 * ggc-common.c (saving_htab):
14036 Change type to hash_table. Update dependent calls and types.
14038 * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
14039 (push_gimplify_context): Likewise.
14040 (pop_gimplify_context): Likewise.
14041 (struct gimple_temp_hash_elt): Added.
14042 (struct gimplify_hasher): Likewise.
14043 (struct gimplify_ctx.temp_htab):
14044 Change type to hash_table. Update dependent calls and types.
14046 * gimple-fold.c: Include gimplify-ctx.h.
14048 * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
14049 Change type to hash_table. Update dependent calls and types.
14050 (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
14051 avoid potential global name collision.
14053 * gimplify.c: Include gimplify-ctx.h.
14054 (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
14055 (htab_t gimplify_ctx::temp_htab):
14056 Update dependent calls and types for new type hash_table.
14057 (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
14058 (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
14060 * gimplify-ctx.h: New.
14061 (struct gimple_temp_hash_elt): Move from gimplify.c.
14062 (class gimplify_hasher): New.
14063 (struct gimplify_ctx): Move from gimple.h.
14064 (htab_t gimplify_ctx::temp_htab):
14065 Change type to hash_table. Update dependent calls and types.
14067 * graphite-clast-to-gimple.c: Include graphite-htab.h.
14068 (htab_t ivs_params::newivs_index):
14069 Change type to hash_table. Update dependent calls and types.
14070 (htab_t ivs_params::params_index): Likewise.
14071 (htab_t print_generated_program::params_index): Likewise.
14072 (htab_t gloog::newivs_index): Likewise.
14073 (htab_t gloog::params_index): Likewise.
14075 * graphite.c: Include graphite-htab.h.
14076 4htab_t graphite_transform_loops::bb_pbb_mapping):
14077 Change type to hash_table. Update dependent calls and types.
14079 * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
14080 (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
14081 (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
14083 * graphite-dependences.c: Include graphite-htab.h.
14084 (loop_is_parallel_p): Change hash table type of parameter.
14086 * graphite-htab.h: New.
14087 (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
14088 (extern find_pbb_via_hash): Move from graphite-poly.h.
14089 (extern loop_is_parallel_p): Move from graphite-poly.h.
14090 (extern get_loop_body_pbbs): Move from graphite-poly.h.
14092 * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
14093 (extern loop_is_parallel_p): Move to graphite-htab.h.
14094 (extern get_loop_body_pbbs): Move to graphite-htab.h.
14096 * haifa-sched.c (htab_t delay_htab):
14097 Change type to hash_table. Update dependent calls and types.
14098 (htab_t delay_htab_i2): Likewise.
14100 * ira-color.c (htab_t allocno_hard_regs_htab):
14101 Change type to hash_table. Update dependent calls and types.
14103 * ira-costs.c (htab_t cost_classes_htab):
14104 Change type to hash_table. Update dependent calls and types.
14106 * loop-invariant.c (htab_t merge_identical_invariants::eq):
14107 Change type to hash_table. Update dependent calls and types.
14109 * loop-iv.c (htab_t bivs):
14110 Change type to hash_table. Update dependent calls and types.
14112 * loop-unroll.c (htab_t opt_info::insns_to_split):
14113 Change type to hash_table. Update dependent calls and types.
14114 (htab_t opt_info::insns_with_var_to_expand): Likewise.
14116 * lto-streamer.h (struct string_slot): Move from data-streamer.h
14117 (struct string_slot_hasher): New.
14118 (htab_t output_block::string_hash_table):
14119 Change type to hash_table. Update dependent calls and types.
14121 * lto-streamer-in.c (freeing_string_slot_hasher): New.
14122 (htab_t file_name_hash_table):
14123 Change type to hash_table. Update dependent calls and types.
14125 * lto-streamer-out.c: Update output_block::string_hash_table dependent
14128 * lto-streamer.c (htab_t tree_htab):
14129 Change type to hash_table. Update dependent calls and types.
14131 * omp-low.c: Include gimplify-ctx.h.
14133 * passes.c (htab_t name_to_pass_map):
14134 Change type to hash_table. Update dependent calls and types.
14135 (pass_traverse): Rename to passes_pass_traverse.
14137 * plugin.c (htab_t event_tab):
14138 Change type to hash_table. Update dependent calls and types.
14140 * postreload-gcse.c (htab_t expr_table):
14141 Change type to hash_table. Update dependent calls and types.
14142 (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
14144 * sese.c (debug_rename_map_1): Make extern.
14145 (htab_t copy_bb_and_scalar_dependences::rename_map):
14146 Change type to hash_table. Update dependent calls and types.
14148 * sese.h (extern debug_rename_map): Move to .c file.
14150 * store-motion.c (htab_t store_motion_mems_table):
14151 Change type to hash_table. Update dependent calls and types.
14153 * trans-mem.c (htab_t tm_new_mem_hash):
14154 Change type to hash_table. Update dependent calls and types.
14156 * tree-browser.c (htab_t TB_up_ht):
14157 Change type to hash_table. Update dependent calls and types.
14159 * tree-cfg.c (htab_t discriminator_per_locus):
14160 Change type to hash_table. Update dependent calls and types.
14162 * tree-complex.c: Include tree-hasher.h
14163 (htab_t complex_variable_components):
14164 Change type to hash_table. Update dependent calls and types.
14166 * tree-eh.c (htab_t finally_tree):
14167 Change type to hash_table. Update dependent calls and types.
14169 * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
14170 struct int_tree_hasher.
14171 (extern int_tree_map_eq): Likewise.
14172 (uid_decl_map_hash): Removed.
14173 (extern decl_tree_map_eq): Likewise.
14175 * tree-hasher.h: New.
14176 (struct int_tree_hasher): New.
14177 (typedef int_tree_htab_type): New.
14179 * tree-inline.c: Include gimplify-ctx.h.
14181 * tree-mudflap.c: Include gimplify-ctx.h.
14183 * tree-parloops.c: Include tree-hasher.h.
14184 (htab_t eliminate_local_variables_stmt::decl_address):
14185 Change type to hash_table. Update dependent calls and types.
14186 (htab_t separate_decls_in_region::decl_copies): Likewise.
14188 * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
14189 Change type to hash_table. Update dependent calls and types.
14191 * tree-sra.c (candidates):
14192 Change type to hash_table. Update dependent calls and types.
14194 * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
14196 (int_tree_map_hash): Likewise.
14198 * tree-ssa-dom.c (htab_t avail_exprs):
14199 Change type to hash_table. Update dependent calls and types.
14201 * tree-ssa-live.c (var_map_base_init::tree_to_index):
14202 Change type to hash_table. Update dependent calls and types.
14204 * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
14205 Change type to hash_table. Update dependent calls and types.
14207 * tree-ssa-phiopt.c (seen_ssa_names):
14208 Change type to hash_table. Update dependent calls and types.
14210 * tree-ssa-strlen.c (decl_to_stridxlist_htab):
14211 Change type to hash_table. Update dependent calls and types.
14213 * tree-ssa-uncprop.c (equiv):
14214 Change type to hash_table. Update dependent calls and types.
14216 2013-04-25 Jakub Jelinek <jakub@redhat.com>
14218 PR rtl-optimization/57003
14219 * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
14220 call note_stores with kill_clobbered_value callback again after
14221 killing regs_invalidated_by_call.
14223 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
14225 * config/aarch64/aarch64-simd.md
14226 (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
14227 (aarch64_simd_bsl<mode>): Likewise.
14228 * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
14230 2013-04-25 Marek Polacek <polacek@redhat.com>
14232 PR tree-optimization/57066
14233 * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
14235 2013-04-25 James Greenhalgh <jame.greenhalgh@arm.com>
14237 * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
14239 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
14241 * config/aarch64/aarch64-builtins.c
14242 (aarch64_fold_builtin): New.
14243 * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
14244 * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
14245 * config/aarch64/aarch64-simd-builtins.def (abs): New.
14246 * config/aarch64/arm_neon.h
14247 (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
14249 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
14250 Tejas Belagod <tejas.belagod@arm.com>
14252 * config/aarch64/aarch64-builtins.c
14253 (aarch64_gimple_fold_builtin): New.
14254 * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
14255 * config/aarch64/aarch64-simd-builtins.def (addv): New.
14256 * config/aarch64/aarch64-simd.md (addpv4sf): New.
14257 (addvv4sf): Update.
14258 * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
14260 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14262 * config/aarch64/aarch64.md
14263 (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
14265 2013-04-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14267 * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
14268 (*ngcsi_uxtw): New pattern.
14270 2013-04-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14271 Julian Brown <julian@codesourcery.com>
14273 * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
14274 (TB_DREG): Add T_V4HF.
14275 (v4hf_UP): New macro.
14276 (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14277 (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14278 Handle initialisation of V4HF. Adjust initialisation of reinterpret
14280 (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
14281 (arm_vector_mode_supported_p): Handle V4HF.
14282 (arm_mangle_map): Handle V4HFmode.
14283 * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
14284 * config/arm/arm_neon_builtins.def: Add entries for
14285 vcvtv4hfv4sf, vcvtv4sfv4hf.
14286 * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
14287 (neon_vcvtv4hfv4sf): Likewise.
14288 * config/arm/neon-gen.ml: Handle half-precision floating point
14290 * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
14291 * config/arm/arm_neon.h: Regenerate.
14292 * config/arm/neon.ml (type elts): Add F16.
14293 (type vectype): Add T_float16x4, T_floatHF.
14294 (type vecmode): Add V4HF.
14295 (type features): Add Requires_FP_bit feature.
14296 (elt_width): Handle F16.
14297 (elt_class): Likewise.
14298 (elt_of_class_width): Likewise.
14299 (mode_of_elt): Refactor.
14300 (type_for_elt): Handle F16, fix error messages.
14301 (vectype_size): Handle T_float16x4.
14302 (vcvt_sh): New function.
14303 (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
14304 (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
14305 (string_of_mode): Handle V4HF.
14306 * doc/arm-neon-intrinsics.texi: Regenerate.
14308 2013-04-25 James Greenhalgh <james.greenhalgh@arm.com>
14310 * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
14311 format specifier in 'X' case.
14313 2013-04-25 Alan Modra <amodra@gmail.com>
14316 * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
14317 rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
14318 (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
14319 Repeat for many other rotate/shift and mask patterns using subregs.
14320 Name lshiftrt insns.
14321 (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
14322 on WORDS_BIG_ENDIAN.
14324 2013-04-25 Alan Modra <amodra@gmail.com>
14326 * config.gcc: Support little-endian powerpc-linux targets.
14327 * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
14328 (LINK_OS_LINUX_SPEC): Define.
14329 * config/rs6000/linuxspe.h (TARGET_DEFAULT):
14330 Preserve MASK_LITTLE_ENDIAN.
14331 * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
14332 * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
14333 * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
14334 (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
14335 (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
14336 * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
14337 Correct fp word order for little-endian. Don't shift toc entries
14338 smaller than a word for little-endian.
14339 * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
14340 (bswapdi2 splits): Correct low-part subreg for little-endian.
14341 Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
14342 low/high where such is correct only for be.
14343 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
14344 little-endian for -mcall-aixdesc.
14346 2013-04-25 Alan Modra <amodra@gmail.com>
14348 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
14349 replace_equiv_address_nv.
14351 2013-04-25 Alan Modra <amodra@gmail.com>
14353 * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
14355 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
14358 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
14359 * rtl.h (struct rtx_def): ...
14361 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
14363 PR rtl-optimizations/57046
14364 * lra-constraints (split_reg): Set up lra_risky_transformations_p
14365 for multi-reg splits.
14367 2013-04-24 H.J. Lu <hongjiu.lu@intel.com>
14369 * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
14371 2013-04-24 Sterling Augustine <saugustine@google.com>
14373 * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
14374 (comp_dir_string, debug_str_dwo_section): New.
14375 (DEBUG_STR_DWO_SECTION): Rename to ...
14376 (DEBUG_DWO_STR_SECTION): ... this.
14377 (DEBUG_NORM_STR_SECTION): Delete.
14378 (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
14379 (DEBUG_STR_DWO_SECTION_FLAGS): New.
14380 (find_AT_string): Move most logic to ...
14381 (find_AT_string_in_table): ... here. New.
14382 (add_top_level_skeleton_die_attrs): Call comp_dir_string and
14383 add_skeleton_AT_string. Delete logic.
14384 (output_skeleton_debug_sections): Remove call to
14385 add_top_level_skeleton_die_attrs.
14386 (add_comp_dir_attribute): Move logic to comp_dir_string.
14387 (dwarf2out_init): Initialize debug_str_dwo_section.
14388 (output_indirect_string): Call find_string_form.
14389 (output_indirect_strings): Rewrite.
14390 (prune_unused_types): Empty skeleton_debug_str_hash.
14391 Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
14392 (dwarf2out_finish): Call output_indirect_strings.
14394 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
14396 * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
14398 2013-04-24 Vladimir Makarov <vmakarov@redhat.com>
14400 * rtl.h (struct rtx_def): Add comment for field jump.
14401 (LRA_SUBREG_P): New macro.
14402 * recog.c (register_operand): Check LRA_SUBREG_P.
14403 * lra.c (lra): Add note at the end of RTL code. Align non-empty
14405 * lra-spills.c (lra_spill): Align stack after spilling pseudos.
14406 (lra_final_code_change): Skip subreg change for operators.
14407 * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
14408 if there are no operand changes.
14409 * lra-constraints.c (curr_insn_set): New.
14410 (match_reload): Set LRA_SUBREG_P.
14411 (emit_spill_move): Ditto.
14412 (check_and_process_move): Use curr_insn_set. Process only single
14413 set insns. Don't initialize sec_mem_p and change_p.
14414 (simplify_operand_subreg): Use LRA_SUBREG_P.
14415 (reg_in_class_p): New function.
14416 (process_alt_operands): Use it. Use #if HAVE_ATTR_enabled instead
14417 of #ifdef. Add code to remove cycling.
14418 (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
14419 non-null disp. Reload inner instead of disp when base and index
14420 are null. Try to put lo_sum into register.
14421 (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
14422 (check_and_process_move): Move code for move cost check to
14423 simple_move_p. Remove equiv_substitution.
14424 (simple_move_p): New function.
14425 (curr_insn_transform): Initialize sec_mem_p and change_p. Set up
14426 curr_insn_set. Call check_and_process_move only for single set
14427 insns. Use the new function. Move call of check_and_process_move
14428 after operand equiv substitution and address process.
14430 2013-04-24 James Greenhalgh <james.greenhalgh@arm.com>
14432 * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
14433 (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
14434 (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
14436 2013-04-24 Paolo Carlini <paolo.carlini@oracle.com>
14438 * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
14440 2013-04-24 Marek Polacek <polacek@redhat.com>
14442 * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
14443 * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
14444 (select_loops_exit_conditions): Likewise.
14445 (number_of_iterations_for_all_loops): Likewise.
14446 (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
14447 (scev_analysis): Likewise.
14449 2013-04-02 Catherine Moore <clm@codesourcery.com>
14450 Chao-ying Fu <fu@mips.com>
14452 * config/mips/micromips.md (jraddiusp): New pattern.
14453 * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
14454 instruction if possible.
14456 2013-04-24 Alan Modra <amodra@gmail.com>
14458 * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
14460 2013-04-24 Julian Brown <julian@codesourcery.com>
14461 Chung-Lin Tang <cltang@codesourcery.com>
14463 * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
14464 dependency behavior in enumeration type DIE generation. Add TODO note
14465 to comments about future DW_FORM_sdata/udata re-work of related code.
14467 2013-04-23 Lawrence Crowl <crowl@google.com>
14469 * Makefile.in: Update as needed below.
14471 * hash-table.h (class hash_table):
14472 Correct many methods with parameter types compare_type to the correct
14473 value_type. (Correct code was unlikely to notice the change.)
14474 (hash_table::elements_with_deleted) New.
14475 (class hashtable::iterator): New.
14476 (hashtable::begin()): New.
14477 (hashtable::end()): New.
14478 (FOR_EACH_HASH_TABLE_ELEMENT): New.
14480 * statistics.c (statistics_hashes):
14481 Change type to hash_table. Update dependent calls and types.
14483 * tree-into-ssa.c (var_infos):
14484 Change type to hash_table. Update dependent calls and types.
14486 * tree-ssa-coalesce.c (struct coalesce_list_d.list):
14487 Change type to hash_table. Update dependent calls and types.
14489 * tree-ssa-loop-im.c (struct mem_ref.refs):
14490 Change type to hash_table. Update dependent calls and types.
14492 * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
14493 Change type to hash_table. Update dependent calls and types.
14495 * tree-ssa-sccvn.c (vn_tables_s::nary):
14496 Change type to hash_table. Update dependent calls and types.
14497 (vn_tables_s::phis): Likewise.
14498 (vn_tables_s::references): Likewise.
14500 * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
14501 (vn_reference_eq): Update parameter and return types.
14503 * tree-ssa-structalias.c (pointer_equiv_class_table):
14504 Change type to hash_table. Update dependent calls and types.
14505 (location_equiv_class_table): Likewise.
14507 * tree-vect-data-refs.c: Consequential changes for making
14508 peeling a hash_table.
14510 * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
14511 (destroy_loop_vec_info): Dependent hash_table update.
14513 * tree-vectorizer.h (peeling_htab):
14514 Change type to hash_table. Update dependent calls and types.
14516 2013-04-23 Shiva Chen <shiva0217@gmail.com>
14518 * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
14519 to check the register content is equal or not.
14520 * lra-constraints.c (match_reload): Use lra_assign_reg_val
14521 to assign register content record.
14522 * lra-eliminations.c (update_reg_eliminate): Use
14523 lra_update_reg_val_offset to update register content offset.
14524 * lra-int.h (struct lra_reg): Add offset member.
14525 (lra_reg_val_equal_p): New static inline function.
14526 (lra_update_reg_val_offset): New static inline function.
14527 (lra_assign_reg_val): New static inline function.
14528 * lra.c (lra_create_new_reg): Use lra_assign_reg_val
14529 to assign register content record.
14530 (initialize_lra_reg_info_element): Initial offset to zero.
14532 2013-04-23 Catherine Moore <clm@codesourcery.com>
14534 * config/mips/mips.md (*movhi_internal, *movqi_internal): New
14535 operands. Record compression.
14537 2013-04-23 Xinliang David Li <davidxl@google.com>
14539 * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
14541 2013-04-23 Richard Biener <rguenther@suse.de>
14543 PR middle-end/57036
14544 * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
14545 parameter, only add abnormal goto edges from the copied body
14546 if the call could perform abnormal gotos.
14547 (copy_cfg_body): Adjust.
14549 2013-04-23 Sofiane Naci <sofiane.naci@arm.com>
14551 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
14553 2013-04-23 Andreas Schwab <schwab@linux-m68k.org>
14555 * coretypes.h (gimple_stmt_iterator): Add struct to make
14558 2013-04-23 Richard Biener <rguenther@suse.de>
14560 PR tree-optimization/57026
14561 * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
14562 from SSA names occuring in abnormal PHI nodes.
14564 2013-04-22 Andi Kleen <ak@linux.intel.com>
14566 * lto/lto.c (print_lto_report_1): Fix LTO report names.
14568 2013-04-22 Andi Kleen <ak@linux.intel.com>
14570 * lto/lto.c (print_lto_report_1): Declare early.
14571 (read_cgraph_and_symbols): Call print_lto_report_1 early.
14573 2013-04-22 Andi Kleen <ak@linux.intel.com>
14575 * common.opt (-flto-report-wpa): Add.
14576 * doc/invoke.texi (-flto-report-wpa): Add.
14577 * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
14580 2013-04-22 Xinliang David Li <davidxl@google.com>
14582 * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
14583 * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
14584 * Makefile.in: New dependency
14586 David Daney <ddaney.cavm@gmail.com>
14588 * configure.ac (gcc_cv_as_micromips_support): Use the
14589 --fatal-warnings option.
14590 * configure: Regenerate.
14592 2013-04-22 Marek Polacek <polacek@redhat.com>
14595 * tsan.c (instrument_expr): Don't instrument expression
14596 in case its size is zero.
14598 2013-04-22 Uros Bizjak <ubizjak@gmail.com>
14602 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
14604 * config/alpha/alpha.c (TARGET_LRA_P): New define.
14606 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
14608 * coretypes.h (gimple_stmt_iterator_d): Forward declare.
14609 (gimple_stmt_iterator): New typedef.
14610 * gimple.h (gimple_stmt_iterator): Rename to...
14611 (gimple_stmt_iterator_d): ... This.
14612 * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
14613 trees be valid for GIMPLE and GENERIC.
14614 (TARGET_GIMPLE_FOLD_BUILTIN): New.
14615 * gimple-fold.c (gimple_fold_call): Call target hook
14616 gimple_fold_builtin.
14617 * hooks.c (hook_bool_gsiptr_false): New.
14618 * hooks.h (hook_bool_gsiptr_false): New.
14619 * target.def (fold_stmt): New.
14620 * doc/tm.texi: Regenerate.
14622 2013-04-22 Vladimir Makarov <vmakarov@redhat.com>
14625 * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
14626 a set sp if no stack realignment.
14628 2013-04-22 Nick Clifton <nickc@redhat.com>
14630 * config.gcc (tilegx-linux): Extend extra_objs rather than
14632 (tilepro-linux): Likewise.
14634 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
14636 * config/aarch64/aarch64-builtins.c
14638 (CF0, CF1, CF2, CF3, CF4, CF10): New.
14639 (VAR<1-12>): Add MAP parameter.
14640 (BUILTIN_*): Likewise.
14641 * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
14642 * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
14643 (aarch64_ushl_n<mode>): Likewise.
14644 (aarch64_sshr_n<mode>): Likewise.
14645 (aarch64_ushr_n<mode>): Likewise.
14646 (aarch64_<maxmin><mode>): Likewise.
14647 (aarch64_sqrt<mode>): Likewise.
14648 * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
14649 (vshr<q>_n_*): Likewise.
14651 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
14653 * config/aarch64/aarch64-builtins.c
14654 (aarch64_simd_builtin_type_mode): Handle SF types.
14656 (BUILTIN_GPF): Define.
14657 (aarch64_init_simd_builtins): Handle SF types.
14658 * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
14659 (frecps): Likewise.
14660 (frecpx): Likewise.
14661 * config/aarch64/aarch64-simd.md
14662 (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
14663 (aarch64_frecpe<mode>): New.
14664 (aarch64_frecps<mode>): Likewise.
14665 * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
14666 (v8type): Add frecp<esx>.
14667 (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
14668 (aarch64_frecps<mode>): Likewise.
14669 * config/aarch64/iterators.md (FRECP): New.
14670 (frecp_suffix): Likewise.
14671 * config/aarch64/arm_neon.h
14672 (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
14674 2013-04-22 Christian Bruel <christian.bruel@st.com>
14677 * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
14678 (REG_CLASS_NAMES): Idem.
14679 (REG_CLASS_CONTENTS): Idem.
14680 (REGCLASS_HAS_FP_REG): Idem.
14681 * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
14682 (sh_conditional_register_usage): Idem.
14684 2013-04-21 Jeff Law <law@redhat.com>
14686 * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
14687 (ssa_forward_propagate_and_combine): Use it.
14689 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
14691 * lra.c: Update the flow chart diagram.
14693 2013-04-19 Vladimir Makarov <vmakarov@redhat.com>
14695 PR rtl-optimization/56847
14696 * lra-constraints.c (process_alt_operands): Discourage alternative
14697 with non-matche doffsettable memory constraint fro memory with
14700 2013-04-19 Richard Biener <rguenther@suse.de>
14702 PR tree-optimization/56982
14703 * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
14705 * gimplify.c (gimplify_call_expr): Notice special calls.
14706 (gimplify_modify_expr): Likewise.
14707 * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
14708 abnormal control flow receivers.
14709 (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
14710 in the same way as cfun->has_nonlocal_labels.
14711 (gimple_purge_dead_abnormal_call_edges): Likewise.
14712 (stmt_starts_bb_p): Make setjmp-like abnormal control flow
14713 receivers start a basic-block.
14715 2013-04-19 Richard Biener <rguenther@suse.de>
14717 * tree-vectorizer.h (struct _slp_instance): Move load_permutation
14719 (struct _slp_tree): ... here. Make it a vector of unsigned ints.
14720 (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
14721 (SLP_TREE_LOAD_PERMUTATION): Add.
14722 (vect_transform_slp_perm_load): Adjust prototype.
14723 * tree-vect-slp.c (vect_free_slp_tree): Adjust.
14724 (vect_free_slp_instance): Likewise.
14725 (vect_create_new_slp_node): Likewise.
14726 (vect_supported_slp_permutation_p): Remove.
14727 (vect_slp_rearrange_stmts): Adjust.
14728 (vect_supported_load_permutation_p): Likewise. Inline
14729 vect_supported_slp_permutation_p here.
14730 (vect_analyze_slp_instance): Compute load permutations per
14731 slp node instead of per instance.
14732 (vect_get_slp_defs): Adjust.
14733 (vect_transform_slp_perm_load): Likewise.
14734 (vect_schedule_slp_instance): Remove redundant code.
14735 (vect_schedule_slp): Remove hack for PR56270, add it ...
14736 * tree-vect-stmts.c (vectorizable_load): ... here, do not
14737 CSE loads for SLP. Adjust.
14739 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
14741 * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
14742 spelling in two comments.
14744 2013-04-19 Greta Yorsh <Greta.Yorsh@arm.com>
14747 * config/arm/arm.c (load_multiple_sequence): Require SP
14748 as base register for loads if SP is in the register list.
14750 2013-04-19 Martin Jambor <mjambor@suse.cz>
14752 PR tree-optimization/56718
14753 * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
14754 * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
14755 and made public. Adjusted all callers.
14756 (ipa_intraprocedural_devirtualization): New function.
14757 * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
14758 (ipa_intraprocedural_devirtualization): Likewise.
14759 * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
14761 2013-04-19 Richard Biener <rguenther@suse.de>
14763 PR tree-optimization/57000
14764 * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
14766 2013-04-19 Terry Guo <terry.guo@arm.com>
14768 * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
14770 (cortex_m4_v_a, cortex_m4_v_b): ... new cpu units.
14771 (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
14772 (cortex_m4_fmacs): Use new reservations.
14773 (cortex_m4_f_load, cortex_m4_f_store): Likewise.
14775 2013-04-18 Vladimir Makarov <vmakarov@redhat.com>
14777 PR rtl-optimization/56999
14778 * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
14780 (lra_coalesce): Remove split_origin_bitmap and related code.
14781 * lra.c (lra): Coalesce after undoing inheritance. Recreate live
14782 ranges if necessary.
14784 2013-04-18 Uros Bizjak <ubizjak@gmail.com>
14786 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
14788 (ix86_expand_call): Remove clobbered_registers array and use
14789 x86_64_ms_sysv_extra_clobbered_registers instead.
14790 * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
14792 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
14794 * config/i386/i386.md (*call_rex64_ms_sysv): Use
14795 call_rex64_ms_sysv_operation predicate. Remove explicit clobbers.
14796 (*call_value_rex64_ms_sysv): Ditto.
14798 2013-04-18 Cary Coutant <ccoutant@google.com>
14800 * dwarf2out.c (output_pubnames): Check die_perennial_p of
14801 parent instead of die_mark.
14803 2013-04-18 Diego Novillo <dnovillo@google.com>
14805 * gimple.c (create_gimple_tmp): New.
14806 (get_expr_type): New.
14807 (build_assign): New.
14808 (build_type_cast): New.
14809 * gimple.h (enum ssa_mode): Define.
14810 (gimple_seq_set_location): New.
14811 * asan.c (build_check_stmt): Change some gimple_build_* calls
14812 to use build_assign and build_type_cast.
14814 2013-04-18 Richard Biener <rguenther@suse.de>
14816 * tree-vect-data-refs.c (vect_analyze_group_access): Properly
14817 handle negative step. Remove redundant checks.
14818 (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
14819 * tree-vect-stmts.c (vectorizable_load): Instead of asserting
14820 for negative step and grouped loads fail to vectorize.
14822 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
14824 * emit-rtl.c (reset_insn_used_flags): New function.
14825 (reset_all_used_flags): Use it.
14826 (verify_insn_sharing): New function.
14827 (verify_rtl_sharing): Fix verification for SEQUENCEs.
14829 2013-04-18 Jakub Jelinek <jakub@redhat.com>
14831 PR tree-optimization/56984
14832 * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
14833 and (x >> M) >= N don't register any assertion if N << M is the
14836 2013-04-18 Steven Bosscher <steven@gcc.gnu.org>
14838 * lower-subreg.c (resolve_simple_move): If called self-recursive,
14839 do not delete_insn insns that have not yet been emitted, only
14840 unlink them with remove_insn.
14841 * df-scan.c (df_insn_delete): Revert r197492.
14843 2013-04-17 Steven Bosscher <steven@gcc.gnu.org>
14845 * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
14846 * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
14848 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
14850 * config/arm/arm.md (movsicc_insn): Convert define_insn into
14851 define_insn_and_split.
14852 (and_scc,ior_scc,negscc): Likewise.
14853 (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
14855 2013-04-17 Greta Yorsh <Greta.Yorsh@arm.com>
14857 * config/arm/arm.c (use_return_insn): Return 0 for targets that
14858 can benefit from using a sequence of LDRD instructions in epilogue
14859 instead of a single LDM instruction.
14861 2013-04-17 Manuel López-Ibáñez <manu@gcc.gnu.org>
14864 * doc/extend.texi: Fix typo.
14866 2013-04-17 Richard Biener <rguenther@suse.de>
14868 * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
14869 (vect_build_slp_tree): ... here.
14870 (vect_build_slp_tree_1): Compute which stmts of the SLP group
14871 match. Remove special-casing of mismatched complex loads.
14872 (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
14873 re-try the match with swapped commutative operands.
14874 (vect_supported_load_permutation_p): Remove special-casing of
14875 mismatched complex loads.
14876 (vect_analyze_slp_instance): Adjust.
14878 2013-04-17 Richard Biener <rguenther@suse.de>
14880 PR rtl-optimization/56921
14881 * cfgloop.h (struct loop): Add simple_loop_desc member.
14882 (struct niter_desc): Mark with GTY(()).
14883 (simple_loop_desc): Do not use aux field but simple_loop_desc.
14884 * loop-iv.c (get_simple_loop_desc): Likewise.
14885 (free_simple_loop_desc): Likewise.
14888 2013-04-16 Richard Biener <rguenther@suse.de>
14890 PR rtl-optimization/56921
14891 * loop-init.c (pass_rtl_move_loop_invariants): Add
14892 TODO_do_not_ggc_collect to todo_flags_finish.
14893 (pass_rtl_unswitch): Same.
14894 (pass_rtl_unroll_and_peel_loops): Same.
14895 (pass_rtl_doloop): Same.
14897 2013-04-17 Eric Botcazou <ebotcazou@adacore.com>
14899 * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
14900 (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
14901 Use nonoverlapping_component_refs_of_decl_p to disambiguate component
14903 (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
14904 * tree-streamer.c (record_common_node): Adjust reference in comment.
14906 2013-04-17 Terry Guo <terry.guo@arm.com>
14908 * config/arm/cortex-m4.md: Add a new bypass.
14910 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14912 * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
14914 (*subs_<optab><mode>_multp2): New pattern.
14915 (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14916 (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
14918 2013-04-16 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
14920 * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
14921 (*subs_mul_imm_<mode>): New pattern.
14923 2013-04-16 David Edelsohn <dje.gcc@gmail.com>
14926 * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
14927 (vsx_movti_64bit): Change j->wa to O->wa. Add n->r alternative.
14928 (vsx_movti_32bit): Change j->wa to O->wa.
14930 2013-04-16 Richard Biener <rguenther@suse.de>
14932 PR rtl-optimization/56921
14933 * loop-init.c (pass_rtl_move_loop_invariants): Add
14934 TODO_do_not_ggc_collect to todo_flags_finish.
14935 (pass_rtl_unswitch): Same.
14936 (pass_rtl_unroll_and_peel_loops): Same.
14937 (pass_rtl_doloop): Same.
14939 2013-04-16 Greta Yorsh <Greta.Yorsh@arm.com>
14941 * config/arm/arm.c (emit_multi_reg_push): New declaration
14942 for an existing function.
14943 (arm_emit_strd_push): New function.
14944 (arm_expand_prologue): Used here.
14945 (arm_emit_ldrd_pop): New function.
14946 (arm_expand_epilogue): Used here.
14947 (arm_get_frame_offsets): Update condition.
14948 (arm_emit_multi_reg_pop): Add a special case for load of a single
14949 register with writeback.
14951 2013-04-16 Uros Bizjak <ubizjak@gmail.com>
14953 * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
14956 2013-04-16 Richard Biener <rguenther@suse.de>
14958 PR tree-optimization/56756
14959 * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
14960 (first_mem_ref_loc): New.
14961 (execute_sm): Place the load temporarily before a previous
14962 access instead of in the latch edge to ensure its SSA dependencies
14963 are defined at points dominating the load.
14965 2013-04-16 Steven Bosscher <steven@gcc.gnu.org>
14967 * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
14968 correct fix by moving header and footer insn to the footer of
14969 the merged basic block. Clear BB_END of the merged-away block.
14971 PR middle-end/43631
14972 * emit-rtl.c (make_note_raw): New function.
14973 (link_insn_into_chain): New static inline function.
14974 (add_insn): Use it.
14975 (add_insn_before, add_insn_after): Factor insn chain linking code...
14976 (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
14977 using link_insn_into_chain.
14978 (note_outside_basic_block_p): New helper function for emit_note_after
14979 and emit_note_before.
14980 (emit_note_after): Use nobb variant of add_insn_after if the note
14981 should not be contained in a basic block.
14982 (emit_note_before): Use nobb variant of add_insn_before if the note
14983 should not be contained in a basic block.
14984 (emit_note_copy): Use make_note_raw.
14985 (emit_note): Likewise.
14986 * bb-reorder.c (insert_section_boundary_note): Remove hack to set
14987 BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
14988 * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
14989 the moved barrier the tail of the basic block it follows.
14990 * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
14992 2013-04-15 Jakub Jelinek <jakub@redhat.com>
14994 PR tree-optimization/56962
14995 * gimple-ssa-strength-reduction.c (record_increment): Only set
14996 initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
14997 either rhs1 or rhs2 is equal to c->base_expr.
14999 2013-04-15 Richard Biener <rguenther@suse.de>
15001 PR tree-optimization/56933
15002 * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
15004 (GROUP_READ_WRITE_DEPENDENCE): Remove.
15005 (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
15006 * tree-vect-data-refs.c (vect_analyze_group_access): Move
15007 dependence check ...
15008 vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
15010 * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
15011 GROUP_READ_WRITE_DEPENDENCE.
15013 2013-04-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
15015 * emit-rtl.c (reset_all_used_flags): New function.
15016 (verify_rtl_sharing): Call reset_all_used_flags before and after
15017 performing the checks.
15019 2013-04-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15021 * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
15022 * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
15023 * config/arm/constraints.md (De): New constraint.
15024 * config/arm/neon.md (anddi3_neon): Delete.
