2013-08-30 Richard Biener <rguenther@suse.de>
[official-gcc.git] / gcc / ChangeLog
blob7a8782b0f13b292e2e392ac1e0a284ff87efafc0
1 2013-08-30  Richard Biener  <rguenther@suse.de>
3         PR tree-optimization/58223
4         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
5         (has_anti_or_output_dependence): ... this and adjust to also
6         look for output dependences.
7         (mark_nodes_having_upstream_mem_writes): Adjust.
8         (rdg_flag_uses): Likewise.
10 2013-08-30  Richard Biener  <rguenther@suse.de>
12         PR tree-optimization/58010
13         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
14         assert that we have a loop-closed PHI.
16 2013-08-29  Jan Hubicka  <jh@suse.cz>
18         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
19         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
20         * lto-section-in.c (lto_free_function_in_decl_state): New function.
21         (lto_free_function_in_decl_state_for_node): New function.
23 2013-08-29  Xinliang David Li  <davidxl@google.com>
25         * loop-unroll.c (report_unroll_peel): Minor message
26         change.
27         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
28         Emit alignment peeling message with default -fopt-info.
29         (vect_loop_versioning): Emit loop version info message.
30         * tree-vectorizer.c (vectorize_loops): Minor message
31         change.
32         (execute_vect_slp): Ditto.
34 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
36         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
37         of the clone from the DECL_NAME of the original function.
39 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
41         * passes.c (register_pass): Add overload.
42         * tree-pass.h (register_pass): Forward declare it.  Add comment.
44 2013-08-29  Jan Hubicka  <jh@suse.cz>
46         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P.
47         DECL_CXX_CONSTRUCTOR_P. DECL_CXX_DESTRUCTOR_P.
48         TYPE_FINAL_P.
49         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields):
50         DECL_FINAL_P. DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
51         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
52         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
53         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P
54         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
56 2013-08-29  Teresa Johnson  <tejohnson@google.com>
58         * dumpfile.c (dump_loc): Output column number.
59         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
60         * doc/invoke.texi: Document optall -fopt-info flag.
61         * profile.c (read_profile_edge_counts): Use new dump framework.
62         (compute_branch_probabilities): Ditto.
63         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
64         when pass not in any opt group.
65         * pass_manager.h (pass_manager::get_pass_profile): New method.
66         * value-prof.c (check_counter): Use new dump framework.
67         (check_ic_target): Ditto.
68         * coverage.c (get_coverage_counts): Ditto.
69         (coverage_init): Setup new dump framework.
71         * testsuite/gcc.dg/pr40209.c: Use -fopt-info.
72         * testsuite/gcc.dg/pr26570.c: Ditto.
73         * testsuite/gcc.dg/pr32773.c: Ditto.
74         * testsuite/g++.dg/tree-ssa/dom-invalid.C: Ditto.
75         * testsuite/gcc.dg/inline-dump.c: New test.
77 2013-08-29  Richard Biener  <rguenther@suse.de>
79         PR tree-optimization/58246
80         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
81         handle the dominance check inside a basic-block.
83 2013-08-29  Richard Biener  <rguenther@suse.de>
85         PR middle-end/57287
86         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
87         of default defs that appear in abnormal PHI nodes.
89 2013-08-29  Richard Biener  <rguenther@suse.de>
91         PR tree-optimization/57685
92         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
93         single-use operands to avoid exponential complexity.
95 2013-08-28  Dehao Chen  <dehao@google.com>
97         * ipa-inline.c (edge_badness): Fix integer underflow.
99 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
101         * gtm-builtins.def (_ITM_free): Declare leaf.
103 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
105         PR target/58067
106         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
107         (*tls_local_dynamic_base_64_largepic): Likewise.
108         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
109         Remove predicate from call operand.
110         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
111         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
113 2013-08-28  Jeff Law  <law@redhat.com>
115         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
116         checks for the number of predecessors and successors allowed.
117         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
118         which require copying a joiner block if there is a request which
119         is a subpath that requires no joiner block copying.
121 2013-08-28  Jan Hubicka  <jh@suse.cz>
123         * lto-streamer-out.c (DFS_write_tree_body): Drop
124         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
125         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
126         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
127         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
128         Do not read DECL_ERROR_ISSUED.
129         (unpack_ts_decl_with_vis_value_fields): Do not read
130         DECL_DEFER_OUTPUT.
131         (lto_input_ts_binfo_tree_pointers): Do not read
132         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
133         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
134         write DECL_ERROR_ISSUED..
135         (pack_ts_decl_with_vis_value_fields): Do not write
136         DECL_DEFER_OUTPUT.
137         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
138         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
139         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
140         * tree.h (tree_decl_common): Update comment.
141         (DECL_ERROR_ISSUED): Remove.
143 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
145         PR middle-end/58257
146         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
148 2013-08-28  Jan Hubicka  <jh@suse.cz>
150         * builtins.def (free): Declare leaf.
152 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
154         * gdbhooks.py: New.
155         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
156         * configure: Regenerate.
158 2013-08-27  Martin Jambor  <mjambor@suse.cz>
160         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
161         (ipa_ancestor_jf_data): Likewise.
162         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
163         (ipa_get_jf_pass_through_type_preserved): New function.
164         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
165         (ipa_get_jf_ancestor_type_preserved): New function.
166         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
167         (ipa_get_jf_ancestor_result): Likewise.
168         (propagate_vals_accross_pass_through): Use
169         ipa_get_jf_pass_through_result to do all the value mappings.
170         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
171         type_preserved flag.
172         (ipa_set_jf_cst_copy): New function.
173         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
174         (ipa_set_jf_arith_pass_through): Likewise.
175         (ipa_set_ancestor_jf): Likewise.
176         (compute_complex_assign_jump_func): Set type_preserved instead of
177         punting.
178         (ipa_compute_jump_functions_for_edge): Likewise.
179         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
180         (update_jump_functions_after_inlining): Update type_preserved.
181         Explicitely create jump functions when combining one with pass_through.
182         (ipa_write_jump_function): Stream the type_preserved flags.
183         (ipa_read_jump_function): Likewise.
185 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
186             Aldy Hernandez  <aldyh@redhat.com>
188         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
189         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
190         * function.h (struct function): Add has_force_vect_loops and
191         has_simduid_loops.
192         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
193         * gimple.c (gimple_build_omp_critical): Add KIND argument and
194         handle it.
195         * gimple.def: Update CLAUSES comments.
196         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
197         (gimple_build_omp_for): Add argument to prototype.
198         (gimple_omp_for_kind): New.
199         (gimple_omp_for_set_kind): New.
200         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
201         GOVD_DATA_SHARE_CLASS.
202         (enum omp_region_type): Add ORT_SIMD.
203         (gimple_add_tmp_var): Handle ORT_SIMD.
204         (gimplify_var_or_parm_decl): Same.
205         (is_gimple_stmt): Same.
206         (omp_firstprivatize_variable): Same.
207         (omp_add_variable): Only use splay_tree_insert if lookup failed.
208         (omp_notice_variable): Handle ORT_SIMD.
209         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
210         (omp_check_private): Handle ORT_SIMD.
211         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
212         OMP_CLAUSE_SAFELEN.
213         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
214         Handle OMP_CLAUSE_LASTPRIVATE.
215         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
216         OMP_CLAUSE_SAFELEN.
217         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
218         (gimplify_expr): Handle OMP_SIMD.
219         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
220         (expand_GOMP_SIMD_VF): New.
221         (expand_GOMP_SIMD_LAST_LANE): New.
222         * internal-fn.def (GOMP_SIMD_LANE): New.
223         (GOMP_SIMD_VF): New.
224         (GOMP_SIMD_LAST_LANE): New.
225         * omp-low.c: Include target.h.
226         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
227         OMP_CLAUSE_SAFELEN.
228         (check_omp_nesting_restrictions): Same.
229         (omp_max_vf): New.
230         (lower_rec_simd_input_clauses): New.
231         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
232         OMP_CLAUSE_LINEAR.
233         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
234         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
235         (expand_omp_build_assign): New.
236         (expand_omp_for_init_counts): New.
237         (expand_omp_for_init_vars): New.
238         (extract_omp_for_update_vars): New.
239         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
240         and rewrite accordingly.
241         (expand_omp_simd): New.
242         (expand_omp_for): Use expand_omp_simd.
243         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
244         (lower_omp_for): Do not lower the body.
245         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
246         in their own loops.
247         * tree-flow.h (find_omp_clause): Remove prototype.
248         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
249         forcing vectorization of the loop, or if flag_tree_vectorize.
250         (gate_tree_if_conversion): Similarly.
251         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
252         gimple_build_omp_for.
253         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
254         * tree-parloops (create_parallel_loop): Pass kind argument to
255         gimple_build_omp_for.
256         * tree-pretty-print.c (dump_omp_clause): Add cases for
257         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
258         OMP_CLAUSE__SIMDUID_.
259         (dump_generic_node): Handle OMP_SIMD.
260         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
261         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
262         unroll OMP_SIMD loops here.
263         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
264         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
265         loop->safelen.
266         (vect_analyze_data_refs): Handle simd loops.
267         * tree-vect-loop.c (vectorizable_live_operation): Handle
268         IFN_GOMP_SIMD*.
269         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
270         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
271         (vectorizable_load): Same.
272         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
273         (struct simduid_to_vf): New.
274         (simduid_to_vf::hash): New.
275         (simduid_to-vf::equal): New.
276         (struct simd_array_to_simduid): New.
277         (simd_array_to_simduid::hash): New.
278         (simd_array_to_simduid::equal): New.
279         (adjust_simduid_builtins): New.
280         (struct note_simd_array_uses_struct): New.
281         (note_simd_array_uses_cb): New.
282         (note_simd_array_uses): New.
283         (vectorize_loops): Handle simd hints and adjust simd builtins
284         accordingly.