15025 (neon_vand<mode>): Expand to standard anddi3 pattern.
15026 * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
15027 Move earlier in the file.
15028 (neon_inv_logic_op2): Likewise.
15029 (arm_anddi_operand_neon): New predicate.
15031 2013-04-15 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15033 * configure.ac (gcc_cv_ld_as_needed): Set
15034 gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
15035 Use -z ignore, -z record on *-*-solaris2*.
15036 (HAVE_LD_AS_NEEDED): Update comment.
15037 (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
15038 * configure: Regenerate.
15039 * config.in: Regenerate.
15040 * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
15041 LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
15042 * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
15043 * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
15044 equivalents. Fix markup.
15045 * doc/tm.texi: Regenerate.
15047 2013-04-15 Andrew Hsieh <andrewhsieh.google.com>
15049 * config/i386/i386.opt: New option mstack-protector-guard=.
15050 * config/i386/i386-opts.h: Add enum stack_protector_guard.
15051 * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
15052 TARGET_SSP_TLS_GUARD.
15053 * config/i386/i386.c (ix86_option_override_internal): Set
15054 ix86_stack_protector_guard.
15055 * config/i386/i386.md (stack_protect_set): Enable for
15056 TARGET_SSP_TLS_GUARD only.
15057 (stack_protect_set_<mode>): Ditto.
15058 (stack_protect_test): Ditto.
15059 (stack_protect_test_<mode>): Ditto.
15060 * doc/invoke.texi (i386 Option): Document.
15062 2013-04-15 Eric Botcazou <ebotcazou@adacore.com>
15065 * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
15066 (S_MODES): Set H_MODE bit.
15067 (SF_MODES): Set only S_MODE and SF_MODE bits.
15068 (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
15069 (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
15070 <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
15071 <MODE_FLOAT>: Likewise.
15073 2013-04-15 Joey Ye <joey.ye@arm.com>
15075 * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
15077 2013-04-15 Joey Ye <joey.ye@arm.com>
15079 * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
15081 (thumb_far_jump_used_p): Count instruction size and set
15084 2013-04-14 Eric Botcazou <ebotcazou@adacore.com>
15086 * reorg.c (fill_simple_delay_slots): Reindent block of code.
15087 * resource.c (mark_target_live_regs): Reformat conditional block.
15089 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
15091 * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
15092 notes, they are emitted only just before final.
15093 * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
15095 2013-04-13 Steven Bosscher <steven@gcc.gnu.org>
15097 * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
15098 * cfgrtl.c (delete_insn): Call it here instead.
15099 * lra-spills.c (lra_final_code_change): Use delete_insn.
15100 * haifa-sched.c (sched_remove_insn): Likewise.
15101 * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
15102 returning to the nop pool.
15103 (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
15104 use delete_insn for definitive removal. Clear BLOCK_FOR_INSN.
15106 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
15108 * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
15109 * doc/tm.texi: Regenerated.
15111 2013-04-12 Uros Bizjak <ubizjak@gmail.com>
15113 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
15116 2013-04-12 Steven Bosscher <steven@gcc.gnu.org>
15118 * df-core.c (df_find_def): Compare register numbers.
15119 (df_find_use): Likewise.
15121 2013-04-12 Vladimir Makarov <vmakarov@redhat.com>
15124 * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
15125 lra_in_progress for return.
15127 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
15129 * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
15130 define_insn into define_insn_and_split and emit movsicc patterns.
15132 2013-04-12 Greta Yorsh <Greta.Yorsh@arm.com>
15134 * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
15136 2013-04-12 Richard Biener <rguenther@suse.de>
15138 * tree-pass.h (TODO_do_not_ggc_collect): New.
15139 * passes.c (execute_one_ipa_transform_pass): Honor
15140 TODO_do_not_ggc_collect.
15141 (execute_one_pass): Likewise.
15144 2013-04-10 Richard Biener <rguenther@suse.de>
15146 * passes.c (init_optimization_passes): Remove reload pass.
15147 * ira.c (do_reload): Merge into ...
15149 (rest_of_handle_reload): Remove.
15150 (pass_reload): Likewise.
15151 * config/i386/i386.c (ix86_option_override): Refer to ira instead
15152 of reload for vzeroupper pass placement.
15154 2013-04-12 Jakub Jelinek <jakub@redhat.com>
15156 PR tree-optimization/56918
15157 PR tree-optimization/56920
15158 * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
15159 instead of op1 - op2. Pass 2 * TYPE_PRECISION (type) as second
15160 argument to rshift method. For 2 * HOST_BITS_PER_WIDE_INT precision
15161 use wide_mul_with_sign method.
15163 2013-04-12 Richard Biener <rguenther@suse.de>
15165 * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
15166 not be considered a gimple constant.
15168 2013-04-12 Marc Glisse <marc.glisse@inria.fr>
15170 * fold-const.c (const_binop): Handle vector shifts by a scalar.
15171 (fold_binary_loc): Call const_binop also for mixed vector-scalar
15174 2013-04-12 Manuel López-Ibáñez <manu@gcc.gnu.org>
15175 Jakub Jelinek <jakub@redhat.com>
15177 * opts.c: Include diagnostic-color.h.
15178 (common_handle_option): Handle OPT_fdiagnostics_color_.
15179 * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
15180 (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
15181 (diagnostic-color.o): New.
15182 * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
15183 (diagnostic_color_rule): New enum.
15184 * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
15185 * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
15186 the location string.
15187 * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
15188 either NULL, or color kind.
15189 * diagnostic-color.c: New file.
15190 * diagnostic-color.h: New file.
15191 * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
15193 * doc/invoke.texi (-fdiagnostics-color): Document.
15194 * pretty-print.h (pp_show_color): Define.
15195 (struct pretty_print_info): Add show_color field.
15196 * diagnostic.c: Include diagnostic-color.h.
15197 (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
15198 macros. Colorize error:, warning: etc. strings and also the location
15200 (diagnostic_show_locus): Colorize the caret line.
15201 * pretty-print.c: Include diagnostic-color.h.
15202 (pp_base_format): Handle %r and %R format specifiers. Colorize strings
15203 inside of %< %> quotes or quoted through q format modifier.
15205 2013-04-12 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
15207 * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
15209 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
15211 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
15212 code in CC_NZ mode.
15213 * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
15216 2013-04-11 Marek Polacek <polacek@redhat.com>
15218 PR tree-optimization/48184
15219 * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
15221 2013-04-11 Eric Botcazou <ebotcazou@adacore.com>
15223 * stor-layout.c (skip_simple_constant_arithmetic): Move to...
15224 * tree.c (skip_simple_constant_arithmetic): ...here and make public.
15225 (skip_simple_arithmetic): Tidy up.
15226 * tree.h (skip_simple_constant_arithmetic): Declare.
15228 2013-04-11 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
15230 * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
15232 2013-04-11 Richard Biener <rguenther@suse.de>
15234 * tree-vect-loop.c (get_initial_def_for_induction): Properly
15235 generate vector constants.
15237 2013-04-11 Richard Biener <rguenther@suse.de>
15239 PR tree-optimization/56878
15240 * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
15241 * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
15243 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15244 Prefer to align the DR with the most invariant base address.
15246 2013-04-11 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
15248 * opts.c (common_handle_option): Fix formatting and add FALLTHRU
15251 2013-04-11 James Greenhalgh <james.greenhalgh@arm.com>
15253 * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
15254 floating-point vector comparisons against 0.
15256 2013-04-11 Jakub Jelinek <jakub@redhat.com>
15258 PR tree-optimization/56899
15259 * fold-const.c (extract_muldiv_1): Apply distributive law
15260 only if TYPE_OVERFLOW_WRAPS (ctype).
15262 2013-04-11 Bin Cheng <bin.cheng@arm.com>
15265 * ira-costs.c (scan_one_insn): Check whether the source rtx of
15266 loading has side effect.
15268 2013-04-10 Steven Bosscher <steven@gcc.gnu.org>
15270 * config/sparc/sparc.c: Include tree-pass.h.
15271 (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
15272 (sparc_reorg): Rename to sparc_do_work_around_errata. Move to
15273 head of file. Change return type. Split off gate function.
15274 (sparc_gate_work_around_errata): New function.
15275 (pass_work_around_errata): New pass definition.
15276 (insert_pass_work_around_errata) New pass insert definition to
15277 insert pass_work_around_errata just after delayed-branch scheduling.
15278 (sparc_option_override): Insert the pass.
15279 * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
15281 2013-04-10 David S. Miller <davem@davemloft.net>
15283 * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
15284 or -mcpu=hypersparc.
15286 * target.def (cstore_mode): New hook.
15287 * target.h: Include insn-codes.h
15288 * targhooks.c: Likewise.
15289 (default_cstore_mode): New function.
15290 * targhooks.h: Declare it.
15291 * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
15292 * doc/tm.texi: Rebuild.
15293 * expmed.c (emit_cstore): Obtain cstore boolean result mode using
15294 target hook, rather than inspecting the insn_data.
15295 * config/sparc/sparc.c (sparc_cstore_mode): New function.
15296 (TARGET_CSTORE_MODE): Redefine.
15297 (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
15299 * config/sparc/predicates.md (cstore_result_operand): New special
15301 * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
15302 Use it for operand 0.
15303 (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
15304 (*snesi_special): Likewise.
15305 (*snesi_zero): Likewise.
15306 (*seqsi_zero): Likewise.
15307 (*sltu_insn): Likewise.
15308 (*sgeu_insn): Likewise.
15309 (*seqdi_special): Make operand 0 and comparison operation be of
15311 (*snedi_special): Likewise.
15312 (*snedi_special_vis3): Likewise.
15313 (*neg_snesi_zero): Rename to *neg_snesisi_zero.
15314 (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
15315 (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
15316 (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
15317 (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
15318 (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
15319 (*sltu_extend_sp64): Likewise.
15320 (*neg_sltu_insn): Rename to *neg_sltusi_insn.
15321 (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
15322 (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
15323 (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
15324 (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
15326 2013-04-10 Yufeng Zhang <yufeng.zhang@arm.com>
15328 * config/aarch64/aarch64.c (aarch64_print_extension): New function.
15329 (aarch64_start_file): Use the new function.
15331 2013-04-10 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
15332 Jason Merrill <jason@redhat.com>
15334 * common.opt: Add -gdwarf.
15335 * opts.c (common_handle_option): Handle it.
15336 * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
15338 2013-04-10 Richard Biener <rguenther@suse.de>
15340 * passes.c (execute_todo): Do not call ggc_collect conditional here.
15341 (execute_one_ipa_transform_pass): But unconditionally here.
15342 (execute_one_pass): And here.
15343 (init_optimization_passes): Remove reload pass.
15344 * tree-pass.h (TODO_ggc_collect): Remove.
15345 (pass_reload): Likewise.
15346 * ira.c (do_reload): Merge into ...
15348 (rest_of_handle_reload): Remove.
15349 (pass_reload): Likewise.
15350 * config/i386/i386.c (ix86_option_override): Refer to ira instead
15351 of reload for vzeroupper pass placement.
15352 * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
15353 and todo_flags_finish of all passes.
15355 2013-04-10 Richard Biener <rguenther@suse.de>
15357 * tree-vectorizer.h (struct _slp_oprnd_info): Remove
15358 first_const_oprnd field, rename first_def_type to first_op_type.
15359 * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
15360 (vect_get_and_check_slp_defs): Always use the type of the
15361 operand. Allow mixed vect_external_def, vect_constant_def types.
15362 (vect_get_constant_vectors): Handle mixed vect_external_def,
15363 vect_constant_def types.
15365 2013-04-10 Joern Rennecke <joern.rennecke@embecosm.com>
15367 PR tree-optimization/55524
15368 * tree-ssa-math-opts.c
15369 (convert_mult_to_fma): Don't use an fms construct
15370 when we don't have an fms operation, but fnma, and it looks
15371 likely that we'll be able to use the latter.
15373 2013-04-10 Zhouyi Zhou <yizhouzhou@ict.ac.cn>
15375 * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
15377 * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
15378 inline fail caused by overwritable functions.
15380 2013-04-10 Chung-Ju Wu <jasonwucj@gmail.com>
15382 * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
15383 unnecessary bits in the constant power of two case.
15385 2013-04-10 Richard Biener <rguenther@suse.de>
15387 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
15388 broken code swapping operands.
15389 (vect_build_slp_tree): Do not compute load permutations here.
15390 (vect_analyze_slp_instance): Compute load permutations here,
15391 after building the SLP tree.
15393 2013-04-09 Christian Bruel <christian.bruel@st.com>
15395 * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
15396 of next/prev_real_insn.
15398 2013-04-09 Jan Hubicka <jh@suse.cz>
15400 * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
15401 Drop aliased parameter.
15402 (function_and_variable_visibility): Do not handle alias pairs.
15403 * cgraph.c (varpool_externally_visible_p): Update prototype.
15404 * varpool.c (varpool_add_new_variable): Update.
15406 2013-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15408 * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
15410 2013-04-09 Steven Bosscher <steven@gcc.gnu.org>
15412 * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
15414 * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
15416 2013-04-09 Marek Polacek <polacek@redhat.com>
15418 PR tree-optimization/48762
15419 * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
15421 2013-04-09 Richard Biener <rguenther@suse.de>
15423 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
15425 (vect_build_slp_tree): Likewise.
15426 (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
15427 calculating the cost of a SLP instance.
15428 (vect_analyze_slp_instance): Use it from here, after building
15431 2013-04-09 Jakub Jelinek <jakub@redhat.com>
15433 PR middle-end/56883
15434 * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
15435 expand_omp_for_static_chunk): Use simple_p = true in
15436 force_gimple_operand_gsi calls when assigning to addressable decls.
15438 2013-04-09 Jeff Law <law@redhat.com>
15440 * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
15441 when the boolean was created by converting a wider object which
15442 had a boolean range.
15444 2013-04-09 Richard Biener <rguenther@suse.de>
15446 * tree-vectorizer.h (slp_void_p): Remove.
15447 (slp_tree): Typedef before _slp_tree declaration.
15448 (struct _slp_tree): Use a vector of slp_tree as children.
15449 (vect_get_place_in_interleaving_chain): Remove.
15450 * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
15452 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
15454 (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
15455 vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
15456 vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
15457 vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
15458 Use slp_node instead of slp_void_p and adjust.
15460 2013-04-09 Richard Biener <rguenther@suse.de>
15462 * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
15463 work that is not necessary.
15465 2013-04-09 Jakub Jelinek <jakub@redhat.com>
15467 PR tree-optimization/56854
15468 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
15469 forward into clobber stmts if it would change MEM_REF lhs into
15472 2013-04-09 Maxim Kuvyrkov <maxim@kugelworks.com>
15474 * tree.c (type_hash_lookup, type_hash_add): Make static.
15475 * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
15477 2013-04-09 Richard Biener <rguenther@suse.de>
15479 * tree.h (unsave_expr_now): Remove.
15480 * tree-inline.c (mark_local_for_remap_r): Remove.
15481 (unsave_expr_1): Likewise.
15482 (unsave_r): Likewise.
15483 (unsave_expr_now): Likewise.
15484 * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
15485 (propagate_tree_value): Likewise.
15487 2013-04-08 Steven Bosscher <steven@gcc.gnu.org>
15489 * doc/rtl.texi (sequence): Rewrite documentation to match the
15490 current use of SEQUENCE rtl objects.
15491 * rtl.def (SEQUENCE): Likewise.
15493 * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
15494 Update documentation.
15495 (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
15496 NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
15498 * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
15500 2013-04-08 Teresa Johnson <tejohnson@google.com>
15502 * basic-block.h (GCOV_COMPUTE_SCALE): Define.
15503 * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
15505 (estimate_edge_size_and_time): Add comment to suggest using rounding
15507 (estimate_node_size_and_time): Ditto.
15508 (remap_edge_change_prob): Use helper rounding divide methods.
15509 * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
15510 (gimple_mod_pow2_value_transform): Ditto.
15511 (gimple_mod_subtract_transform): Ditto.
15512 (gimple_ic_transform): Ditto.
15513 (gimple_stringops_transform): Ditto.
15514 * stmt.c (conditional_probability): Ditto.
15515 (emit_case_dispatch_table): Ditto.
15516 * lto-cgraph.c (merge_profile_summaries): Ditto.
15517 * tree-optimize.c (execute_fixup_cfg): Ditto.
15518 * cfgcleanup.c (try_forward_edges): Ditto.
15519 * cfgloopmanip.c (scale_loop_profile): Ditto.
15521 (duplicate_loop_to_header_edge): Ditto.
15522 (lv_adjust_loop_entry_edge): Ditto.
15523 * tree-vect-loop.c (vect_transform_loop): Ditto.
15524 * profile.c (compute_branch_probabilities): Ditto.
15525 * cfgbuild.c (compute_outgoing_frequencies): Ditto.
15526 * lto-streamer-in.c (input_cfg): Ditto.
15527 * gimple-streamer-in.c (input_bb): Ditto.
15528 * ipa-cp.c (update_profiling_info): Ditto.
15529 (update_specialized_profile): Ditto.
15530 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
15531 * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
15533 * sched-rgn.c (compute_dom_prob_ps): Ditto.
15534 (compute_trg_info): Ditto.
15535 * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
15536 (purge_dead_edges): Ditto.
15537 * loop-unswitch.c (unswitch_loop): Ditto.
15538 * cgraphclones.c (cgraph_clone_edge): Ditto.
15539 (cgraph_clone_node): Ditto.
15540 * tree-inline.c (copy_bb): Ditto.
15541 (copy_edges_for_bb): Ditto.
15542 (initialize_cfun): Ditto.
15543 (copy_cfg_body): Ditto.
15544 (expand_call_inline): Ditto.
15546 2013-04-08 Kai Tietz <ktietz@redhat.com>
15548 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
15549 TARGET_CYGWIN64 by TARGET_64BIT.
15551 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
15553 * config/epiphany/epiphany.md (GPR_1): New constant.
15554 (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
15555 * config/epiphany/epiphany.c (gen_compare_reg):
15556 For flag_finite_math_only, avoid swapping operands when r0 and/or r1
15557 is already in place.
15558 Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
15559 Don't require being called during rtl expansion; If y operlaps r0,
15561 (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
15562 (epiphany_expand_epilogue): Likewise.
15564 * config/epiphany/epiphany.c (epiphany_select_cc_mode):
15565 Don't use CC_FPmode for ORDERED / UNORDERED.
15566 * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
15568 * config/epiphany/constraints.md (CnL): New constraint.
15569 * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
15570 * config/epiphany/predicates.md (add_operand): Allow 1024.
15572 * config/epiphany/epiphany.md (logical_op): New code iterator.
15573 (op_mnc): New code attribute.
15574 (<op_mnc>_f, mov_f, cstoresi4): New patterns.
15575 (mov_f+1, mov_f+2): New peephole2 patterns.
15577 * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
15578 (cstoresi4): Also allow re-use of zero result when doing a NE
15579 comparison to a non-zero operand.
15580 Use (clobber (scratch)) for first insn if the gpr output is not needed.
15582 * config/epiphany/epiphany.md (<insn_opname>v2si3):
15583 Use gen_addsi3_i / gen_subsi3_i.
15585 2013-04-08 Jakub Jelinek <jakub@redhat.com>
15589 * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
15590 contain anything but clobbers, at most one __builtin_stack_restore,
15591 optionally debug stmts and final resx, and if it has at least one
15592 incoming EH edge. Don't check for SSA_NAME on LHS of a clobber.
15593 (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
15594 Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
15595 which isn't defaut definition, remove them.
15596 (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
15597 instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
15598 * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
15599 with MEM_REF LHS with SSA_NAME address.
15601 2013-04-08 Jeff Law <law@redhat.com>
15603 * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
15605 2013-04-08 Richard Biener <rguenther@suse.de>
15607 * gimple-pretty-print.c (debug_gimple_stmt): Do not print
15609 * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
15610 determined vector type.
15611 (vect_analyze_data_refs): Likewise.
15612 (vect_get_new_vect_var): Adjust.
15613 (vect_create_destination_var): Preserve SSA name versions.
15614 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
15615 not dump anything here.
15617 2013-04-08 Joern Rennecke <joern.rennecke@embecosm.com>
15619 * config/epiphany/epiphany.h (struct GTY (()) machine_function):
15620 Add member lr_slot_known.
15621 * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
15623 * config/epiphany/epiphany.c (epiphany_compute_frame_size):
15624 Remove code that sets lr_slot_offset according to what a previous
15625 version of epiphany_emit_save_restore used to do.
15626 (epiphany_emit_save_restore): When doing an lr save or restore,
15627 set/verify lr_slot_known and lr_slot_offset.
15629 2013-04-08 Xinyu Qi <xyqi@marvell.com>
15632 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
15635 2013-04-08 Richard Biener <rguenther@suse.de>
15637 * alias.c (find_base_term): Fix thinko in previous change.
15639 2013-04-08 Jakub Jelinek <jakub@redhat.com>
15641 * tree-loop-distribution.c (const_with_all_bytes_same): New function.
15642 (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
15643 TYPE_PRECISION is equal to mode bitsize. Use const_with_all_bytes_same
15644 if possible to compute val.
15645 (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
15646 For QImode integers don't require anything about precision. Use
15647 const_with_all_bytes_same to find out if the constant doesn't have
15648 repeated bytes in it.
15650 2013-04-08 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
15652 * config/s390/s390.c (s390_expand_insv): Only accept insertions
15655 2013-04-08 Marek Polacek <polacek@redhat.com>
15657 PR rtl-optimization/48182
15658 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
15661 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
15664 * config/pa/pa.c (legitimize_pic_address): Before incrementing label
15665 nuses, make sure we have a label.
15667 2013-04-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15670 * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
15671 (rs6000_emit_swdiv_low_precision): Remove.
15672 (rs6000_emit_swdiv): Rewrite to handle between one and four
15673 iterations of Newton-Raphson generally; modify required number of
15674 iterations for some cases.
15675 * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
15677 2013-04-05 Steven Bosscher <steven@gcc.gnu.org>
15679 * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
15680 set-but-unused variable.
15682 * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
15683 basic blocks of released function bodies garbage-collectable.
15685 * ree.c (find_and_remove_re): Do not call df_finish_pass here.
15686 (struct rtl_opt_pass): Add TODO_df_finish.
15688 * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
15690 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15692 * config/arm/constraints.md (q): New constraint.
15693 * config/arm/ldrdstrd.md: New file.
15694 * config/arm/arm.md (ldrdstrd.md) New include.
15695 (arm_movdi): Use "q" instead of "r" constraint
15696 for double-word memory access.
15697 (movdf_soft_insn): Likewise.
15698 * config/arm/vfp.md (movdi_vfp): Likewise.
15699 * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
15700 * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
15701 * config/arm/arm.c (gen_operands_ldrd_strd): New function.
15702 (mem_ok_for_ldrd_strd): Likewise.
15703 (output_move_double): Update assertion.
15705 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15707 * config/arm/arm.md: Comment on splitting Thumb1 patterns.
15709 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15711 * config/arm/arm.md (arm_smax_insn): Convert define_insn into
15712 define_insn_and_split.
15713 (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
15715 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15717 * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
15718 define_insn_and_split.
15719 (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
15720 (shiftsi3_compare): New pattern.
15721 (rrx): New pattern.
15722 * config/arm/unspecs.md (UNSPEC_RRX): New.
15724 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15726 * config/arm/arm.md (negdi_extendsidi): New pattern.
15727 (negdi_zero_extendsidi): Likewise.
15729 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15731 * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
15732 define_insn_and_split.
15733 (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
15734 (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
15736 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15738 * config/arm/arm.md (arm_subdi3): Convert define_insn into
15739 define_insn_and_split.
15740 (subdi_di_zesidi,subdi_di_sesidi): Likewise.
15741 (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
15743 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15745 * config/arm/arm.md (subsi3_carryin): New pattern.
15746 (subsi3_carryin_const): Likewise.
15747 (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
15748 (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
15750 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15752 * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
15754 2013-04-05 Greta Yorsh <Greta.Yorsh@arm.com>
15756 * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
15757 (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
15759 2013-04-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15761 * config/arm/arm.c (arm_expand_builtin): Change fcode
15762 type to unsigned int.
15764 2013-04-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15766 * doc/invoke.texi (ARM Options): Document cortex-a53 support.
15768 2013-04-04 Ian Lance Taylor <iant@google.com>
15770 * doc/standards.texi (Standards): The Go frontend supports the Go 1
15773 2013-04-04 Steven Bosscher <steven@gcc.gnu.org>
15775 PR middle-end/56729
15776 * df-scan.c (df_insn_delete): Disable failing assert.
15778 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15780 * config/arm/arm-protos.h (arm_builtin_vectorized_function):
15781 New function prototype.
15782 * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
15783 (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
15784 (arm_builtin_vectorized_function): New function.
15786 2013-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15788 * config/arm/arm_neon_builtins.def: New file.
15789 * config/arm/arm.c (neon_builtin_data): Move contents to
15790 arm_neon_builtins.def.
15791 (enum arm_builtins): Include neon builtin definitions.
15792 (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
15793 * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
15795 2013-04-04 Marek Polacek <polacek@redhat.com>
15797 PR tree-optimization/48186
15798 * predict.c (maybe_hot_frequency_p): Return false if
15799 HOT_BB_FREQUENCY_FRACTION is 0.
15800 (cgraph_maybe_hot_edge_p): Likewise.
15802 2013-04-04 Richard Biener <rguenther@suse.de>
15804 PR tree-optimization/56826
15805 * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
15808 2013-04-04 Richard Biener <rguenther@suse.de>
15810 PR tree-optimization/56213
15811 * tree-vect-data-refs.c (vect_check_strided_load): Remove.
15812 (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
15814 2013-04-04 Richard Biener <rguenther@suse.de>
15816 PR tree-optimization/56837
15817 * tree-loop-distribution.c (classify_partition): For non-zero
15818 values require that the value has the same precision as its
15819 mode to be useful as memset value.
15821 2013-04-03 Nick Clifton <nickc@redhat.com>
15823 * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
15824 (fmssf4): Use fmsf.s on E3V5 architectures.
15825 (fnmasf4): Use fnmaf.s on E3V5 architectures.
15826 (fnmssf4): Use fnmsf.s on E3V5 architectures.
15828 2013-04-03 Jeff Law <law@redhat.com>
15830 * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
15831 (lra-eliminations.o): Likewise.
15833 2013-04-03 Teresa Johnson <tejohnson@google.com>
15835 * gcov-io.c (compute_working_sets): Moved most of body of old
15836 compute_working_sets here from profile.c.
15837 * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
15838 (gcov_working_set_t): Moved typedef here from basic-block.h
15839 (compute_working_set): Declare.
15840 * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
15841 (get_working_sets): Renamed from compute_working_set,
15842 replace most of body with call to new compute_working_sets.
15843 (get_exec_counts): Replace call to compute_working_sets
15844 to get_working_sets.
15845 * profile.h (get_working_sets): Renamed from compute_working_set.
15846 * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
15847 to get_working_sets.
15848 * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
15849 * gcov-dump.c (dump_working_sets): New function.
15851 2013-04-03 Kenneth Zadeck <zadeck@naturalbridge.com>
15853 * hwint.c (sext_hwi, zext_hwi): New functions.
15854 * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
15855 HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
15856 HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
15857 HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
15858 HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
15859 (sext_hwi, zext_hwi): New functions.
15861 2013-04-03 Jeff Law <law@redhat.com>
15863 PR tree-optimization/56799
15864 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
15865 back test for widening conversion erroneously dropped in prior change.
15867 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15870 * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
15871 instead of next_real_insn.
15873 2013-04-03 Marek Polacek <polacek@redhat.com>
15876 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
15878 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15881 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
15883 (thumb1_output_casesi): Likewise.
15884 (thumb2_output_casesi): Likewise.
15886 2013-04-03 Richard Biener <rguenther@suse.de>
15888 PR tree-optimization/56817
15889 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
15891 (tree_unroll_loops_completely_1): ... new function to manually
15892 walk the loop tree, properly defering outer loops of unrolled
15893 loops to later iterations.
15895 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
15897 * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
15898 (vectorizable_load): Likewise.
15899 * tree-vect-slp.c (vect_build_slp_tree): Likewise.
15900 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
15902 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
15904 * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
15907 2013-04-03 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
15909 * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
15911 2013-04-03 Bin Cheng <bin.cheng@arm.com>
15913 * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
15915 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
15917 PR tree-optimization/56790
15918 * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
15921 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
15923 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
15925 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
15926 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
15929 2013-04-03 Jakub Jelinek <jakub@redhat.com>
15932 * tree.h (force_folding_builtin_constant_p): New decl.
15933 * builtins.c (force_folding_builtin_constant_p): New variable.
15934 (fold_builtin_constant_p): Fold immediately also if
15935 force_folding_builtin_constant_p.
15937 2013-04-03 Richard Biener <rguenther@suse.de>
15939 PR tree-optimization/56812
15940 * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
15941 DRs of the same interleaving chain are independent.
15943 2013-04-02 Jason Merrill <jason@redhat.com>
15945 * gdbinit.in (pbb): Use debug fn.
15947 2013-04-02 Lawrence Crowl <crowl@google.com>
15949 * sese.h (struct ivtype_map_elt_s): Remove unused.
15950 (extern debug_ivtype_map): Remove unused.
15951 (extern eq_ivtype_map_elts): Remove unused.
15952 * sese.c (debug_ivtype_map): Removed unused.
15953 (debug_ivtype_map_1): Removed unused.
15954 (debug_ivtype_elt): Remove unused.
15955 (eq_ivtype_map_elts): Remove unused.
15958 2013-04-02 Kai Tietz <ktietz@redhat.com>
15961 * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
15962 * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
15963 * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
15965 (legitimize_pe_coff_symbol): Likewise.
15966 (is_imported_p): New helper-function.
15967 (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
15968 for Windows x64 targets.
15969 (ix86_expand_prologue): Optimize for pe-coff targets.
15970 (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
15971 (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
15972 medium/large code-model.
15973 (legitimize_pic_address): Likewise.
15974 (legitimize_tls_address): Likewise.
15975 (ix86_expand_call): Likewise.
15976 (x86_output_mi_thunk): Likewise.
15977 (get_dllimport_decl): Add new beimport argument.
15978 (construct_plt_address): Don't assert for x64 pe-coff targets.
15979 * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
15981 (SYMBOL_FLAG_STUBVAR): New macro.
15982 (SYMBOL_REF_STUBVAR_P): Likewise.
15983 * config/i386/winnt.c (stub_list): New structure.
15984 (stub_head): New local variable.
15985 (i386_pe_record_stub): New function.
15986 (i386_pe_file_end): Emit refptr-stubs.
15988 2013-04-02 Jakub Jelinek <jakub@redhat.com>
15990 PR rtl-optimization/56745
15991 * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
15992 if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
15995 * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
15996 and both of them are MEM_REFs, just compare first argument for
15997 equality and attempt to deal even with differing offsets.
16000 * tree-cfg.c (verify_gimple_assign_single): Allow lhs
16001 of gimple_clobber_p to be MEM_REF.
16002 * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
16003 an assignment from TREE_CLOBBER_P. Allow it to be MEM_REF
16004 after gimplification.
16005 * asan.c (get_mem_ref_of_assignment): Don't instrument
16006 gimple_clobber_p stmts.
16007 * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
16008 gimple_clobber_p stmt if they have MEM_REF lhs and
16009 are dead because of another gimple_clobber_p stmt.
16010 * tree-ssa-live.c (clear_unused_block_pointer): Treat
16011 gimple_clobber_p stmts like debug stmts.
16012 (remove_unused_locals): Remove clobbers with MEM_REF lhs
16013 that refer to unused VAR_DECLs or uninitialized values.
16014 * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
16015 gimple_clobber_p stmts if they refer to removed parameters.
16016 (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
16019 2013-04-02 Uros Bizjak <ubizjak@gmail.com>
16021 * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
16022 using SWI48 mode attribute.
16024 2013-04-02 Wei Mi <wmi@google.com>
16026 * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
16027 ashl<mode>3_mask, *<shift_insn><mode>3_mask and
16028 *<rotate_insn><mode>3_mask in i386.md.
16030 2013-04-02 Alexander Ivchenko <alexander.ivchenko@intel.com>
16032 * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
16034 2013-04-02 Richard Biener <rguenther@suse.de>
16036 PR tree-optimization/56778
16037 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
16038 Runtime alias tests are not supported for gather loads.
16039 * tree-vect-loop-manip.c (vect_loop_versioning): Insert
16040 stmts referenced from SSA operands before updating SSA form.
16042 2013-04-02 Ian Caulfield <ian.caulfield@arm.com>
16043 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16045 * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
16046 * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
16047 * config/arm/cortex-a53.md: New file.
16048 * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
16049 * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
16050 * config/arm/arm.c (arm_issue_rate): Likewise.
16051 * config/arm/arm-tune.md: Regenerate
16052 * config/arm/arm-tables.opt: Regenerate.
16053 * config/arm/arm-cores.def: Add cortex-a53.
16055 2013-04-02 Zhenqiang Chen <zhenqiang.chen@arm.com>
16057 * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
16060 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
16062 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
16063 scalar load/store operations using B/H registers.
16064 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
16066 2013-04-02 Sofiane Naci <sofiane.naci@arm.com>
16068 * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
16070 * config/aarch64/aarch64.c
16071 (aarch64_simd_scalar_immediate_valid_for_move): New.
16072 * config/aarch64/aarch64-protos.h
16073 (aarch64_simd_scalar_immediate_valid_for_move): New.
16074 * config/aarch64/constraints.md (Dh, Dq): New.
16075 * config/aarch64/iterators.md (hq): New.
16077 2013-04-02 Eric Botcazou <ebotcazou@adacore.com>
16079 * reorg.c (get_branch_condition): Deal with conditional returns.
16080 (fill_simple_delay_slots): Remove dead code dealing with jumps.
16082 2013-04-01 Wei Mi <wmi@google.com>
16084 * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
16085 Truncate operand 2 using %b asm operand modifier.
16086 (*<shift_insn><mode>3_mask): Ditto.
16087 (*<rotate_insn><mode>3_mask): Ditto.
16089 2013-04-01 Steven Bosscher <steven@gcc.gnu.org>
16091 PR middle-end/56798
16092 * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
16094 2013-03-31 Kaz Kojima <kkojima@gcc.gnu.org>
16096 * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
16098 (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
16100 2013-03-30 Lawrence Crowl <crowl@google.com>
16102 * dse.c (clear_alias_sets): Remove never set.
16103 (disqualified_clear_alias_sets): Remove never set.
16104 (clear_alias_mode_pool): Remove never set.
16105 (dse_step0): Remove condition that is never true.
16106 (canon_address): Remove condition that is never true.
16107 (dse_step7): Remove condition that is never true.
16108 (rest_of_handle_dse): Remove condition that is never true.
16109 (rest_of_handle_dse::did_global): Remove never read from above.
16110 (dse_step2_spill): Remove never called from above.
16111 (dse_step5_spill): Remove never called from above.
16113 2013-03-30 Steven Bosscher <steven@gcc.gnu.org>
16115 * doc/md.texi (Standard Names) <casesi>: Update documentation for
16116 JUMP_TABLE_DATA changes.
16117 * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
16118 * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
16119 (Insns) <jump_table_data>: New entry.
16120 * doc/tm.texi: Regenerate.
16122 * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
16124 * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
16125 for table jump at the end of a basic block using tablejump_p.
16126 * targhooks.c (default_invalid_within_doloop): Likewise.
16127 * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
16128 target hook implementation that is identical to the default hook.
16129 (rs6000_invalid_within_doloop): Remove.
16131 * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
16132 unused variable from tablejump_p call.
16134 * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
16135 * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
16136 (INSN_DELETED_P): Likewise.
16137 (emit_jump_table_data): New prototype.
16138 * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
16139 after 4th as unused.
16140 * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
16141 * sched-vis.c (print_insn): Likewise.
16142 * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
16143 insn for compatibility with back ends that use next_active_insn to
16144 identify jump table data.
16145 (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
16146 (remove_insn): Likewise.
16147 (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
16149 (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
16150 (emit_jump_table_data): New function.
16152 * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
16153 basic block, a JUMP_TABLE_DATA never is.
16154 (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
16155 * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
16156 off from code handling real insns.
16157 * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
16158 * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
16159 test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
16160 * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
16161 is not a NONDEBUG_INSN_P.
16162 * ira-costs.c (scan_one_insn): Likewise.
16163 * jump.c (mark_all_labels): Likewise.
16164 (mark_jump_label_1): Likewise.
16165 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
16166 * lra.c (get_insn_freq): Expect all insns reaching here to be in
16168 (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
16169 * predict.c (expensive_function_p): Use FOR_BB_INSNS.
16170 * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
16171 JUMP_TABLE_DATA_P insns.
16172 (calculate_elim_costs_all_insns): Likewise.
16173 (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
16174 (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
16175 (delete_output_reload): Code style fixups.
16176 * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
16177 insn flags on this non-insn.
16178 * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
16179 as scheduling barriers, for pre-change compatibility.
16180 * stmt.c (emit_case_dispatch_table): Emit jump table data not as
16181 JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
16183 * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
16184 redundant JUMP_TABLE_DATA_P test.
16185 * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
16186 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
16187 (frv_for_each_packet): Likewise.
16188 * config/i386/i386.c (min_insn_size): Likewise.
16189 (ix86_avoid_jump_mispredicts): Likewise.
16190 * config/m32r/m32r.c (m32r_is_insn): Likewise.
16191 * config/mep/mep.c (mep_reorg_erepeat): Likewise.
16192 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
16193 (mips16_insn_length): Robustify.
16194 (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
16195 (mips16_split_long_branches): Likewise.
16196 * config/pa/pa.c (pa_combine_instructions): Likewise.
16197 * config/rs6000/rs6000.c (get_next_active_insn): Treat
16198 JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
16199 * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
16200 as contributing to pool range lengths.
16201 * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
16202 Remove redundant JUMP_TABLE_DATA_P test.
16203 (sh_loop_align): Likewise.
16204 (split_branches): Likewise.
16205 (sh_insn_length_adjustment): Likewise.
16206 * config/spu/spu.c (get_branch_target): Likewise.
16208 2013-03-29 Jan Hubicka <jh@suse.cz>
16210 * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
16211 gcov streaming; stream hot bb threshold to ltrans.
16212 * predict.c (get_hot_bb_threshold): Break out from ....
16213 (maybe_hot_count_p): ... here.
16214 (set_hot_bb_threshold): New function.
16215 * lto-section-in.c (lto_section_name): Add profile.
16216 * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
16217 * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
16218 and data-streamer.h
16219 (histogram_entry): New structure.
16220 (histogram, histogram_pool): New global vars.
16221 (histogram_hash): New structure.
16222 (histogram_hash::hash): New method.
16223 (histogram_hash::equal): Likewise.
16224 (account_time_size): New function.
16225 (cmp_counts): New function.
16226 (dump_histogram): New function.
16227 (ipa_profile_generate_summary): New function.
16228 (ipa_profile_write_summary): New function.
16229 (ipa_profile_read_summary): New function.
16230 (ipa_profile): Decide on threshold.
16231 (pass_ipa_profile): Add ipa_profile_write_summary and
16232 ipa_profile_read_summary.
16233 * Makefile.in (ipa.o): Update dependencies.
16234 * lto-streamer.h (LTO_section_ipa_profile): New section.
16236 2013-03-29 Gabriel Dos Reis <gdr@integrable-solutions.net>
16238 * tree.h (VAR_P): New.
16240 2013-03-29 Paolo Carlini <paolo.carlini@oracle.com>
16243 * doc/invoke.texi ([-fwhole-program]): Fix typo.
16245 2013-03-29 Steven Bosscher <steven@gcc.gnu.org>
16247 * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
16248 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
16249 (control_flow_insn_p): Likewise.
16250 * cfgrtl.c (duplicate_insn_chain): Likewise.
16251 * final.c (get_attr_length_1): Likewise.
16252 (shorten_branches): Likewise.
16253 (final_scan_insn): Likewise.
16254 * function.c (instantiate_virtual_regs): Likewise.
16255 * gcse.c (insert_insn_end_basic_block): Likewise.
16256 * ira-costs.c (scan_one_insn): Likewise.
16257 * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
16258 * lra.c (check_rtl): Likewise.
16259 * reload1.c (elimination_costs_in_insn): Likewise.
16260 * reorg.c (follow_jumps): Likewise.
16262 * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
16263 of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
16264 (thumb_far_jump_used_p): Likewise.
16265 * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
16266 (workaround_speculation): Likewise.
16267 (add_sched_insns_for_speculation): Likewise.
16268 * config/c6x/c6x.c (reorg_emit_nops): Likewise.
16269 * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
16270 (frv_for_each_packet): Likewise.
16271 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
16272 * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
16273 (final_emit_insn_group_barriers): Likewise.
16274 * config/m32r/m32r.c (m32r_is_insn): Likewise.
16275 * config/mips/mips.c (USEFUL_INSN_P): Likewise.
16276 (mips16_insn_length): Likewise.
16277 * config/pa/pa.c (pa_reorg): Likewise.
16278 (pa_combine_instructions): Likewise.
16279 * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
16280 * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
16281 (sh_reorg): Likewise.
16282 (split_branches): Likewise.
16283 * config/spu/spu.c (get_branch_target): Likewise.
16285 * config/s390/s390.c (s390_chunkify_start): Simplify logic using
16288 2013-03-29 Kirill Yukhin <kirill.yukhin@intel.com>
16290 * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
16291 Fix declaration name.
16293 2013-03-28 Lawrence Crowl <crowl@google.com>
16295 * graphds.h (struct graph.indicies): Remove unused.
16296 * graphite-poly.h (struct graph.original_pddrs): Remove unused.
16297 (SCOP_ORIGINAL_PDDRS): Remove unused.
16298 * sese.h (extern insert_loop_close_phis): Removed unused.
16299 (extern insert_guard_phis): Removed unused.
16300 (extern ivtype_map_elt_info): Removed unused.
16301 (new_ivtype_map_elt): Removed unused.
16302 * sese.c (ivtype_map_elt_info): Removed unused.
16304 2013-03-28 Lawrence Crowl <crowl@google.com>
16306 * Makefile.in: Add several missing include dependences.
16308 (test-dump.o): New. This object is not added to any executable,
16309 but is present for ad-hoc testing.
16311 (debug (const bitmap_head_def &)): New.
16312 (debug (const bitmap_head_def *)): New.
16314 (extern debug (const bitmap_head_def &)): New.
16315 (extern debug (const bitmap_head_def *)): New.
16317 (debug (edge_def &)): New.
16318 (debug (edge_def *)): New.
16320 (debug (basic_block_def &)): New.
16321 (debug (basic_block_def *)): New.
16323 (dump_node (const_tree, int, FILE *)): Correct source file.
16325 (debug (die_struct &)): New.
16326 (debug (die_struct *)): New.
16328 (extern debug (die_struct &)): New.
16329 (extern debug (die_struct *)): New.
16330 * gimple-pretty-print.c
16331 (debug (gimple_statement_d &)): New.
16332 (debug (gimple_statement_d *)): New.
16333 * gimple-pretty-print.h
16334 (extern debug (gimple_statement_d &)): New.
16335 (extern debug (gimple_statement_d *)): New.
16337 (debug (ira_allocno_copy &)): New.
16338 (debug (ira_allocno_copy *)): New.
16339 (debug (ira_allocno &)): New.
16340 (debug (ira_allocno *)): New.
16342 (extern debug (ira_allocno_copy &)): New.
16343 (extern debug (ira_allocno_copy *)): New.
16344 (extern debug (ira_allocno &)): New.
16345 (extern debug (ira_allocno *)): New.
16347 (debug (live_range &)): New.
16348 (debug (live_range *)): New.
16350 (debug (lra_live_range &)): New.
16351 (debug (lra_live_range *)): New.
16353 (debug (lra_live_range &)): New.
16354 (debug (lra_live_range *)): New.
16356 (debug (omega_pb_d &)): New.
16357 (debug (omega_pb_d *)): New.
16359 (extern debug (omega_pb_d &)): New.
16360 (extern debug (omega_pb_d *)): New.
16362 (debug (const rtx_def &)): New.
16363 (debug (const rtx_def *)): New.
16365 (debug_tree (tree): Move within file.
16366 (debug_raw (const tree_node &)): New.
16367 (debug_raw (const tree_node *)): New.
16368 (dump_tree_via_hooks (const tree_node *, int)): New.
16369 (debug (const tree_node &)): New.
16370 (debug (const tree_node *)): New.
16371 (debug_verbose (const tree_node &)): New.
16372 (debug_verbose (const tree_node *)): New.
16373 (debug_head (const tree_node &)): New.
16374 (debug_head (const tree_node *)): New.
16375 (debug_body (const tree_node &)): New.
16376 (debug_body (const tree_node *)): New.
16377 (debug_vec_tree (tree): Move and reimplement in terms of dump.
16378 (debug (vec<tree, va_gc> &)): New.
16379 (debug (vec<tree, va_gc> *)): New.
16381 (extern debug (const rtx_def &)): New.
16382 (extern debug (const rtx_def *)): New.
16384 (debug_raw (simple_bitmap_def &)): New.
16385 (debug_raw (simple_bitmap_def *)): New.
16386 (debug (simple_bitmap_def &)): New.
16387 (debug (simple_bitmap_def *)): New.
16389 (extern debug (simple_bitmap_def &)): New.
16390 (extern debug (simple_bitmap_def *)): New.
16391 (extern debug_raw (simple_bitmap_def &)): New.
16392 (extern debug_raw (simple_bitmap_def *)): New.
16394 (debug (vinsn_def &)): New.
16395 (debug (vinsn_def *)): New.
16396 (debug_verbose (vinsn_def &)): New.
16397 (debug_verbose (vinsn_def *)): New.
16398 (debug (expr_def &)): New.
16399 (debug (expr_def *)): New.
16400 (debug_verbose (expr_def &)): New.
16401 (debug_verbose (expr_def *)): New.
16402 (debug (vec<rtx> &)): New.
16403 (debug (vec<rtx> *)): New.
16405 (extern debug (vinsn_def &)): New.
16406 (extern debug (vinsn_def *)): New.
16407 (extern debug_verbose (vinsn_def &)): New.
16408 (extern debug_verbose (vinsn_def *)): New.
16409 (extern debug (expr_def &)): New.
16410 (extern debug (expr_def *)): New.
16411 (extern debug_verbose (expr_def &)): New.
16412 (extern debug_verbose (expr_def *)): New.
16413 (extern debug (vec<rtx> &)): New.
16414 (extern debug (vec<rtx> *)): New.
16416 (_list_iter_cond_expr): Make inline instead of static.
16418 (debug (sreal &)): New.
16419 (debug (sreal *)): New.
16421 (extern debug (sreal &)): New.
16422 (extern debug (sreal *)): New.
16424 (extern debug_raw (const tree_node &)): New.
16425 (extern debug_raw (const tree_node *)): New.
16426 (extern debug (const tree_node &)): New.
16427 (extern debug (const tree_node *)): New.
16428 (extern debug_verbose (const tree_node &)): New.
16429 (extern debug_verbose (const tree_node *)): New.
16430 (extern debug_head (const tree_node &)): New.
16431 (extern debug_head (const tree_node *)): New.
16432 (extern debug_body (const tree_node &)): New.
16433 (extern debug_body (const tree_node *)): New.
16434 (extern debug (vec<tree, va_gc> &)): New.
16435 (extern debug (vec<tree, va_gc> *)): New.
16437 (debug (struct loop &)): New.
16438 (debug (struct loop *)): New.
16439 (debug_verbose (struct loop &)): New.
16440 (debug_verbose (struct loop *)): New.
16441 * tree-dump.c: Add header dependence.
16443 (extern debug (struct loop &)): New.
16444 (extern debug (struct loop *)): New.
16445 (extern debug_verbose (struct loop &)): New.
16446 (extern debug_verbose (struct loop *)): New.
16448 (debug (data_reference &)): New.
16449 (debug (data_reference *)): New.
16450 (debug (vec<data_reference_p> &)): New.
16451 (debug (vec<data_reference_p> *)): New.
16452 (debug (vec<ddr_p> &)): New.
16453 (debug (vec<ddr_p> *)): New.
16455 (extern debug (data_reference &)): New.
16456 (extern debug (data_reference *)): New.
16457 (extern debug (vec<data_reference_p> &)): New.
16458 (extern debug (vec<data_reference_p> *)): New.
16459 (extern debug (vec<ddr_p> &)): New.
16460 (extern debug (vec<ddr_p> *)): New.
16462 (debug (pt_solution &)): New.
16463 (debug (pt_solution *)): New.
16465 (extern debug (pt_solution &)): New.
16466 (extern debug (pt_solution *)): New.
16468 (debug (_var_map &)): New.
16469 (debug (_var_map *)): New.
16470 (debug (tree_live_info_d &)): New.
16471 (debug (tree_live_info_d *)): New.
16473 (extern debug (_var_map &)): New.
16474 (extern debug (_var_map *)): New.
16475 (extern debug (tree_live_info_d &)): New.
16476 (extern debug (tree_live_info_d *)): New.
16478 2013-03-28 Jan Hubicka <jh@suse.cz>
16480 * lto-cgraph.c (merge_profile_summaries): Fix overflows.
16482 2013-03-28 Ian Bolton <ian.bolton@arm.com>
16484 * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
16485 record only when desired or required.
16487 2013-03-28 Uros Bizjak <ubizjak@gmail.com>
16489 * config/i386/i386.md (*vec_extract2vdi_1): Merge with
16490 *vec_extractv2di_1_rex64. Use x64 isa attribute.
16492 2013-03-28 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
16494 * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
16495 (*andsi3_compare0_uxtw): New pattern.
16496 (*and_<SHIFT:optab><mode>3_compare0): New pattern.
16497 (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
16499 2013-03-28 Jan Hubicka <jh@suse.cz>
16501 * data-streamer-in.c (streamer_read_gcov_count): New function.
16502 * gimple-streamer-out.c: Include value-prof.h.
16503 (output_gimple_stmt): Output histogram.
16504 (output_bb): Use streamer_write_gcov_count.
16505 * value-prof.c: Include data-streamer.h
16506 (dump_histogram_value): Add HIST_TYPE_MAX.
16507 (stream_out_histogram_value): New function.
16508 (stream_in_histogram_value): New function.
16509 * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
16510 (stream_out_histogram_value, stream_in_histogram_value): Declare.
16511 * data-streamer-out.c (streamer_write_gcov_count): New function.
16512 (streamer_write_gcov_count_stream): New function.
16513 * lto-cgraph.c (lto_output_edge): Update counter streaming.
16514 (lto_output_node): Likewise.
16515 (input_node, input_edge): Likewise.
16516 * lto-streamer-out.c (output_cfg): Update streaming.
16517 * lto-streamer-in.c (input_cfg): Likewise.
16518 * data-streamer.h (streamer_write_gcov_count,
16519 streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
16520 * gimple-streamer-in.c: Include value-prof.h
16521 (input_gimple_stmt): Input histograms.
16522 (input_bb): Update profile streaming.
16524 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
16526 * genmodes.c (emit_max_int): New function.
16527 (emit_insn_modes_h): Added call to emit_max_function.
16528 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
16530 * machmode.def: Fixed comment.
16532 2013-03-28 Kenneth Zadeck <zadeck@naturalbridge.com>
16534 * combine.c (try_combine): Removed useless assert.
16535 * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
16537 2013-03-28 Marek Polacek <polacek@redhat.com>
16538 Richard Biener <rguenther@suse.de>
16540 PR tree-optimization/56695
16541 * tree-vect-stmts.c (vectorizable_condition): Unconditionally
16542 build signed result of a vector comparison.
16543 * tree-cfg.c (verify_gimple_comparison): Check that a result
16544 of a vector comparison has signed type.
16546 2013-03-28 Richard Biener <rguenther@suse.de>
16548 PR tree-optimization/37021
16549 * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
16550 do not restrict gaps between groups.
16551 * tree-vect-stmts.c (vectorizable_load): Properly account for
16552 a gap between groups.
16554 2013-03-28 Eric Botcazou <ebotcazou@adacore.com>
16556 * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
16557 general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
16560 2013-03-27 Gerald Pfeifer <gerald@pfeifer.com>
16562 * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
16563 * doc/extend.texi (Named Address Spaces): Ditto.
16564 (Variable Attributes): Ditto.
16566 2013-03-27 Kai Tietz <ktietz@redhat.com>
16568 * config.build: Add support for cygwin x64 target.
16569 * config.gcc: Likewise.
16570 * config.host: Likewise.
16571 * configure.ac: Likewise
16572 * configure: Regenerated.
16574 2013-03-27 Kai Tietz <ktietz@redhat.com>
16576 * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
16577 * config/i386/t-cygwin-w64: New file.
16578 * config/i386/cygwin-w64.h: New file.
16579 * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
16580 and add support for x64-cygwin target.
16581 (CPP_SPEC): Likewise.
16582 (CXX_WRAP_SPEC_LIST): Undefine before define.
16583 (LIBGCJ_SONAME): Use 15 as version.
16585 2013-03-27 Richard Biener <rguenther@suse.de>
16587 PR tree-optimization/56716
16588 * tree-ssa-structalias.c (perform_var_substitution): Adjust
16589 dumping for ref nodes.
16591 2013-03-27 Martin Jambor <mjambor@suse.cz>
16593 PR tree-optimization/55334
16594 * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
16595 restricted pointers to arrays.
16597 2013-03-27 Gabriel Dos Reis <gdr@integrable-solutions.net>
16599 * Makefile.in (.SUFFIXES): Add .cc.
16600 (.c.o): Apply same recipe for implicit rule .cc.o.
16602 2013-03-27 Richard Biener <rguenther@suse.de>
16604 PR tree-optimization/37021
16605 * tree-vect-data-refs.c (vect_check_strided_load): Allow
16606 REALPART/IMAGPART_EXPRs around the supported refs.
16607 * tree-ssa-structalias.c (find_func_aliases): Assume that
16608 floating-point values are not used to transfer pointers.
16610 2013-03-27 Alexander Ivchenko <alexander.ivchenko@intel.com>
16612 * target.def (TARGET_HAS_IFUNC_P): New target hook.
16613 * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
16614 * doc/tm.texi: Regenerate.
16615 * targhooks.h (default_has_ifunc_p): New.
16616 * targhooks.c (default_has_ifunc_p): Ditto.
16617 * config/linux-protos.h: New file.
16618 * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
16619 hook for linux which disables support of indirect functions in android.
16620 * config/linux-android.c: New file.
16621 * config/t-linux-android.c: Ditto.
16622 * config.gcc: Added new object file linux-android.o.
16623 * config/i386/i386.c (ix86_get_function_versions_dispatcher):
16624 Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
16625 * varasm.c (do_assemble_alias): Likewise.
16626 * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
16627 doesn't support indirect functions.
16628 * configure: Regenerate.
16630 2013-03-27 Bin Cheng <bin.cheng@arm.com>
16633 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
16634 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
16637 2013-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
16639 * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
16641 2013-03-27 Terry Guo <terry.guo@arm.com>
16643 * config/arm/arm-cores.def: Added core cortex-r7.
16644 * config/arm/arm-tune.md: Regenerated.
16645 * config/arm/arm-tables.opt: Regenerated.
16646 * doc/invoke.texi: Added entry for core cortex-r7.
16648 2013-03-27 Walter Lee <walt@tilera.com>
16650 * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
16651 double-decrement of next_scratch_regno.
16653 2013-03-27 Walter Lee <walt@tilera.com>
16655 * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
16657 (insn_v1mulus): Ditto.
16658 (insn_v2muls): Ditto.
16660 2013-03-27 Walter Lee <walt@tilera.com>
16662 * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
16663 (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
16665 2013-03-27 Walter Lee <walt@tilera.com>
16667 * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
16668 (*sibcall_value): Ditto.
16670 2013-03-27 Walter Lee <walt@tilera.com>
16672 * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
16673 (insn_mnz_v8qi): ... this ...
16674 (insn_mnz_v4hi): ... and this. Replace (const_int 0) with the
16676 (insn_v<n>mnz): Replaced by ...
16677 (insn_v1mnz): ... this ...
16678 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
16680 (insn_mz_<mode>): Replaced by ...
16681 (insn_mz_v8qi): ... this ...
16682 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
16684 (insn_v<n>mz): Replaced by ...
16685 (insn_v1mz): ... this ...
16686 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
16689 2013-03-26 Eric Botcazou <ebotcazou@adacore.com>
16691 * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
16693 2013-03-26 Roland McGrath <mcgrathr@google.com>
16695 * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
16696 than fprintf with a non-constant, non-format string.
16698 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
16700 * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
16701 using nox64 isa attribute. Use nonimmediate_x86nomem_operand as
16702 operand 0 predicate.
16703 (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
16704 attribute. Use general_x64nomem_operand as operand 1 predicate.
16705 (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
16706 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
16707 (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
16708 attribute. Use nonimmediate_x64nomem_operand as operand 0 predicate.
16709 (mov<mode>_insv_1): Remove expander. Merge insn with
16710 movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
16711 Use general_x64nomem_operand as operand 1 predicate.
16712 (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
16713 (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
16714 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
16715 (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
16716 attribute. Use nonimmediate_x64nomem_operand as operand 2 predicate.
16717 (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
16718 attribute. Use nonimmediate_x64nomem_operand as operand 1 predicate.
16719 (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
16720 isa attribute. Use general_x64nomem_operand as operand 2 predicate.
16721 * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
16722 (general_x64nomem_operand): Ditto.
16724 2013-03-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16726 * config/rtems.opt: Add -pthread option.
16728 2013-03-26 Richard Biener <rguenther@suse.de>
16730 * alias.c (find_base_term): Avoid redundant and not used recursion.
16731 (base_alias_check): Get the initial base term from the caller.
16732 (true_dependence_1): Compute and pass base terms to base_alias_check.
16733 (write_dependence_p): Likewise.
16734 (may_alias_p): Likewise.
16736 2013-03-26 Sofiane Naci <sofiane.naci@arm.com>
16738 * config/aarch64/aarch64.c (aarch64_classify_address): Support
16739 PC-relative load in SI modes and above only.
16741 2013-03-26 Xinyu Qi <xyqi@marvell.com>
16743 * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
16744 * config/arm/iwmmxt.md (WCGR0): Update.
16745 (WCGR1, WCGR2, WCGR3): Likewise.
16747 2013-03-26 Uros Bizjak <ubizjak@gmail.com>
16749 * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
16750 Use x64 and nox64 isa attributes.
16752 2013-03-26 Richard Biener <rguenther@suse.de>
16754 * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
16755 alignment computations and rely on get_object_alignment_1
16756 for the !TYPE_P case.
16757 Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
16759 2013-03-26 Walter Lee <walt@tilera.com>
16761 * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
16762 * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
16764 2013-03-25 Jeff Law <law@redhat.com>
16766 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
16767 check for INTEGRAL_TYPE_P that was missing due to checking in
16768 wrong version of prior patch.
16770 2013-03-25 Walter Lee <walt@tilera.com>
16772 * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
16773 TILEGX_INSN_SHUFFLEBYTES1.
16774 * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
16776 (tilegx_builtins): Ditto.
16777 * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
16779 2013-03-25 Walter Lee <walt@tilera.com>
16781 * config/tilegx/tilegx.md (floatsisf2): New pattern.
16782 (floatunssisf2): New pattern.
16783 (floatsidf2): New pattern.
16784 (floatunssidf2): New pattern.
16786 2013-03-25 Walter Lee <walt@tilera.com>
16788 * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
16789 tests for constraint J, K, N, P.
16791 2013-03-25 Walter Lee <walt@tilera.com>
16793 * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
16794 Use indirect/pcrel encoding.
16795 * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
16798 2013-03-25 Steve Ellcey <sellcey@mips.com>
16800 * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
16801 74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
16802 * config/mips/mips.c (mips_option_override): Set IMADD default.
16803 * config/mips/mips.h (PTF_AVOID_IMADD): New.
16804 (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
16805 (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
16806 * config/mips/mips.md (mimadd): New flag for integer madd/msub.
16807 * doc/invoke.texi (-mimadd/-mno-imadd): New.
16809 2013-03-25 Jeff Law <law@redhat.com>
16811 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
16812 slightly to avoid creating and folding useless trees. Simplify
16813 slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
16815 2013-03-25 Uros Bizjak <ubizjak@gmail.com>
16817 * config/i386/i386.md (*zero_extendsidi2): Merge with
16818 *zero_extendsidi2_rex64. Use x64 and nox64 isa attributes.
16819 * config/i386/predicates.md (x86_64_zext_operand): Rename from
16820 x86_64_zext_general_operand. Use nonimmediate_operand on 32bit
16821 targets. Clarify comment.
16823 2013-03-25 Martin Jambor <mjambor@suse.cz>
16825 * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
16826 pass-through jump functions differently.
16827 (ipa_read_jump_function): Likewise. Also use setter functions to set
16830 2013-03-25 Martin Jambor <mjambor@suse.cz>
16832 * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
16833 ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
16835 (ipa_get_indirect_edge_target): New function.
16836 (devirtualization_time_bonus): New parameter known_aggs, pass it to
16837 ipa_get_indirect_edge_target. Update all callers.
16838 (ipcp_discover_new_direct_edges): New parameter aggvals. Pass it to
16839 ipa_get_indirect_edge_target_1 instead of calling
16840 ipa_get_indirect_edge_target.
16841 (create_specialized_node): Pass aggvlas to
16842 ipcp_discover_new_direct_edges.
16844 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16846 * config/arm/arm.md (f_sels, f_seld): New types.
16847 (*cmov<mode>): New pattern.
16848 * config/arm/predicates.md (arm_vsel_comparison_operator): New
16851 2013-03-25 Kai Tietz <ktietz@redhat.com>
16853 * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
16854 POSIX-printf for mingw-hosted builds.
16856 2013-03-25 Richard Biener <rguenther@suse.de>
16858 PR middle-end/56694
16859 * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
16860 must-not-throw stmt location.
16862 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16864 * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
16865 Emit load-acquire versions when acq is true.
16866 (arm_emit_store_exclusive): Add rel parameter.
16867 Emit store-release versions when rel is true.
16868 (arm_split_compare_and_swap): Use acquire-release instructions
16870 of barriers when appropriate.
16871 (arm_split_atomic_op): Likewise.
16872 * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
16873 * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
16874 (VUNSPEC_SLX): Likewise.
16875 (VUNSPEC_LDA): Likewise.
16876 (VUNSPEC_STL): Likewise.
16877 * config/arm/sync.md (atomic_load<mode>): New pattern.
16878 (atomic_store<mode>): Likewise.
16879 (arm_load_acquire_exclusive<mode>): Likewise.
16880 (arm_load_acquire_exclusivesi): Likewise.
16881 (arm_load_acquire_exclusivedi): Likewise.
16882 (arm_store_release_exclusive<mode>): Likewise.
16884 2013-03-25 Catherine Moore <clm@codesourcery.com>
16886 * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
16887 Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
16888 * config/mip/predicates.md (lwsp_swsp_operand,
16889 lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
16890 sb16_operand, db4_operand, db7_operand, ib3_operand,
16891 sb4_operand, ub4_operand, uh4_operand, uw4_operand,
16892 uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
16893 andi16_operand): New predicates.
16894 * config/mips/mips.md (compression): New attribute.
16895 (enabled): New attribute.
16896 (length): Consider compression in computing length.
16897 (shift_compression): New code attribute.
16898 (*add<mode>3): New operands. Record compression.
16899 (sub<mode>3): Likewise.
16900 (one_cmpl<mode>2): Likewise.
16901 (*and<mode>3): Likewise.
16902 (*ior<mode>3): Likewise.
16903 (unnamed pattern for xor): Likewise.
16904 (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
16905 (*<optab><mode>3): Likewise.
16906 (*mov<mode>_internal: Likewise.
16907 * config/mips/mips-protos.h (mips_signed_immediate_p): New.
16908 (mips_unsigned_immediate_p): New.
16909 (umips_lwsp_swsp_address_p): New.
16910 (m16_based_address_p): New.
16911 * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
16912 (mips_unsigned_immediate_p): New prototype.
16913 (lwsp_swsp_address_p): New prototype.
16914 (m16_based_address_p): New prototype.
16915 * config/mips/mips.c (mips_unsigned_immediate_p): New function.
16916 (mips_signed_immediate_p): New function.
16917 (m16_based_address_p): New function.
16918 (lwsp_swsp_address_p): New function.
16919 (mips_print_operand_punctuation): Recognize short delay slot insns
16920 for microMIPS.add<mode>3"
16922 2013-03-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16925 * config/arm/iterators.md (v_cmp_result): New mode attribute.
16926 * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
16928 2013-03-25 Richard Biener <rguenther@suse.de>
16930 PR tree-optimization/56689
16931 * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
16934 2013-03-25 Richard Biener <rguenther@suse.de>
16936 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
16938 (memory_references): Likewise.
16939 (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
16940 gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
16941 ref_indep_loop_p_2, find_refs_for_sm): Adjust.
16942 (gather_mem_refs_in_loops): Fold into ...
16943 (analyze_memory_references): ... this. Move initialization
16944 to tree_ssa_lim_initialize.
16945 (fill_always_executed_in): Rename to ...
16946 (fill_always_executed_in_1): ... this.
16947 (fill_always_executed_in): Move contains_call computation to
16948 this new function from ...
16949 (tree_ssa_lim_initialize): ... here.
16950 (tree_ssa_lim): Call fill_always_executed_in.
16952 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
16954 * postreload.c (reload_combine): Fix code detecting returns.
16956 2013-03-25 Eric Botcazou <ebotcazou@adacore.com>
16958 * function.c (emit_use_return_register_into_block): On cc0 targets,
16959 do not emit the sequence between cc0 setter and user.
16961 2013-03-25 Kai Tietz <ktietz@redhat.com>
16963 * config/i386/predicates.md (local_symbolic_operand): Interpret
16964 dll-imported symbols as none-local.
16966 2013-03-25 Richard Biener <rguenther@suse.de>
16968 * tree-ssa-loop-im.c (struct depend): Remove.
16969 (struct lim_aux_data): Make depends a vec of gimples.
16970 (free_lim_aux_data): Adjust.
16971 (add_dependency): Likewise.
16972 (set_level): Likewise.
16974 2013-03-25 Richard Biener <rguenther@suse.de>
16976 PR middle-end/56434
16977 * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
16978 the pointer returned by calls with ECF_MALLOC set.
16980 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
16982 * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
16984 2013-03-24 Uros Bizjak <ubizjak@gmail.com>
16986 * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
16987 using MMXMODE mode iterator.
16988 (*move<mode>_internal): Merge with *movv2sf_internal and
16989 *movv2sf_internal_rex64 using MMXMODE mode iterator.
16991 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
16993 * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
16994 (record_last_mem_set_info): Likewise.
16996 * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
16997 of XNEWVEC followed by memset.
16998 (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
17000 2013-03-23 Steven Bosscher <steven@gcc.gnu.org>
17002 * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
17003 config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
17004 config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
17005 config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
17006 config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
17007 config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
17008 dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
17009 the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
17010 BARRIER_P instead of GET_CODE.
17012 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
17014 * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
17015 inaccuracy in the probing code.
17017 * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
17018 (ctrapdi4): Likewise.
17020 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
17022 * calls.c (expand_call): Add missing guard to code handling return
17023 of non-BLKmode structures in MSB.
17024 * function.c (expand_function_end): Likewise.
17026 2013-03-23 Eric Botcazou <ebotcazou@adacore.com>
17028 * combine.c (try_combine): Adjust comment. Do not add the set of
17029 insn #0 if the destination indirectly is set or dies in insn #2.
17030 Tidy up code to distribute a new note.
17032 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
17034 * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
17035 also for alternatives 16 and 17.
17037 2013-03-22 Uros Bizjak <ubizjak@gmail.com>
17039 * config/i386/sse.md (*mov<mode>_internal): Merge with
17040 *mov<mode>_internal_rex64. Use x64 and nox64 isa attributes.
17041 Emit insn template depending on type attribute. Use
17042 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17043 movd instead of movq mnemonic for interunit moves. Rewrite mode
17044 attribute calculation. Remove unit attribute calculation.
17045 Set prefix attribute to maybe_vex for sselog1 and ssemov types.
17046 Set prefix_data16 attribute for DImode ssemov types.
17047 Use Ym instead of y for SSE-MMX conversion alternatives.
17048 Reorder operand constraints.
17050 2013-03-22 Steven Bosscher <steven@gcc.gnu.org>
17052 * df.h (df_insn_delete): Adjust prototype.
17053 * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
17054 and let it decide whether mark the basic block dirty.
17055 (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
17056 * df-scan.c (df_insn_info_delete): New helper function, split
17057 off from df_insn_delete.
17058 (df_scan_free_bb_info): Use it.
17059 (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
17061 (df_insn_delete): Likewise. Take insn rtx as argument. Verify
17062 that the insn is actually an insn and it has a non-NULL basic block.
17063 Do not mark basic block dirty if only deleting a DEBUG_INSN.
17065 2013-03-22 Richard Biener <rguenther@suse.de>
17067 * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
17069 (mem_ref_alloc): Do not allocate them.
17070 (refs_independent_p): Do not query or maintain a cache.
17072 2013-03-22 Richard Biener <rguenther@suse.de>
17074 * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
17075 (gather_mem_refs_in_loops): Do not compute it.
17076 (analyze_memory_references): Do not allocate it.
17077 (tree_ssa_lim_finalize): Do not free it.
17078 (for_all_locs_in_loop): Do not query all_refs_in_loop.
17080 2013-03-22 Richard Biener <rguenther@suse.de>
17082 * is-a.h (as_a): Use gcc_checking_assert.
17084 2013-03-22 Ian Bolton <ian.bolton@arm.com>
17086 * config/aarch64/aarch64.c (aarch64_print_operand): New
17087 format specifier for printing a constant in hex.
17088 * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
17089 format specifier for printing second operand.
17091 2013-03-22 Richard Biener <rguenther@suse.de>
17093 * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
17095 (gather_mem_refs_in_loops): Perform store accumulation here.