285         * tree-vectorizer.h (struct _stmt_vec_info): Add
286         simd_lane_access_p field.
287         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
288         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
289         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
290         (omp_clause_code_name): Same.
291         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
292         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
293         * tree.def (OMP_SIMD): New entry.
294         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
295         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
296         (OMP_CLAUSE_DECL): Adjust range for new clauses.
297         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
298         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
299         (OMP_CLAUSE_LINEAR_STEP): New.
300         (OMP_CLAUSE_SAFELEN_EXPR): New.
301         (OMP_CLAUSE__SIMDUID__DECL): New.
302         (find_omp_clause): New prototype.
304 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
306         * config/i386/driver-i386.c (host_detect_local_cpu): Update
307         Haswell processor detection.
309 2013-08-27  Christian Widmer  <shadow@umbrox.de>
311         PR target/57927
312         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
313         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
314         AVX2 capable processors.
316 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
318         * config/aarch64/arm_neon.h: Replace all inline asm implementations
319         of vget_low_* with implementations in terms of other intrinsics.
321 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
323         PR middle-end/57219
324         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
325         values to -1, 0 and 1.
327 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
329         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
330         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
331         (<optab><mode>3_insn): Remove.
332         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
333         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
334         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
335         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
336         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
337         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
338         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
339         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
340         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
341         (ror<mode>3_insn): Likewise.
342         * config/aarch64/predicates.md (aarch64_simd_register): New.
344 2013-08-27  Richard Biener  <rguenther@suse.de>
346         PR tree-optimization/57521
347         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
348         one edge is non-critical.
349         (find_phi_replacement_condition): Make sure to use a non-critical
350         edge.  Cleanup and remove old bug workarounds.
351         (bb_postdominates_preds): Remove.
352         (if_convertible_loop_p_1): Do not compute post-dominators.
353         (combine_blocks): Do not free post-dominators.
354         (main_tree_if_conversion): Likewise.
355         (pass_data_if_conversion): Add TODO_verify_ssa.
357 2013-08-27  DJ Delorie  <dj@redhat.com>
359         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
361 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
363         * function.c (assign_parm_find_data_types): Set passed_mode and
364         nominal_mode to the TYPE_MODE of nominal_type for the built
365         pointer type in case of the struct-pass-by-reference.
367 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
369         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
370         (UINT16_TYPE): Change default to "unsigned int".
372         * config/avr/avr.opt (mfract-convert-truncate): New option.
373         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
374         is set, round negative fractional integers according to n1169
375         when converting to integer types.
377 2013-08-26  Jan Hubicka  <jh@suse.cz>
379         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
380         methods can not be called indirectly when their address is not taken.
382 2013-08-26  Jan Hubicka  <jh@suse.cz>
384         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
385         ctor_for_folding.
387 2013-08-26  Jan Hubicka  <jh@suse.cz>
389         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
390         can be unshared.
392 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
394         * reload.c (find_valid_class): Allow classes that do not include
395         FIRST_PSEUDO_REGISTER - 1.
397 2013-08-26  Jan Hubicka  <jh@suse.cz>
399         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
400         fix edge count/frequency when speculation failed; fix type check
401         for the direct call.
403 2013-08-26  Jan Hubicka  <jh@suse.cz>
405         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
407 2013-08-26  Jan Hubicka  <jh@suse.cz>
409         * ipa-inline-transform.c (inline_transform): Be ready for basic block
410         to be changed by edge redirection.
412 2013-08-26  Jan Hubicka  <jh@suse.cz>
414         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
415         formating; add sanity check.
416         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
417         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
418         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
419         (resolve_noninline_speculation): Update callee keys, too.
421 2013-08-26  Jan Hubicka  <jh@suse.cz>
423         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
424         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
426 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
428         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
429         into proper place.
431 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
433         * config/i386/i386.c (ix86_debug_options): Remove prototype.
434         (x86_64_elf_select_section): Ditto.
435         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
436         arguments.
437         (ix86_pass_by_reference): Ditto.
438         (output_set_got): Ditto.
439         (ix86_unary_operator_ok): Ditto.
440         (ix86_expand_builtin): Ditto.
442 2013-08-23  Jan Hubicka  <jh@suse.cz>
444         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
446 2013-08-23  Jan Hubicka  <jh@suse.cz>
448         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
449         (tree_decl_with_vis): Add FINAL field.
451 2013-08-23  Jeff Law  <law@redhat.com>
453         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
454         the debugging dump when the expression is fully redundant.
456 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
458         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
459         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
460         * pretty-print.c (pp_formatted_text_data): Likewise.
461         (pp_write_text_to_stream): Likewise.
462         (pp_write_text_as_dot_label_to_stream): Likewise.
463         (pp_append_r): Likewise.
464         (pp_format): Likewise.
465         (pp_flush): Likewise.
466         (pp_clear_output_area): Likewise.
467         (pp_append_text): Likewise.
468         (pp_formatted_text): Likewise.
469         (pp_remaining_character_count_for_line): Likewise.
470         (pp_newline): Likewise.
471         (pp_character): Likewise.
472         (output_buffer::~output_buffer): Define.
473         (pretty_printer::~pretty_printer): Destruct output buffer.
474         * pretty-print.h (output_buffer::~output_buffer): Declare.
475         (pretty_printer::~pretty_printer): Declare virtual.
477 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
479         PR other/57324
480         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
481         HOST_WIDE_INT_M1U): New macros.
482         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
483         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
484         unsigned -1 for lshift.
485         * cse.c (cse_insn): Likewise.
486         * double-int.c (rshift_double, lshift_double): Likewise.
487         * builtins.c (fold_builtin_bitop): Likewise.
488         * combine.c (force_to_mode): Likewise.
489         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
490         * simplify-rtx.c (simplify_const_unary_operation,
491         simplify_const_binary_operation): Likewise.
492         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
493         check_va_list_escapes): Likewise.
494         * rtlanal.c (nonzero_bits1): Likewise.
495         * expmed.c (expand_smod_pow2): Likewise.
496         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
498 2013-08-23  Jan Hubicka  <jh@suse.cz>
500         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
501         as having address taken.
503 2013-08-23  Jan Hubicka  <jh@suse.cz>
505         * ipa-utils.h (method_class_type): Declare.
506         * ipa-devirt.c (method_class_type): Export.
508         * cgraphunit.c (analyze_functions): Do basic devirtualization;
509         do not walk base classes of anonymous types.
511 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
513         PR rtl-optimization/58220
514         PR regression/58221
515         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
516         handle SEQUENCE insns properly.
518 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
520         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
521         definition.
522         (pp_newline_and_indent): Likewise.
523         (pp_separate_with): Likewise.
524         * pretty-print.c (pp_newline_and_flush): Define.
525         (pp_newline_and_indent): Likewise.
526         (pp_separate_with): Likewise.
528 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
530         PR target/58218
531         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
532         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
534 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
536         * gcc/config/i386/predicates.md (ext_sse_reg_operand): New.
537         * gcc/config/i386/i386.md (*movti_internal): Use
538         predicate to determine if EVEX is needed.
539         (*movsi_internal): Ditto.
540         (*movdf_internal): Ditto.
541         (*movsf_internal): Ditto.
542         * gcc/config/i386/mmx.md (*mov<mode>_internal): Ditto.
544 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
546         PR tree-optimization/58209
547         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
548         (find_tail_calls): Give up for pointer result types if m is non-NULL.
549         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
550         emit POINTER_PLUS_EXPR.
551         (create_tailcall_accumulator): For pointer result type accumulate in
552         sizetype type.
554 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
556         * configure.ac: Add backslashes missing from the last change.
557         * configure: Regenerate.
559 2013-08-22  Jan Hubicka  <jh@susue.cz>
561         * ipa.c (function_and_variable_visibility): First remember function
562         was global and then make it local.
564 2013-08-22  Julian Brown  <julian@codesourcery.com>
566         * configure.ac: Add aarch64 to list of arches which use "nop" in
567         debug_line test.
568         * configure: Regenerate.
570 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
572         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
573         gnu_libc_has_function.
574         * config/s390/tpf.h: Likewise.
576 2013-08-22  Jan Hubicka  <jh@susue.cz>
578         * timevar.c (validate_phases): Add cast.
580 2013-08-22  Jan Hubicka  <jh@susue.cz>
582         * timevar.c (validate_phases): Use size_t for memory.
583         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
585 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
587         * pretty-print.h (output_buffer::output_buffer): Declare.
588         (pretty_printer::pretty_printer): Likewise.
589         (pp_construct): Remove.
590         * pretty-print.c (output_buffer::output_buffer): Define.
591         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
592         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
593         (print_gimple_expr): Likewise.
594         (print_gimple_seq): Likewise.
595         (gimple_dump_bb): Likewise.
596         * sched-vis.c (dump_value_slim): Likewise.
597         (dump_insn_slim): Likewise.
598         (dump_rtl_slim): Likewise.
599         (str_pattern_slim): Likewise.
600         * tree-mudflap.c (mf_varname_tree): Likewise.
601         * graph.c (print_graph_cfg): Likewise.
602         (start_graph_dump): Likewise.
603         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
604         placement-new.
605         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
606         pretty printer initialization.
607         * coretypes.h (diagnostic_context): Remove superflous type alias
608         declaration.
609         (pretty_printer): Likewise.  Declare directly as a class.
610         (pretty_print_info): Remove declaration as class.
611         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
612         and pp_clear_output_area.
613         (asan_add_global): Likewise.
615 2013-08-22  Jan Hubicka  <jh@suse.cz>
617         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
618         * ipa-utils.h (update_type_inheritance_graph): Declare.
619         (possible_polymorphic_call_target_p): Declare.
620         (possible_polymorphic_call_target_p): New.
621         * ipa-devirt.c: Update toplevel comments.
622         (cached_polymorphic_call_targets): Move up.
623         (odr_type_d): Move ID down.