17096 (create_vop_ref_mapping_loop): Remove.
17097 (create_vop_ref_mapping): Likewise.
17098 (analyze_memory_references): Initialize refs_stored_in_loop.
17099 (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
17100 (record_indep_loop): Remove.
17101 (record_dep_loop): New function.
17102 (ref_indep_loop_p_1): Adjust to only walk over references
17103 in the loop, not its subloops.
17104 (ref_indep_loop_p): Rename to ...
17105 (ref_indep_loop_p_2): ... this and recurse over the loop tree,
17106 maintaining a more fine-grained cache.
17107 (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
17108 (tree_ssa_lim_finalize): Free refs_stored_in_loop.
17110 2013-03-22 Richard Biener <rguenther@suse.de>
17112 * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
17113 (struct mem_ref): Make accesses_in_loop a vec of a vec of
17114 aggregate mem_ref_loc.
17115 (free_mem_ref_locs): Inline into ...
17116 (memref_free): ... this and adjust.
17117 (mem_ref_alloc): Adjust.
17118 (mem_ref_locs_alloc): Remove.
17119 (record_mem_ref_loc): Adjust.
17120 (get_all_locs_in_loop): Rewrite into ...
17121 (for_all_locs_in_loop): ... this iterator.
17122 (rewrite_mem_ref_loc): New functor.
17123 (rewrite_mem_refs): Use for_all_locs_in_loop.
17124 (sm_set_flag_if_changed): New functor.
17125 (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
17126 (ref_always_accessed): New functor.
17127 (ref_always_accessed_p): Use for_all_locs_in_loop.
17129 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
17131 * tree-pass.h (PROP_gimple_lvec): New.
17132 * passes.c (dump_properties): Handle PROP_gimple_lvec.
17133 (init_optimization_passes): Move pass_lower_vector.
17134 * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
17136 (pass_lower_vector): Provide PROP_gimple_lvec.
17137 (pass_lower_vector_ssa): Likewise.
17138 * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
17140 2013-03-21 Mark Wielaard <mjw@redhat.com>
17142 * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
17144 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
17146 * config/i386/i386.md (*movdi_internal): Disparage slightly
17147 all MMX moves to/from memory. Use Yi instead of x for SSE-MMX
17148 conversion alternatives.
17150 2013-03-21 Jakub Jelinek <jakub@redhat.com>
17152 PR middle-end/48087
17153 * diagnostic.def (DK_WERROR): New kind.
17154 * diagnostic.h (werrorcount): Define.
17155 * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
17156 promoted to DK_ERROR, increment DK_WERROR counter instead of
17158 * toplev.c (toplev_main): Call print_ignored_options even if
17159 just werrorcount is non-zero. Exit with FATAL_EXIT_CODE
17160 even if just werrorcount is non-zero.
17163 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
17165 (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
17166 (string_cst_pool_decl): New function.
17167 (optimize_one_addr_into_implicit_ptr): New function.
17168 (resolve_addr_in_expr): Optimize DWARF location expression
17169 DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
17170 which doesn't live in memory, but has DW_AT_location or
17171 DW_AT_const_value, or refers to a string literal, into
17172 DW_OP_GNU_implicit_pointer.
17173 (optimize_location_into_implicit_ptr): New function.
17174 (resolve_addr): If removing DW_AT_location of a variable because
17175 it was DW_OP_addr of address of the variable, but the variable doesn't
17176 live in memory, try to emit const value attribute for the initializer.
17178 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
17180 * tree.h (VECTOR_TYPE_P): New macro.
17181 (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
17182 TYPE_MODE): Use it.
17183 * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
17184 VEC_COND_EXPR cannot be lvalues.
17185 (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
17187 2013-03-21 Marc Glisse <marc.glisse@inria.fr>
17189 * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
17190 Restrict the transformation to equal modes.
17192 2013-03-21 Richard Biener <rguenther@suse.de>
17194 PR tree-optimization/39326
17195 * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
17196 (MEM_ANALYZABLE): Adjust.
17197 (record_mem_ref_loc): Move bitmap ops ...
17198 (gather_mem_refs_stmt): ... here. Use the shared mem-ref for
17199 unanalyzable refs, do not record locations for it.
17200 (analyze_memory_references): Allocate ref zero as shared
17202 (refs_independent_p): Do not test for unanalyzed mems here.
17203 (ref_indep_loop_p_1): Special-case disambiguation against
17204 the unanalyzed ref.
17205 (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
17207 2013-03-21 Christophe Lyon <christophe.lyon@linaro.org>
17209 * config/arm/arm-protos.h (tune_params): Add
17210 prefer_neon_for_64bits field.
17211 * config/arm/arm.c (prefer_neon_for_64bits): New variable.
17212 (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
17213 (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
17214 (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
17215 (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
17216 (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
17217 (arm_option_override): Handle -mneon-for-64bits new option.
17218 * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
17219 (prefer_neon_for_64bits): Declare new variable.
17220 * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
17221 avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
17222 (arch_enabled): Handle new arch types. Remove support for onlya8
17224 (one_cmpldi2): Use new arch names.
17225 (zero_extend<mode>di2, extend<mode>di2): Ditto.
17226 * config/arm/arm.opt (mneon-for-64bits): Add option.
17227 * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
17228 (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
17229 neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
17231 * doc/invoke.texi (-mneon-for-64bits): Document.
17233 2013-03-21 Richard Biener <rguenther@suse.de>
17235 PR tree-optimization/39326
17236 * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
17237 (sort_bbs_in_loop_postorder_cmp): New function.
17238 (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
17240 2013-03-21 Richard Biener <rguenther@suse.de>
17242 * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
17243 (vect_insert_into_interleaving_chain): Likewise.
17244 (vect_drs_dependent_in_basic_block): Inline ...
17245 (vect_slp_analyze_data_ref_dependence): ... here. New function,
17247 (vect_analyze_data_ref_dependence): ... here. Simplify.
17248 (vect_check_interleaving): Simplify.
17249 (vect_analyze_data_ref_dependences): Likewise. Split out ...
17250 (vect_slp_analyze_data_ref_dependences): ... this new function.
17251 (dr_group_sort_cmp): New function.
17252 (vect_analyze_data_ref_accesses): Compute data-reference groups
17253 here instead of in vect_analyze_data_ref_dependence. Use
17254 a more efficient algorithm.
17255 * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
17256 vect_slp_analyze_data_ref_dependences. Call
17257 vect_analyze_data_ref_accesses earlier.
17258 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
17259 * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
17260 (vect_slp_analyze_data_ref_dependences): New prototype.
17262 2013-03-21 Richard Biener <rguenther@suse.de>
17264 * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
17265 ref is stored in the loop.
17266 (find_refs_for_sm): Walk only over all stores.
17267 (store_motion_loop): Allocate from lim_bitmap_obstack.
17268 (store_motion): Likewise.
17270 2013-03-21 Richard Biener <rguenther@suse.de>
17272 * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
17273 Update virtual SSA form.
17275 2013-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17277 * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
17278 * configure: Regenerate.
17279 * config.in: Regenerate.
17280 * config/sol2.c (solaris_override_options): Only enforce DWARF 2
17281 if !HAVE_LD_EH_FRAME_CIEV3.
17283 2013-03-21 Richard Biener <rguenther@suse.de>
17285 * tree-cfg.c (verify_expr_no_block): New function.
17286 (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
17287 nor DECL_VALUE_EXPR have locations with associated blocks.
17288 * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
17289 (clear_unused_block_pointer): Remove code dealing with
17290 blocks in DECL_DEBUG_EXPR locations.
17292 2013-03-21 Richard Biener <rguenther@suse.de>
17294 * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
17295 (DECL_HAS_DEBUG_EXPR_P): ... this. Guard properly.
17296 * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
17297 * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
17298 instead of DECL_DEBUG_EXPR_IS_FROM.
17299 * gimplify.c (gimplify_modify_expr): Likewise.
17300 * tree-cfg.c (verify_expr_location_1): Likewise.
17301 * tree-complex.c (create_one_component_var): Likewise.
17302 * tree-sra.c (create_access_replacement): Likewise.
17303 * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
17304 (clear_unused_block_pointer): Likewise.
17305 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
17306 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
17307 * var-tracking.c (var_debug_decl): Likewise.
17308 (track_expr_p): Likewise.
17309 * tree-inline.c (add_local_variables): Likewise. Set
17310 DECL_HAS_DEBUG_EXPR_P after copying it.
17311 * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
17312 instead of DECL_DEBUG_EXPR_IS_FROM. Guard properly.
17314 2013-03-21 Uros Bizjak <ubizjak@gmail.com>
17317 * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
17318 * configure: Regenerate.
17319 * config.in: Regenerate.
17320 * config/i386/i386.md (*movdf_internal): Use
17321 HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
17322 movd instead of movq mnemonic for interunit moves.
17323 (*movdi_internal): Ditto.
17325 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
17327 * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
17328 (abd<mode>_3): New pattern.
17329 (aba<mode>_3): New pattern.
17330 (fabd<mode>_3): New pattern.
17332 2013-03-21 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
17334 * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
17335 * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
17336 occurrence of REGISTER_PREFIX as its empty string.
17338 2013-03-20 Jeff Law <law@redhat.com>
17340 * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
17341 addititional equivalences for equality comparisons between an SSA_NAME
17342 and a constant where the SSA_NAME was set from a widening conversion.
17344 2013-03-20 Walter Lee <walt@tilera.com>
17346 * config/tilegx/sync.md (atomic_test_and_set): New pattern.
17348 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
17350 * config/i386/i386.md (*movoi_internal_avx): Emit insn template
17351 depending on type attribute.
17352 (*movti_internal): Ditto.
17353 (*movtf_internal): Ditto.
17354 (*movxf_internal): Ditto.
17355 (*movdf_internal): Ditto.
17356 (*movsf_internal): Ditto.
17358 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
17360 * config/i386/i386.md (*movti_internal): Set prefix attribute to
17361 maybe_vex for sselog1 and ssemov types.
17362 (*movdi_internal): Reorder operand constraints.
17363 (*movsi_internal): Ditto. Set prefix attribute to
17364 maybe_vex for sselog1 and ssemov types.
17365 (*movtf_internal): Set prefix attribute to maybe_vex
17366 for sselog1 and ssemov types.
17367 (*movdf_internal): Ditto. Set prefix_data16 attribute for
17368 DImode ssemov types. Reorder operand constraints.
17369 (*movsf_internal): Set type of alternatives 3,4 to imov. Set prefix
17370 attribute to maybe_vex for sselog1 and ssemov types. Set prefix_data16
17371 attribute for SImode ssemov types. Reorder operand constraints.
17373 2013-03-20 Martin Jambor <mjambor@suse.cz>
17375 * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
17376 * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
17378 2013-03-20 Pat Haugen <pthaugen@us.ibm.com>
17380 * config/rs6000/predicates.md (indexed_address, update_address_mem
17381 update_indexed_address_mem): New predicates.
17382 * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
17383 attribute for load/store instructions.
17384 * config/rs6000/dfp.md (movsd_store): Likewise.
17385 (movsd_load): Likewise.
17386 * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
17387 (unnamed HI->DI extend define_insn): Likewise.
17388 (unnamed SI->DI extend define_insn): Likewise.
17389 (unnamed QI->SI extend define_insn): Likewise.
17390 (unnamed QI->HI extend define_insn): Likewise.
17391 (unnamed HI->SI extend define_insn): Likewise.
17392 (unnamed HI->SI extend define_insn): Likewise.
17393 (extendsfdf2_fpr): Likewise.
17394 (movsi_internal1): Likewise.
17395 (movsi_internal1_single): Likewise.
17396 (movhi_internal): Likewise.
17397 (movqi_internal): Likewise.
17398 (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
17399 attribute for load/store instructions.
17400 (mov<mode>_hardfloat): Set correct "type" attribute for load/store
17402 (mov<mode>_softfloat): Likewise.
17403 (mov<mode>_hardfloat32): Likewise.
17404 (mov<mode>_hardfloat64): Likewise.
17405 (mov<mode>_softfloat64): Likewise.
17406 (movdi_internal32): Likewise.
17407 (movdi_internal64): Likewise.
17408 (probe_stack_<mode>): Likewise.
17410 2013-03-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17412 * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
17413 floating point, and decimal floating point to reload iterator.
17415 * config/rs6000/constraints.md (wl constraint): New constraints to
17416 return FLOAT_REGS if certain options are used to reduce the number
17417 of separate patterns that exist in the file.
17418 (wx constraint): Likewise.
17419 (wz constraint): Likewise.
17421 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17422 -mdebug=reg, print wg, wl, wx, and wz constraints.
17423 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17424 Initialize the reload functions for 64-bit binary/decimal floating
17426 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17427 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17428 create the buffer on the stack to overcome not having a 32-bit
17430 (rs6000_emit_move): Likewise.
17431 (rs6000_secondary_memory_needed_rtx): Likewise.
17432 (rs6000_alloc_sdmode_stack_slot): Likewise.
17433 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17434 via xxlxor, just like DFmode 0.0.
17436 * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
17437 define as 1 if we are running on a power7 or newer.
17438 (enum r6000_reg_class_enum): Add new constraints.
17440 * config/rs6000/dfp.md (movsd): Delete, combine with binary
17441 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
17442 with other moves by using conditional constraits (wg). Use LFIWZX
17443 and STFIWX for loading SDmode on power7. Use xxlxor to create 0.0f.
17444 (movsd splitter): Likewise.
17445 (movsd_hardfloat): Likewise.
17446 (movsd_softfloat): Likewise.
17448 * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
17449 binary and decimal floating point moves.
17450 (fmove_ok): New attributes to combine binary and decimal floating
17451 point moves, and to combine power6x (mfpgpr) moves along normal
17453 (real_value_to_target): Likewise.
17454 (f32_lr): Likewise.
17455 (f32_lm): Likewise.
17456 (f32_li): Likewise.
17457 (f32_sr): Likewise.
17458 (f32_sm): Likewise.
17459 (f32_si): Likewise.
17460 (movsf): Combine binary and decimal floating point moves. Combine
17461 power6x (mfpgpr) moves with other moves by using conditional
17462 constraits (wg). Use LFIWZX and STFIWX for loading SDmode on power7.
17463 (mov<mode> for SFmode/SDmode); Likewise.
17464 (SFmode/SDmode splitters): Likewise.
17465 (movsf_hardfloat): Likewise.
17466 (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
17467 (movsf_softfloat): Likewise.
17468 (mov<mode>_softfloat for SFmode/SDmode): Likewise.
17470 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
17471 wx and wz constraints.
17473 * config/rs6000/constraints.md (wg constraint): New constraint to
17474 return FLOAT_REGS if -mmfpgpr (power6x) was used.
17476 * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
17479 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
17480 -mdebug=reg, print wg, wl, wx, and wz constraints.
17481 (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
17482 Initialize the reload functions for 64-bit binary/decimal floating
17484 (reg_offset_addressing_ok_p): If we are on a power7 or later, use
17485 LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
17486 create the buffer on the stack to overcome not having a 32-bit
17488 (rs6000_emit_move): Likewise.
17489 (rs6000_secondary_memory_needed_rtx): Likewise.
17490 (rs6000_alloc_sdmode_stack_slot): Likewise.
17491 (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
17492 via xxlxor, just like DFmode 0.0.
17494 * config/rs6000/dfp.md (movdd): Delete, combine with binary
17495 floating point moves in rs6000.md. Combine power6x (mfpgpr) moves
17496 with other moves by using conditional constraits (wg). Use LFIWZX
17497 and STFIWX for loading SDmode on power7.
17498 (movdd splitters): Likewise.
17499 (movdd_hardfloat32): Likewise.
17500 (movdd_softfloat32): Likewise.
17501 (movdd_hardfloat64_mfpgpr): Likewise.
17502 (movdd_hardfloat64): Likewise.
17503 (movdd_softfloat64): Likewise.
17505 * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
17506 64-bit binary and decimal floating point moves.
17507 (FMOVE64X): Likewise.
17508 (movdf): Combine 64-bit binary and decimal floating point moves.
17509 Combine power6x (mfpgpr) moves with other moves by using
17510 conditional constraits (wg).
17511 (mov<mode> for DFmode/DDmode): Likewise.
17512 (DFmode/DDmode splitters): Likewise.
17513 (movdf_hardfloat32): Likewise.
17514 (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
17515 (movdf_softfloat32): Likewise.
17516 (movdf_hardfloat64_mfpgpr): Likewise.
17517 (movdf_hardfloat64): Likewise.
17518 (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
17519 (movdf_softfloat64): Likewise.
17520 (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
17521 (reload_<mode>_load): Move to later in the file so they aren't in
17522 the middle of the floating point move insns.
17523 (reload_<mode>_store): Likewise.
17525 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
17528 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
17529 constraint if -mdebug=reg.
17530 (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
17531 Enable using dd reload support if needed.
17533 * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
17534 binary and decimal floating point moves in rs6000.md.
17535 (movtd_internal): Likewise.
17537 * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
17538 decimal floating point moves.
17540 (movtf_internal): Likewise.
17541 (mov<mode>_internal, TDmode/TFmode): Likewise.
17542 (movtf_softfloat): Likewise.
17543 (mov<mode>_softfloat, TDmode/TFmode): Likewise.
17545 * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
17546 movdi_internal64, using wg constraint for move direct operations.
17547 (movdi_internal64): Likewise.
17549 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
17550 MODES_TIEABLE_P for selected modes. Print the numerical value of
17551 the various virtual registers. Use GPR/FPR first/last values,
17552 instead of hard coding the register numbers. Print which modes
17553 have reload functions registered.
17554 (rs6000_option_override_internal): If -mdebug=reg, trace the options
17555 settings before/after setting cpu, target and subtarget settings.
17556 (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
17557 and for secondary reload failures in rs6000_secondary_reload_inner.
17558 (rs6000_secondary_reload_fail): Likewise.
17559 (rs6000_secondary_reload_inner): Likewise.
17561 * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
17562 macros for first/last GPR and FPR registers.
17563 (LAST_GPR_REGNO): Likewise.
17564 (FIRST_FPR_REGNO): Likewise.
17565 (LAST_FPR_REGNO): Likewise.
17567 * config/rs6000/vector.md (mul<mode>3): Use the combined macro
17568 VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
17569 VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
17570 (vcond<mode><mode>): Likewise.
17571 (vcondu<mode><mode>): Likewise.
17572 (vector_gtu<mode>): Likewise.
17573 (vector_gte<mode>): Likewise.
17574 (xor<mode>3): Don't allow logical operations on TImode in 32-bit
17575 to prevent the compiler from converting DImode operations to TImode.
17576 (ior<mode>3): Likewise.
17577 (and<mode>3): Likewise.
17578 (one_cmpl<mode>2): Likewise.
17579 (nor<mode>3): Likewise.
17580 (andc<mode>3): Likewise.
17582 * config/rs6000/constraints.md (wt constraint): New constraint
17583 that returns VSX_REGS if TImode is allowed in VSX registers.
17585 * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
17586 constant under VSX.
17588 * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
17589 similar to TImode, but it is restricted to being in the GPRs.
17591 * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
17592 TImode to occupy a single VSX register.
17594 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
17595 -mvsx-timode for power7/power8.
17596 (power7 cpu): Likewise.
17597 (power8 cpu): Likewise.
17599 * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
17600 sure that TFmode/TDmode take up two registers if they are ever
17601 allowed in the upper VSX registers.
17602 (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
17604 (rs6000_init_hard_regno_mode_ok): Likewise.
17605 (rs6000_debug_reg_global): Add debugging for PTImode and wt
17606 constraint. Print if LRA is turned on.
17607 (rs6000_option_override_internal): Give an error if -mvsx-timode
17608 and VSX is not enabled.
17609 (invalid_e500_subreg): Handle PTImode, restricting it to GPRs. If
17610 -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
17611 to reg+offset addressing. Use PTImode when checking offset
17612 addresses for validity.
17613 (reg_offset_addressing_ok_p): Likewise.
17614 (rs6000_legitimate_offset_address_p): Likewise.
17615 (rs6000_legitimize_address): Likewise.
17616 (rs6000_legitimize_reload_address): Likewise.
17617 (rs6000_legitimate_address_p): Likewise.
17618 (rs6000_eliminate_indexed_memrefs): Likewise.
17619 (rs6000_emit_move): Likewise.
17620 (rs6000_secondary_reload): Likewise.
17621 (rs6000_secondary_reload_inner): Handle PTImode. Allow 64-bit
17622 reloads to fpr registers to continue to use reg+offset addressing,
17623 but 64-bit reloads to altivec registers need reg+reg addressing.
17624 Drop test for PRE_MODIFY, since VSX loads/stores no longer support
17625 it. Treat LO_SUM like a PLUS operation.
17626 (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
17627 FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
17628 (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
17629 registers to share a register with a smaller sized type, since VSX
17630 puts scalars in the upper 64-bits.
17631 (print_operand): Add support for PTImode.
17632 (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
17633 VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
17634 registers, but don't have arithmetic support.
17635 (rs6000_memory_move_cost): Add test for VSX.
17636 (rs6000_opt_masks): Add -mvsx-timode.
17638 * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
17641 (VSr): Use wt constraint for TImode.
17642 (VSv): Drop TImode support.
17643 (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
17644 (vsx_movti_64bit): Likewise.
17645 (vsx_movti_32bit): Likewise.
17646 (vec_store_<mode>): Use VSX iterator instead of vector iterator.
17647 (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
17648 one '?' on the appropriate output constraint. Do not allow TImode
17649 logical operations on 32-bit systems.
17650 (vsx_ior<mode>3): Likewise.
17651 (vsx_xor<mode>3): Likewise.
17652 (vsx_one_cmpl<mode>2): Likewise.
17653 (vsx_nor<mode>3): Likewise.
17654 (vsx_andc<mode>3): Likewise.
17655 (vsx_concat_<mode>): Likewise.
17656 (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
17658 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
17659 OPTION_MASK_VSX_TIMODE.
17660 (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
17661 (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
17663 * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
17664 (TI2 iterator): New iterator for TImode, PTImode.
17665 (wd mode attribute): Add values for vector types.
17666 (movti_string): Replace TI move operations with operations for TImode
17667 and PTImode. Add support for TImode being allowed in VSX registers.
17668 (mov<mode>_string, TImode/PTImode): Likewise.
17669 (movti_ppc64): Likewise.
17670 (mov<mode>_ppc64, TImode/PTImode): Likewise.
17671 (TI mode splitters): Likewise.
17673 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
17676 2013-03-20 Marc Glisse <marc.glisse@inria.fr>
17678 PR tree-optimization/56355
17679 * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
17680 Also handle integers with undefined overflow.
17682 2013-03-20 Catherine Moore <clm@codesourcery.com>
17683 Maciej W. Rozycki <macro@codesourcery.com>
17684 Tom de Vries <tom@codesourcery.com>
17685 Nathan Sidwell <nathan@codesourcery.com>
17686 Iain Sandoe <iain@codesourcery.com>
17687 Nathan Froyd <froydnj@codesourcery.com>
17688 Chao-ying Fu <fu@mips.com>
17690 * doc/extend.texi: (micromips, nomicromips, nocompression):
17691 Document new function attributes.
17692 * doc/invoke.texi (minterlink-compressed, mmicromips,
17693 m14k, m14ke, m14kec): Document new options.
17694 (minterlink-mips16): Update documentation.
17695 * doc/md.texi (ZC, ZD): Document new constraints.
17696 * configure.ac (gcc_cv_as_micromips): Check if linker
17697 supports the .set micromips directive.
17698 * configure: Regenerate.
17699 * config.in: Regenerate.
17700 * config/mips/mips-tables.opt: Regenerate.
17701 * config/mips/micromips.md: New file.
17702 * constraints.md (ZC, ZD): New constraints.
17703 * config/mips/predicates.md (movep_src_register): New predicate.
17704 (movep_src_operand): New predicate.
17705 (non_volatile_mem_operand): New predicate.
17706 * config/mips/mips.md (multimem): New type.
17707 (length): Differentiate between 17-bit and 18-bit branch offsets.
17708 (MOVEP1, MOVEP2): New mode iterator.
17709 (mov_<load>l): Use ZC constraint.
17710 (mov_<load>r): Likewise.
17711 (mov_<store>l): Likewise.
17712 (mov_<store>r): Likewise.
17713 (*branch_equality<mode>_inverted): Add microMIPS support.
17714 (*branch_equality<mode>): Likewise.
17715 (*jump_absolute): Likewise.
17716 (indirect_jump_<mode>): Likewise.
17717 (tablejump_<mode>): Likewise.
17718 (<optab>_internal): Likewise.
17719 (sibcall_internal): Likewise.
17720 (sibcall_value_internal): Likewise.
17721 (prefetch): Use constraint ZD.
17722 * config/mips/mips.opt (minterlink-compressed): New option.
17723 (minterlink-mips16): Now an alias for minterlink-compressed.
17724 (mmicromips): New option.
17725 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
17726 (compare_and_swap_12): Likewise.
17727 (sync_add<mode>): Likewise.
17728 (sync_<optab>_12): Likewise.
17729 (sync_old_<optab>_12): Likewise.
17730 (sync_new_<optab>_12): Likewise.
17731 (sync_nand_12): Likewise.
17732 (sync_old_nand_12): Likewise.
17733 (sync_new_nand_12): Likewise.
17734 (sync_sub<mode>): Likewise.
17735 (sync_old_add<mode>): Likewise.
17736 (sync_old_sub<mode>): Likewise.
17737 (sync_new_add<mode>): Likewise.
17738 (sync_new_sub<mode>): Likewise.
17739 (sync_<optab><mode>): Likewise.
17740 (sync_old_<optab><mode>): Likewise.
17741 (sync_new_<optab><mode>): Likewise.
17742 (sync_nand<mode>): Likewise.
17743 (sync_old_nand<mode>): Likewise.
17744 (sync_new_nand<mode>): Likewise.
17745 (sync_lock_test_and_set<mode>): Likewise.
17746 (test_and_set_12): Likewise.
17747 (atomic_compare_and_swap<mode>): Likewise.
17748 (atomic_exchange<mode>_llsc): Likewise.
17749 (atomic_fetch_add<mode>_llsc): Likewise.
17750 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
17751 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
17752 (umips_save_restore_pattern_p): Likewise.
17753 (umips_load_store_pair_p): Likewise.
17754 (umips_output_load_store_pair): Likewise.
17755 (umips_movep_target_p): Likewise.
17756 (umips_12bit_offset_address_p): Likewise.
17757 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
17758 (mips_base_mips16): Rename this...
17759 (mips_base_compression_flags): ...to this. Update all uses.
17760 (mips_attribute_table): Add micromips, nomicromips and nocompression.
17761 (mips_mips16_decl_p): Delete.
17762 (mips_nomips16_decl_p): Delete.
17763 (mips_get_compress_on_flags): New function.
17764 (mips_get_compress_off_flags): New function.
17765 (mips_get_compress_mode): New function.
17766 (mips_get_compress_on_name): New function.
17767 (mips_get_compress_off_name): New function.
17768 (mips_insert_attributes): Support multiple compression types.
17769 (mips_merge_decl_attributes): Likewise.
17770 (umips_12bit_offset_address_p): New function.
17771 (mips_start_function_definition): Emit .set micromips directive.
17772 (mips_call_may_need_jalx_p): New function.
17773 (mips_function_ok_for_sibcall): Add microMIPS support.
17774 (mips_print_operand_punctuation): Support short delay slots and
17776 (umips_swm_mask, umips_swm_encoding): New.
17777 (umips_build_save_restore): New function.
17778 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
17779 (was_mips16_p): Remove.
17780 (old_compression_mode): New.
17781 (mips_set_compression_mode): New function.
17782 (mips_set_current_function): Add microMIPS support.
17783 (mips_option_override): Likewise.
17784 (umips_save_restore_pattern_p): New function.
17785 (umips_output_save_restore): New function.
17786 (umips_load_store_pair_p_1): New function.
17787 (umips_load_store_pair_p): New function.
17788 (umips_output_load_store_pair_1): New function.
17789 (umips_output_load_store_pair): New function.
17790 (umips_movep_target_p) New function.
17791 (mips_prepare_pch_save): Add microMIPS support.
17792 * config/mips/mips.h (TARGET_COMPRESSION): New.
17793 (TARGET_CPU_CPP_BUILTINS): Update macro
17794 to use new compression flags and to support microMIPS.
17795 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
17796 (MIPS_ARCH_FLOAT_SPEC): Likewise.
17797 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
17798 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
17799 (ASM_SPEC): Support mmicromips and mno-micromips.
17800 (M16STORE_REG_P): New macro.
17801 (MIPS_CALL): Support TARGET_MICROMIPS.
17802 (MICROMIPS_J): New macro.
17803 (mips_base_mips16): Rename this...
17804 (mips_base_compression_flags): ...to this.
17805 (UMIPS_12BIT_OFFSET_P): New macro.
17806 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
17807 (MULTILIB_DIRNAMES): Likewise.
17808 2013-03-20 Richard Biener <rguenther@suse.de>
17810 PR tree-optimization/56661
17811 * tree-ssa-sccvn.c (visit_use): Only value-number calls if
17812 the result does not have to be distinct.
17814 2013-03-20 Richard Biener <rguenther@suse.de>
17816 * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
17819 2013-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17820 Steven Bosscher <steven@gcc.gnu.org>
17822 PR rtl-optimization/56605
17823 * loop-iv.c (implies_p): Handle equal RTXs and subregs.
17825 2013-03-20 Uros Bizjak <ubizjak@gmail.com>
17828 * config/i386/i386.md (*movdi_internal): Handle broken assemblers
17829 that require movd instead of movq.
17831 2013-03-20 Richard Biener <rguenther@suse.de>
17833 * tree-ssa-structalias.c (struct variable_info): Add pointer
17834 to the first field of an aggregate with sub-vars. Make
17835 this and the pointer to the next subfield its ID.
17836 (vi_next): New function.
17837 (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
17838 storedanything_id, integer_id): Increment by one.
17839 (new_var_info, get_call_vi, lookup_call_clobber_vi,
17840 get_call_clobber_vi): Adjust.
17841 (solution_set_expand): Simplify and speedup.
17842 (solution_set_add): Inline into ...
17843 (set_union_with_increment): ... this. Adjust accordingly.
17844 (do_sd_constraint): Likewise.
17845 (do_ds_constraint): Likewise.
17846 (do_complex_constraint): Simplify.
17847 (build_pred_graph): Adjust.
17848 (solve_graph): Likewise. Simplify and speedup.
17849 (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
17850 get_constraint_for_component_ref, get_constraint_for_1,
17851 first_vi_for_offset, first_or_preceding_vi_for_offset,
17852 create_function_info_for, create_variable_info_for_1,
17853 create_variable_info_for, intra_create_variable_infos): Adjust.
17854 (init_base_vars): Push NULL for ID zero.
17855 (compute_points_to_sets): Adjust.
17857 2013-03-20 Richard Biener <rguenther@suse.de>
17859 * cfgloop.c (verify_loop_structure): Streamline and avoid
17860 ICEing on corrupt loop tree.
17861 * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
17864 2013-03-20 Richard Biener <rguenther@suse.de>
17866 * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
17867 check whether an SSA update is needed.
17869 2013-03-20 Richard Sandiford <rdsandiford@googlemail.com>
17871 * config/mips/constraints.md (T): Rename to...
17875 * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
17876 (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
17878 2013-03-19 Ian Bolton <ian.bolton@arm.com>
17880 * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
17881 (*subsi3_carryin_uxtw): Likewise.
17883 2013-03-19 Ian Bolton <ian.bolton@arm.com>
17885 * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
17886 (*rorsi3_insn_uxtw): Likewise.
17888 2013-03-19 Ian Bolton <ian.bolton@arm.com>
17890 * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
17891 (*extrsi5_insn_uxtw): Likewise.
17893 2013-03-19 Richard Biener <rguenther@suse.de>
17895 PR tree-optimization/56273
17896 * passes.c (init_optimization_passes): Move second VRP after DOM.
17898 2013-03-19 Uros Bizjak <ubizjak@gmail.com>
17900 * config/i386/i386.md (*movti_internal): Merge from
17901 *movti_internal_rex64 and *movti_internal_sse. Use x64 isa attribute.
17902 (*movdi_internal): Merge with *movdi_internal_rex64. Use x64 and
17903 nox64 isa attributes.
17905 2013-03-18 Richard Biener <rguenther@suse.de>
17907 * tree-ssa-structalias.c (find): Use gcc_checking_assert.
17909 (merge_node_constraints): Likewise.
17910 (build_succ_graph): Likewise.
17911 (valid_graph_edge): Inline into single caller.
17912 (unify_nodes): Likewise. Use bitmap_set_bit return value
17914 (scc_visit): Fix formatting and variable use.
17915 (do_sd_constraint): Use gcc_checking_assert.
17916 (do_ds_constraint): Likewise.
17917 (do_complex_constraint): Likewise.
17918 (condense_visit): Likewise. Cleanup.
17919 (dump_pred_graph): New function.
17920 (perform_var_substitution): Dump the pred-graph before
17921 variable substitution.
17922 (find_equivalent_node): Use gcc_checking_assert.
17923 (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
17925 2013-03-18 Richard Biener <rguenther@suse.de>
17927 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
17928 Remove cond_expr_stmt_list argument and do not gimplify the
17930 (vect_loop_versioning): Adjust.
17931 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
17932 Cleanup to use less temporaries.
17933 (vect_create_data_ref_ptr): Cleanup.
17935 2013-03-18 Jakub Jelinek <jakub@redhat.com>
17937 PR tree-optimization/56635
17938 * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
17939 require types_compatible_p types.
17941 2013-03-18 Nick Clifton <nickc@redhat.com>
17943 * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
17944 spurious backslash.
17946 * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
17947 Add missing line to comment describing function.
17949 2013-03-18 Richard Biener <rguenther@suse.de>
17951 PR tree-optimization/56210
17952 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17953 Handle string / character search functions.
17954 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
17956 2013-03-18 Richard Biener <rguenther@suse.de>
17958 PR middle-end/56483
17959 * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
17960 and implement properly.
17961 * gimple.h (gimple_cond_single_var_p): Remove.
17963 2013-03-18 Richard Biener <rguenther@suse.de>
17965 * tree-data-ref.h (find_data_references_in_loop): Declare.
17966 * tree-data-ref.c (get_references_in_stmt): Use a stack
17967 vector pre-allocated in the callers.
17968 (find_data_references_in_stmt): Adjust.
17969 (graphite_find_data_references_in_stmt): Likewise.
17970 (create_rdg_vertices): Likewise.
17971 (find_data_references_in_loop): Export.
17972 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17973 Compute dependences here...
17974 (vect_analyze_data_refs): ...not here. When we encounter
17975 a non-vectorizable data reference in basic-block vectorization
17976 truncate the data reference vector. Do not bother to
17977 fixup data-dependence information for gather loads.
17978 * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
17979 of data references, as reported.
17981 2013-03-18 Richard Biener <rguenther@suse.de>
17983 PR tree-optimization/3713
17984 * tree-ssa-sccvn.c (visit_copy): Simplify. Always propagate
17985 has_constants and expr.