624         (polymorphic_type_binfo_p): Update comment.
625         (odr_hasher::remove): Likewise;
626         (get_odr_type): Set anonymous_namespace.
627         (dump_odr_type): Dump it.
628         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
629         (maybe_record_node): Record node in cached_polymorphic_call_targets.
630         (record_binfo): Add comment.
631         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
632         built.
633         (devirt_node_removal_hook): Do not iCE when cache is freed.
634         (possible_polymorphic_call_target_p): New predicate.
635         (update_type_inheritance_graph): New function.
637 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
638             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
639             Sergey Lega  <sergey.s.lega@intel.com>
640             Anna Tikhonova  <anna.tikhonova@intel.com>
641             Ilya Tocar  <ilya.tocar@intel.com>
642             Andrey Turetskiy  <andrey.turetskiy@intel.com>
643             Ilya Verbin  <ilya.verbin@intel.com>
644             Kirill Yukhin  <kirill.yukhin@intel.com>
645             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
647         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
648         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
649         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
650         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
651         (OPTION_MASK_ISA_AVX2_UNSET): Update.
652         (OPTION_MASK_ISA_AVX512F_UNSET): New.
653         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
654         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
655         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
656         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
657         OPT_mavx512pf, OPT_mavx512er cases.
658         * config/i386/constraints.md (v): New constraint.
659         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
660         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
661         (bit_AVX512CD): New.
662         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
663         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
664         * config/i386/i386-c.c (ix86_target_macros_internal):
665         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
666         __AVX512PF__.
667         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
668         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
669         * config/i386/i386.c (regclass_map, dbx_register_map)
670         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
671         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
672         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
673         -mavx512pf options.
674         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
675         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
676         -mavx512pf options.  Fix formatting.
677         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
678         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
679         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
680         -mavx512cd, -mavx512pf options.
681         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
682         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
683         (ix86_preferred_output_reload_class): Replace SSE_REGS with
684         ALL_SSE_REGS.
685         (ix86_hard_regno_mode_ok): Support 512-bit registers.
686         (ix86_set_reg_reg_cost): Ditto.
687         (x86_order_regs_for_local_alloc): Ditto.
688         (MAX_VECT_LEN): Extend to 64-byte.
689         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
690         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
691         (TARGET_AVX512ER, TARGET_AVX512CD): New.
692         (BIGGEST_ALIGNMENT): Extend to 512-bits.
693         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
694         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
695         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
696         (SSE_REG_MODE_P): Support new modes.
697         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
698         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
699         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
700         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
701         (REG_CLASS_CONTENTS): Add new registers.
702         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
703         (EXT_REX_SSE_REGNO_P): New.
704         (HI_REGISTER_NAMES): Add new registers.
705         * config/i386/i386.md: Define constants for new registers.
706         (mode): Add new 512-bit modes.
707         (prefix): Support evex prefix.
708         (isa): Support avx512f, noavx512f, fma_avx512f.
709         (ssemodesuffix): Add new 512-bit modes.
710         (movxi): New.
711         (*movxi_internal_avx512f): Ditto.
712         (*movdi_internal): Replace constraint "x" with the new constraint "v".
713         Support MODE_XI.
714         (*movsi_internal): Likewise.
715         (*movdf_internal): Likewise.
716         (*movsf_internal): Likewise.
717         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
718         (<code><mode>3): Likewise.
719         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
720         New.
721         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
722         with the new constraint "v".
723         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
724         modes.
725         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
726         with the new constraint "v".
727         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
728         (<sse2>_storedqu<avxsizesuffix>): Likewise.
729         (*<plusminus_insn><mode>3): Likewise.
730         (<sse>_vm<plusminus_insn><mode>3): Likewise.
731         (*mul<mode>3): Likewise.
732         (<sse>_vmmul<mode>3): Likewise.
733         (<sse>_div<mode>3): Likewise.
734         (<sse>_vmdiv<mode>3): Likewise.
735         (<sse>_sqrt<mode>2): Likewise.
736         (<sse>_vmsqrt<mode>2): Likewise.
737         (*<code><mode>3_finite): Likewise.
738         (*<code><mode>3) <smaxmin>: Likewise.
739         (<sse>_vm<code><mode>3): Likewise.
740         (*<code><mode>3) <any_logic>: Likewise.
741         (*fma_fmadd_<mode>): Likewise.
742         (*fma_fmsub_<mode>): Likewise.
743         (*fma_fnmadd_<mode>): Likewise.
744         (*fma_fnmsub_<mode>): Likewise.
745         (*fma_fmaddsub_<mode>): Likewise.
746         (*fma_fmsubadd_<mode>): Likewise.
747         (*fmai_fmadd_<mode>): Likewise.
748         (*fmai_fmsub_<mode>): Likewise.
749         (*fmai_fnmadd_<mode>): Likewise.
750         (*fmai_fnmsub_<mode>): Likewise.
751         (sse_cvtsi2ss): Likewise.
752         (sse_cvtsi2ssq): Likewise.
753         (sse_cvtss2si): Likewise.
754         (sse_cvtss2si_2): Likewise.
755         (sse_cvtss2siq): Likewise.
756         (sse_cvtss2siq_2): Likewise.
757         (sse_cvttss2si): Likewise.
758         (sse_cvtss2siq_2): Likewise.
759         (float<sseintvecmodelower><mode>2): Likewise.
760         (sse2_cvtsd2si_2): Likewise.
761         (sse2_cvtsd2siq_2): Likewise.
762         (*<plusminus_insn><mode>3): Likewise.
763         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
764         (*<sse4_1_avx2>_mul<mode>3): Likewise.
765         (ashr<mode>3): Likewise.
766         (<shift_insn><mode>3): Likewise.
767         (avx2_<code><mode>3): Likewise.
768         (*avx2_<code><mode>3): Likewise.
769         (*andnot<mode>3): Likewise.
770         (*<code><mode>3) <any_logic>: Likewise.
771         (abs<mode>2): Likewise.
772         (avx2_permvar<mode>): Likewise.
773         (avx2_perm<mode>_1): Likewise.
774         (*avx_vpermilp<mode>): Likewise.
775         (avx_vpermilvar<mode>3): Likewise.
776         (avx2_ashrv<mode>): Likewise.
777         (avx2_<shift_insn>v<mode>): Likewise.
778         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
779         -mavx512cd.
780         * doc/rtl.texi: Document XImode.
782 2013-08-21  Jeff Law  <law@redhat.com>
784         * tree-flow.h (register_jump_thread): Pass vector of edges
785         instead of each important edge.
786         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
787         thread path into a vector and pass that to register_jump_thread.
788         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
789         passed in edge vector to the current 3-edge form.
791         Revert:
792         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
794         * dce.c (fini_dce): Call df_analyze again just in case
795         delete_unmarked_insns removed anything.
797 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
799         * reload.h (struct reg_equivs): Rename to ..
800         (struct reg_equivs_s): .. this.
802 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
804         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
806 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
808         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
810 2013-08-21  Jeff Law  <law@redhat.com>
812         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
813         simplify assignments too.  If the RHS collapses to a singleton
814         range, then return the value for the range.
816 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
818         * config/i386/sse.md (V16): Rename to...
819         (VMOVE): this.
820         (mov<mode>): Update iterator name.
821         (*mov<mode>_internal): Ditto.
822         (push<mode>1): Ditto.
823         (movmisalign<mode>): Ditto.
825 2013-08-20  Jan Hubicka  <jh@suse.cz>
827         PR bootstrap/58186
828         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
829         entry for direct edges.
830         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
832 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
834         Revert my last two changes, r201865 and r201864:
836         Revert r201865:
837         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
839         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
840         instances can own GC refs.
842         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
843         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
844         (gcc::context::gt_pch_nx): Likewise.
845         (gcc::context::gt_pch_nx):  Likewise.
846         * ggc.h (gt_ggc_mx <T>): New.
847         (gt_pch_nx_with_op <T>): New.
848         (gt_pch_nx <T>): New.
849         * passes.c (opt_pass::gt_ggc_mx): New.
850         (opt_pass::gt_pch_nx): New.
851         (opt_pass::gt_pch_nx_with_op): New.
852         (pass_manager::gt_ggc_mx): New.
853         (pass_manager::gt_pch_nx): New.
854         (pass_manager::gt_pch_nx_with_op): New.
855         (pass_manager::operator new): Use
856         ggc_internal_cleared_alloc_stat rather than xcalloc.
857         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
858         (pass_manager::gt_ggc_mx): New.
859         (pass_manager::gt_pch_nx): New.
860         (pass_manager::gt_pch_nx_with_op): New.
861         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
862         (opt_pass::operator new): New.
863         (opt_pass::gt_ggc_mx): New.
864         (opt_pass::gt_pch_nx): New.
865         (opt_pass::gt_pch_nx_with_op): New.
867         Revert r201864:
868         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
870         * Makefile.in (GTFILES): Add context.h.
871         * context.c (gcc::context::operator new): New.
872         (gcc::context::gt_ggc_mx): New.
873         (gcc::context::gt_pch_nx): New.
874         (gcc::context::gt_pch_nx): New.
875         * context.h (gcc::context): Add GTY((user)) marking.
876         (gcc::context::operator new): New.
877         (gcc::context::gt_ggc_mx): New.
878         (gcc::context::gt_pch_nx): New.
879         (gcc::context::gt_pch_nx): New.
880         (g): Add GTY marking.
881         (gt_ggc_mx (gcc::context *)): New.
882         (gt_pch_nx (gcc::context *)): New.
883         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
884         void *cookie)): New.
885         * gengtype.c (open_base_files) <ifiles>: Add context.h.
887 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
889         * dce.c (fini_dce): Call df_analyze again just in case
890         delete_unmarked_insns removed anything.
892 2013-08-20  Teresa Johnson  <tejohnson@google.com>
894         PR rtl-optimizations/57451
895         * final.c (reemit_insn_block_notes): Prevent lexical blocks
896         from crossing split section boundaries.
898 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
900         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
901         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
902         with MULTLIB_DEFAULTS.
904 2013-08-20  Nick Clifton  <nickc@redhat.com>
906         * target.def (narrow_volatile_bitfield): Note that the default
907         value is false, not !TARGET_STRICT_ALIGN.
908         * doc/tm.texi: Regenerate.
910 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
912         Fix LIB_SPEC for systems without libpthread.
914         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
915         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
916         for Android.
917         * config/i386/linux-common.h: Likewise.
918         * config/mips/linux-common.h: Likewise.
920 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
922         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
923         checks.
925 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
927         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
928         instances can own GC refs.
930         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
931         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
932         (gcc::context::gt_pch_nx): Likewise.
933         (gcc::context::gt_pch_nx):  Likewise.
934         * ggc.h (gt_ggc_mx <T>): New.
935         (gt_pch_nx_with_op <T>): New.
936         (gt_pch_nx <T>): New.
937         * passes.c (opt_pass::gt_ggc_mx): New.
938         (opt_pass::gt_pch_nx): New.
939         (opt_pass::gt_pch_nx_with_op): New.
940         (pass_manager::gt_ggc_mx): New.
941         (pass_manager::gt_pch_nx): New.
942         (pass_manager::gt_pch_nx_with_op): New.
943         (pass_manager::operator new): Use
944         ggc_internal_cleared_alloc_stat rather than xcalloc.
945         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
946         (pass_manager::gt_ggc_mx): New.
947         (pass_manager::gt_pch_nx): New.
948         (pass_manager::gt_pch_nx_with_op): New.
949         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
950         (opt_pass::operator new): New.
951         (opt_pass::gt_ggc_mx): New.
952         (opt_pass::gt_pch_nx): New.
953         (opt_pass::gt_pch_nx_with_op): New.
955 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
957         * Makefile.in (GTFILES): Add context.h.
958         * context.c (gcc::context::operator new): New.
959         (gcc::context::gt_ggc_mx): New.
960         (gcc::context::gt_pch_nx): New.
961         (gcc::context::gt_pch_nx): New.
962         * context.h (gcc::context): Add GTY((user)) marking.
963         (gcc::context::operator new): New.
964         (gcc::context::gt_ggc_mx): New.
965         (gcc::context::gt_pch_nx): New.
966         (gcc::context::gt_pch_nx): New.
967         (g): Add GTY marking.
968         (gt_ggc_mx (gcc::context *)): New.
969         (gt_pch_nx (gcc::context *)): New.
970         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
971         void *cookie)): New.
972         * gengtype.c (open_base_files) <ifiles>: Add context.h.
974 2013-08-20  Alan Modra  <amodra@gmail.com>
976         PR target/57865
977         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
978         (rs6000_emit_epilogue): Likewise.
980 2013-08-19  Dehao Chen  <dehao@google.com>
982         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
984 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
985             Jakub Jelinek  <jakub@redhat.com>
987         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
988         (BUILT_IN_FABSD64): Likewise.
989         (BUILT_IN_FABSD128): Likewise.
990         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
991         (fold_builtin_1): Likewise.
992         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
993         destination and source operands.
994         (*abstd2_fpr): Likewise.
995         (*nabstd2_fpr): Likewise.
997 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
999         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
1000         JUMP_P and INSN_P.
1002 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
1004         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
1005         incomplete.
1007 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1009         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
1010         * builtins.c (default_libc_has_function): New.
1011         (gnu_libc_has_function): Ditto.
1012         (no_c99_libc_has_function): Ditto.
1013         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
1014         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
1015         (fold_builtin_sincos): Likewise.
1016         (fold_builtin_cexp): Likewise.
1017         * builtins.def (DEF_C94_BUILTIN): Likewise.
1018         (DEF_C99_BUILTIN): Likewise.
1019         (DEF_C99_C90RES_BUILTIN): Likewise.
1020         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
1021         definitions to using this define.
1022         * config/darwin-protos.h (darwin_libc_has_function): New.
1023         * config/darwin.c (darwin_libc_has_function): Ditto.
1024         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
1025         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
1026         * config/darwin.h: Ditto.
1027         * config/elfos.h: Ditto.
1028         * config/freebsd.h: Ditto.
1029         * config/i386/cygming.h: Ditto.
1030         * config/i386/djgpp.h: Ditto.
1031         * config/i386/i386-interix.h: Ditto.
1032         * config/microblaze/microblaze.h: Ditto.
1033         * config/mmix/mmix.h: Ditto.
1034         * config/gnu-user.h: Ditto.
1035         * config/ia64/hpux.h: Ditto.
1036         * config/pa/pa-hpux.h: Ditto.
1037         * config/pdp11/pdp11.h: Ditto.
1038         * config/picochip/picochip.h: Ditto.
1039         * config/linux.h: Ditto.
1040         * config/netbsd.h: Ditto.
1041         * config/openbsd.h: Ditto.
1042         * config/rs6000/aix43.h: Ditto.
1043         * config/rs6000/aix51.h: Ditto.
1044         * config/rs6000/aix52.h: Ditto.
1045         * config/rs6000/aix53.h: Ditto.
1046         * config/rs6000/aix61.h: Ditto.
1047         * config/rs6000/darwin.h: Ditto.
1048         * config/rs6000/linux.h: Ditto.
1049         * config/rs6000/linux64.h: Ditto.
1050         * config/s390/tpf.h: Ditto.
1051         * config/sol2-10.h: Ditto.
1052         * config/sol2.h: Ditto.
1053         * config/vms/vms.h: Ditto.
1054         * config/vxworks.h: Ditto.
1055         * config/linux-android.c (linux_android_libc_has_function):
1056         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
1057         * config/linux-protos.h (linux_android_libc_has_function):
1058         New declaration.
1059         * config/i386/i386.c (ix86_libc_has_function): New.
1060         * config/i386/i386-protos.h
1061         (ix86_libc_has_function): New declaration.
1062         * config/i386/i386.md
1063         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
1064         ("isinf<mode>2): Likewise.
1065         * convert.c (convert_to_integer): Using new target hook
1066         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
1067         TARGET_C99_FUNCTIONS.
1068         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
1069         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
1070         * coretypes.h (function_class): New enum for different
1071         classes of functions.
1072         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
1073         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
1074         (TARGET_HAS_SINCOS): Likewise.
1075         (TARGET_LIBC_HAS_FUNCTION): New.
1076         * doc/tm.texi: Regenerated.
1077         * targhooks.h (default_libc_has_function): New declaration.
1078         (no_c99_libc_has_function): Ditto.
1079         (gnu_libc_has_function): Ditto.
1080         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
1081         and TARGET_HAS_SINCOS.
1083 2013-08-18  Jan Hubicka  <jh@suse.cz>
1085         * Makeifle-in (ipa-devirt.o): New.
1086         (GTFILES): Add ipa-utils.h and ipa-devirt.c
1087         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
1088         (analyze_functions): Look into possible targets of polymorphic call.
1089         * dumpfile.c (dump_files): Add type-inheritance dump.
1090         * dumpfile.h (TDI_inheritance): New.
1091         * ipa-devirt.c: New file.
1092         * ipa-utils.h (odr_type_d): Forward declare.
1093         (odr_type): New type.
1094         (build_type_inheritance_graph): Declare.
1095         (possible_polymorphic_call_targets): Declare and introduce inline
1096         variant when only edge is pased.
1097         (dump_possible_polymorphic_call_targets): Likewise.
1098         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
1099         * tree.c (type_in_anonymous_namespace_p): Break out from ...
1100         (types_same_for_odr): ... here.
1101         * tree.h (type_in_anonymous_namespace_p): Declare.
1103 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
1105         PR tree-optimization/58006
1106         * tree-parloops.c (take_address_of): Don't ICE if get_name
1107         returns NULL.
1108         (eliminate_local_variables_stmt): Remove clobber stmts.
1110 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
1112         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
1113         error message is issued for an alias to undefined symbol.
1115 2013-08-18  Jan Hubicka  <jh@suse.cz>
1117         * cgraph.c (cgraph_create_indirect_edge): Discover
1118         polymorphic calls and record basic info into indirect_info.
1119         * gimple-fold.c (gimple_fold_call): When doing BINFO based
1120         devirtualization, ignore objc function calls.
1121         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
1122         call with no parm index info.
1123         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
1124         * tree.c (virtual_method_call_p): New function.
1125         * tree.h (virtual_method_call_p): Declare.
1127 2013-08-16  Jan Hubicka  <jh@suse.cz>
1129         PR middle-end/58179
1130         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
1132 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
1134         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
1135         attribute.
1137 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1139         * gengtype.c (type_for_name): Add special-case support for
1140         locating types within the "gcc::" namespace.
1141         (open_base_files): Emit a "using namespace gcc" directive.
1143 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
1145         PR target/58160
1146         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
1147         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
1149         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
1150         array instead of each individual operand as a separate argument.
1151         (emit_fusion_gpr_load): Likewise.
1152         (expand_fusion_gpr_load): Add new function declaration.
1154         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
1155         signature to have the operands passed as an array, instead of as
1156         separate arguments.  Allow ZERO_EXTEND to be in the memory
1157         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
1158         depend on the register live/dead flags when peepholes are run.
1159         (expand_fusion_gpr_load): New function to be called from the
1160         peephole2 pass, to change the register that addis sets to be the
1161         target register.
1162         (emit_fusion_gpr_load): Change the calling signature to have the
1163         operands passed as an array, instead of as separate arguments.
1164         Allow ZERO_EXTEND to be in the memory address, and also
1165         SIGN_EXTEND if -mpower8-fusion-sign.
1167         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
1168         unspec enumeration.
1169         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
1170         adjust the register addis loads up in the peephole2 pass.  Do not
1171         depend on the register live/dead state when the peephole pass is done.