17986 (stmt_has_constants): Properly valueize SSA names when deciding
17987 whether the stmt has constants.
17989 2013-03-18 Richard Biener <rguenther@suse.de>
17991 * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
17992 whole function when there is nothing to do.
17993 * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
17994 * tree-vectorizer.c (vectorize_loops): Update virtual and
17995 loop-closed SSA once.
17996 * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
17998 2013-03-18 Richard Biener <rguenther@suse.de>
18000 PR middle-end/56113
18001 * domwalk.c (bb_postorder): New global static.
18002 (cmp_bb_postorder): New function.
18003 (walk_dominator_tree): Replace scheme imposing an order for
18004 visiting dominator sons by one sorting them at the time they
18005 are pushed on the stack.
18007 2013-03-18 Richard Biener <rguenther@suse.de>
18009 PR tree-optimization/39326
18010 * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
18011 (struct mem_ref): Replace mem member with ao_ref typed member.
18012 (MEM_ANALYZABLE): Adjust.
18013 (memref_eq): Likewise.
18014 (mem_ref_alloc): Likewise.
18015 (gather_mem_refs_stmt): Likewise.
18016 (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
18017 (execute_sm_if_changed_flag_set): Adjust.
18018 (execute_sm): Likewise.
18019 (ref_always_accessed_p): Likewise.
18020 (refs_independent_p): Likewise.
18021 (can_sm_ref_p): Likewise.
18023 2013-03-18 Jakub Jelinek <jakub@redhat.com>
18026 * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
18027 return 1 even for !unsignedp.
18029 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
18031 * config/i386/i386.md (isa): Add x64 and nox64.
18032 (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
18033 (*pushtf): Enable *roF alternative for x64 isa only.
18034 (*pushxf): Merge with *pushxf_nointeger. Use Yx*r constraint. Set
18035 mode attribute of integer alternatives to DImode for TARGET_64BIT.
18036 (*pushdf): Merge with *pushdf_rex64. Use x64 and nox64 isa attributes.
18037 (*movtf_internal): Merge from *movtf_internal_rex64 and
18038 *movtf_internal_sse. Use x64 and nox64 isa attributes.
18039 (*movxf_internal): Merge with *movxf_internal_rex64. Use x64 and
18040 nox64 isa attributes.
18041 (*movdf_internal): Merge with *movdf_internal_rex64. Use x64 and
18042 nox64 isa attributes.
18043 * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
18045 2013-03-17 Uros Bizjak <ubizjak@gmail.com>
18047 * config/alpha/alpha.c (TARGET_LRA_P): New define.
18049 2013-03-17 Jakub Jelinek <jakub@redhat.com>
18052 * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
18053 class names. Remove trailing comma after "ALL_REGS".
18055 2013-03-16 Jan Hubicka <jh@suse.cz>
18057 * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
18058 * cgraph.c (cgraph_get_create_real_symbol_node): New function.
18059 * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
18060 of cgraph_get_create_node.
18061 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18063 2013-03-16 Jason Merrill <jason@redhat.com>
18066 * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
18067 with DW_AT_default_value.
18069 2013-03-16 Jakub Jelinek <jakub@redhat.com>
18071 * BASE-VER: Set to 4.9.0.
18073 2013-03-14 Andi Kleen <ak@linux.intel.com>
18076 * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
18077 __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
18078 Document _x* TSX intrinsics.
18080 2013-03-14 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
18081 David Holsgrove <david.holsgrove@xilinx.com>
18083 * configure.ac: Add MicroBlaze TLS support detection.
18084 * configure: Regenerate.
18085 * config/microblaze/microblaze-protos.h
18086 (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
18087 symbol_mentioned_p, label_mentioned_p): Add prototypes.
18088 * config/microblaze/microblaze.c (microblaze_address_type): Add
18089 ADDRESS_TLS and tls_reloc address types.
18090 (microblaze_address_info): Add tls_reloc.
18091 (TARGET_HAVE_TLS): Define.
18092 (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
18093 microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
18094 symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
18095 load_tls_operand, microblaze_call_tls_get_addr,
18096 microblaze_legitimize_tls_address): New functions.
18097 (microblaze_classify_unspec): Handle UNSPEC_TLS.
18098 (get_base_reg): Use microblaze_tls_symbol_p.
18099 (microblaze_classify_address): Handle TLS.
18100 (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
18101 label_mentioned_p and microblaze_tls_referenced_p.
18102 (microblaze_legitimize_address): Handle TLS.
18103 (microblaze_address_insns): Handle ADDRESS_TLS.
18104 (pic_address_needs_scratch): Handle TLS.
18105 (print_operand_address): Handle TLS.
18106 (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
18107 (microblaze_expand_move): Handle TLS.
18108 (microblaze_legitimate_constant_p): Check
18109 microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
18110 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
18111 * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
18112 (PIC_OFFSET_TABLE_REGNUM): Set.
18113 * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
18114 * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
18115 (addsi3, movsi_internal2, movdf_internal): Update constraints
18116 * config/microblaze/predicates.md (arith_plus_operand): Define
18117 (move_operand): Redefine as move_src_operand,
18118 check microblaze_tls_referenced_p.
18120 2013-03-14 Ian Bolton <ian.bolton@arm.com>
18122 * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
18123 (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
18125 2013-03-14 Ian Bolton <ian.bolton@arm.com>
18127 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
18130 2013-03-14 Jakub Jelinek <jakub@redhat.com>
18132 PR tree-optimization/53265
18133 * common.opt (Waggressive-loop-optimizations): New option.
18134 * tree-ssa-loop-niter.c: Include tree-pass.h.
18135 (do_warn_aggressive_loop_optimizations): New function.
18136 (record_estimate): Call it. Don't add !is_exit bounds to loop->bounds
18137 if number_of_latch_executions returned constant.
18138 (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
18139 early. If number_of_latch_executions returned constant, set
18140 nb_iterations_upper_bound back to it.
18141 * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
18143 * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
18144 * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
18146 * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
18147 (MULTILIB_OSDIRNAMES): Set.
18148 * genmultilib: If defaultosdirname doesn't start with :: , set
18149 defaultosdirname2 instead, clear it and emit two . multilib_raw
18150 entries instead of just one.
18152 2013-03-14 Kaz Kojima <kkojima@gcc.gnu.org>
18154 * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
18155 (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
18156 * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
18157 (SUBTARGET_OVERRIDE_OPTIONS): New.
18159 2013-03-13 Oleg Endo <olegendo@gcc.gnu.org>
18162 * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
18163 (musermode): Convert to Var(TARGET_USERMODE).
18164 * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
18165 MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
18166 * config/sh/sh.c (sh_option_override): Use
18167 TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
18168 * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
18170 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
18172 (udivsi3_i4_single, divsi3_i4_single): Use
18173 TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
18175 2013-03-13 Dave Korn <dave.korn.cygwin@....>
18177 * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
18180 2013-03-13 Richard Biener <rguenther@suse.de>
18182 PR tree-optimization/56608
18183 * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
18184 calls when vectorizing basic-blocks.
18186 2013-03-13 Jakub Jelinek <jakub@redhat.com>
18189 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
18192 2013-03-12 Jakub Jelinek <jakub@redhat.com>
18194 * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
18196 2013-03-11 Jan Hubicka <jh@suse.cz>
18199 * lto-streamer-out.c (output_symbol_p): Skip references from
18200 constructors of external variables.
18202 2013-03-11 Jan Hubicka <jh@suse.cz>
18204 PR middle-end/56571
18205 * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
18207 * emit-rtl.c (verify_rtx_sharing): Likewise.
18208 (copy_insn_1): Likewise.
18209 * rtl.c (copy_rtx): Likewise.
18211 2013-03-11 Georg-Johann Lay <avr@gjlay.de>
18214 * config/avr/avr.c (avr_print_operand): Add space after '%c' in
18215 output_operand_lossage message.
18217 2013-03-11 Richard Earnshaw <rearnsha@arm.com>
18220 * arm.c (shift_op): Validate RTL pattern on the fly.
18221 (arm_print_operand, case 'S'): Don't use shift_operator to validate
18224 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
18227 * config/pa/pa.md (call_value): Check for calls to powf and direct to
18228 new call patterns that clobber %fr12.
18229 (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
18230 split and postreload patterns.
18231 * config/pa/pa.c (pa_conditional_register_usage): Revert marking
18232 registers %fr12 and %fr12R as call used.
18234 2013-03-09 Steven Bosscher <steven@gcc.gnu.org>
18236 * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
18237 (canon_address, record_store, replace_read, check_mem_read_rtx,
18238 scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
18239 dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
18240 rest_of_handle_dse): Likewise.
18242 2013-03-09 Richard Sandiford <rdsandiford@googlemail.com>
18244 PR middle-end/56524
18245 * tree.h (tree_optimization_option): Rename target_optabs to optabs.
18247 (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
18248 (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
18249 (save_optabs_if_changed): Replace with...
18250 (init_tree_optimization_optabs): ...this.
18251 * optabs.c (save_optabs_if_changed): Rename to...
18252 (init_tree_optimization_optabs): ...this. Take the optimization node
18253 as argument. Do nothing if the base optabs are already correct.
18254 Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
18255 to recompute optabs.
18256 * function.h (function): Remove optabs field.
18257 * function.c (invoke_set_current_function_hook): Call
18258 init_tree_optimization_optabs. Use the result to initialize
18261 2013-02-27 Aldy Hernandez <aldyh@redhat.com>
18263 * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
18264 if GTMA_HAS_NO_INSTRUMENTATION.
18265 (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
18266 (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
18267 * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
18268 * gimple-pretty-print.c (dump_gimple_transaction): Handle
18269 GTMA_HAS_NO_INSTRUMENTATION.
18271 2013-03-08 Jakub Jelinek <jakub@redhat.com>
18273 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
18276 2013-03-08 Marek Polacek <polacek@redhat.com>
18277 Jakub Jelinek <jakub@redhat.com>
18279 PR tree-optimization/56478
18280 * predict.c (is_comparison_with_loop_invariant_p): Change the
18281 type of loop_step to tree.
18282 (predict_loops): Adjust.
18283 (predict_iv_comparison): Perform the computations on double_ints.
18285 2013-03-08 Richard Biener <rguenther@suse.de>
18287 PR tree-optimization/56570
18288 * tree-cfg.c (verify_expr_location_1): Verify locations for
18290 * tree-sra.c (create_access_replacement): Strip locations
18291 from DECL_DEBUG_EXPRs.
18293 2013-03-08 Richard Biener <rguenther@suse.de>
18295 * tree-inline.c (expand_call_inline): Do not associate
18296 a BLOCK with the location in BLOCK_SOURCE_LOCATION.
18297 * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
18299 2013-03-08 Richard Biener <rguenther@suse.de>
18301 * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
18302 or block changes with -Og. Fix for location / block encoding
18303 changes and PHI arguments with locations.
18305 2013-03-07 Steven Bosscher <steven@gcc.gnu.org>
18307 * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
18309 (struct output_info): Likewise.
18310 (register_overhead): Remove bad gcc_assert.
18311 (bitmap_find_bit): If there is only a single bitmap element, do not
18312 count a miss as a search.
18313 (print_statistics): Update for counter type changes.
18314 (dump_bitmap_statistics): Likewise. Print headers such that they
18315 are properly lined up with the printed counters.
18317 2013-03-07 Jakub Jelinek <jakub@redhat.com>
18319 PR tree-optimization/56559
18320 * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
18321 check that it has only a single use.
18323 2013-03-07 Richard Biener <rguenther@suse.de>
18325 * doc/invoke.texi (fwhole-program): Discourage use in combination
18328 2013-03-06 Jakub Jelinek <jakub@redhat.com>
18330 * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
18332 PR tree-optimization/56539
18333 * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
18334 instead of GSI_CONTINUE_LINKING as last argument to
18335 force_gimple_operand_gsi. Adjust function comment.
18337 * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
18340 PR middle-end/56548
18341 * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
18342 promoted mode, convert the result back to the original mode.
18344 2013-03-06 Richard Biener <rguenther@suse.de>
18346 PR middle-end/56294
18347 * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
18348 (insert_updated_phi_nodes_compare_uids): New function.
18349 (update_ssa): Sort symbols_to_rename after UID before
18350 traversing it to insert PHI nodes.
18352 2013-03-06 Richard Biener <rguenther@suse.de>
18354 PR middle-end/50494
18355 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
18356 Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
18359 2013-02-13 Richard Biener <rguenther@suse.de>
18362 * varasm.c (output_constant_def_1): Get the decl representing
18363 the constant as argument.
18364 (output_constant_def): Wrap output_constant_def_1.
18365 (make_decl_rtl): Use output_constant_def_1 with the decl
18366 representing the constant.
18367 (build_constant_desc): Optionally re-use a decl already
18368 representing the constant.
18369 (tree_output_constant_def): Adjust.
18371 2013-03-06 Joey Ye <joey.ye@arm.com>
18374 * gcc.c (convert_white_space): New function.
18375 (main): Handles white space in function name.
18377 2013-03-06 Oleg Endo <olegendo@gcc.gnu.org>
18380 * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
18381 instead of TARGET_SH2 for call-table case. Do not set sh_div_strategy
18382 to SH_DIV_CALL_TABLE for TARGET_SH2.
18383 * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
18385 * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
18386 call-table options.
18388 2013-03-05 Sterling Augustine <saugustine@google.com>
18389 Cary Coutant <ccoutant@google.com>
18392 * dwarf2out.c (resolve_addr): Don't call
18393 remove_loc_list_addr_table_entries a second time for the same
18396 2013-03-05 Jakub Jelinek <jakub@redhat.com>
18399 * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
18400 (avoid_complex_debug_insns): New function.
18401 (expand_debug_locations): Call it.
18403 PR rtl-optimization/56484
18404 * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
18405 lifetimes of hard registers on small register class machines.
18407 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
18409 * config/microblaze/microblaze-protos.h: Rename
18410 microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
18411 * config/microblaze/microblaze.c (microblaze_attribute_table): Add
18413 (microblaze_fast_interrupt_function_p): New function.
18414 (microblaze_is_interrupt_handler): Rename to
18415 microblaze_is_interrupt_variant and add fast_interrupt check.
18416 (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
18417 (save_restore_insns): Likewise.
18418 (compute_frame_size): Likewise.
18419 (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
18420 (microblaze_globalize_label): Likewise.
18421 * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
18422 * config/microblaze/microblaze.md: Use wrapper
18423 microblaze_is_interrupt_variant.
18425 2013-03-05 Kai Tietz <ktietz@redhat.com>
18427 * sdbout.c (sdbout_one_type): Switch to current function's section
18428 supporting cold/hot.
18430 2013-03-05 David Holsgrove <david.holsgrove@xilinx.com>
18432 * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
18435 2013-03-05 Jakub Jelinek <jakub@redhat.com>
18437 PR middle-end/56461
18438 * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
18439 if VALGRIND_GET_VBITS is defined, temporarily make object
18440 memory all defined, and restore previous valgrind addressability
18441 and definability afterwards. Free this_object at the end.
18443 PR middle-end/56461
18444 * lra.c (lra): Call lra_clear_live_ranges if live_p,
18445 right before calling lra_create_live_ranges, also call it
18446 when clearing live_p. Only call lra_clear_live_ranges
18447 at the end if live_p.
18449 PR middle-end/56461
18450 * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
18452 2013-03-05 Richard Biener <rguenther@suse.de>
18454 PR tree-optimization/56521
18455 * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
18458 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
18461 * except.h (remove_unreachable_eh_regions): New prototype.
18462 * except.c (remove_eh_handler_splicer): New function, split out
18463 of remove_eh_handler.
18464 (remove_eh_handler): Use remove_eh_handler_splicer. Add comment
18465 warning about running it on many EH regions one at a time.
18466 (remove_unreachable_eh_regions_worker): New function, walk the
18467 EH tree in depth-first order and remove non-marked regions.
18468 (remove_unreachable_eh_regions): New function.
18469 * tree-eh.c (mark_reachable_handlers): New function, split out
18470 from remove_unreachable_handlers.
18471 (remove_unreachable_handlers): Use mark_reachable_handlers and
18472 remove_unreachable_eh_regions.
18473 (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
18474 and remove_unreachable_eh_regions.
18476 2013-03-05 Richard Biener <rguenther@suse.de>
18478 PR middle-end/56525
18479 * loop-init.c (fix_loop_structure): Remove loops in two stages,
18480 not freeing them until the end.
18482 2013-03-05 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
18484 * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
18486 2013-03-05 Richard Biener <rguenther@suse.de>
18488 PR tree-optimization/56270
18489 * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
18490 of loads after scheduling an SLP instance.
18492 2013-03-05 Jakub Jelinek <jakub@redhat.com>
18494 * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
18496 (check_gcc_parallelize): Run guality.exp as a separate job from
18497 vect.exp with unsorted.exp and $(dg_target_exps) separately from
18498 struct-layout-1.exp with stackalign.exp.
18500 * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
18502 PR middle-end/56461
18503 * tree-vect-slp.c (vect_supported_load_permutation_p): Free
18504 load_index sbitmap even if some bit in it isn't set.
18506 PR middle-end/56461
18507 * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
18508 (discover_iteration_bound_by_body_walk): Change queues to
18509 vec<vec<basic_block> > and queue to vec<basic_block>. Fix up
18510 spelling in comment. Call safe_push on queues[bound_index] directly.
18511 Release queues[queue_index] in every iteration unconditionally.
18512 Release bounds vector.
18514 PR middle-end/56461
18515 * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
18516 free_stmt_vec_info on any left-over stmt_vec_info in the vector.
18517 * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
18520 2013-03-05 Richard Biener <rguenther@suse.de>
18523 * tree-inline.c (remap_blocks_to_null): New function.
18524 (expand_call_inline): When expanding a call stmt without
18525 an associated block inline remap all callee blocks to NULL.
18527 2013-03-05 Jakub Jelinek <jakub@redhat.com>
18529 PR rtl-optimization/56494
18530 * simplify-rtx.c (simplify_truncation): If C is narrower than A,
18531 optimize (truncate:A (subreg:B (truncate:C X) 0)) into
18532 (subreg:A (truncate:C X) 0) instead of (truncate:A X).
18534 PR middle-end/56461
18535 * sel-sched-ir.c (free_sched_pools): Release
18536 succs_info_pool.stack[succs_info_pool.max_top] vectors too
18537 if succs_info_pool.max_top isn't -1.
18540 * opts.c (opts_obstack, opts_concat): Moved to...
18541 * opts-common.c (opts_obstack, opts_concat): ... here.
18543 2013-03-04 Jakub Jelinek <jakub@redhat.com>
18545 PR middle-end/56461
18546 * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
18548 2013-03-04 Martin Jambor <mjambor@suse.cz>
18550 * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
18551 all appropriate places.
18553 2013-01-04 Eric Botcazou <ebotcazou@adacore.com>
18555 PR tree-optimization/56424
18556 * ipa-split.c (split_function): Do not set the RSO flag if result is
18557 not by reference and its type is a register type.
18559 2013-03-04 David Holsgrove <david.holsgrove@xilinx.com>
18561 * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
18562 (microblaze_legitimate_pic_operand): Likewise
18563 * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
18564 new function microblaze_legitimate_pic_operand
18565 * config/microblaze/microblaze-protos.h
18566 (microblaze_legitimate_pic_operand): Declare.
18568 2013-03-04 Edgar E. Iglesias <edgar.iglesias@gmail.com>
18570 * config/microblaze/predicates.md (call_insn_simple_operand):
18571 New predicate for supported rtx code types.
18572 * config/microblaze/microblaze.md (call_internal1): Use
18573 call_insn_simple_operand predicate.
18575 2013-03-04 Jakub Jelinek <jakub@redhat.com>
18577 PR middle-end/56461
18578 * tree-loop-distribution.c (ldist_gen): Call partition_free after each
18579 partitions.ordered_remove.
18581 PR middle-end/56461
18582 * tree-vect-stmts.c (vectorizable_conversion): Don't call
18583 vec_oprnds0.create (1) for modifier == NONE.
18585 PR middle-end/56461
18586 * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
18587 on vec_oprnds0 or vec_oprnds1 before loop, only call it on
18588 vec_oprnds1 right before pushing anything to it for
18591 PR middle-end/56461
18592 * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
18593 set nbbs to 0 instead of having separate code path.
18594 (vect_analyze_loop_form): Call destroy_loop_vec_info with true
18595 instead of false as last argument if returning NULL.
18597 2013-03-03 Sandra Loosemore <sandra@codesourcery.com>
18599 * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
18600 the attribute is now called "target" instead of "option".
18601 (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
18602 * doc/tm.texi.in (Target Attributes): Likewise document the correct
18603 attribute/pragma name for TARGET_OPTION_VALID_P and
18604 TARGET_OPTION_PRAGMA_PARSE. Also copy-edit and correct markup.
18605 * doc/tm.texi: Regenerated.
18607 2013-03-02 David Holsgrove <david.holsgrove@xilinx.com>
18609 * config/microblaze/microblaze.c:
18610 Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
18611 * config/microblaze/microblaze.h: Add -mxl-reorder to
18613 * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
18614 instructions emitted if TARGET_REORDER.
18615 * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
18616 or 0 for -m/-mno case, but initialises as 2 to detect default use case
18619 2013-03-01 Xinliang David Li <davidxl@google.com>
18621 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
18624 2013-03-01 Jakub Jelinek <jakub@redhat.com>
18626 PR middle-end/56461
18627 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
18628 vector even when returning true. Fix up function comment formatting.
18630 PR middle-end/56461
18631 * ira-build.c (ira_loop_nodes_count): New variable.
18632 (create_loop_tree_nodes): Initialize it.
18633 (finish_loop_tree_nodes): Use it instead of looking at current_loops.
18635 PR middle-end/56461
18636 * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
18637 method on dr_chain and result_chain.
18638 * tree-vect-stmts.c (vectorizable_store): Only call
18639 result_chain.create if j == 0.
18641 PR middle-end/56461
18642 * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
18643 vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
18644 before overwriting it.
18646 2013-03-01 Tobias Burnus <burnus@net-b.de>
18648 * doc/extended.texi (C Extensions): Change order in @menu
18650 (Other MIPS Built-in Functions): Move last MIPS entry before
18651 "picoChip Built-in Functions".
18652 (SH Built-in Functions): Move after RX Built-in Functions.
18653 * doc/gcc.texi (Introduction): Change order in @menu
18655 * doc/md.texi (Constraints): Ditto.
18656 * gty.texi (Type Information): Ditto.
18657 (User-provided marking routines for template types): Make
18659 * doc/invoke.texi (AArch64 Options): Move before
18660 "Adapteva Epiphany Options".
18662 2013-02-28 Konstantin Serebryany <konstantin.s.serebryany@gmail.com>
18663 Jakub Jelinek <jakub@redhat.com>
18666 * asan.c (gate_asan): Lookup no_sanitize_address instead of
18667 no_address_safety_analysis attribute.
18668 * doc/extend.texi (no_address_safety_attribute): Rename to
18669 no_sanitize_address attribute, mention no_address_safety_analysis
18670 attribute as deprecated alias.
18672 2013-02-28 Jakub Jelinek <jakub@redhat.com>
18674 PR middle-end/56461
18675 * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
18676 type to vec<vec<tree> > *.
18677 * tree-vect-slp.c (vect_get_slp_defs): Likewise. Change vec_defs
18678 to be vec<tree> instead of vec<tree> *, set vec_defs
18679 to vNULL and call vec_defs.create (number_of_vects), adjust other
18681 * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
18682 vectorizable_condition): Adjust vect_get_slp_defs callers.
18684 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
18686 * config/aarch64/aarch64.c
18687 (aarch64_float_const_representable): Remove unused variable.
18689 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
18691 * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
18693 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
18695 * config/aarch64/aarch64-builtins.c
18696 (aarch64_init_simd_builtins): Make static.
18698 2013-02-28 James Greenhalgh <james.greenhalgh@arm.com>
18700 * config/aarch64/aarch64.c
18701 (aarch64_simd_make_constant): Make static.
18703 2013-02-28 Martin Jambor <mjambor@suse.cz>
18705 * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
18706 with no initialization to the RHS of debug statements.
18708 2013-02-28 Martin Jambor <mjambor@suse.cz>
18710 PR tree-optimization/56294
18711 * tree-sra.c (analyze_access_subtree): Create replacement declarations.
18713 (get_access_replacement): Do not call create_access_replacement.
18714 Assert a replacement exists.
18715 (get_repl_default_def_ssa_name): Create the replacement declaration
18718 2013-02-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18720 * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
18721 final_end_function.
18723 2013-02-28 Marek Polacek <polacek@redhat.com>
18725 PR rtl-optimization/56466
18726 * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
18727 if we're changing a loop.
18728 (peel_loops_completely): Likewise.
18730 2013-02-28 Paolo Carlini <paolo.carlini@oracle.com>
18733 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
18735 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
18738 * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
18739 macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
18740 INTX_FTYPE_FX, FX_FTYPE_INTX.
18741 * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
18743 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
18745 * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
18746 (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
18747 (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
18748 (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
18749 (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
18750 (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
18751 (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
18752 (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
18753 (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
18754 (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
18755 (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
18756 (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
18757 (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
18758 (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
18759 (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
18760 (avrxmega6): Increase max flash segments from 5 to 6.
18761 * config/avr/t-multilib: Regenerate.
18762 * config/avr/avr-tables.opt: Regenerate.
18763 * doc/avr-mmcu.texi: Regenerate.
18765 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
18767 * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
18768 (avr_device_to_arch): Rename to avr_device_to_ld.
18769 (avr_device_to_as): New prototype.
18770 (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
18771 (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
18772 * config/avr/driver-avr.c (avr_device_to_as): New.
18773 (avr_device_to_arch): Rename to avr_device_to_ld.
18775 2013-02-27 Jakub Jelinek <jakub@redhat.com>
18777 PR middle-end/56461
18778 * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
18779 method on dr_chain and result_chain.
18781 PR middle-end/56461
18782 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
18783 pointer_set_destroy on not_executed_last_iteration.
18785 PR middle-end/56461
18786 * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
18788 PR middle-end/56461
18789 * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
18790 FOR_EACH_DEFINED_FUNCTION when freeing state.
18792 PR middle-end/56461
18793 * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
18795 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
18798 PR middle-end/56461
18799 * ipa-cp.c (decide_whether_version_node): Call vec_free on
18800 known_aggs[i].items and release known_aggs vector.
18802 PR middle-end/56461
18803 * ipa-reference.c (propagate): Free node_info even for alias nodes.
18805 2013-02-27 Edgar E. Iglesias <edgar.iglesias@gmail.com>
18807 * config/microblaze/microblaze.c (microblaze_emit_compare):
18808 Use xor for EQ/NE comparisions.
18809 * config/microblaze/microblaze.md (cstoresf4): Add constraints
18810 (cbranchsf4): Adjust operator to comparison_operator.
18812 2013-02-27 Jakub Jelinek <jakub@redhat.com>
18814 PR middle-end/56461
18815 * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
18817 * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
18818 vec_safe_push, always update *slot.
18819 (redirect_edge_var_map_clear): Use vec_free.
18820 (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
18821 (free_var_map_entry): Use vec_free.
18822 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
18823 FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
18825 2013-02-27 Andrey Belevantsev <abel@ispras.ru>
18827 PR middle-end/45472
18828 * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
18829 when the may_trap_p bit of the exprs being merged differs.
18830 Reorder tests for speculativeness in the logical and operator.
18832 2013-02-27 Jakub Jelinek <jakub@redhat.com>
18834 * incpath.c (add_standard_paths): Use reconcat instead of concat
18835 where appropriate and avoid leaking memory.
18837 * opts.h: Include obstack.h.
18838 (opts_concat): New prototype.
18839 (opts_obstack): New declaration.
18840 * opts.c (opts_concat): New function.
18841 (opts_obstack): New variable.
18842 (init_options_struct): Call gcc_init_obstack on opts_obstack.
18843 (finish_options): Use opts_concat instead of concat
18844 and XOBNEWVEC instead of XNEWVEC.
18845 * opts-common.c (generate_canonical_option, decode_cmdline_option,
18846 generate_option): Likewise.
18847 * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
18848 * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
18851 * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
18852 and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
18854 2013-02-26 Jakub Jelinek <jakub@redhat.com>
18856 PR middle-end/56461
18857 * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
18859 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
18861 * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
18862 (arm_block_move_unaligned_straight): Likewise.
18863 (arm_adjust_block_mem): Likewise.
18865 2013-02-26 Joern Rennecke <joern.rennecke@embecosm.com>
18868 * config/lm32/lm32.c (gen_int_relational): Remove unused variables
18869 temp, cond and label.
18870 * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
18873 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
18874 * config/c6x/c6x.h (dbx_register_map): Update declaration.
18877 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
18878 of prologue/epilogue functions.
18881 * config/tilegx/tilegx.c (tilegx_expand_prologue):
18882 Remove unused variable cfa_offset.
18883 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
18886 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
18887 type promotion to unsigned.
18890 * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
18891 for HOST_WIDE_INT of 32 bit / same size as int.
18892 (arm_block_move_unaligned_straight): Likewise.
18893 (arm_adjust_block_mem): Likewise.
18896 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
18899 2013-02-26 Marek Polacek <polacek@redhat.com>
18901 PR tree-optimization/56426
18902 * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
18904 2013-02-26 Richard Biener <rguenther@suse.de>
18907 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
18908 unused variable loops.
18910 2013-02-26 Jakub Jelinek <jakub@redhat.com>
18912 PR tree-optimization/56448
18913 * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
18914 TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
18915 Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
18916 later operands of the references, or even first operand for
18917 INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
18919 PR tree-optimization/56443
18920 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
18921 overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
18922 to type_for_mode langhook.
18924 2013-02-25 Matt Turner <mattst88@gmail.com>
18926 * doc/invoke.texi: Document r4700.
18928 2013-02-25 Richard Biener <rguenther@suse.de>
18930 PR tree-optimization/56175
18931 * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
18933 (simplify_bitwise_binary): ... here. Also guard the conversion
18934 of (type) X op CST to (type) (X op ((type-x) CST)) with it.
18936 2013-02-25 Catherine Moore <clm@codesourcery.com>
18939 2013-02-24 Catherine Moore <clm@codesourcery.com>
18940 Maciej W. Rozycki <macro@codesourcery.com>
18941 Tom de Vries <tom@codesourcery.com>
18942 Nathan Sidwell <nathan@codesourcery.com>
18943 Iain Sandoe <iain@codesourcery.com>
18944 Nathan Froyd <froydnj@codesourcery.com>
18945 Chao-ying Fu <fu@mips.com>
18947 * doc/extend.texi: (micromips, nomicromips, nocompression):
18948 Document new function attributes.
18949 * doc/invoke.texi (minterlink-compressed, mmicromips,
18950 m14k, m14ke, m14kec): Document new options.
18951 (minterlink-mips16): Update documentation.
18952 * doc/md.texi (ZC, ZD): Document new constraints.
18953 * configure.ac (gcc_cv_as_micromips): Check if linker
18954 supports the .set micromips directive.
18955 * configure: Regenerate.
18956 * config.in: Regenerate.
18957 * config/mips/mips-tables.opt: Regenerate.
18958 * config/mips/micromips.md: New file.
18959 * constraints.md (ZC, AD): New constraints.
18960 * config/mips/predicates.md (movep_src_register): New predicate.
18961 (movep_src_operand): New predicate.
18962 (non_volatile_mem_operand): New predicate.
18963 * config/mips/mips.md (multimem): New type.
18964 (length): Differentiate between 17-bit and 18-bit branch offsets.
18965 (MOVEP1, MOVEP2): New mode iterator.
18966 (mov_<load>l): Use ZC constraint.
18967 (mov_<load>r): Likewise.
18968 (mov_<store>l): Likewise.
18969 (mov_<store>r): Likewise.
18970 (*branch_equality<mode>_inverted): Add microMIPS support.
18971 (*branch_equality<mode>): Likewise.
18972 (*jump_absolute): Likewise.
18973 (indirect_jump_<mode>): Likewise.
18974 (tablejump_<mode>): Likewise.
18975 (<optab>_internal): Likewise.
18976 (sibcall_internal): Likewise.
18977 (sibcall_value_internal): Likewise.
18978 (prefetch): Use constraint ZD.
18979 * config/mips/mips.opt (minterlink-compressed): New option.
18980 (minterlink-mips16): Now an alias for minterlink-compressed.
18981 (mmicromips): New option.
18982 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
18983 (compare_and_swap_12): Likewise.
18984 (sync_add<mode>): Likewise.
18985 (sync_<optab>_12): Likewise.
18986 (sync_old_<optab>_12): Likewise.
18987 (sync_new_<optab>_12): Likewise.
18988 (sync_nand_12): Likewise.
18989 (sync_old_nand_12): Likewise.
18990 (sync_new_nand_12): Likewise.
18991 (sync_sub<mode>): Likewise.
18992 (sync_old_add<mode>): Likewise.
18993 (sync_old_sub<mode>): Likewise.
18994 (sync_new_add<mode>): Likewise.
18995 (sync_new_sub<mode>): Likewise.
18996 (sync_<optab><mode>): Likewise.
18997 (sync_old_<optab><mode>): Likewise.
18998 (sync_new_<optab><mode>): Likewise.
18999 (sync_nand<mode>): Likewise.
19000 (sync_old_nand<mode>): Likewise.
19001 (sync_new_nand<mode>): Likewise.
19002 (sync_lock_test_and_set<mode>): Likewise.
19003 (test_and_set_12): Likewise.
19004 (atomic_compare_and_swap<mode>): Likewise.
19005 (atomic_exchange<mode>_llsc): Likewise.
19006 (atomic_fetch_add<mode>_llsc): Likewise.
19007 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19008 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19009 (umips_save_restore_pattern_p): Likewise.
19010 (umips_load_store_pair_p): Likewise.
19011 (umips_output_load_store_pair): Likewise.
19012 (umips_movep_target_p): Likewise.
19013 (umips_12bit_offset_address_p): Likewise.
19014 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19015 (mips_base_mips16): Rename this...
19016 (mips_base_compression_flags): ...to this. Update all uses.
19017 (mips_attribute_table): Add micromips, nomicromips and nocompression.
19018 (mips_mips16_decl_p): Delete.
19019 (mips_nomips16_decl_p): Delete.
19020 (mips_get_compress_on_flags): New function.
19021 (mips_get_compress_off_flags): New function.
19022 (mips_get_compress_mode): New function.
19023 (mips_get_compress_on_name): New function.
19024 (mips_get_compress_off_name): New function.
19025 (mips_insert_attributes): Support multiple compression types.
19026 (mips_merge_decl_attributes): Likewise.
19027 (umips_12bit_offset_address_p): New function.
19028 (mips_start_function_definition): Emit .set micromips directive.
19029 (mips_call_may_need_jalx_p): New function.
19030 (mips_function_ok_for_sibcall): Add microMIPS support.
19031 (mips_print_operand_punctuation): Support short delay slots and
19033 (umips_swm_mask, umips_swm_encoding): New.
19034 (umips_build_save_restore): New function.