1173 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
1175         * gengtype.c (create_user_defined_type): Ensure that the kind
1176         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
1177         declaration is seen before the GTY((user)) marking.
1179 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1181         PR target/58105
1182         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
1184 2013-08-16  Jan Hubicka  <jh@suse.cz>
1186         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
1187         arugment expected_type.
1188         (gimple_fold_call): Use it.
1189         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
1190         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
1191         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
1192         (try_make_edge_direct_virtual_call): Likewise.
1193         * tree.c (obj_type_ref_class): New.
1194         * tree.h (obj_type_ref_class): Use it.
1196 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
1198         * sched-vis.c (rtl_slim_pp_initialized): Remove.
1199         (rtl_slim_pp): Likewise.
1200         (init_rtl_slim_pretty_print): Likewise.
1201         (dump_value_slim):  Don't call it.  Use local pretty printer.
1202         (dump_insn_slim): Likewise.
1203         (dump_rtl_slim): Likewise.
1204         (str_pattern_slim): Likewise.
1205         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
1206         Simplify.
1208 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
1210         PR tree-optimization/58164
1211         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
1212         walk gimple_goto_dest of GIMPLE_GOTO.
1214         PR tree-optimization/58165
1215         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
1216         bi_call must be the last stmt in a bb, don't split_block, instead
1217         use fallthru edge from it and give up if there is none.
1218         Release conds vector when returning early.
1220 2013-08-14  Xinliang David Li  <davidxl@google.com>
1222         * config/i386/i386.c (ix86_option_override_internal):
1223         Remove unused variable and field.
1225 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1227         PR target/57949
1228         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
1229         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
1230         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
1231         and Linux, correct BLKmode alignment when 128-bit alignment is
1232         required and compatibility flag is not set.
1233         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
1234         for zero-size arguments when compatibility flag is not set.
1236 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1238         PR tree-optimization/58145
1239         * tree-sra.c (build_ref_for_offset): If prev_base has
1240         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
1242 2013-08-14  Xinliang David Li  <davidxl@google.com>
1244         * config/i386/i386.c (ix86_option_override_internal):
1245         Fix uninitialized variable error.
1247 2013-08-14  Xinliang David Li  <davidxl@google.com>
1249         * config/i386/i386.opt: Define two new options.
1250         * config/i386/x86-tune.def: Add arch selector field in macros.
1251         * config/i386/i386.h: Adjust macro definition.
1252         * config/i386/i386.c (ix86_option_override_internal):
1253         Refactor the code.
1254         (parse_mtune_ctrl_str): New function.
1255         (set_ix86_tune_features): New function.
1256         (ix86_function_specific_restore): Call the new helper function.
1258 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
1260         PR rtl-optimization/57662
1261         * sel-sched.c (code_motion_process_successors): When the current insn
1262         is removed after the recursive traversal, break from the loop.
1263         Add comments and debug printouts.
1265 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
1266             Alexandre Oliva  <aoliva@redhat.com>
1268         PR target/58067
1269         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
1270         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
1271         there also UNSPEC_PLTOFF.
1273 2013-08-14  Marek Polacek  <polacek@redhat.com>
1275         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
1276         NUM_CONDITIONS bit positions.
1278 2013-08-13  Cary Coutant  <ccoutant@google.com>
1280         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
1281         (attr_checksum): Hash vector contents instead of pointer.
1282         (attr_checksum_ordered): Likewise.
1284 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
1286         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
1287         when Pmode != word_mode.  Add length_address attribute.
1288         (sse3_monitor_<mode>): Merge from sse3_monitor and
1289         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
1290         Pmode != word_mode.  Update insn length attribute.
1291         * config/i386/i386.c (ix86_option_override_internal): Update
1292         ix86_gen_monitor selection for merged sse3_monitor insn.
1294 2013-08-13  Julian Brown  <julian@codesourcery.com>
1296         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
1297         perform invalid legitimization on greater-than-word-size modes for
1298         TARGET_E500_DOUBLE.
1300 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
1302         * ira.c (setup_class_translate_array): Use aclass instead of cl
1303         for classes not fully covered by allocno classes.
1305 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
1307         PR tree-optimization/57661
1308         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
1309         * tree-inline.c (tree_function_versioning): Initialize it.
1310         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
1311         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
1312         that is not being copied.
1314         PR sanitizer/56417
1315         * asan.c (instrument_strlen_call): Fix typo in comment.
1316         Use char * type even for the lhs of POINTER_PLUS_EXPR.
1318 2013-08-13  Steve Ellcey  <sellcey@mips.com>
1320         * config/mips/mips.md (prefetch): Use lw instead of ld on
1321         loongson in 32bit mode.
1323 2013-08-13  Nick Clifton  <nickc@redhat.com>
1325         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
1327 2013-08-13  Jan Hubicka  <jh@suse.cz>
1329         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
1330         introduced edge; fix typo in sanity check.
1331         (cgraph_resolve_speculation): Export; improve diagnostic.
1332         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
1333         speculation at type mismatch.
1334         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
1335         (cgraph_resolve_speculation): Declare.
1336         (symtab_can_be_discarded): New function.
1337         * value-prof.c (gimple_ic_transform): Remove actual transform code.
1338         * ipa-inline-transform.c (speculation_removed): New global var.
1339         (clone_inlined_nodes): See if speculation can be removed.
1340         (inline_call): If speculations was removed, we growths may not match.
1341         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
1342         (speculation_useful_p): New function.
1343         (resolve_noninline_speculation): New function.
1344         (inline_small_functions): Resolve useless speculations.
1345         * ipa-inline.h (speculation_useful_p): Declare
1346         * ipa.c (can_replace_by_local_alias): Simplify.
1347         (ipa_profile): Produce speculative calls in non-lto, too;
1348         add simple cost model; produce local aliases.
1350 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
1352         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
1353         PASS_MANAGER_H.
1355 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
1357         * config/i386/i386.c (ix86_function_versions): Use error + inform.
1359 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
1361         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
1362         iterator instead of X87MODEF.
1364 2013-08-12  Perez Read  <netfirewall@gmail.com>
1366         PR target/58132
1367         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
1368         operand 0 for intel asm alternative.
1369         (*movabs<mode>_2): Ditto for operand 1.
1371 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
1373         * config/aarch64/arm_none.h
1374         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
1376 2013-08-12  Nick Clifton  <nickc@redhat.com>
1378         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
1380 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
1382         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
1383         expand for QI/HImode operand to produce more effictive code for
1384         unsigned char(short) --> float(double) conversion.
1386 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
1388         * doc/invoke.texi: Mention that -ftls-model does not force the final
1389         model.
1391 2013-08-12  Marek Polacek  <polacek@redhat.com>
1392             Marc Glisse  <marc.glisse@inria.fr>
1394         PR tree-optimization/57980
1395         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
1396         when creating -1 constant.
1398 2013-08-10  Jan Hubicka  <jh@suse.cz>
1400         Workaround binutils PR14342.
1401         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
1402         (gimple_init_edge_profiler): Likewise.
1403         (gimple_gen_ic_func_profiler): Likewise.
1405 2013-08-09  Jan Hubicka  <jh@suse.cz>
1407         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
1409 2013-08-09  Xinliang David Li  <davidxl@google.com>
1411         * config/i386/stringop.def: New file.
1412         * config/i386/stringop.opt: New file.
1413         * config/i386/i386-opts.h: Include stringopt.def.
1414         * config/i386/i386.opt: Include stringopt.opt.
1415         * config/i386/i386.c (ix86_option_override_internal):
1416         Override default size based stringop inline strategies with options.
1417         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
1418         New function.
1420 2013-08-09  Jan Hubicka  <jh@suse.cz>
1422         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
1424 2013-08-09  Jan Hubicka  <jh@suse.cz>
1426         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
1427         CGRAPH_FREQ_MAX.
1428         (dump_cgraph_node): Dump profile-id.
1429         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
1430         and common_target_probability.
1431         * lto-cgraph.c (lto_output_edge): Stream common targets.
1432         (lto_output_node): Stream profile ids.
1433         (input_node): Stream profile ids.
1434         (input_edge): Stream common targets.
1435         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
1436         * ipa.c: Include value-prof.h
1437         (ipa_profile_generate_summary): Turn indirect call statement histograms
1438         into common targets.
1439         (ipa_profile): Turn common targets into speculative edges.
1441 2013-08-09  Jan Hubicka  <jh@suse.cz>
1443         * cgraph.h (cgraph_node): Add profile_id.
1444         * value-prof.c (cgraph_node_map): Turn into pointer_map.
1445         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
1446         (del_node_map): Update.
1447         (find_func_by_funcdef_no): Replace by ...
1448         (find_func_by_profile_id): ... this one.
1449         (gimple_ic_transform): Do not remove useful histograms when
1450         speculation is not done; dump info when indirect call removal
1451         can happen at LTO.
1452         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
1453         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
1454         (__gcov_indirect_call_profiler_v2): .. this one.
1455         * profile.h (init_node_map): Update.
1456         * coverage.c (coverage_compute_profile_id): New function.
1457         * coverage.h (coverage_compute_profile_id): Declare.
1458         * tree-profile.c (init_ic_make_global_vars): Make
1459         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
1460         (gimple_init_edge_profiler): Update prototype of
1461         __gcov_indirect_call_profiler.
1462         (gimple_gen_ic_func_profiler): Simplify.
1463         (tree_profiling): Use init_node_map
1465 2013-08-09  Jan Hubicka  <jh@suse.cz>
1467         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
1468         non-speculative refs.
1469         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
1470         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
1471         (cgraph_set_call_stmt): Likewise.
1472         (cgraph_create_edge_1): Fix release checking compilatoin;
1473         clear lto_stmt_uid.
1474         (cgraph_free_edge): Free indirect info.
1475         (cgraph_turn_edge_to_speculative): New function.