19035 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19036 (was_mips16_p): Remove.
19037 (old_compression_mode): New.
19038 (mips_set_compression_mode): New function.
19039 (mips_set_current_function): Add microMIPS support.
19040 (mips_option_override): Likewise.
19041 (umips_save_restore_pattern_p): New function.
19042 (umips_output_save_restore): New function.
19043 (umips_load_store_pair_p_1): New function.
19044 (umips_load_store_pair_p): New function.
19045 (umips_output_load_store_pair_1): New function.
19046 (umips_output_load_store_pair): New function.
19047 (umips_movep_target_p) New function.
19048 (mips_prepare_pch_save): Add microMIPS support.
19049 * config/mips/mips.h (TARGET_COMPRESSION): New.
19050 (TARGET_CPU_CPP_BUILTINS): Update macro
19051 to use new compression flags and to support microMIPS.
19052 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19053 (MIPS_ARCH_FLOAT_SPEC): Likewise.
19054 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19055 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19056 (ASM_SPEC): Support mmicromips and mno-micromips.
19057 (M16STORE_REG_P): New macro.
19058 (MIPS_CALL): Support TARGET_MICROMIPS.
19059 (MICROMIPS_J): New macro.
19060 (mips_base_mips16): Rename this...
19061 (mips_base_compression_flags): ...to this.
19062 (UMIPS_12BIT_OFFSET_P): New macro.
19063 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19064 (MULTILIB_DIRNAMES): Likewise.
19066 2013-02-25 Tom de Vries <tom@codesourcery.com>
19068 PR rtl-optimization/56131
19069 * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
19070 * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
19071 NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
19073 2013-02-25 Tobias Burnus <burnus@net-b.de>
19075 * doc/invoke.texi (-fsanitize=): Move from optimization
19076 to debugging options.
19078 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
19080 * sched-deps.c (sched_analyze_insn): Fix typo in comment.
19082 2013-02-25 Andrey Belevantsev <abel@ispras.ru>
19083 Alexander Monakov <amonakov@ispras.ru>
19085 PR middle-end/56077
19086 * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
19087 flush pending lists also on non-jumps. Adjust comment.
19089 2013-02-24 Catherine Moore <clm@codesourcery.com>
19090 Maciej W. Rozycki <macro@codesourcery.com>
19091 Tom de Vries <tom@codesourcery.com>
19092 Nathan Sidwell <nathan@codesourcery.com>
19093 Iain Sandoe <iain@codesourcery.com>
19094 Nathan Froyd <froydnj@codesourcery.com>
19095 Chao-ying Fu <fu@mips.com>
19097 * doc/extend.texi: (micromips, nomicromips, nocompression):
19098 Document new function attributes.
19099 * doc/invoke.texi (minterlink-compressed, mmicromips,
19100 m14k, m14ke, m14kec): Document new options.
19101 (minterlink-mips16): Update documentation.
19102 * doc/md.texi (ZC, ZD): Document new constraints.
19103 * configure.ac (gcc_cv_as_micromips): Check if linker
19104 supports the .set micromips directive.
19105 * configure: Regenerate.
19106 * config.in: Regenerate.
19107 * config/mips/mips-tables.opt: Regenerate.
19108 * config/mips/micromips.md: New file.
19109 * constraints.md (ZC, AD): New constraints.
19110 * config/mips/predicates.md (movep_src_register): New predicate.
19111 (movep_src_operand): New predicate.
19112 (non_volatile_mem_operand): New predicate.
19113 * config/mips/mips.md (multimem): New type.
19114 (length): Differentiate between 17-bit and 18-bit branch offsets.
19115 (MOVEP1, MOVEP2): New mode iterator.
19116 (mov_<load>l): Use ZC constraint.
19117 (mov_<load>r): Likewise.
19118 (mov_<store>l): Likewise.
19119 (mov_<store>r): Likewise.
19120 (*branch_equality<mode>_inverted): Add microMIPS support.
19121 (*branch_equality<mode>): Likewise.
19122 (*jump_absolute): Likewise.
19123 (indirect_jump_<mode>): Likewise.
19124 (tablejump_<mode>): Likewise.
19125 (<optab>_internal): Likewise.
19126 (sibcall_internal): Likewise.
19127 (sibcall_value_internal): Likewise.
19128 (prefetch): Use constraint ZD.
19129 * config/mips/mips.opt (minterlink-compressed): New option.
19130 (minterlink-mips16): Now an alias for minterlink-compressed.
19131 (mmicromips): New option.
19132 * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
19133 (compare_and_swap_12): Likewise.
19134 (sync_add<mode>): Likewise.
19135 (sync_<optab>_12): Likewise.
19136 (sync_old_<optab>_12): Likewise.
19137 (sync_new_<optab>_12): Likewise.
19138 (sync_nand_12): Likewise.
19139 (sync_old_nand_12): Likewise.
19140 (sync_new_nand_12): Likewise.
19141 (sync_sub<mode>): Likewise.
19142 (sync_old_add<mode>): Likewise.
19143 (sync_old_sub<mode>): Likewise.
19144 (sync_new_add<mode>): Likewise.
19145 (sync_new_sub<mode>): Likewise.
19146 (sync_<optab><mode>): Likewise.
19147 (sync_old_<optab><mode>): Likewise.
19148 (sync_new_<optab><mode>): Likewise.
19149 (sync_nand<mode>): Likewise.
19150 (sync_old_nand<mode>): Likewise.
19151 (sync_new_nand<mode>): Likewise.
19152 (sync_lock_test_and_set<mode>): Likewise.
19153 (test_and_set_12): Likewise.
19154 (atomic_compare_and_swap<mode>): Likewise.
19155 (atomic_exchange<mode>_llsc): Likewise.
19156 (atomic_fetch_add<mode>_llsc): Likewise.
19157 * config/mips/mips-cpus.def (m14kc, m14k): New processors.
19158 * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
19159 (umips_save_restore_pattern_p): Likewise.
19160 (umips_load_store_pair_p): Likewise.
19161 (umips_output_load_store_pair): Likewise.
19162 (umips_movep_target_p): Likewise.
19163 (umips_12bit_offset_address_p): Likewise.
19164 * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
19165 (mips_base_mips16): Rename this...
19166 (mips_base_compression_flags): ...to this. Update all uses.
19167 (mips_attribute_table): Add micromips, nomicromips and nocompression.
19168 (mips_mips16_decl_p): Delete.
19169 (mips_nomips16_decl_p): Delete.
19170 (mips_get_compress_on_flags): New function.
19171 (mips_get_compress_off_flags): New function.
19172 (mips_get_compress_mode): New function.
19173 (mips_get_compress_on_name): New function.
19174 (mips_get_compress_off_name): New function.
19175 (mips_insert_attributes): Support multiple compression types.
19176 (mips_merge_decl_attributes): Likewise.
19177 (umips_12bit_offset_address_p): New function.
19178 (mips_start_function_definition): Emit .set micromips directive.
19179 (mips_call_may_need_jalx_p): New function.
19180 (mips_function_ok_for_sibcall): Add microMIPS support.
19181 (mips_print_operand_punctuation): Support short delay slots and
19183 (umips_swm_mask, umips_swm_encoding): New.
19184 (umips_build_save_restore): New function.
19185 (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
19186 (was_mips16_p): Remove.
19187 (old_compression_mode): New.
19188 (mips_set_compression_mode): New function.
19189 (mips_set_current_function): Add microMIPS support.
19190 (mips_option_override): Likewise.
19191 (umips_save_restore_pattern_p): New function.
19192 (umips_output_save_restore): New function.
19193 (umips_load_store_pair_p_1): New function.
19194 (umips_load_store_pair_p): New function.
19195 (umips_output_load_store_pair_1): New function.
19196 (umips_output_load_store_pair): New function.
19197 (umips_movep_target_p) New function.
19198 (mips_prepare_pch_save): Add microMIPS support.
19199 * config/mips/mips.h (TARGET_COMPRESSION): New.
19200 (TARGET_CPU_CPP_BUILTINS): Update macro
19201 to use new compression flags and to support microMIPS.
19202 (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
19203 (MIPS_ARCH_FLOAT_SPEC): Likewise.
19204 (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
19205 (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
19206 (ASM_SPEC): Support mmicromips and mno-micromips.
19207 (M16STORE_REG_P): New macro.
19208 (MIPS_CALL): Support TARGET_MICROMIPS.
19209 (MICROMIPS_J): New macro.
19210 (mips_base_mips16): Rename this...
19211 (mips_base_compression_flags): ...to this.
19212 (UMIPS_12BIT_OFFSET_P): New macro.
19213 * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
19214 (MULTILIB_DIRNAMES): Likewise.
19216 2013-02-24 Jakub Jelinek <jakub@redhat.com>
19219 * target-globals.c (save_target_globals): For init_reg_sets and
19220 target_reinit remporarily set this_fn_optabs to this_target_optabs.
19222 2013-02-22 James Grennahlgh <james.greenhalgh@arm.com>
19224 * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
19225 * config/aarch64/t-aarch64
19226 (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
19228 2013-02-22 Vladimir Makarov <vmakarov@redhat.com>
19230 PR inline-asm/56148
19231 * lra-constraints.c (process_alt_operands): Reload operand
19232 conflicting with earlier clobber only if no more other conflicting
19235 2013-02-22 Jakub Jelinek <jakub@redhat.com>
19238 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
19239 if not linking a shared library.
19241 2013-02-22 Seth LaForge <sethml@google.com>
19243 * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
19245 2013-02-22 Greta Yorsh <Greta.Yorsh@arm.com>
19247 * config/arm/arm.md (split for extendsidi): Update condition.
19248 (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
19249 * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
19250 (qhs_zextenddi_cstr): Likewise.
19252 2013-02-21 Jakub Jelinek <jakub@redhat.com>
19254 PR middle-end/56420
19255 * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
19256 avoid signed wrapping.
19257 (expand_mult): Handle properly multiplication by
19258 ((dword_type) -1) << (BITS_PER_WORD - 1). Improve multiplication by
19259 ((dword_type) 1) << (BITS_PER_WORD - 1). Avoid undefined behavior
19260 in the compiler if coeff is HOST_WIDE_INT_MIN.
19261 (expand_divmod): Don't make ext_op1 static, change it's type to
19262 uhwi. Avoid undefined behavior in -INTVAL (op1).
19264 PR rtl-optimization/50339
19265 * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
19267 * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
19268 (compute_costs): Call compute_splitting_shift also for ASHIFTRT
19269 into splitting_ashiftrt field.
19270 (find_decomposable_shift_zext, resolve_shift_zext): Handle also
19272 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
19275 2013-02-20 Aldy Hernandez <aldyh@redhat.com>
19277 PR middle-end/56108
19278 * trans-mem.c (execute_tm_mark): Do not expand transactions that
19279 are sure to go irrevocable.
19281 2013-02-21 Hans-Peter Nilsson <hp@axis.com>
19283 * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
19284 scalars are valid operands.
19286 2013-02-21 Martin Jambor <mjambor@suse.cz>
19288 PR tree-optimization/56310
19289 * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
19290 only matching indices and non-negative final offsets.
19291 (intersect_aggregates_with_edge): Pass src_idx to
19292 agg_replacements_to_vector. Pass src_idx insstead of index to
19293 intersect_with_agg_replacements.
19295 2013-02-21 Martin Jambor <mjambor@suse.cz>
19297 * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
19298 instead of hard-wired defaults.
19300 2013-02-21 Maciej W. Rozycki <macro@codesourcery.com>
19302 * doc/invoke.texi (MIPS Options): Update documentation of the
19303 floating-point multiply-accumulate instruction restrictions.
19305 2013-02-21 Kostya Serebryany <kcc@google.com>
19307 * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
19308 asan_shadow_offset on x86_64 linux.
19310 2013-02-21 Richard Biener <rguenther@suse.de>
19312 PR tree-optimization/56415
19314 2013-02-11 Richard Biener <rguenther@suse.de>
19316 PR tree-optimization/56273
19317 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19320 2013-02-21 Jakub Jelinek <jakub@redhat.com>
19323 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
19326 PR inline-asm/56405
19327 * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
19328 use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
19330 2013-02-20 Jan Hubicka <jh@suse.cz>
19332 PR tree-optimization/56265
19333 * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
19334 when target is referenced for first time.
19336 2013-02-20 Richard Biener <rguenther@suse.de>
19338 * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
19339 * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
19340 * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
19341 * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
19342 not return anything.
19343 (rename_ssa_copies): Do not remove unused locals.
19344 * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
19345 * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
19346 * passes.c (execute_function_todo): Do not schedule unused locals
19347 removal if cleanup_tree_cfg did something.
19348 * tree-ssa-live.c (remove_unused_locals): Dump statistics
19349 about the number of removed locals.
19351 2013-02-20 Richard Biener <rguenther@suse.de>
19353 PR tree-optimization/56398
19354 * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
19356 2013-02-20 Martin Jambor <mjambor@suse.cz>
19358 PR tree-optimization/55334
19359 * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
19360 restricted pointers to arrays.
19362 2013-02-20 Richard Biener <rguenther@suse.de>
19363 Jakub Jelinek <jakub@redhat.com>
19365 PR tree-optimization/56396
19366 * tree-ssa-ccp.c (n_const_val): New static variable.
19367 (get_value): Return NULL for SSA names we don't have a lattice
19369 (ccp_initialize): Initialize n_const_val.
19370 * tree-ssa-copy.c (n_copy_of): New static variable.
19371 (init_copy_prop): Initialize n_copy_of.
19372 (get_value): Return NULL_TREE for SSA names we don't have a
19375 2013-02-20 Martin Jambor <mjambor@suse.cz>
19377 * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
19379 2013-02-20 Richard Biener <rguenther@suse.de>
19381 * genpreds.c (write_lookup_constraint): Do not compare first
19382 letter of the constraint again.
19384 2013-02-20 Richard Biener <rguenther@suse.de>
19386 * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
19388 (get_use_iv_cost): Terminate hashtable walk when coming across
19391 2013-02-20 Igor Zamyatin <igor.zamyatin@intel.com>
19393 * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
19394 reassociation for avx2 targets.
19396 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
19398 * config/microblaze/microblaze.c: microblaze_has_clz = 0
19399 Add version check for v8.10.a to enable microblaze_has_clz
19400 * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
19401 version and TARGET_PATTERN_COMPARE check
19402 * config/microblaze/microblaze.md: New clzsi2 instruction
19404 2012-02-19 Edgar E. Iglesias <edgar.iglesias@gmail.com>
19406 * config/microblaze/microblaze.md (call_value_intern): Check symbol is
19407 function before branching.
19409 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
19411 * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
19413 (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
19415 2012-02-19 Andrey Belevantsev <abel@ispras.ru>
19417 PR middle-end/55889
19418 * sel-sched.c: Include ira.h.
19419 (implicit_clobber_conflict_p): New function.
19420 (moveup_expr): Use it.
19421 * Makefile.in (sel-sched.o): Depend on ira.h.
19423 2013-02-19 Richard Biener <rguenther@suse.de>
19425 PR tree-optimization/56384
19426 * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
19427 (vn_hash_type): Split out from ...
19428 (vn_hash_constant_with_type): ... here.
19429 * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
19430 (vn_phi_eq): Compare types from vn_phi_s structure.
19431 (vn_phi_lookup): Populate vn_phi_s type.
19432 (vn_phi_insert): Likewise.
19434 2013-02-19 Jakub Jelinek <jakub@redhat.com>
19436 PR tree-optimization/56350
19437 * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
19438 if haven't found reduction or nested cycle operand, rather than
19439 asserting we must find it.
19441 PR tree-optimization/56381
19442 * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
19445 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
19446 Jakub Jelinek <jakub@redhat.com>
19449 * genopinit.c (raw_optab_handler): Use this_fn_optabs.
19450 (swap_optab_enable): Same.
19451 (init_all_optabs): Use argument instead of global.
19452 * tree.h (struct tree_optimization_option): New field target_optabs.
19453 * expr.h (init_all_optabs): Add argument to prototype.
19454 (TREE_OPTIMIZATION_OPTABS): New.
19455 (save_optabs_if_changed): Protoize.
19456 * optabs.h: Declare this_fn_optabs.
19457 * optabs.c (save_optabs_if_changed): New.
19458 Declare this_fn_optabs.
19459 (init_optabs): Add argument to init_all_optabs() call.
19460 * function.c (invoke_set_current_function_hook): Handle per
19462 * function.h (struct function): New field optabs.
19463 * config/mips/mips.c (mips_set_mips16_mode): Handle when
19464 optimization_current_node has changed.
19465 * target-globals.h (save_target_globals_default_opts): Protoize.
19466 * target-globals.c (save_target_globals_default_opts): New.
19468 2013-02-18 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19471 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
19472 registers %fr12 and %fr12R as call used.
19475 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
19476 and HImode, require all displacements to be an integer multiple of
19478 * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
19479 only allow QImode and HImode when reload is in progress and strict is
19480 true. Likewise for symbolic addresses. Use base14_operand to check
19481 displacements in REG+BASE addresses.
19483 2013-02-18 Richard Biener <rguenther@suse.de>
19485 PR tree-optimization/56366
19486 * tree-vect-loop.c (get_initial_def_for_induction): Properly
19487 handle sign-conversion of outer-loop initial induction value.
19489 2013-02-18 Richard Biener <rguenther@suse.de>
19491 PR middle-end/56349
19492 * cfghooks.c (merge_blocks): If we merge a latch into another
19493 block adjust references to it.
19494 * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
19495 (verify_loop_structure): Verify that a recorded latch is in fact
19498 2013-02-18 Richard Biener <rguenther@suse.de>
19500 PR tree-optimization/56321
19501 * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
19502 order SSA name release and virtual operand unlinking.
19504 2013-02-17 Edgar E. Iglesias <edgar.iglesias@gmail.com>
19506 * config/microblaze/microblaze.md (save_stack_block): Define.
19507 (restore_stack_block): Likewise.
19509 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
19511 * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
19512 * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
19513 * config/microblaze/microblaze.c (microblaze_option_override):
19514 Bail out early for PIC modes when target does not support PIC.
19516 2013-02-16 Edgar E. Iglesias <edgar.iglesias@gmail.com>
19518 * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
19519 Replace with a microblaze version.
19520 (microblaze_trampoline_init): Adapt for microblaze.
19521 * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
19524 2013-02-16 Jakub Jelinek <jakub@redhat.com>
19525 Dodji Seketeli <dodji@redhat.com>
19528 * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
19529 (instrument_mem_region_access): Do not forget to always put
19530 instrumentation of the of 'base' and 'base + len' in a "if (len !=
19531 0) statement, even for cases where either 'base' or 'base + len'
19532 are not instrumented -- because they have been previously
19533 instrumented. Simplify the logic by putting all the statements
19534 instrument 'base + len' inside a sequence, and then insert that
19535 sequence right before the current insertion point. Then, to
19536 instrument 'base + len', just get an iterator on that statement.
19537 And do not forget to update the pointer to iterator the function
19538 received as argument.
19540 2013-02-15 Vladimir Makarov <vmakarov@redhat.com>
19542 PR rtl-optimization/56348
19543 * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
19545 2013-02-15 Steven Bosscher <steven@gcc.gnu.org>
19547 * graph.c (start_graph_dump): Print dumpfile base as digraph label.
19548 (clean_graph_dump_file): Pass base to start_graph_dump.
19550 2013-02-14 Richard Henderson <rth@redhat.com>
19553 * lower-subreg.c (simple_move): Check dest mode instead of src mode.
19555 2013-02-14 Steven Bosscher <steven@gcc.gnu.org>
19557 * collect2-aix.h: Define F_LOADONLY.
19559 2013-02-14 Richard Biener <rguenther@suse.de>
19562 * varasm.c (output_constant_def_1): Get the decl representing
19563 the constant as argument.
19564 (output_constant_def): Wrap output_constant_def_1.
19565 (make_decl_rtl): Use output_constant_def_1 with the decl
19566 representing the constant.
19567 (build_constant_desc): Optionally re-use a decl already
19568 representing the constant.
19569 (tree_output_constant_def): Adjust.
19571 2013-02-14 Dodji Seketeli <dodji@redhat.com>
19574 * asan.c (instrument_builtin_call): Really put the length of the
19575 second source argument into src1_len.
19577 2013-02-13 Jakub Jelinek <jakub@redhat.com>
19579 * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
19580 argument. If it is false, don't create edge from then_bb to
19582 (insert_if_then_before_iter): Pass true to it.
19583 (build_check_stmt): Pass false to it.
19584 (transform_statements): Flush hash table only on extended basic
19585 block boundaries, rather than at the beginning of every bb.
19586 Don't flush hash table on nonfreeing_call_p calls.
19587 * tree-flow.h (nonfreeing_call_p): New prototype.
19588 * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
19590 2013-02-13 David S. Miller <davem@davemloft.net>
19592 * expmed.c (expand_shift_1): Only strip scalar integer subregs.
19594 2013-02-13 Vladimir Makarov <vmakarov@redhat.com>
19597 * ira.c (max_regno_before_ira): Move from ...
19599 (fix_reg_equiv_init): Use max_regno_before_ira instead of
19602 2013-02-13 Jakub Jelinek <jakub@redhat.com>
19604 * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
19606 2013-02-13 Richard Biener <rguenther@suse.de>
19609 * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
19610 globals in MEM_REFs.
19612 2013-02-13 Richard Biener <rguenther@suse.de>
19614 * loop-init.c (loop_optimizer_init): Clear loop state when
19615 re-initializing preserved loops.
19616 * loop-unswitch.c (unswitch_single_loop): Return whether
19617 we unswitched the loop. Do not verify loop state here.
19618 (unswitch_loops): When we unswitched a loop discover new loops.
19620 2013-02-13 Kostya Serebryany <kcc@google.com>
19622 * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
19624 * sanitizer.def: Rename __asan_init to __asan_init_v1.
19626 2013-02-12 Dodji Seketeli <dodji@redhat.com>
19628 Avoid instrumenting duplicated memory access in the same basic block
19629 * Makefile.in (asan.o): Add new dependency on hash-table.h
19630 * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
19631 (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
19632 (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
19633 (free_mem_ref_resources, has_mem_ref_been_instrumented)
19634 (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
19635 (get_mem_ref_of_assignment): New functions.
19636 (get_mem_refs_of_builtin_call): Extract from
19637 instrument_builtin_call and tweak a little bit to make it fit with
19639 (instrument_builtin_call): Use the new
19640 get_mem_refs_of_builtin_call. Use gimple_call_builtin_p instead
19641 of is_gimple_builtin_call.
19642 (instrument_derefs, instrument_mem_region_access): Insert the
19643 instrumented memory reference into the hash table.
19644 (maybe_instrument_assignment): Renamed instrument_assignment into
19645 this, and change it to advance the iterator when instrumentation
19646 actually happened and return true in that case. This makes it
19647 homogeneous with maybe_instrument_assignment, and thus give a
19648 chance to callers to be more 'regular'.
19649 (transform_statements): Clear the memory reference hash table
19650 whenever we enter a new BB, when we cross a function call, or when
19651 we are done transforming statements. Use
19652 maybe_instrument_assignment instead of instrumentation. No more
19653 need to special case maybe_instrument_assignment and advance the
19654 iterator after calling it; it's now handled just like
19655 maybe_instrument_call. Update comment.
19657 2013-02-13 Richard Biener <rguenther@suse.de>
19659 * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
19660 Fix loop discovery code.
19662 2013-02-12 Vladimir Makarov <vmakarov@redhat.com>
19664 PR inline-asm/56148
19665 * lra-constraints.c (process_alt_operands): Match early clobber
19666 operand with itself. Check conflicts with earlyclobber only if
19667 the operand is not reloaded. Prefer to reload conflicting operand
19668 if earlyclobber and matching operands are the same.
19670 2013-02-12 Richard Biener <rguenther@suse.de>
19673 * lto-streamer-out.c (write_symbol): Do not output symbols
19674 for hard register variables.
19676 2013-02-12 Georg-Johann Lay <avr@gjlay.de>
19679 * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
19680 (umulsidi3_insn, mulsidi3_insn): New insns.
19682 2013-02-12 Christophe Lyon <christophe.lyon@linaro.org>
19684 * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
19685 (struct tune_params): Add vec_costs field.
19686 * config/arm/arm.c (arm_builtin_vectorization_cost)
19687 (arm_add_stmt_cost): New functions.
19688 (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
19689 (TARGET_VECTORIZE_ADD_STMT_COST): Define.
19690 (arm_default_vec_cost): New struct of type cpu_vec_costs.
19691 (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
19692 (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
19693 (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
19694 (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
19696 2013-02-12 Richard Biener <rguenther@suse.de>
19699 * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
19700 decls again if possible.
19702 2013-02-12 Richard Biener <rguenther@suse.de>
19704 PR middle-end/56288
19705 * tree-ssa.c (verify_ssa_name): Fix check, move
19706 SSA_NAME_IN_FREE_LIST check up.
19708 2013-02-12 Jakub Jelinek <jakub@redhat.com>
19709 Steven Bosscher <steven@gcc.gnu.org>
19711 PR rtl-optimization/56151
19712 * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
19713 equal to op0 or op1, and last_insn pattern is CODE operation
19714 with MEM dest and one of the operands matches that MEM.
19716 2013-02-11 Sriraman Tallam <tmsriramgoogle.com>
19718 * doc/extend.texi: Document Function Multiversioning and "default"
19719 parameter string to target attribute.
19720 * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
19721 target attribute parameter is "default".
19722 (ix86_compare_version_priority): Remove checks for target attribute.
19723 (ix86_mangle_function_version_assembler_name): Change error to sorry.
19724 Remove check for target attribute equal to NULL. Add assert.
19725 (ix86_generate_version_dispatcher_body): Change error to sorry.
19727 2013-02-11 Iain Sandoe <iain@codesourcery.com>
19728 Jack Howarth <howarth@bromo.med.uc.edu>
19729 Patrick Marlier <patrick.marlier@gmail.com>
19732 * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
19733 define ENDFILE_SPEC as TM_DESTRUCTOR.
19734 * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
19736 2013-02-11 Alexander Potapenko <glider@google.com>
19737 Jack Howarth <howarth@bromo.med.uc.edu>
19738 Jakub Jelinek <jakub@redhat.com>
19741 * config/darwin.c (cdtor_record): Rename ctor_record.
19742 (sort_cdtor_records): Rename sort_ctor_records.
19743 (finalize_dtors): New routine to sort destructors by
19744 priority before use in assemble_integer.
19745 (machopic_asm_out_destructor): Use finalize_dtors if needed.
19747 2013-02-11 Uros Bizjak <ubizjak@gmail.com>
19749 PR rtl-optimization/56275
19750 * simplify-rtx.c (avoid_constant_pool_reference): Check that
19751 offset is non-negative and less than cmode size before
19752 calling simplify_subreg.
19754 2013-02-11 Richard Biener <rguenther@suse.de>
19756 PR tree-optimization/56264
19757 * cfgloop.h (fix_loop_structure): Adjust prototype.
19758 * loop-init.c (fix_loop_structure): Return the number of
19759 newly discovered loops.
19760 * tree-cfgcleanup.c (repair_loop_structures): When new loops
19761 are discovered, do a full loop-closed SSA rewrite.
19763 2013-02-11 Richard Biener <rguenther@suse.de>
19765 PR tree-optimization/56273
19766 * tree-vrp.c (simplify_cond_using_ranges): Disable for the
19768 (check_array_ref): Fix missing newline in dumps.
19769 (search_for_addr_array): Likewise.
19771 2013-02-09 David Edelsohn <dje.gcc@gmail.com>
19773 * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
19775 2013-02-09 Jakub Jelinek <jakub@redhat.com>
19778 * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
19780 2013-02-08 Vladimir Makarov <vmakarov@redhat.com>
19782 PR rtl-optimization/56246
19783 * lra-constraints.c (simplify_operand_subreg): Try to reuse
19785 * lra.c (lra): Clear lra_optional_reload_pseudos only when all
19786 constraints are satisfied.
19788 2013-02-08 Jeff Law <law@redhat.com>
19791 * emit-rtl.c (reg_is_parm_p): New function.
19792 * regs.h (reg_is_parm_p): New prototype.
19793 * ira-conflicts.c (ira_build_conflicts): Allow parameters in
19794 callee-clobbered registers.
19796 2013-02-08 Michael Meissner <meissner@linux.vnet.ibm.com>
19799 * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
19800 If there is no implicit builtin declaration, just return NULL.
19802 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
19804 * config/i386/sse.md (FMAMODEM): New mode iterator.
19805 (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
19806 mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
19808 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
19810 * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
19811 when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
19812 * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
19814 2013-02-08 Edgar E. Iglesias <edgar.iglesias@gmail.com>
19816 * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
19817 (microblaze*-*-elf): Likewise.
19818 * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
19820 * config/microblaze/microblaze-c.c: Add builtin defines for
19821 _LITTLE_ENDIAN and _BIG_ENDIAN.
19822 * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
19823 add to TARGET_DEFAULT flags.
19824 Expand ASM_SPEC and LINK_SPEC.
19825 Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
19826 * config/microblaze/microblaze.md: Update extendsidi2 and
19827 movdi_internal instructions to use low-order / high-order reg
19829 * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
19830 options and inversemask / mask of LITTLE_ENDIAN.
19831 * config/microblaze/t-microblaze: Expand multilib options to
19832 include mlittle-endian (le) and update exceptions patterns.
19834 2013-02-08 Jakub Jelinek <jakub@redhat.com>
19836 PR rtl-optimization/56195
19837 * lra-constraints.c (get_reload_reg): Don't reuse regs
19838 if they have smaller mode than requested, if they have
19839 wider mode than requested, try to return a SUBREG.
19841 PR tree-optimization/56250
19842 * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
19843 if type is unsigned and code isn't MULT_EXPR.
19845 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
19847 PR tree-optimization/56064
19848 * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
19849 bits according to mode.
19850 * fixed-value.h (fixed_from_double_int)
19851 (const_fixed_from_double_int): Adjust comments.
19853 2013-02-08 Richard Biener <rguenther@suse.de>
19856 * lto-streamer.h (struct data_in): Remove current_file, current_line
19857 and current_col members.
19858 * lto-streamer-out.c (lto_output_location): Stream changed bits
19859 en-block for efficiency.
19860 * lto-streamer-in.c (clear_line_info): Remove.
19861 (lto_input_location): Cache current file, line and column
19862 globally via local statics. Read changed bits en-block.
19863 (input_function): Do not call clear_line_info.
19864 (lto_read_body): Likewise.
19865 (lto_input_toplevel_asms): Likewise.
19867 2013-02-08 Michael Matz <matz@suse.de>
19869 PR tree-optimization/52448
19870 * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
19871 (nt_call_phase): New static.
19872 (add_or_mark_expr): Only mark accesses with newer phase than any
19874 (nonfreeing_call_p): New.
19875 (nt_init_block): Update nt_call_phase, mark blocks as visited.
19876 (nt_fini_block): Keep blocks marked as visited.
19877 (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
19879 2013-02-08 Richard Biener <rguenther@suse.de>
19881 * ira.c (ira): Free broken dominator information.
19883 2013-02-08 Uros Bizjak <ubizjak@gmail.com>
19885 * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
19887 2013-02-08 Marek Polacek <polacek@redhat.com>
19889 * cfgloop.c (verify_loop_structure): Add more checking of headers.
19891 2013-02-08 Richard Biener <rguenther@suse.de>
19893 PR middle-end/56181
19894 * cfgloop.h (flow_loops_find): Adjust.
19895 (bb_loop_header_p): Declare.
19896 * cfgloop.c (bb_loop_header_p): New function split out from ...
19897 (flow_loops_find): ... here. Adjust function signature,
19898 support incremental loop structure update.
19899 (verify_loop_structure): Cleanup. Verify a loop is a loop.
19900 * cfgloopmanip.c (fix_loop_structure): Move ...
19901 * loop-init.c (fix_loop_structure): ... here.
19902 (apply_loop_flags): Split out from ...
19903 (loop_optimizer_init): ... here.
19904 (fix_loop_structure): Use apply_loop_flags. Use flow_loops_find
19905 in incremental mode, only remove dead loops here.
19907 2013-02-08 Georg-Johann Lay <avr@gjlay.de>
19910 * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
19911 * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
19912 (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
19913 (*round<mode>3.libgcc): New insns for fixed-modes.
19914 * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
19915 (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
19916 (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
19917 * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
19918 implementations. Define to __builtin_avr_absFX,
19919 __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
19920 (roundFX, countlsFX): Define to __builtin_avr_roundFX,
19921 __builtin_avr_countlsFX, respectively.
19922 * config/avr/avr-c.c (target.h): Include it.
19923 (enum avr_builtin_id): New enum.
19924 (avr_resolve_overloaded_builtin): New static function.
19925 (avr_register_target_pragmas): Use it to set
19926 targetm.resolve_overloaded_builtin.
19927 * config/avr/avr.c (avr_init_builtins): Supply myriads of local
19928 tree nodes used by DEF_BUILTIN.
19929 (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
19930 (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
19931 <AVR_BUILTIN_xxBITS>: Same.
19933 2013-02-08 Richard Biener <rguenther@suse.de>
19935 * cfgloop.c (verify_loop_structure): Properly handle
19936 a loop exiting to another loop header.
19937 * ira-int.h (ira_loops): Remove.
19938 * ira.c (ira_loops): Remove.
19939 (ira): Use loop_optimizer_init and loop_optimizer_finalize.
19940 (do_reload): Use loop_optimizer_finalize.
19941 * ira-build.c (create_loop_tree_nodes): Use get_loops and
19942 number_of_loops to access the loop tree.
19943 (more_one_region_p): Likewise.
19944 (finish_loop_tree_nodes): Likewise.
19945 (rebuild_regno_allocno_maps): Likewise.
19946 (mark_loops_for_removal): Likewise.
19947 (mark_all_loops_for_removal): Likewise.
19948 (remove_unnecessary_regions): Likewise.
19949 (ira_build): Likewise.
19950 * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
19952 2013-02-08 Richard Biener <rguenther@suse.de>
19954 * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
19955 * ipa-pure-const.c (analyze_function): Avoid calling
19956 mark_irreducible_loops twice.
19957 * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
19959 2013-02-07 David S. Miller <davem@davemloft.net>
19961 * dwarf2out.c (based_loc_descr): Perform leaf register remapping
19963 * var-tracking.c (vt_add_function_parameter): Test the presence of
19964 HAVE_window_save properly and do not remap argument registers when
19965 we have a leaf function.
19967 2013-02-07 Uros Bizjak <ubizjak@gmail.com>
19970 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
19972 * config/i386/i386.c (ix86_print_operand): Ditto.
19974 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
19976 * lra-constraints.c (process_alt_operands): Fix recently added comment.
19978 2013-02-07 Vladimir Makarov <vmakarov@redhat.com>
19980 PR rtl-optimization/56225
19981 * lra-constraints.c (process_alt_operands): Check that reload hard
19982 reg can hold value for strict_low_part.
19984 2013-02-07 Jakub Jelinek <jakub@redhat.com>
19987 * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
19988 dwarf2out_end_function.
19989 (in_first_function_p, maybe_at_text_label_p,
19990 first_loclabel_num_not_at_text_label): New variables.