1476         (cgraph_speculative_call_info): New function.
1477         (cgraph_make_edge_direct): Return direct edge; handle speculation.
1478         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
1479         (dump_cgraph_node): Dump speculation.
1480         (verify_edge_count_and_frequency): Accept speculative edges.
1481         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
1482         (verify_cgraph_node): Handle speculation.
1483         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
1484         (cgraph_set_call_stmt): Update prototype.
1485         (cgraph_make_edge_direct): Update prototype.
1486         (cgraph_speculative_call_info): Declare.
1487         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
1488         to change; update call of ipa_find_references.
1489         * ipa-ref.c (ipa_record_reference): Fix return value; clear
1490         lto_stmt_uid and speculative flags.
1491         (ipa_dump_references): Dump speculation.
1492         (ipa_clone_references): Clone speculative flag.
1493         (ipa_clone_referring): Likewise.
1494         (ipa_clone_ref): New function.
1495         (ipa_find_reference): Look into lto_stmt_uids
1496         (ipa_clear_stmts_in_references): Do not clear speculative calls.
1497         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
1498         (ipa_find_reference): Update declaration.
1499         (ipa_clone_ref): Declare.
1500         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
1501         stream speculative flag.
1502         (lto_output_ref): Stream statements uids and speculation.
1503         (input_ref): Likewise.
1504         (input_edge): Stream speuclation.
1505         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
1506         (cgraph_set_call_stmt_including_clones): Handle speculation.
1507         * ipa-inline.c (heap_edge_removal_hook): New function.
1508         (inline_small_functions): Register it.
1509         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
1510         also initialize refs.
1511         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
1512         edge to change.
1513         (try_make_edge_direct_simple_call): Likewise.
1514         (try_make_edge_direct_simple_call): Likewise.
1515         (update_indirect_edges_after_inlining): Likewise.
1516         (remove_described_reference): Look proper lto_stmt_uid.
1517         (propagate_controlled_uses): Likewise.
1518         (propagate_controlled_uses): Liekwise.
1519         * tree-inline.c (copy_bb): Copy speculative edges.
1520         (redirect_all_calls): New function.
1521         (copy_cfg_body): Do redirection after loop info is updated.
1522         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
1524 2013-08-09  Jan Hubicka  <jh@suse.cz>
1526         * lto-streamer-out.c (output_function): Renumber PHIs.
1527         * lto-streamer-in.c (input_function): Likewise.
1529 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
1531         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
1532         (get_lane_unsigned): Likewise.
1533         (dup_lane_scalar): Likewise.
1534         (get_lane): enable for VALL.
1535         * config/aarch64/aarch64-simd.md
1536         (aarch64_dup_lane_scalar<mode>): Remove.
1537         (aarch64_get_lane_signed<mode>): Likewise.
1538         (aarch64_get_lane_unsigned<mode>): Likewise.
1539         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
1540         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
1541         (aarch64_get_lane<mode>): Enable for all vector modes.
1542         (aarch64_get_lanedi): Remove misleading constraints.
1543         * config/aarch64/arm_neon.h
1544         (__aarch64_vget_lane_any): Define.
1545         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
1546         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
1547         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
1548         * config/aarch64/iterators.md (VDQQH): New.
1549         (VDQQHS): Likewise.
1550         (vwcore): Likewise.
1552 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
1554         * configure.ac: Add GAS check for LEON instructions on SPARC.
1555         * configure: Regenerate.
1556         * config.in: Likewise.
1557         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
1558         sparc*-*-* block.
1559         * config/sparc/sparc.opt (LEON, LEON3): New masks.
1560         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
1561         for LEON or LEON3.
1562         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
1563         (AS_LEON_FLAG): New macro.
1564         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
1565         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
1566         Deal with LEON and LEON3 for the memory model.
1567         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
1568         (atomic_compare_and_swap<mode>_1): Likewise.
1569         (*atomic_compare_and_swap<mode>_1): Likewise.
1571 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
1573         * config/arm/neon.md (vcond): Fix floating-point vector
1574         comparisons against 0.
1576 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
1578         * lra-constraints.c (emit_spill_move): Remove assert.
1579         (process_alt_operands): Add more debugging
1580         output.  Increase reject for spilling into memory.  Decrease
1581         reject for reloading scratch.
1582         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
1584 2013-08-08  Steve Ellcey  <sellcey@mips.com>
1586         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
1587         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
1588         micromips incompatible.  Add nan2008.
1589         (MULTILIB_DIRNAMES): Add nan2008.
1590         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1591         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
1592         and micromips incompatible.  Add nan2008.
1593         (MULTILIB_DIRNAMES): Add nan2008.
1594         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
1596 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
1598         PR rtl-optimization/58079
1599         * combine.c (combine_simplify_rtx): Avoid using SUBST if
1600         simplify_comparison has widened a comparison with an integer.
1602 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1604         * config/arm/neon.md (movmisalign<mode>): Disable when we
1605         don't allow unaligned accesses.
1606         (*movmisalign<mode>_neon_store): Likewise.
1607         (*movmisalign<mode>_neon_load): Likewise.
1608         (*movmisalign<mode>_neon_store): Likewise.
1609         (*movmisalign<mode>_neon_load): Likewise.
1611 2013-08-08  Jan Hubicka  <jh@suse.cz>
1613         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
1614         (make_pass_rebuild_cgraph_edges): Also clear references.
1615         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
1616         * ipa-inline-transform.c (inline_transform): Remove all references
1617         after inlining.
1618         * cgraphunit.c (expand_function): Remove all references after
1619         expansion.
1620         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
1621         (ipa_find_reference): Rewrite to iterator.
1622         (remove_stmt_references): Likewise.
1623         (ipa_clear_stmts_in_references): New function.
1624         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
1625         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
1626         clear references.
1627         * ipa-split.c (split_function): Remove references in split function.
1629 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1631         PR target/57431
1632         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
1633         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
1635 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
1637         PR target/56979
1638         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1639         suggested mode for the assignment isn't compatible with the
1640         registers required.
1642 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
1644         PR target/58065
1645         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
1647 2013-08-07  Xinliang David Li  <davidxl@google.com>
1649         * config/i386/i386.opt: New option -mtune-ctrl=.
1650         * config/i386/x86-tune.def: New file.
1651         * config/i386/i386.h: include x86-tune.def.
1652         * config/i386/i386.c (ix86_option_override_internal):
1653         Parsing -mtune-ctrl= option and set tune features.
1655 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
1657         PR other/12081
1658         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
1659         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
1660         to gen_2arg_fn_t.
1662 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1664         * rtl.h (update_alignments): Declare.
1665         * final.c (grow_label_align): New function extracted from...
1666         (shorten_branches): ...here.  Call it.
1667         (update_alignments): New function.
1668         * reorg.c (sibling_labels): New variable.
1669         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
1670         the new label along with it onto the sibling_labels vector.
1671         (fill_simple_delay_slots): Adjust call to get_label_before.
1672         (fill_slots_from_thread): Likewise.
1673         (relax_delay_slots): Likewise.
1674         (make_return_insns): Likewise.
1675         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
1677 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
1679         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
1680         document its semantics.
1681         (diagnostic_report_diagnostic): Adjust accordingly.
1683 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1685         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
1686         (sparc_option_override): ...and port to new C++ pass API.
1687         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
1689 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
1691         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
1693 2013-08-06  Caroline Tice  <cmtice@google.com>
1695         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
1696         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
1697         * tree-pass.h: Add pass_vtable_verify.
1698         * varasm.c (assemble_variable): Add code to properly set the comdat
1699         section and name for the .vtable_map_vars section.
1700         (assemble_vtyv_preinit_initializer): New function.
1701         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
1702         LINK_ONCE flag.
1703         * output.h: Add function decl for assemble_vtv_preinit_initializer.
1704         * vtable-verify.c: New file.
1705         * vtable-verify.h: New file.
1706         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
1707         initialiation levels.
1708         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
1709         * passes.def: Insert pass_vtable_verify.
1710         * aclocal.m4: Reorder includes.
1711         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
1712         -fvtv-counts options.
1713         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
1714         as appropriate, if -fvtable-verify=... is used.
1715         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
1716         -fvtable-verify=... is used.
1717         * Makefile.in (OBJS):  Add vtable-verify.o to list.
1718         (vtable-verify.o): Add new build rule.
1719         (GTFILES): Add vtable-verify.c to list.
1720         * common.opt (fvtable-verify=): New flag.
1721         (vtv_priority): Values for fvtable-verify= flag.
1722         (fvtv-counts): New flag.
1723         (fvtv-debug): New flag.
1724         * tree.h (save_vtable_map_decl): New extern function decl.
1726 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1728         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
1729         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
1730         (pass_data_rl78_devirt): ...new pass_data instance and...
1731         (make_pass_rl78_devirt): ...new function.
1732         (rl78_asm_file_start): Port pass registration to new C++ API.
1734 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
1736         * coretypes.h (rtl_opt_pass): Add.
1737         (gcc::context): Add.
1738         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
1739         (epiphany_init): Port to new C++ pass API.
1740         (epiphany_optimize_mode_switching): Likewise.
1741         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
1742         (pass_manager::get_pass_mode_switching): New.
1743         (pass_manager::get_pass_peephole2): New.
1744         * mode-switching.c (pass_mode_switching): Add clone method.
1745         * recog.c (pass_peephole2): Add clone method.
1746         (pass_split_all_insns): Add clone method.
1748 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
1750         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
1751         (mips_option_override): ...here, porting to new C++ API for passes.
1753 2013-08-06  Jan Hubicka  <jh@suse.cz>
1755         * cgraph.c (cgraph_get_body): New function based on lto.c
1756         implementation.
1757         * cgraph.h (cgraph_get_body): Declare.
1758         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
1759         LTO paths.
1760         * cgraphunit.c (expand_function): Get body prior expanding.
1761         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
1762         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
1763         really need.