19991 (dwarf2out_var_location): In the first function find out
19992 lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
19993 (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
19996 2013-02-07 Eric Botcazou <ebotcazou@adacore.com>
19998 PR rtl-optimization/56178
19999 * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
20000 SUBREG of a register. Tidy up related block of code.
20001 * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
20002 note if the source is a register or a SUBREG of a register.
20004 2013-02-07 Jakub Jelinek <jakub@redhat.com>
20007 * config/rs6000/rs6000.md (ptrm): New mode attr.
20008 (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
20009 call_value_indirect_aix<pttrsize>,
20010 call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
20013 2013-02-07 Michael Haubenwallner <michael.haubenwallner@salomon.at>
20015 * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
20016 if -bnortl. Convert to strcmp and strncmp.
20018 2013-02-07 Alan Modra <amodra@gmail.com>
20021 * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
20022 addresses won't wrap when offsetting.
20023 (rs6000_secondary_reload): Provide secondary reloads needed for
20024 wrapping LO_SUM addresses.
20026 2013-02-06 Thomas Schwinge <thomas@codesourcery.com>
20028 * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
20029 MACH, just __MACH__.
20031 2013-02-06 Richard Biener <rguenther@suse.de>
20033 * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
20034 instead of calling fix_loop_structure.
20036 2013-02-06 Jakub Jelinek <jakub@redhat.com>
20038 PR middle-end/56217
20039 * omp-low.c (use_pointer_for_field): Return false if
20040 lower_send_shared_vars doesn't generate any copy-out code.
20042 2013-02-06 Tom de Vries <tom@codesourcery.com>
20044 PR rtl-optimization/56131
20045 * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
20046 to get the bb of a NOTE_INSN_BASIC_BLOCK. Handle the case that the bb
20047 of the label is NULL. Add comment.
20049 2013-02-05 Jakub Jelinek <jakub@redhat.com>
20051 * tree.h (struct tree_decl_with_vis): Remove thread_local field.
20054 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
20055 (STATIC_LIBTSAN_LIBS): Likewise.
20056 * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
20057 (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
20058 is defined, don't add anything else beyond that.
20059 (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
20060 (LINK_COMMAND_SPEC): Use them.
20062 PR tree-optimization/56205
20063 * tree-stdarg.c (check_all_va_list_escapes): Return true if
20064 there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
20065 and some va_list_escape_vars SSA_NAME appears in some PHI argument.
20067 2013-02-05 Richard Biener <rguenther@suse.de>
20069 PR tree-optimization/53342
20070 PR tree-optimization/53185
20071 * tree-vectorizer.h (vect_check_strided_load): Remove.
20072 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
20073 not disallow peeling for vectorized strided loads.
20074 (vect_check_strided_load): Make static and simplify.
20075 (vect_analyze_data_refs): Adjust.
20076 * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
20077 correctly when vectorizing strided loads.
20079 2013-02-05 Richard Biener <rguenther@suse.de>
20081 * doc/install.texi: Refer to ISL, not PPL.
20083 2013-02-05 Jan Hubicka <jh@suse.cz>
20085 PR tree-optimization/55789
20086 * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
20088 2013-02-05 Jan Hubicka <jh@suse.cz>
20090 PR tree-optimization/55789
20091 * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
20092 the dead call anyway.
20094 2013-02-05 Eric Botcazou <ebotcazou@adacore.com>
20097 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
20099 2013-02-04 Alexander Potapenko <glider@google.com>
20100 Jack Howarth <howarth@bromo.med.uc.edu>
20101 Jakub Jelinek <jakub@redhat.com>
20104 * config/darwin.c (sort_ctor_records): Stabilized qsort
20105 on constructor priority by using original position.
20106 (finalize_ctors): New routine to sort constructors by
20107 priority before use in assemble_integer.
20108 (machopic_asm_out_constructor): Use finalize_ctors if needed.
20110 2013-02-04 Jakub Jelinek <jakub@redhat.com>
20113 * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
20114 about visibility on artificial decls.
20115 * config/sol2.c (solaris_assemble_visibility): Likewise.
20117 2013-02-04 Kai Tietz <ktietz@redhat.com>
20120 * config/i386/i386.c (function_value_ms_64): Add additional valtype
20121 argument and improve checking of return-argument types for 16-byte
20123 (ix86_function_value_1): Add additional valtype argument on call
20124 of function_value_64.
20125 (return_in_memory_ms_64): Sync 16-byte sized mode handling with
20126 handling infunction_value_64 function.
20128 2013-02-04 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
20130 * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
20132 2013-02-04 Richard Biener <rguenther@suse.de>
20134 PR tree-optimization/56188
20135 * tree-ssa-structalias.c (label_visit): Consider case with
20136 initially non-empty points-to set.
20137 (perform_var_substitution): Dump node mapping and clean up.
20139 2013-02-04 Richard Guenther <rguenther@suse.de>
20142 * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
20143 node prevail as last resort.
20144 (lto_symtab_merge_decls): Remove guard on LTRANS here.
20145 (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
20147 2013-02-04 Richard Biener <rguenther@suse.de>
20149 PR tree-optimization/56113
20150 * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
20152 (equiv_class_lookup_or_add): ... this.
20153 (label_visit): Adjust and fix error in previous patch.
20154 (perform_var_substitution): Adjust.
20156 2013-02-03 Oleg Endo <olegendo@gcc.gnu.org>
20158 * config/sh/divtab.c: Fix formatting and comments throughout the file.
20159 * config/sh/sh4-300.md: Likewise.
20160 * config/sh/sh4a.md: Likewise.
20161 * config/sh/constraints.md: Likewise.
20162 * config/sh/sh.md: Likewise.
20163 * config/sh/netbsd-elf.h: Likewise.
20164 * config/sh/predicates.md: Likewise.
20165 * config/sh/sh-protos.h: Likewise.
20166 * config/sh/ushmedia.h: Likewise.
20167 * config/sh/linux.h: Likewise.
20168 * config/sh/sh.c: Likewise.
20169 * config/sh/superh.h: Likewise.
20170 * config/sh/elf.h: Likewise.
20171 * config/sh/sh4.md: Likewise.
20172 * config/sh/sh.h: Likewise.
20174 2013-02-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
20176 * config/pa/constraints.md: Adjust unused letters. Change "T"
20177 constraint to match_test floating_point_store_memory_operand().
20178 * config/pa/predicates.md (reg_plus_base_memory_operand): New.
20179 (base14_operand): New.
20180 (floating_point_store_memory_operand): New.
20181 (integer_store_memory_operand): Revise to use base14_operand and
20182 reg_plus_base_memory_operand.
20183 (move_dest_operand): Allow symbolic_memory_operands.
20184 (symbolic_memory_operand): Check for LO_SOM.
20185 (symbolic_operand): Change default case to break.
20186 * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
20187 CONST_DOUBLE values to be reloaded by putting them into memory when
20188 the destination is a floating point register.
20189 (movdf): Remove code to handle CONST_DOUBLE.
20191 (reload_indf_r1): New.
20192 (reload_insf_r1): New.
20193 Consistently use "Q" and "T" constraints with integer and floating
20194 point move instructions, respectively.
20195 (movdi): Remove FAIL.
20196 Change predicate for source operand unamed DImode move from
20197 general_operand to move_src_operand.
20198 (umulsidi3): Change predicate for destination operand to
20200 Likewise for similar unamed patterns.
20201 * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
20202 * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
20203 (hppa_legitimize_address): Simplify mask calculation.
20204 (pa_emit_move_sequence): Revised handling of secondary reloads from
20205 REG+D addresses for floating point loads and stores. Directly handle
20206 loading CONST0_RTX (mode) to a floating point register.
20207 (pa_secondary_reload): Handle reloading DF and SFmode constant values
20208 to floating point registers. Don't restrict secondary reloads to
20209 floating point registers to integer modes. Revise some comments and
20211 (TARGET_LEGITIMATE_ADDRESS_P): Define.
20212 (pa_legitimate_address_p): New.
20213 (pa_legitimize_reload_address): New.
20214 * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
20215 (STRICT_REG_OK_FOR_BASE_P): New.
20216 (GO_IF_LEGITIMATE_ADDRESS): Delete. Update some related comments.
20217 (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
20219 2013-02-03 David Edelsohn <dje.gcc@gmail.com>
20220 Andrew Dixie <andrewd@gentrack.com>
20222 * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
20225 2013-02-03 Richard Sandiford <rdsandiford@googlemail.com>
20227 * expmed.c (extract_bit_field_1): Pass the full width of the
20228 structure to get_best_reg_extraction_insn.
20230 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
20233 * configure.ac (use_cxa_atexit): Add AIX.
20234 * configure: Regenerate.
20236 * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
20238 2013-02-01 Jakub Jelinek <jakub@redhat.com>
20241 * final.c (need_profile_function): New variable.
20242 (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
20243 If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
20244 is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
20245 notes, targetm.asm_out.function_prologue doesn't emit anything,
20246 HAVE_prologue and profiler should be emitted before prologue,
20247 set need_profile_function instead of emitting it.
20248 (final_scan_insn): If need_profile_function, emit
20249 profile_function on the first NOTE_INSN_BASIC_BLOCK or
20250 NOTE_INSN_FUNCTION_BEG note.
20252 2013-02-01 Richard Henderson <rth@redhat.com>
20254 * config/rs6000/rs6000.md (smulditi3): New.
20257 * config/alpha/alpha.md (umulditi3): New.
20259 2013-02-01 David Edelsohn <dje.gcc@gmail.com>
20261 * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
20262 (ASM_OUTPUT_ALIGNED_LOCAL): New.
20264 2013-02-01 Richard Biener <rguenther@suse.de>
20266 PR tree-optimization/56113
20267 * tree-ssa-structalias.c (label_visit): Reduce work for
20268 single-predecessor nodes.
20270 2013-02-01 Eric Botcazou <ebotcazou@adacore.com>
20272 * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
20273 range isn't testing for zero.
20275 2013-01-31 Steven Bosscher <steven@gcc.gnu.org>
20277 PR middle-end/56113
20278 * fwprop.c (fwprop_init): Set up loops without CFG modifications.
20280 2013-01-31 Hiroyuki Ono <hiroyuki.ono.jc@renesas.com>
20281 Nick Clifton <nickc@redhat.com>
20283 * config/v850/constraints.md (Q): Define as a memory constraint.
20284 * config/v850/predicates.md (label_ref_operand): New predicate.
20285 (e3v5_shift_operand): New predicate.
20286 (ior_operator): New predicate.
20287 * config/v850/t-v850: Add e3v5 multilib.
20288 * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
20289 (v850_gen_movdi): Prototype.
20290 * config/v850/v850.c: Add support for e3v5 architecture.
20291 Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
20293 (construct_save_jarl): Add e3v5 long JARL support.
20294 (v850_adjust_insn_length): New function. Adjust length of call
20295 insns when using e3v5 instructions.
20296 (v850_gen_movdi): New function: Generate instructions to move a
20298 * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
20299 (CPP_SPEC): Define __v850e3v5__ as appropriate.
20300 (TARGET_USE_FPU): Enable for e3v5.
20301 (CONST_OK_FOR_W): New macro.
20302 (ADJUST_INSN_LENGTH): Define.
20303 * config/v850/v850.md (UNSPEC_LOOP): Define.
20304 (attr cpu): Add v850e3v5.
20305 Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
20306 (movdi): New pattern.
20307 (movdi_internal): New pattern.
20308 (cbranchsf4): Conditionalize on TARGET_USE_FPU.
20309 (cbranchdf4): Conditionalize on TARGET_USE_FPU.
20310 (cstoresf4): Likewise.
20311 (cstoredf4): Likewise.
20312 (insv): New pattern.
20313 (rotlso3_a): New pattern.
20314 (rotlsi3_b): New pattern
20315 (rotlsi3_v850e3v5): New pattern.
20316 (doloop_begin): New pattern.
20317 (fix_loop_counter): New pattern.
20318 (doloop_end): New pattern.
20319 (branch_normal): Add e3v5 long branch support.
20320 (branch_invert): Likewise.
20321 (branch_z_normal): Likewise.
20322 (branch_z_invert): Likewise.
20323 (branch_nz_normal): Likewise.
20324 (branch_nz_invert): Likewise.
20325 (call_internal_short): Add e3v5 register-indirect JARL support.
20326 (call_internal_long): Likewise.
20327 (call_value_internal_short): Likewise.
20328 (call_value_internal_long): Likewise.
20329 * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
20330 (mloop): New option.
20331 * config.gcc: Add support for configuring v840e3v5 target.
20332 * doc/invoke.texi: Document new v850 specific command line options.
20334 2013-01-31 Paul Koning <ni1d@arrl.net>
20338 * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
20339 children if parent is a class.
20340 (prune_unused_types_prune): Don't add DW_AT_declaration.
20342 2013-01-31 Richard Biener <rguenther@suse.de>
20344 PR tree-optimization/56157
20345 * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
20346 match up operand with SLP child.
20348 2013-01-31 Jason Merrill <jason@redhat.com>
20351 * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
20352 parameters the first time.
20353 (gen_scheduled_generic_parms_dies): Check completeness here.
20355 2013-01-31 Richard Biener <rguenther@suse.de>
20357 PR middle-end/53073
20358 * common.opt (faggressive-loop-optimizations): New flag,
20359 enabled by default.
20360 * doc/invoke.texi (faggressive-loop-optimizations): Document.
20361 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
20362 infer_loop_bounds_from_undefined by it.
20364 2013-01-31 Richard Biener <rguenther@suse.de>
20366 PR tree-optimization/56150
20367 * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
20368 visit virtual operands.
20369 (find_uses_to_rename_bb): Likewise.
20371 2013-01-31 Richard Biener <rguenther@suse.de>
20373 PR tree-optimization/56150
20374 * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
20375 mixed store non-store stmts.
20377 2013-01-30 Jakub Jelinek <jakub@redhat.com>
20380 * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
20381 LIBASAN_EARLY_SPEC is defined.
20382 (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
20383 (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
20385 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
20388 * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
20389 invalid args instead of ICEing on it.
20390 (ix86_valid_target_attribute_tree): Return error_mark_node if
20391 ix86_valid_target_attribute_inner_p failed.
20392 (ix86_valid_target_attribute_p): Return false only if
20393 ix86_valid_target_attribute_tree returned error_mark_node. Allow
20394 target("default") attribute.
20395 (sorted_attr_string): Change argument from const char * to tree,
20396 merge in all target attribute arguments rather than just one.
20397 Formatting fix. Use XNEWVEC instead of xmalloc and XDELETEVEC
20398 instead of free. Avoid using strcat.
20399 (ix86_mangle_function_version_assembler_name): Mangle
20400 target("default") as if no target attribute is present. Adjust
20401 sorted_attr_string caller. Avoid leaking memory. Use XNEWVEC
20402 instead of xmalloc and XDELETEVEC instead of free.
20403 (ix86_function_versions): Don't return true if one of the decls
20404 doesn't have target attribute. If they don't and one of the decls
20405 is DECL_FUNCTION_VERSIONED, report an error. Adjust
20406 sorted_attr_string caller. Use XDELETEVEC instead of free.
20407 (ix86_supports_function_versions): Remove.
20408 (make_name): Fix up formatting.
20409 (make_dispatcher_decl): Remove resolver_name and its initialization.
20410 Avoid leaking memory.
20411 (is_function_default_version): Return true if there is
20412 target("default") attribute rather than no target attribute at all.
20413 (make_resolver_func): Avoid leaking memory.
20414 (ix86_generate_version_dispatcher_body): Likewise.
20415 (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
20416 * target.def (supports_function_versions): Remove.
20417 * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
20418 * doc/tm.texi: Regenerated.
20420 2013-01-30 Vladimir Makarov <vmakarov@redhat.com>
20422 PR rtl-optimization/56144
20423 * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
20424 for values with side effects.
20426 2013-01-30 Richard Biener <rguenther@suse.de>
20428 * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
20429 (sparseset_pop): Likewise.
20430 * cfganal.c (compute_idf): Likewise. Increase work-stack size
20431 to be able to use quick_push in the worker loop.
20433 2013-01-30 Marek Polacek <polacek@redhat.com>
20435 * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
20437 2013-01-30 Richard Biener <rguenther@suse.de>
20440 * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
20442 2013-01-30 Georg-Johann Lay <avr@gjlay.de>
20444 PR tree-optimization/56064
20445 * fixed-value.c (fixed_from_double_int): New function.
20446 * fixed-value.h (fixed_from_double_int): New prototype.
20447 (const_fixed_from_double_int): New static inline function.
20448 * fold-const.c (native_interpret_fixed): New static function.
20449 (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
20450 (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
20451 (native_encode_fixed): New static function.
20452 (native_encode_expr) <FIXED_CST>: Use it.
20453 (native_interpret_int): Move double_int worker code to...
20454 * double-int.c (double_int::from_buffer): ...this new static method.
20455 * double-int.h (double_int::from_buffer): Prototype it.
20457 2013-01-30 Richard Biener <rguenther@suse.de>
20459 * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
20460 New pointer-map and obstack.
20461 (init_alias_vars): Allocate pointer-map and obstack.
20462 (delete_points_to_sets): Free them.
20463 (find_what_var_points_to): Cache result.
20464 (find_what_p_points_to): Adjust for changed interface of
20465 find_what_var_points_to.
20466 (compute_points_to_sets): Likewise.
20467 (ipa_pta_execute): Likewise.
20469 2013-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20471 * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
20472 * configure: Regenerate.
20473 * config.in: Regenerate.
20474 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
20475 #nobits/#progbits if supported.
20477 2013-01-29 Oleg Endo <olegendo@gcc.gnu.org>
20480 * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
20481 bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
20482 bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
20484 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
20486 * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
20487 (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
20489 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
20491 * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
20492 * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
20494 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
20496 * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
20498 * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
20499 * config/arm/cortex-a7.md: New bypasses using
20500 arm_mac_accumulator_is_result.
20502 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
20504 * config/arm/cortex-a7.md (cortex_a7_neon_mul): New reservation.
20505 (cortex_a7_neon_mla): Likewise.
20506 (cortex_a7_fpfmad): New reservation.
20507 (cortex_a7_fpmacs): Use ffmas and update required units.
20508 (cortex_a7_fpmuld): Update required units and latency.
20509 (cortex_a7_fpmacd): Likewise.
20510 (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
20511 (cortex_a7_neon). Likewise.
20512 (bypass) Update participating units.
20514 2013-01-29 Greta Yorsh <Greta.Yorsh@arm.com>
20516 * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
20517 * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
20519 * config/arm/vfp11.md (vfp_farith): Use ffmas.
20520 (vfp_fmul): Use ffmad.
20521 * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
20522 (cortex_r4_fmacd): Use ffmad.
20523 * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
20524 * config/arm/cortex-a9.md (cortex_a9_fmacs): Use ffmas.
20525 (cortex_a9_fmacd): Use ffmad.
20526 * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
20527 (cortex_a8_vfp_macd): Use ffmad.
20528 * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
20529 (cortex_a5_fpmacd): Use ffmad.
20530 * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
20531 (cortex_a15_vfp_macd): Use ffmad.
20532 * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
20534 2013-01-29 Jason Merrill <jason@redhat.com>
20537 * varasm.c (default_assemble_visibility): Don't warn about
20538 visibility on artificial decls.
20540 2013-01-29 Richard Biener <rguenther@suse.de>
20542 PR tree-optimization/56113
20543 * tree-ssa-structalias.c (equiv_class_lookup): Also return
20545 (label_visit): Free duplicate bitmaps and record the leader instead.
20546 (perform_var_substitution): Adjust.
20548 2013-01-29 Richard Biener <rguenther@suse.de>
20550 PR tree-optimization/55270
20551 * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
20552 the CFG, schedule loops for fixup.
20554 2013-01-29 Nick Clifton <nickc@redhat.com>
20556 * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
20559 2013-01-28 Leif Ekblad <leif@rdos.net>
20561 * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
20562 * config/i386/i386.h (TARGET_RDOS): New macro.
20563 (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
20564 * config/i386/i386.c (ix86_option_override_internal): For 64bit
20565 TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
20566 * config/i386/i386.opt (mlarge-data-threshold): Initialize to
20567 DEFAULT_LARGE_SECTION_THRESHOLD.
20568 * config/i386/i386.md (R14_REG, R15_REG): New constants.
20569 * config/i386/rdos.h: New file.
20570 * config/i386/rdos64.h: New file.
20572 2013-01-28 Bernd Schmidt <bernds@codesourcery.com>
20575 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
20578 2013-01-28 Jakub Jelinek <jakub@redhat.com>
20580 PR rtl-optimization/56117
20581 * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
20582 call cselib_lookup_from_insn on the MEM before calling
20583 add_insn_mem_dependence.
20585 2013-01-28 Richard Biener <rguenther@suse.de>
20587 * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
20588 to a stmt that didn't have one.
20589 (copy_phis_for_bb): Likewise for PHI arguments.
20590 (copy_debug_stmt): Likewise for debug stmts.
20592 2013-01-28 Richard Biener <rguenther@suse.de>
20594 PR tree-optimization/56034
20595 * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
20596 (partition_builtin_p): Adjust.
20597 (generate_code_for_partition): Handle PKIND_REDUCTION. Assert
20598 it is the last partition.
20599 (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
20600 up the vertex for the definition.
20601 (classify_partition): Classify whether a partition is a
20602 PKIND_REDUCTION, thus has uses outside of the loop.
20603 (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
20604 Merge all PKIND_REDUCTION partitions into the last partition.
20605 (tree_loop_distribution): Seed partitions from reductions as well.
20607 2013-01-28 Jakub Jelinek <jakub@redhat.com>
20609 PR tree-optimization/56125
20610 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
20611 pow(x,c) into sqrt(x) * powi(x, n/2) or
20612 1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
20613 optimizing for size.
20614 Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
20615 1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
20618 PR tree-optimization/56094
20619 * gimplify.c (force_gimple_operand_1): Temporarily set input_location
20620 to UNKNOWN_LOCATION while gimplifying expr.
20622 2013-01-27 Uros Bizjak <ubizjak@gmail.com>
20625 * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
20626 operand 0 in movabs insn template for -masm=intel asm alternative.
20627 (*movabs<mode>_2): Ditto for operand 1.
20629 2013-01-26 David Holsgrove <david.holsgrove@xilinx.com>
20632 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
20635 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
20637 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
20640 2013-01-25 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
20642 * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
20643 Undef to avoid warning.
20645 2013-01-25 Michael Haubenwallner <michael.haubenwallner@salomon.at>
20647 * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
20648 * configure: Regenerate.
20650 2013-01-25 Jakub Jelinek <jakub@redhat.com>
20652 PR tree-optimization/56098
20653 * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
20654 for stmts with volatile ops.
20655 (cond_store_replacement): Don't optimize if assign has volatile ops.
20656 (cond_if_else_store_replacement_1): Don't optimize if either
20657 then_assign or else_assign have volatile ops.
20658 (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
20661 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
20663 * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
20665 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
20667 * doc/extend.texi (Example of asm with clobbered asm reg): Fix
20668 missing ':' in asm example.
20670 2013-01-25 Tejas Belagod <tejas.belagod@arm.com>
20672 * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
20673 entries into lane and laneq entries.
20674 * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
20675 Remove AdvSIMD scalar modes.
20676 (aarch64_sq<r>dmulh_laneq<mode>): New.
20677 (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
20679 * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
20680 builtin implementations to relfect changes in RTL in aarch64-simd.md.
20681 * config/aarch64/iterators.md (VCOND): New.
20684 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
20687 * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
20688 Add NULL LIBNAME argument to existing definitions.
20689 (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
20690 * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
20691 * config/avr/avr.c (DEF_BUILTIN): Same.
20692 (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
20693 (avr_expand_builtin): Expand to a vanilla call if a libgcc
20694 implementation is available (DECL_ASSEMBLER_NAME is set).
20695 (avr_fold_absfx): New static function.
20696 (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
20697 AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
20698 AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
20699 AVR_BUILTIN_ABSLLK.
20700 * config/avr/stdfix.h (abshr, absr, abslr, absllr)
20701 (abshk, absk, abslk, absllk): Provide as static inline functions.
20703 2013-01-25 Marek Polacek <polacek@redhat.com>
20705 PR tree-optimization/56035
20706 * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
20708 2012-01-24 Uros Bizjak <ubizjak@gmail.com>
20710 * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
20711 (*movtf_internal_rex64): Add (!o,C) alternative
20712 (*movxf_internal_rex64): Ditto.
20713 (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
20715 2013-01-24 Shenghou Ma <minux.ma@gmail.com>
20717 * doc/invoke.texi: fix typo.
20718 * doc/objc.texi: fix typo.
20720 2013-01-24 Richard Sandiford <rdsandiford@googlemail.com>
20722 * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
20723 for the first two alternatives.
20725 2013-01-24 Diego Novillo <dnovillo@google.com>
20727 * Makefile.in (GGC): Remove. Replace all instances with ggc-page.o.
20728 (ggc-zone.o): Remove.
20729 * configure.ac: Remove option --with-gc.
20730 * configure: Re-generate.
20731 * doc/install.texi: Remove documentation for --with-gc.
20732 * gengtype.c (write_enum_defn): Remove. Update all users.
20733 (write_Types_process_field): Remove generation of gt_e_* argument.
20734 (output_type_enum): Remove. Update all users.
20735 (write_enum_defn): Remove. Update all users.
20736 (enum alloc_zone): Remove. Update all users.
20737 (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
20738 * ggc-common.c (ggc_splay_alloc): Remove first argument.
20739 Update all callers.
20740 (struct ptr_data): Remove field TYPE. Update all users.
20741 (gt_pch_note_object): Remove argument TYPE. Update all users.
20742 * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
20744 * ggc-none.c (ggc_alloc_typed_stat): Remove.
20745 (struct alloc_zone): Remove.
20746 (ggc_internal_alloc_zone_stat): Remove.
20747 (ggc_internal_cleared_alloc_zone_stat): Remove.
20748 * ggc-page.c (ggc_alloc_typed_stat): Remove.
20749 (ggc_pch_count_object): Remove last argument. Update all users.
20750 (ggc_pch_alloc_object): Remove last argument. Update all users.
20751 (struct alloc_zone): Remove.
20752 * ggc-zone.c: Remove.
20753 * ggc.h (gt_pch_note_object): Remove last argument. Update all users.
20754 (struct alloc_zone): Remove.
20755 (ggc_alloc_typed_stat): Remove.
20756 (ggc_alloc_typed): Remove.
20757 (ggc_splay_alloc): Remove first argument.
20758 (rtl_zone): Remove. Update all users.
20759 (tree_zone): Remove. Update all users.
20760 (tree_id_zone): Remove. Update all users.
20761 (ggc_internal_zone_alloc_stat): Remove. Update all users.
20762 (ggc_internal_zone_cleared_alloc_stat): Remove. Update all users.
20763 (ggc_internal_zone_vec_alloc_stat): Remove. Update all users.
20764 * tree-ssanames.c: Remove references to zone allocator in comments.
20766 2013-01-24 Georg-Johann Lay <avr@gjlay.de>
20768 * config/avr/avr.c (avr_out_fract): Make register numbers that
20769 might be outside of source operand signed.
20771 2013-01-24 Uros Bizjak <ubizjak@gmail.com>
20773 * config/i386/constraints.md (Yf): New constraint.
20774 * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
20775 of f constraint to conditionaly disable x87 register preferences.
20776 (*movdf_internal): Ditto.
20777 (*movsf_internal): Ditto.
20779 2013-01-24 Steven Bosscher <steven@gcc.gnu.org>
20781 PR inline-asm/55934
20782 * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
20783 that have operands with impossible constraints.
20784 Add a FIXME for a speed-up opportunity.
20785 * lra-constraints.c (process_alt_operands): Verify that a class
20786 selected from constraints on asms is valid for the operand mode.
20787 (curr_insn_transform): Remove incorrect comment.
20789 2013-01-23 David Edelsohn <dje.gcc@gmail.com>
20791 * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
20792 TOC operand is a valid symbol ref in the constant pool.
20794 2013-01-23 Edgar E. Iglesias <edgar.iglesias@gmail.com>
20796 * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
20798 2013-01-23 Georg-Johann Lay <avr@gjlay.de>
20801 * config/avr/stdfix.h: New file.
20802 * t-avr (stdfix-gcc.h): New rule to build it.
20803 (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
20805 2013-01-23 Kostya Serebryany <kcc@google.com>
20807 * config/darwin.h: remove dependency on
20808 CoreFoundation (asan on Mac OS).
20810 2013-01-23 Jakub Jelinek <jakub@redhat.com>
20813 * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
20814 instead of cmpdi_operand for first comparison operand.
20815 Don't assert that comparison operands aren't both constants.
20817 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
20819 * doc/install.texi (Downloading the Source): Update references to
20820 downloading separate components.
20822 2013-01-22 Jonathan Wakely <jwakely.gcc@gmail.com>
20824 * doc/extend.texi (__int128): Improve grammar.
20826 2013-01-22 Uros Bizjak <ubizjak@gmail.com>
20829 * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
20830 alternative to (o,r).
20831 (*movdi_internal_rex64): Remove (!o,n) alternative.
20832 (DImode immediate->memory splitter): Remove.
20833 (DImode immediate->memory peephole2): Remove.
20834 (movtf): Enable for TARGET_64BIT || TARGET_SSE.
20835 (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
20836 alternative to (!o,*r).
20837 (*movtf_internal_sse): New pattern.
20838 (*movxf_internal_rex64): New pattern.
20839 (*movxf_internal): Disable for TARGET_64BIT.
20840 (*movdf_internal_rex64): Remove (!o,F) alternative.
20842 2013-01-22 Jakub Jelinek <jakub@redhat.com>
20844 PR middle-end/56074
20845 * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
20846 isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
20847 * tree-vect-loop-manip.c (find_loop_location): Also ignore
20848 stmt locations where LOCATION_LOCUS of the stmt location is
20849 UNKNOWN_LOCATION or BUILTINS_LOCATION.
20852 * config/i386/i386.md (UNSPEC_STOS): New.
20853 (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
20854 *strsetqi_1): Add UNSPEC_STOS.
20856 2013-01-22 Paolo Carlini <paolo.carlini@oracle.com>
20859 * doc/invoke.texi: Remove left over -Wsynth example.
20861 2013-01-21 Jakub Jelinek <jakub@redhat.com>
20863 PR tree-optimization/56051
20864 * fold-const.c (fold_binary_loc): Don't fold
20865 X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
20866 a narrowing conversion, or widening conversion from signed
20869 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
20871 PR rtl-optimization/56023
20872 * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
20873 dependent on debug instruction.
20875 2013-01-21 Martin Jambor <mjambor@suse.cz>
20877 PR middle-end/56022
20878 * function.c (allocate_struct_function): Call
20879 invoke_set_current_function_hook earlier.
20881 2013-01-21 Jakub Jelinek <jakub@redhat.com>
20883 * reload1.c (init_reload): Only initialize reload_obstack
20884 during the first call.
20886 2013-01-21 Marek Polacek <polacek@redhat.com>
20888 * cfgloop.c (verify_loop_structure): Fix up grammar.
20890 2013-01-21 Yi-Hsiu Hsu <ahsu@marvell.com>
20892 * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
20893 pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
20895 2013-01-21 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20898 * config/arm/marvell-pj4.md: Update copyright year.
20899 Fix up use of alu to alu_reg and simple_alu_imm.
20901 2013-01-21 Uros Bizjak <ubizjak@gmail.com>
20903 * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
20905 2013-01-20 Vladimir Makarov <vmakarov@redhat.com>
20908 * lra-constraints.c (curr_insn_transform): Don't reuse original
20909 insn for secondary memory move when memory mode should be different.
20911 2013-01-20 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
20913 * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
20914 atomic_storedi_1): New patterns.
20916 2013-01-20 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
20918 btver2 pipeline descriptions.
20919 * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
20921 * config/i386/i386.md (btver2_decode): New type attributes.
20922 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
20924 * config/i386/btver2.md: New file describing btver2 pipelines.
20926 2013-01-19 Andrew Pinski <apinski@cavium.com>
20928 PR tree-optimization/52631
20929 * tree-ssa-sccvn (visit_use): Before looking up the original
20930 statement, try looking up the simplified expression.
20932 2013-01-19 Anthony Green <green@moxielogic.com>
20934 * config/moxie/moxie.c (moxie_expand_prologue): Set
20935 current_function_static_stack_size.
20937 2013-01-18 Jakub Jelinek <jakub@redhat.com>
20939 PR tree-optimization/56029
20940 * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
20941 gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
20943 2013-01-18 Sharad Singhai <singhai@google.com>
20945 PR tree-optimization/55995
20946 * dumpfile.c (dump_loc): Print location only if available.
20947 * tree-vectorizer.c (increase_alignment): Intialize vect_location.
20949 2013-01-18 Vladimir Makarov <vmakarov@redhat.com>
20952 * lra-constraints.c (curr_insn_transform): Reuse original insn for
20953 secondary memory move.
20954 (inherit_reload_reg): Use rclass instead of cl for
20955 check_secondary_memory_needed_p.
20957 2013-01-18 Jakub Jelinek <jakub@redhat.com>
20959 PR middle-end/56015
20960 * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
20961 the case where writing real complex part of target modifies op1.
20963 2013-01-18 James Greenhalgh <james.greenhalgh@arm.com>
20965 * config/aarch64/aarch64-simd.md
20966 (aarch64_vcond_internal<mode>): Handle unordered cases.
20967 * config/aarch64/iterators.md (v_cmp_result): New.
20969 2013-01-18 Yi-Hsiu Hsu <ahsu@marvell.com>
20970 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20972 * config/arm/marvell-pj4.md: New file.
20973 * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
20974 * config/arm/arm.md (generic_sched): Add marvell_pj4.
20975 (generic_vfp): Likewise.
20976 * config/arm/arm-cores.def: Add marvell-pj4.
20977 * config/arm/arm-tune.md: Regenerate.
20978 * config/arm/arm-tables.opt: Regenerate.
20979 * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
20980 * doc/invoke.texi: Document marvell-pj4.
20982 2013-01-18 Tejas Belagod <tejas.belagod@arm.com>
20984 * config/aarch64/arm_neon.h: Map scalar types to standard types.
20986 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
20991 * var-tracking.c (negative_power_of_two_p): New.
20992 (global_get_addr_cache, local_get_addr_cache): New.
20993 (get_addr_from_global_cache, get_addr_from_local_cache): New.
20994 (vt_canonicalize_addr): Rewrite using the above. Adjust the
20996 (vt_stack_offset_p): Remove.
20997 (vt_canon_true_dep): Always canonicalize loc's address.
20998 (clobber_overlapping_mems): Make sure we have a MEM.
20999 (local_get_addr_clear_given_value): New.
21000 (val_reset): Clear local cached entries.
21001 (compute_bb_dataflow): Create and release the local cache.
21002 Disable duplicate MEMs clobbering.
21003 (emit_notes_in_bb): Clobber MEMs likewise.
21004 (vt_emit_notes): Create and release the local cache.
21005 (vt_initialize, vt_finalize): Create and release the global
21006 cache, respectively.