1764         * passes.c (do_per_function_toporder): Get body.
1765         * tree-inline.c (expand_call_inline): Get body prior inlining it.
1766         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
1768 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1770         PR fortran/57987
1771         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
1772         is not re-finalized.  Rename second parameter to no_collect.
1774 2013-08-06  Martin Jambor  <mjambor@suse.cz>
1776         PR middle-end/58041
1777         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
1778         MEM_REF has proper alignment information.
1780 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
1782         PR other/12081
1783         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
1784         class insn_gen_fn.
1785         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
1786         rtx (*) (rtx, ...) with insn_gen_fn.
1787         * genoutput.c (output_insn_data): Cast gen_? function pointers to
1788         insn_gen_fn::stored_funcptr.  Add initializer braces.
1790 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1792         Rewrite how instances of passes are cloned to remove assumptions
1793         about their sizes (thus allowing pass subclasses to have
1794         additional data fields, albeit non-GC-managed ones at this point).
1796         * passes.c (make_pass_instance): Now that passes have clone
1797         methods, rewrite this function to eliminate XNEW and memcpy
1798         calls that used hardcoded sizes.  Since this function no longer
1799         creates pass instances, rename it to...
1800         (add_pass_instance): ...this.  Document the old way that passes were
1801         numbered and flagged, and rework this function to continue using it.
1802         (next_pass_1): Add an initial_pass argument for use by
1803         add_pass_instance.
1804         (position_pass): When adding multiple instances of a pass, use
1805         the pass's clone method, rather than relying on the XNEW/memcpy
1806         within the former make_pass_instance (now add_pass_instance).
1807         (pass_manager::pass_manager): When invoking next_pass_1, also supply
1808         the initial instance of the current pass within the pass manager.
1810 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
1812         This is the automated part of the conversion of passes from C
1813         structs to C++ classes.
1815         Patch autogenerated by refactor_passes.py from
1816         https://github.com/davidmalcolm/gcc-refactoring-scripts
1817         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
1819         * asan.c (pass_asan): Convert from a global struct to a subclass of
1820         gimple_opt_pass along with...
1821         (pass_data_asan): ...new pass_data instance and...
1822         (make_pass_asan): ...new function.
1823         (pass_asan_O0): Convert from a global struct to a subclass of
1824         gimple_opt_pass along with...
1825         (pass_data_asan_O0): ...new pass_data instance and...
1826         (make_pass_asan_O0): ...new function.
1827         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
1828         subclass of rtl_opt_pass along with...
1829         (pass_data_inc_dec): ...new pass_data instance and...
1830         (make_pass_inc_dec): ...new function.
1831         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
1832         a subclass of rtl_opt_pass along with...
1833         (pass_data_reorder_blocks): ...new pass_data instance and...
1834         (make_pass_reorder_blocks): ...new function.
1835         (pass_duplicate_computed_gotos): Convert from a global struct to a
1836         subclass of rtl_opt_pass along with...
1837         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
1838         (make_pass_duplicate_computed_gotos): ...new function.
1839         (pass_partition_blocks): Convert from a global struct to a subclass of
1840         rtl_opt_pass along with...
1841         (pass_data_partition_blocks): ...new pass_data instance and...
1842         (make_pass_partition_blocks): ...new function.
1843         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
1844         struct to a subclass of rtl_opt_pass along with...
1845         (pass_data_branch_target_load_optimize1): ...new pass_data instance
1846         and...
1847         (make_pass_branch_target_load_optimize1): ...new function.
1848         (pass_branch_target_load_optimize2): Convert from a global struct to a
1849         subclass of rtl_opt_pass along with...
1850         (pass_data_branch_target_load_optimize2): ...new pass_data instance
1851         and...
1852         (make_pass_branch_target_load_optimize2): ...new function.
1853         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
1854         of rtl_opt_pass along with...
1855         (pass_data_jump): ...new pass_data instance and...
1856         (make_pass_jump): ...new function.
1857         (pass_jump2): Convert from a global struct to a subclass of
1858         rtl_opt_pass along with...
1859         (pass_data_jump2): ...new pass_data instance and...
1860         (make_pass_jump2): ...new function.
1861         * cfgexpand.c (pass_expand): Convert from a global struct to a
1862         subclass of rtl_opt_pass along with...
1863         (pass_data_expand): ...new pass_data instance and...
1864         (make_pass_expand): ...new function.
1865         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
1866         of rtl_opt_pass along with...
1867         (pass_data_free_cfg): ...new pass_data instance and...
1868         (make_pass_free_cfg): ...new function.
1869         (pass_into_cfg_layout_mode): Convert from a global struct to a
1870         subclass of rtl_opt_pass along with...
1871         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
1872         (make_pass_into_cfg_layout_mode): ...new function.
1873         (pass_outof_cfg_layout_mode): Convert from a global struct to a
1874         subclass of rtl_opt_pass along with...
1875         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
1876         (make_pass_outof_cfg_layout_mode): ...new function.
1877         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
1878         struct to a subclass of gimple_opt_pass along with...
1879         (pass_data_build_cgraph_edges): ...new pass_data instance and...
1880         (make_pass_build_cgraph_edges): ...new function.
1881         (pass_rebuild_cgraph_edges): Convert from a global struct to a
1882         subclass of gimple_opt_pass along with...
1883         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
1884         (make_pass_rebuild_cgraph_edges): ...new function.
1885         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
1886         subclass of gimple_opt_pass along with...
1887         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
1888         and...
1889         (make_pass_remove_cgraph_callee_edges): ...new function.
1890         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
1891         struct to a subclass of rtl_opt_pass along with...
1892         (pass_data_stack_adjustments): ...new pass_data instance and...
1893         (make_pass_stack_adjustments): ...new function.
1894         * combine.c (pass_combine): Convert from a global struct to a subclass
1895         of rtl_opt_pass along with...
1896         (pass_data_combine): ...new pass_data instance and...
1897         (make_pass_combine): ...new function.
1898         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
1899         global struct to a subclass of rtl_opt_pass along with...
1900         (pass_data_compare_elim_after_reload): ...new pass_data instance
1901         and...
1902         (make_pass_compare_elim_after_reload): ...new function.
1903         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
1904         of rtl_opt_pass along with...
1905         (pass_data_rtl_cprop): ...new pass_data instance and...
1906         (make_pass_rtl_cprop): ...new function.
1907         * cse.c (pass_cse): Convert from a global struct to a subclass of
1908         rtl_opt_pass along with...
1909         (pass_data_cse): ...new pass_data instance and...
1910         (make_pass_cse): ...new function.
1911         (pass_cse2): Convert from a global struct to a subclass of
1912         rtl_opt_pass along with...
1913         (pass_data_cse2): ...new pass_data instance and...
1914         (make_pass_cse2): ...new function.
1915         (pass_cse_after_global_opts): Convert from a global struct to a
1916         subclass of rtl_opt_pass along with...
1917         (pass_data_cse_after_global_opts): ...new pass_data instance and...
1918         (make_pass_cse_after_global_opts): ...new function.
1919         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
1920         of rtl_opt_pass along with...
1921         (pass_data_ud_rtl_dce): ...new pass_data instance and...
1922         (make_pass_ud_rtl_dce): ...new function.
1923         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
1924         rtl_opt_pass along with...
1925         (pass_data_fast_rtl_dce): ...new pass_data instance and...
1926         (make_pass_fast_rtl_dce): ...new function.
1927         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
1928         a subclass of rtl_opt_pass along with...
1929         (pass_data_df_initialize_opt): ...new pass_data instance and...
1930         (make_pass_df_initialize_opt): ...new function.
1931         (pass_df_initialize_no_opt): Convert from a global struct to a
1932         subclass of rtl_opt_pass along with...
1933         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
1934         (make_pass_df_initialize_no_opt): ...new function.
1935         (pass_df_finish): Convert from a global struct to a subclass of
1936         rtl_opt_pass along with...
1937         (pass_data_df_finish): ...new pass_data instance and...
1938         (make_pass_df_finish): ...new function.
1939         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
1940         rtl_opt_pass along with...
1941         (pass_data_rtl_dse1): ...new pass_data instance and...
1942         (make_pass_rtl_dse1): ...new function.
1943         (pass_rtl_dse2): Convert from a global struct to a subclass of
1944         rtl_opt_pass along with...
1945         (pass_data_rtl_dse2): ...new pass_data instance and...
1946         (make_pass_rtl_dse2): ...new function.
1947         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
1948         subclass of rtl_opt_pass along with...
1949         (pass_data_dwarf2_frame): ...new pass_data instance and...
1950         (make_pass_dwarf2_frame): ...new function.
1951         * except.c (pass_set_nothrow_function_flags): Convert from a global
1952         struct to a subclass of rtl_opt_pass along with...
1953         (pass_data_set_nothrow_function_flags): ...new pass_data instance
1954         and...
1955         (make_pass_set_nothrow_function_flags): ...new function.
1956         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
1957         subclass of rtl_opt_pass along with...
1958         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
1959         and...
1960         (make_pass_convert_to_eh_region_ranges): ...new function.
1961         * final.c (pass_compute_alignments): Convert from a global struct to a
1962         subclass of rtl_opt_pass along with...
1963         (pass_data_compute_alignments): ...new pass_data instance and...
1964         (make_pass_compute_alignments): ...new function.
1965         (pass_final): Convert from a global struct to a subclass of
1966         rtl_opt_pass along with...
1967         (pass_data_final): ...new pass_data instance and...
1968         (make_pass_final): ...new function.
1969         (pass_shorten_branches): Convert from a global struct to a subclass of
1970         rtl_opt_pass along with...
1971         (pass_data_shorten_branches): ...new pass_data instance and...
1972         (make_pass_shorten_branches): ...new function.
1973         (pass_clean_state): Convert from a global struct to a subclass of
1974         rtl_opt_pass along with...