21007 * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
21009 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
21011 PR libmudflap/53359
21012 * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
21013 not found in the symtab.
21015 2013-01-18 Alexandre Oliva <aoliva@redhat.com>
21018 PR rtl-optimization/55547
21019 PR rtl-optimization/53827
21022 * alias.c (offset_overlap_p): New, factored out of...
21023 (memrefs_conflict_p): ... this. Use absolute sizes. Retain
21024 the conservative special case for symbolic constants. Don't
21025 adjust zero sizes on alignment.
21027 2013-01-18 Bernd Schmidt <bernds@codesourcery.com>
21029 PR rtl-optimization/52573
21030 * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
21031 REG_UNUSED for the same register.
21033 2013-01-17 Richard Biener <rguenther@suse.de>
21034 Marek Polacek <polacek@redhat.com>
21036 PR rtl-optimization/55833
21037 * loop-unswitch.c (unswitch_loops): Move loop verification...
21038 (unswitch_single_loop): ...here. Call mark_irreducible_loops.
21039 * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
21040 Set it to true when we're removing a loop from hierarchy tree in
21041 an irreducible region.
21042 (fix_bb_placements): Adjust caller.
21043 (fix_loop_placements): Likewise.
21045 2013-01-17 Georg-Johann Lay <avr@gjlay.de>
21047 * config/avr/builtins.def (DEF_BUILTIN): Factor out
21048 "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
21049 Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
21050 Remove ID. Adjust comments.
21051 * config/avr/avr-c.c (avr_builtin_name): Remove.
21052 (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
21053 * config/avr/avr.c (avr_tolower): New static function.
21054 (DEF_BUILTIN): Remove parameter ID. Prefix ICODE by 'CODE_FOR_'.
21055 Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
21056 (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
21059 2013-01-17 Jan Hubicka <jh@suse.cz>
21061 PR tree-optimization/55273
21062 * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
21064 2013-01-17 Uros Bizjak <ubizjak@gmail.com>
21067 * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
21068 store through atomic_store<mode>_1.
21069 (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
21071 2013-01-17 Martin Jambor <mjambor@suse.cz>
21073 PR tree-optimizations/55264
21074 * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
21075 for virtual methods.
21076 * ipa.c (symtab_remove_unreachable_nodes): Never return true for
21077 virtual methods before inlining is over.
21078 * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
21080 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
21083 2013-01-16 Vladimir Makarov <vmakarov@redhat.com>
21085 PR rtl-optimization/56005
21086 * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
21087 pending reads for prefetch.
21089 2013-01-16 Ian Bolton <ian.bolton@arm.com>
21091 * config/aarch64/aarch64.md
21092 (*cstoresi_neg_uxtw): New pattern.
21093 (*cmovsi_insn_uxtw): New pattern.
21094 (*<optab>si3_uxtw): New pattern.
21095 (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
21096 (*<optab>si3_insn_uxtw): New pattern.
21097 (*bswapsi2_uxtw): New pattern.
21099 2013-01-16 Richard Biener <rguenther@suse.de>
21101 * tree-inline.c (tree_function_versioning): Remove set but
21102 never used variable.
21104 2013-01-16 Richard Biener <rguenther@suse.de>
21106 PR tree-optimization/55964
21107 * tree-flow.h (rename_variables_in_loop): Remove.
21108 (rename_variables_in_bb): Likewise.
21109 * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
21110 (copy_loop_before): Adjust and delete update-ssa status.
21111 * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
21112 (rename_variables_in_bb): Likewise. Properly walk over predecessors.
21113 (rename_variables_in_loop): Remove.
21114 (slpeel_update_phis_for_duplicate_loop): Likewise.
21115 (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
21116 use available cfg machinery instead of duplicating it.
21117 Update PHI nodes and perform poor-mans SSA update here.
21118 (slpeel_tree_peel_loop_to_edge): Adjust.
21120 2013-01-16 Richard Biener <rguenther@suse.de>
21122 PR tree-optimization/54767
21123 PR tree-optimization/53465
21124 * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
21125 (vrp_visit_phi_node): For PHI arguments coming via backedges
21126 drop all symbolical range information.
21127 (execute_vrp): Compute backedges.
21129 2013-01-16 Richard Biener <rguenther@suse.de>
21131 * doc/install.texi: Update CLooG and ISL requirements to
21134 2013-01-16 Christian Bruel <christian.bruel@st.com>
21137 * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
21138 (broken_move): Handle UNSPECV_SP_SWITCH_B.
21139 * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
21141 2013-01-16 DJ Delorie <dj@redhat.com>
21143 * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
21144 (UNSPECV_SP_SWITCH_E): New.
21145 (sp_switch_1): Change to an unspec.
21146 (sp_switch_2): Change to an unspec. Don't use post-inc when we
21149 2013-01-16 Uros Bizjak <ubizjak@gmail.com>
21151 * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
21152 with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
21153 * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
21154 (expand_mem_thread_fence): Ditto.
21155 (expand_mem_signal_fence): Ditto.
21156 (expand_atomic_load): Ditto.
21157 (expand_atomic_store): Ditto.
21159 2013-01-16 Alexandre Oliva <aoliva@redhat.com>
21161 PR rtl-optimization/55547
21162 PR rtl-optimization/53827
21165 * alias.c (memrefs_conflict_p): Set sizes to negative after
21168 2013-01-15 Jakub Jelinek <jakub@redhat.com>
21171 * function.c (thread_prologue_and_epilogue_insns): Always
21172 add crtl->drap_reg to set_up_by_prologue.set, even if
21173 stack_realign_drap is false.
21175 2013-01-15 Jan-Benedict Glaw <jbglaw@lug-owl.de>
21177 * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
21178 and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
21179 *call): Fix indention.
21181 2013-01-15 Tom de Vries <tom@codesourcery.com>
21184 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
21187 2013-01-15 Vladimir Makarov <vmakarov@redhat.com>
21189 PR rtl-optimization/55153
21190 * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
21192 2013-01-15 Martin Jambor <mjambor@suse.cz>
21194 PR tree-optimization/55920
21195 * tree-sra.c (analyze_access_subtree): Do not mark non-removable
21196 accesses as grp_to_be_debug_replaced.
21198 2013-01-15 Jakub Jelinek <jakub@redhat.com>
21200 PR tree-optimization/55920
21201 * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
21202 there is non-useless type conversion needed from debug rhs to lhs,
21203 use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
21205 2013-01-15 Joseph Myers <joseph@codesourcery.com>
21206 Mikael Pettersson <mikpe@it.uu.se>
21209 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
21211 (ASM_OUTPUT_CASE_LABEL): Remove.
21212 (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
21213 * final.c (shorten_branches): Update alignment of labels before
21214 jump tables if CASE_VECTOR_SHORTEN_MODE.
21216 2013-01-15 Richard Biener <rguenther@suse.de>
21219 * system.h: Do not include gmp.h for building host tools.
21221 2013-01-15 Richard Biener <rguenther@suse.de>
21223 PR middle-end/55882
21224 * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
21225 account for bitpos when computing alignment.
21227 2013-01-15 Vladimir Yakovlev <vladimir.b.yakovlev@intel.com>
21229 * config/i386/i386-c.c (ix86_target_macros_internal): New case.
21230 (ix86_target_macros_internal): Likewise.
21232 * config/i386/i386.c (m_CORE2I7): Removed.
21233 (m_CORE_HASWELL): New macro.
21234 (m_CORE_ALL): Likewise.
21235 (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
21236 (initial_ix86_arch_features): Likewise.
21237 (processor_target_table): Initializations for Core avx2.
21238 (cpu_names): New names "core-avx2".
21239 (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
21240 PROCESSOR_CORE_HASWELL.
21241 (ix86_issue_rate): New case.
21242 (ia32_multipass_dfa_lookahead): Likewise.
21243 (ix86_sched_init_global): Likewise.
21245 * config/i386/i386.h (TARGET_HASWELL): New macro.
21246 (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
21247 (processor_type): New PROCESSOR_HASWELL.
21249 2013-01-15 Jakub Jelinek <jakub@redhat.com>
21251 PR tree-optimization/55955
21252 * tree-vect-loop.c (vectorizable_reduction): Give up early on
21253 *SHIFT_EXPR and *ROTATE_EXPR codes.
21255 PR tree-optimization/48766
21256 * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
21257 -ftrapv disable -fwrapv.
21259 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
21262 * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
21263 etc. to 1 and not to __flash.
21264 Use LL suffix for __INT24_MAX__ with -mint8.
21265 Use ULL suffix for __UINT24_MAX__ with -mint8.
21267 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
21269 * config/avr/avr-arch.h
21270 (struct base_arch_s): Use typedef avr_arch_t instead.
21271 (struct arch_info_s): Use typedef avr_arch_info_t instead.
21272 (struct mcu_type_s): Use typedef avr_mcu_t instead.
21273 * config/avr/avr.c: Same.
21274 * config/avr/avr-devices.c: Same.
21275 * config/avr/driver-avr.c: Same.
21276 * config/avr/gen-avr-mmcu-texi.c: Same.
21277 * config/avr/avr-mcus.def: Adjust comment.
21279 2013-01-14 Tejas Belagod <tejas.belagod@arm.com>
21281 * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
21282 * config/aarch64/iterators.md (VALLDI): New.
21284 2013-01-14 Uros Bizjak <ubizjak@gmail.com>
21285 Andi Kleen <ak@linux.intel.com>
21288 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
21289 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
21292 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
21294 * config/avr/avr-stdint.h: Remove trailing blanks.
21295 * config/avr/avr-log.h: Same.
21296 * config/avr/avr-arch.h: Same.
21297 * config/avr/avr-devices.c: Same.
21298 * config/avr/avr-dimode.md: Same.
21299 * config/avr/predicates.md: Same.
21300 * config/avr/avr-c.c: Same. And fix typo.
21302 * config/avr/avr-protos.h: Same. And:
21303 (function_arg_regno_p): Rename to avr_function_arg_regno_p.
21304 (init_cumulative_args): Rename to avr_init_cumulative_args.
21305 (expand_prologue): Rename to avr_expand_prologue.
21306 (expand_epilogue): Rename to avr_expand_epilogue.
21307 (adjust_insn_length): Rename to avr_adjust_insn_length.
21308 (notice_update_cc): Rename to avr_notice_update_cc.
21309 (final_prescan_insn): Rename to avr_final_prescan_insn.
21310 * config/avr/avr.c: Same.
21311 * config/avr/avr.h: Same.
21312 * config/avr/avr.md: Remove trailing blanks.
21313 (prologue): Use avr_expand_prologue.
21314 (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
21316 2013-01-14 Richard Biener <rguenther@suse.de>
21318 * tree-cfg.c (verify_expr_location, verify_expr_location_1,
21319 verify_location, collect_subblocks): New functions.
21320 (verify_gimple_in_cfg): Verify that locations only reference
21321 BLOCKs in the functions BLOCK tree.
21323 2013-01-14 Richard Biener <rguenther@suse.de>
21325 * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
21327 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
21329 (insert_out_of_ssa_copy_on_edge): Likewise.
21330 (rewrite_close_phi_out_of_ssa): Likewise.
21331 * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
21333 * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
21334 propagated constants.
21335 * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
21338 2013-01-14 Georg-Johann Lay <avr@gjlay.de>
21340 * config/avr/avr-modes.def: Add GPL copyright notice.
21342 2013-01-13 Uros Bizjak <ubizjak@gmail.com>
21344 * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
21345 MEMMODEL_MASK to determine memory model.
21346 (atomic_store<mode>): Ditto from operands[2].
21347 * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
21349 2013-01-13 Jakub Jelinek <jakub@redhat.com>
21352 * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
21353 (fold_gimple_assign): Don't call unshare_expr here.
21354 (fold_ctor_reference): Call unshare_expr.
21356 2013-01-13 Terry Guo <terry.guo@arm.com>
21358 * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
21359 * doc/fragments.texi: Document MULTILIB_REUSE.
21360 * gcc.c (multilib_reuse): New internal spec.
21361 (set_multilib_dir): Also search multilib from multilib_reuse.
21362 * genmultilib (tmpmultilib3): Refactor code.
21363 (tmpmultilib4): Ditto.
21364 (multilib_reuse): New multilib argument.
21366 2013-01-13 Richard Sandiford <rdsandiford@googlemail.com>
21368 * Makefile.in: Update copyright.
21370 2013-01-12 Tom de Vries <tom@codesourcery.com>
21372 PR middle-end/55890
21373 * calls.c (expand_call): Check if arg_nr is valid.
21375 2013-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21377 * doc/extend.texi (X86 Built-in Functions): Add whitespace in
21378 __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
21379 documentation. Add missing '__' in front of
21380 __builtin_ia32_packssdw256.
21382 2013-01-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
21385 * config/s390/s390.c (s390_preferred_reload_class): Do not return
21386 NO_REGS for larl operands.
21387 (s390_reload_larl_operand): Use s390_load_address instead of
21390 2013-01-11 Richard Biener <rguenther@suse.de>
21392 * tree-cfg.c (verify_node_sharing_1): Split out from ...
21393 (verify_node_sharing): ... here.
21394 (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
21396 2013-01-11 Eric Botcazou <ebotcazou@adacore.com>
21398 * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
21399 Substitute TREECHECKING.
21400 * configure: Regenerate.
21401 * Makefile.in (TREECHECKING): New.
21403 2013-01-11 Richard Guenther <rguenther@suse.de>
21405 PR tree-optimization/44061
21406 * tree-vrp.c (extract_range_basic): Compute zero as
21407 value-range for __builtin_constant_p of function parameters.
21409 2013-01-10 Richard Sandiford <rdsandiford@googlemail.com>
21411 Update copyright years.
21413 2013-01-10 Vladimir Makarov <vmakarov@redhat.com>
21415 PR rtl-optimization/55672
21416 * lra-eliminations.c (mark_not_eliminable): Permit addition with
21417 const to be eliminable.
21419 2013-01-10 David Edelsohn <dje.gcc@gmail.com>
21421 * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
21422 * configure: Regenerate.
21424 2013-01-10 Richard Biener <rguenther@suse.de>
21426 * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
21428 2013-01-10 Richard Biener <rguenther@suse.de>
21431 * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
21432 locations for virtual PHI arguments.
21433 (rewrite_update_phi_arguments): Likewise.
21435 2013-01-10 Joel Sherrill <joel.sherrill@OARcorp.com>
21437 * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
21440 2013-01-10 Jakub Jelinek <jakub@redhat.com>
21442 PR tree-optimization/55921
21443 * tree-complex.c (expand_complex_asm): New function.
21444 (expand_complex_operations_1): Call it for GIMPLE_ASM.
21446 2013-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
21449 * config/s390/s390.c (s390_symref_operand_p)
21450 (s390_loadrelative_operand_p): Merge the two functions.
21451 (s390_check_qrst_address, print_operand_address): Add parameters
21452 to s390_loadrelative_operand_p invokation.
21453 (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
21454 (s390_reload_larl_operand, s390_secondary_reload): Use
21455 s390_loadrelative_operand_p instead of s390_symref_operand_p.
21456 (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
21458 2013-01-09 Mike Stump <mikestump@comcast.net>
21460 * dse.c (record_store): Remove unnecessary assert.
21462 2013-01-09 Jan Hubicka <jh@suse.cz>
21464 PR tree-optimization/55569
21465 * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
21466 * cfgloop.h (scale_loop_profile): Likewise.
21468 2013-01-09 Jan Hubicka <jh@suse.cz>
21471 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
21473 * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
21475 2013-01-09 Richard Sandiford <rdsandiford@googlemail.com>
21477 PR middle-end/55114
21478 * expr.h (maybe_emit_group_store): Declare.
21479 * expr.c (maybe_emit_group_store): New function.
21480 * builtins.c (expand_builtin_int_roundingfn): Call it.
21481 (expand_builtin_int_roundingfn_2): Likewise.
21483 2013-01-09 Vladimir Makarov <vmakarov@redhat.com>
21485 PR rtl-optimization/55829
21486 * lra-constraints.c (match_reload): Add code for absent output.
21487 (curr_insn_transform): Add code for reloads of matched inputs
21490 2013-01-09 Uros Bizjak <ubizjak@gmail.com>
21492 * config/i386/sse.md (*vec_interleave_highv2df): Change mode
21493 attribute of movddup insn to DF.
21494 (*vec_interleave_lowv2df): Ditto.
21495 (vec_dupv2df): Ditto.
21497 2013-01-09 Jan Hubicka <jh@suse.cz>
21499 PR tree-optimiation/55875
21500 * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
21501 EVERY_ITERATION parameter.
21502 (number_of_iterations_exit): Check if exit is executed every iteration.
21503 (idx_infer_loop_bounds): Similarly here.
21504 (n_of_executions_at_most): Simplify
21505 to only test for cases where statement is dominated by the
21506 particular bound; handle correctly the "postdominance" test.
21507 (scev_probably_wraps_p): Use max loop iterations info
21508 as a global bound first.
21510 2013-01-09 Nguyen Duy Dat <dat.nguyen.yn@rvc.renesas.com>
21511 Nick Clifton <nickc@redhat.com>
21513 * config/v850/v850.md (cbranchsf4): New pattern.
21514 (cstoresf4): New pattern.
21515 (cbranchdf4): New pattern.
21516 (cstoredf4): New pattern.
21517 (movsicc): Disallow floating point comparisons.
21518 (cmpsf_le_insn): Fix order of operators.
21519 (cmpsf_lt_insn): Likewise.
21520 (cmpsf_eq_insn): Likewise.
21521 (cmpdf_le_insn): Likewise.
21522 (cmpdf_lt_insn): Likewise.
21523 (cmpdf_eq_insn): Likewise.
21524 (cmpsf_ge_insn): Use LE comparison.
21525 (cmpdf_ge_insn): Likewise.
21526 (cmpsf_gt_insn): Use LT comparison.
21527 (cmpdf_gt_insn): Likewise.
21528 (cmpsf_ne_insn): Delete pattern.
21529 (cmpdf_ne_insn): Delete pattern.
21530 * config/v850/v850.c (v850_gen_float_compare): Use
21531 gen_cmpdf_eq_insn for NE comparison.
21532 (v850_float_z_comparison_operator)
21533 (v850_float_nz_comparison_operator): Move from here ...
21534 * config/v850/predicates.md: ... to here. Move GT and GE
21535 comparisons into v850_float_z_comparison_operator.
21536 * config/v850/v850-protos.h (v850_float_z_comparison_operator):
21538 (v850_float_nz_comparison_operator): Likewise.
21540 2013-01-09 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
21542 * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
21543 with calls to gen_insvsi/gen_insvdi.
21545 2013-01-09 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
21547 * config/i386/i386.c (initial_ix86_tune_features): Set up
21548 X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
21550 2013-01-09 Steven Bosscher <steven@gcc.gnu.org>
21551 Jakub Jelinek <jakub@redhat.com>
21553 PR tree-optimization/48189
21554 * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
21555 If nitercst is 0, don't predict the exit edge.
21557 2013-01-08 Naveen H.S <Naveen.Hurugalawadi@caviumnetworks.com>
21559 * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
21560 in asm_fprintf with reg_names.
21561 (aarch64_print_operand_address): Likewise.
21562 (aarch64_return_addr): Likewise.
21563 * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
21565 2013-01-08 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
21567 * config/pa/pa.h (VAL_U6_BITS_P): Define.
21568 (INT_U6_BITS): Likewise.
21569 * config/pa/predicates.md (uint6_operand): New predicate.
21570 (shift5_operand, shift6_operand): Likewise.
21571 * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
21573 (lshrdi3): Use shift6_operand.
21574 (shrpsi4, shrpdi4): New insn patterns.
21575 (extzv): Delete expander.
21576 (extzvsi, extzvdi): New expanders. Use uint5_operand and uint6_operand
21577 predicates in unamed zero extract patterns. Tighten common constraint.
21578 (extv): Delete expander.
21579 (extvsi, extvdi): New expanders. Use uint5_operand and uint6_operand
21580 predicates in unamed sign extract patterns. Tighten common constraint.
21581 (insv): Delete expander.
21582 (insvsi, insvdi): New expanders. Use uint5_operand and uint6_operand
21583 predicates in unamed insert patterns. Tighten common constraint.
21584 Change uint32_operand predicate to uint6_operand predicate in unamed
21585 DImode pattern to insert constant values of type 1...1xxxx.
21587 2013-01-04 Jan Hubicka <jh@suse.cz>
21589 PR tree-optimization/55823
21590 * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
21593 2013-01-08 Jakub Jelinek <jakub@redhat.com>
21594 Uros Bizjak <ubizjak@gmail.com>
21596 PR rtl-optimization/55845
21597 * df-problems.c (can_move_insns_across): Stop scanning at
21598 volatile_insn_p source instruction or give up if
21599 across_from .. across_to range contains any volatile_insn_p
21602 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
21604 * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
21605 * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
21607 * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
21608 aarch64_simd_make_constant, aarch64_expand_vector_init): New.
21610 2013-01-08 Jakub Jelinek <jakub@redhat.com>
21613 * asan.c (asan_clear_shadow): New function.
21614 (asan_emit_stack_protection): Use it.
21616 2013-01-08 Tejas Belagod <tejas.belagod@arm.com>
21618 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
21619 aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
21620 with tab instead of space.
21622 2013-01-08 Nick Clifton <nickc@redhat.com>
21624 * config/rl78/rl78.c (rl78_expand_prologue): Always select
21625 register bank 0 at the start of an interrupt handler.
21626 * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
21629 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
21631 * config/aarch64/aarch64-simd.md
21632 (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
21633 (aarch64_simd_bsl): Likewise.
21634 (aarch64_vcond_internal<mode>): Likewise.
21635 (vcond<mode><mode>): Likewise.
21636 (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
21637 * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
21639 2013-01-08 James Greenhalgh <james.greenhalgh@arm.com>
21641 * config/aarch64/aarch64-builtins.c
21642 (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
21644 2013-01-08 Martin Jambor <mjambor@suse.cz>
21647 * tree-sra.c (analyze_access_subtree): Return true also after
21648 potentially creating a debug-only replacement.
21650 2013-01-08 Jakub Jelinek <jakub@redhat.com>
21652 PR middle-end/55890
21653 * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
21655 PR tree-optimization/54120
21656 * tree-vrp.c (range_fits_type_p): Don't allow
21657 src_precision < precision from signed vr to unsigned_p
21658 if vr->min or vr->max is negative.
21659 (simplify_float_conversion_using_ranges): Test can_float_p
21660 against CODE_FOR_nothing.
21662 2013-01-08 Jakub Jelinek <jakub@redhat.com>
21663 Richard Biener <rguenther@suse.de>
21665 PR middle-end/55851
21666 * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
21667 types instead of just INTEGER_TYPE types.
21669 2013-01-07 Mark Kettenis <kettenis@openbsd.org>
21671 * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
21674 2013-01-07 Steve Ellcey <sellcey@mips.com>
21677 * config/mips/mips.opt: Change mad to mmad to match documentation.
21679 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
21682 * doc/extend.texi (AVR Named Address Spaces): __memx goes into
21683 .progmemx.data now.
21685 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
21688 * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
21689 (avr_addrspace_t): Add .section_name field.
21690 * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
21692 (avr_addrspace): Same. Initialize .section_name. Remove last
21693 NULL entry. Put __memx into .progmemx.data.
21694 (progmem_section_prefix): Remove.
21695 (avr_asm_init_sections): No need to initialize progmem_section.
21696 (avr_asm_named_section): Use avr_addrspace[].section_name to get
21697 section name prefix.
21698 (avr_asm_select_section): Ditto. And use get_unnamed_section to
21699 retrieve the progmem section.
21700 * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
21701 boundary to run over avr_addrspace[].
21702 (avr_register_target_pragmas): Ditto.
21704 2013-01-06 Jakub Jelinek <jakub@redhat.com>
21706 * varasm.c (output_constant_def_contents): For asan_protect_global
21707 protected strings, adjust DECL_ALIGN if needed, before testing for
21709 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
21710 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
21712 (output_object_block): For asan protected decls, emit asan padding
21713 after their contents.
21714 * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
21715 (asan_finish_file): Test it here instead.
21717 2013-01-07 Nick Clifton <nickc@redhat.com>
21718 Matthias Klose <doko@debian.org>
21719 Doug Kwan <dougkwan@google.com>
21720 H.J. Lu <hongjiu.lu@intel.com>
21723 * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
21725 * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
21727 * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
21729 * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
21731 * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
21733 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
21736 * doc/install.texi (Cross-Compiler-Specific Options): Document
21739 2013-01-07 Tejas Belagod <tejas.belagod@arm.com>
21741 * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
21742 vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
21743 vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
21744 vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
21745 vqmovun_high_s64): Fix source operand number and update copyright.
21747 2013-01-07 Richard Biener <rguenther@suse.de>
21749 PR middle-end/55890
21750 * gimple.h (gimple_call_builtin_p): New overload.
21751 * gimple.c (validate_call): New function.
21752 (gimple_call_builtin_p): Likewise.
21753 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21754 Use gimple_call_builtin_p.
21755 (find_func_clobbers): Likewise.
21756 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
21757 (strlen_optimize_stmt): Likewise.
21759 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
21761 * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
21762 (vld1q_dup_*): Likewise.
21763 (vld1_*): Likewise.
21764 (vld1q_*): Likewise.
21765 (vld1_lane_*): Likewise.
21766 (vld1q_lane_*): Likewise.
21768 2013-01-07 Richard Biener <rguenther@suse.de>
21770 * lto-streamer.h (LTO_minor_version): Bump to 2.
21772 2013-01-07 James Greenhalgh <james.greenhalgh@arm.com>
21774 * config/aarch64/aarch64-protos.h
21775 (aarch64_const_double_zero_rtx_p): Rename to...
21776 (aarch64_float_const_zero_rtx_p): ...this.
21777 (aarch64_float_const_representable_p): New.
21778 (aarch64_output_simd_mov_immediate): Likewise.
21779 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
21780 move immediate case.
21781 * config/aarch64/aarch64.c
21782 (aarch64_const_double_zero_rtx_p): Rename to...
21783 (aarch64_float_const_zero_rtx_p): ...this.
21784 (aarch64_print_operand): Allow printing of new constants.
21785 (aarch64_valid_floating_const): New.
21786 (aarch64_legitimate_constant_p): Check for valid floating-point
21788 (aarch64_simd_valid_immediate): Likewise.
21789 (aarch64_vect_float_const_representable_p): New.
21790 (aarch64_float_const_representable_p): Likewise.
21791 (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
21792 (aarch64_output_simd_mov_immediate): New.
21793 * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
21794 (*movdf_aarch64): Likewise.
21795 * config/aarch64/constraints.md (Ufc): New.
21796 (Y): call aarch64_float_const_zero_rtx.
21797 * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
21799 2013-01-07 Richard Biener <rguenther@suse.de>
21801 PR tree-optimization/55888
21802 PR tree-optimization/55862
21803 * tree-ssa-pre.c (phi_translate_1): Revert previous change.
21804 (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
21805 not if it is contained therein.
21807 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
21809 * config/avr/t-avr: Typo.
21811 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
21814 * config/avr/t-avr: Don't automatically rebuild
21815 $(srcdir)/config/avr/t-multilib
21816 $(srcdir)/config/avr/avr-tables.opt
21817 $(srcdir)/doc/avr-mmcu.texi
21818 (avr-mcus): New phony target to build them on request.
21819 (s-avr-mlib, s-avr-mmcu-texi): Remove.
21820 * avr/avr-mcus.def: Adjust comments.
21822 2013-01-07 Uros Bizjak <ubizjak@gmail.com>
21824 * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
21826 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
21828 * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
21830 2013-01-06 Richard Sandiford <rdsandiford@googlemail.com>
21832 * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
21834 2013-01-05 David Edelsohn <dje.gcc@gmail.com>
21836 * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
21837 to generate profiling.
21838 * config/rs6000/aix64.h (LIB_SPEC): Same.
21840 2013-01-04 Andrew Pinski <apinski@cavium.com>
21842 * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
21844 (TARGET_FIXED_CONDITION_CODE_REGS): Define.
21846 2013-01-04 Uros Bizjak <ubizjak@gmail.com>
21848 * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
21850 (ix86_expand_move): Ditto.
21851 (ix86_zero_extend_to_Pmode): Ditto.
21852 (ix86_expand_call): Ditto.
21853 (ix86_expand_special_args_builtin): Ditto.
21854 (ix86_expand_builtin): Ditto.
21856 2013-01-04 Richard Biener <rguenther@suse.de>
21858 PR tree-optimization/55862
21859 * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
21860 translating them through PHI nodes.
21862 2013-01-04 Martin Jambor <mjambor@suse.cz>
21864 PR tree-optimization/55755
21865 * tree-sra.c (sra_modify_assign): Do not check that an access has no
21866 children when trying to avoid producing a VIEW_CONVERT_EXPR.
21868 2013-01-04 Marek Polacek <polacek@redhat.com>
21870 PR middle-end/55859
21871 * opts.c (default_options_optimization): Clarify error message.
21873 2013-01-04 Richard Biener <rguenther@suse.de>
21875 PR middle-end/55863
21876 * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
21879 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
21882 * config/pa/pa.md (movsi): Revert previous change.
21883 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
21886 2013-01-03 Richard Henderson <rth@redhat.com>
21888 * config/i386/i386.c (ix86_expand_move): Always assign to op1
21889 after eliminating TLS symbols.
21891 2013-01-03 Marc Glisse <marc.glisse@inria.fr>
21894 * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
21895 * graphite-poly.c (debug_gmp_value): Likewise.
21897 2013-01-03 Uros Bizjak <ubizjak@gmail.com>
21900 * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
21901 selected code model, define __code_mode_small__, __code_model_medium__,
21902 __code_model_large__, __code_model_32__ or __code_model_kernel__.
21903 * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
21904 xchg temporary register with %k. Declare temporary register as
21906 [__x86_64__]: For medium and large code models, preserve %rbx register.
21908 2013-01-03 Richard Biener <rguenther@suse.de>
21910 * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
21911 (dump_subscript): Adjust.
21912 (finalize_ddr_dependent): Do not dump redundant info.
21913 (analyze_siv_subscript): Adjust.
21914 (subscript_dependence_tester): Likewise.
21915 (compute_affine_dependence): Likewise.
21917 2013-01-03 Richard Biener <rguenther@suse.de>
21920 2013-01-03 Richard Biener <rguenther@suse.de>
21922 PR tree-optimization/55857
21923 * tree-vect-stmts.c (vectorizable_load): Do not setup
21924 re-alignment for invariant loads.
21926 2013-01-02 Richard Biener <rguenther@suse.de>
21928 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
21929 invariant load do not generate a vector load from the scalar location.
21931 2013-01-03 Richard Biener <rguenther@suse.de>
21933 * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
21934 for not vectorizing.
21935 * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
21936 not build INDIRECT_REFs, call get_name once only.
21937 (vect_create_data_ref_ptr): Likewise. Dump base object kind
21938 based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
21940 2013-01-03 Richard Biener <rguenther@suse.de>
21942 PR tree-optimization/55857
21943 * tree-vect-stmts.c (vectorizable_load): Do not setup
21944 re-alignment for invariant loads.
21946 2013-01-03 Richard Biener <rguenther@suse.de>
21949 * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
21950 prefer a built-in decl.
21952 2013-01-03 Jakub Jelinek <jakub@redhat.com>
21954 * gcc.c (process_command): Update copyright notice dates.
21955 * gcov.c (print_version): Likewise.
21956 * gcov-dump.c (print_version): Likewise.
21958 PR rtl-optimization/55838
21959 * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
21960 iv0.step, iv1.step and step.
21962 2013-01-03 Jakub Jelinek <jakub@redhat.com>
21963 Marc Glisse <marc.glisse@inria.fr>
21965 PR tree-optimization/55832
21966 * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
21967 ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
21968 integer_{one,zero}_node.
21970 2013-01-03 Jakub Jelinek <jakub@redhat.com>
21973 * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
21974 * var-tracking.c (reverse_op): Don't add reverse ops to
21975 VALUEs that have already
21976 PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
21978 2013-01-02 Gerald Pfeifer <gerald@pfeifer.com>
21980 * doc/contrib.texi: Note years as release manager for Mark Mitchell.
21982 2013-01-02 Teresa Johnson <tejohnson@google.com>
21984 * dumpfile.c (dump_loc): Print filename with location.
21985 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
21986 new location_t parameter to emit complete unroll message with
21987 new dump framework.
21988 (canonicalize_loop_induction_variables): Compute loops location
21989 and pass to try_unroll_loop_completely.
21990 * loop-unroll.c (report_unroll_peel): New function.
21991 (peel_loops_completely): Use new dump format with location
21992 for main dumpfile message, and invoke report_unroll_peel on success.
21993 (decide_unrolling_and_peeling): Ditto.
21994 (decide_peel_once_rolling): Remove old dumpfile message subsumed
21995 by report_unroll_peel.
21996 (decide_peel_completely): Ditto.
21997 (decide_unroll_constant_iterations): Ditto.
21998 (decide_unroll_runtime_iterations): Ditto.
21999 (decide_peel_simple): Ditto.
22000 (decide_unroll_stupid): Ditto.
22001 * cfgloop.c (get_loop_location): New function.
22002 * cfgloop.h (get_loop_location): Declare.
22004 2013-01-02 Sriraman Tallam <tmsriram@google.com>
22006 * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
22009 2013-01-02 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
22011 PR middle-end/55198
22012 * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
22013 BLKmode objects when EXPAND_MEMORY is specified.
22015 2013-01-02 Sriraman Tallam <tmsriram@google.com>
22017 * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
22019 (fold_builtin_cpu): Do not share cpu model decls across statements.
22021 2013-01-02 Jason Merrill <jason@redhat.com>
22024 * tree.c (build_array_type_1): Revert earlier change.
22026 2013-01-02 Yufeng Zhang <yufeng.zhang@arm.com>
22028 * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
22030 * config/aarch64/aarch64-tune.md: Re-generate.
22032 2013-01-02 Richard Biener <rguenther@suse.de>
22034 * tree-vect-stmts.c (vectorizable_load): When vectorizing an
22035 invariant load do not generate a vector load from the scalar location.
22037 2013-01-02 Richard Biener <rguenther@suse.de>
22040 * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
22041 * configure: Regenerate.
22043 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
22045 * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
22046 (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
22047 (expand_builtin_int_roundingfn_2): Keep the original target around
22048 for the fallback case.
22050 2013-01-02 Richard Sandiford <rdsandiford@googlemail.com>
22052 * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
22053 to be clear for sign changes.
22055 2013-01-01 Jan Hubicka <jh@suse.cz>
22057 * ipa-inline-analysis.c: Fix formatting.
22059 2013-01-01 Jakub Jelinek <jakub@redhat.com>
22061 PR tree-optimization/55831
22062 * tree-vect-loop.c (get_initial_def_for_induction): Use
22063 gsi_after_labels instead of gsi_start_bb.
22065 Copyright (C) 2013 Free Software Foundation, Inc.
22067 Copying and distribution of this file, with or without modification,
22068 are permitted in any medium without royalty provided the copyright
22069 notice and this notice are preserved.