1975         (pass_data_clean_state): ...new pass_data instance and...
1976         (make_pass_clean_state): ...new function.
1977         * function.c (pass_instantiate_virtual_regs): Convert from a global
1978         struct to a subclass of rtl_opt_pass along with...
1979         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
1980         (make_pass_instantiate_virtual_regs): ...new function.
1981         (pass_leaf_regs): Convert from a global struct to a subclass of
1982         rtl_opt_pass along with...
1983         (pass_data_leaf_regs): ...new pass_data instance and...
1984         (make_pass_leaf_regs): ...new function.
1985         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
1986         subclass of rtl_opt_pass along with...
1987         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
1988         and...
1989         (make_pass_thread_prologue_and_epilogue): ...new function.
1990         (pass_match_asm_constraints): Convert from a global struct to a
1991         subclass of rtl_opt_pass along with...
1992         (pass_data_match_asm_constraints): ...new pass_data instance and...
1993         (make_pass_match_asm_constraints): ...new function.
1994         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
1995         subclass of rtl_opt_pass along with...
1996         (pass_data_rtl_fwprop): ...new pass_data instance and...
1997         (make_pass_rtl_fwprop): ...new function.
1998         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
1999         rtl_opt_pass along with...
2000         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
2001         (make_pass_rtl_fwprop_addr): ...new function.
2002         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
2003         rtl_opt_pass along with...
2004         (pass_data_rtl_pre): ...new pass_data instance and...
2005         (make_pass_rtl_pre): ...new function.
2006         (pass_rtl_hoist): Convert from a global struct to a subclass of
2007         rtl_opt_pass along with...
2008         (pass_data_rtl_hoist): ...new pass_data instance and...
2009         (make_pass_rtl_hoist): ...new function.
2010         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
2011         subclass of gimple_opt_pass along with...
2012         (pass_data_lower_cf): ...new pass_data instance and...
2013         (make_pass_lower_cf): ...new function.
2014         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
2015         from a global struct to a subclass of gimple_opt_pass along with...
2016         (pass_data_strength_reduction): ...new pass_data instance and...
2017         (make_pass_strength_reduction): ...new function.
2018         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
2019         of rtl_opt_pass along with...
2020         (pass_data_rtl_ifcvt): ...new pass_data instance and...
2021         (make_pass_rtl_ifcvt): ...new function.
2022         (pass_if_after_combine): Convert from a global struct to a subclass of
2023         rtl_opt_pass along with...
2024         (pass_data_if_after_combine): ...new pass_data instance and...
2025         (make_pass_if_after_combine): ...new function.
2026         (pass_if_after_reload): Convert from a global struct to a subclass of
2027         rtl_opt_pass along with...
2028         (pass_data_if_after_reload): ...new pass_data instance and...
2029         (make_pass_if_after_reload): ...new function.
2030         * init-regs.c (pass_initialize_regs): Convert from a global struct to
2031         a subclass of rtl_opt_pass along with...
2032         (pass_data_initialize_regs): ...new pass_data instance and...
2033         (make_pass_initialize_regs): ...new function.
2034         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
2035         of ipa_opt_pass_d along with...
2036         (pass_data_ipa_cp): ...new pass_data instance and...
2037         (make_pass_ipa_cp): ...new function.
2038         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
2039         global struct to a subclass of gimple_opt_pass along with...
2040         (pass_data_inline_parameters): ...new pass_data instance and...
2041         (make_pass_inline_parameters): ...new function.
2042         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
2043         subclass of gimple_opt_pass along with...
2044         (pass_data_early_inline): ...new pass_data instance and...
2045         (make_pass_early_inline): ...new function.
2046         (pass_ipa_inline): Convert from a global struct to a subclass of
2047         ipa_opt_pass_d along with...
2048         (pass_data_ipa_inline): ...new pass_data instance and...
2049         (make_pass_ipa_inline): ...new function.
2050         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
2051         struct to a subclass of gimple_opt_pass along with...
2052         (pass_data_local_pure_const): ...new pass_data instance and...
2053         (make_pass_local_pure_const): ...new function.
2054         (pass_ipa_pure_const): Convert from a global struct to a subclass of
2055         ipa_opt_pass_d along with...
2056         (pass_data_ipa_pure_const): ...new pass_data instance and...
2057         (make_pass_ipa_pure_const): ...new function.
2058         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
2059         to a subclass of ipa_opt_pass_d along with...
2060         (pass_data_ipa_reference): ...new pass_data instance and...
2061         (make_pass_ipa_reference): ...new function.
2062         * ipa-split.c (pass_split_functions): Convert from a global struct to
2063         a subclass of gimple_opt_pass along with...
2064         (pass_data_split_functions): ...new pass_data instance and...
2065         (make_pass_split_functions): ...new function.
2066         (pass_feedback_split_functions): Convert from a global struct to a
2067         subclass of gimple_opt_pass along with...
2068         (pass_data_feedback_split_functions): ...new pass_data instance and...
2069         (make_pass_feedback_split_functions): ...new function.
2070         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
2071         global struct to a subclass of simple_ipa_opt_pass along with...
2072         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
2073         instance and...
2074         (make_pass_ipa_function_and_variable_visibility): ...new function.
2075         (pass_ipa_free_inline_summary): Convert from a global struct to a
2076         subclass of simple_ipa_opt_pass along with...
2077         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
2078         (make_pass_ipa_free_inline_summary): ...new function.
2079         (pass_ipa_whole_program_visibility): Convert from a global struct to a
2080         subclass of ipa_opt_pass_d along with...
2081         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
2082         and...
2083         (make_pass_ipa_whole_program_visibility): ...new function.
2084         (pass_ipa_profile): Convert from a global struct to a subclass of
2085         ipa_opt_pass_d along with...
2086         (pass_data_ipa_profile): ...new pass_data instance and...
2087         (make_pass_ipa_profile): ...new function.
2088         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
2089         ipa_opt_pass_d along with...
2090         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
2091         (make_pass_ipa_cdtor_merge): ...new function.
2092         * ira.c (pass_ira): Convert from a global struct to a subclass of
2093         rtl_opt_pass along with...
2094         (pass_data_ira): ...new pass_data instance and...
2095         (make_pass_ira): ...new function.
2096         (pass_reload): Convert from a global struct to a subclass of
2097         rtl_opt_pass along with...
2098         (pass_data_reload): ...new pass_data instance and...
2099         (make_pass_reload): ...new function.
2100         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
2101         subclass of rtl_opt_pass along with...
2102         (pass_data_cleanup_barriers): ...new pass_data instance and...
2103         (make_pass_cleanup_barriers): ...new function.
2104         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
2105         of rtl_opt_pass along with...
2106         (pass_data_loop2): ...new pass_data instance and...
2107         (make_pass_loop2): ...new function.
2108         (pass_rtl_loop_init): Convert from a global struct to a subclass of
2109         rtl_opt_pass along with...
2110         (pass_data_rtl_loop_init): ...new pass_data instance and...
2111         (make_pass_rtl_loop_init): ...new function.
2112         (pass_rtl_loop_done): Convert from a global struct to a subclass of
2113         rtl_opt_pass along with...
2114         (pass_data_rtl_loop_done): ...new pass_data instance and...
2115         (make_pass_rtl_loop_done): ...new function.
2116         (pass_rtl_move_loop_invariants): Convert from a global struct to a
2117         subclass of rtl_opt_pass along with...
2118         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
2119         (make_pass_rtl_move_loop_invariants): ...new function.
2120         (pass_rtl_unswitch): Convert from a global struct to a subclass of
2121         rtl_opt_pass along with...
2122         (pass_data_rtl_unswitch): ...new pass_data instance and...
2123         (make_pass_rtl_unswitch): ...new function.
2124         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
2125         subclass of rtl_opt_pass along with...
2126         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
2127         and...
2128         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2129         (pass_rtl_doloop): Convert from a global struct to a subclass of
2130         rtl_opt_pass along with...
2131         (pass_data_rtl_doloop): ...new pass_data instance and...
2132         (make_pass_rtl_doloop): ...new function.
2133         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
2134         a subclass of rtl_opt_pass along with...
2135         (pass_data_lower_subreg): ...new pass_data instance and...
2136         (make_pass_lower_subreg): ...new function.
2137         (pass_lower_subreg2): Convert from a global struct to a subclass of
2138         rtl_opt_pass along with...
2139         (pass_data_lower_subreg2): ...new pass_data instance and...
2140         (make_pass_lower_subreg2): ...new function.
2141         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
2142         struct to a subclass of ipa_opt_pass_d along with...
2143         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
2144         (make_pass_ipa_lto_gimple_out): ...new function.
2145         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
2146         of ipa_opt_pass_d along with...
2147         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
2148         (make_pass_ipa_lto_finish_out): ...new function.
2149         * mode-switching.c (pass_mode_switching): Convert from a global struct
2150         to a subclass of rtl_opt_pass along with...
2151         (pass_data_mode_switching): ...new pass_data instance and...
2152         (make_pass_mode_switching): ...new function.
2153         * modulo-sched.c (pass_sms): Convert from a global struct to a
2154         subclass of rtl_opt_pass along with...
2155         (pass_data_sms): ...new pass_data instance and...
2156         (make_pass_sms): ...new function.
2157         * omp-low.c (pass_expand_omp): Convert from a global struct to a
2158         subclass of gimple_opt_pass along with...
2159         (pass_data_expand_omp): ...new pass_data instance and...
2160         (make_pass_expand_omp): ...new function.
2161         (pass_lower_omp): Convert from a global struct to a subclass of
2162         gimple_opt_pass along with...
2163         (pass_data_lower_omp): ...new pass_data instance and...
2164         (make_pass_lower_omp): ...new function.
2165         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
2166         of gimple_opt_pass along with...
2167         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
2168         (make_pass_diagnose_omp_blocks): ...new function.
2169         * passes.c (pass_early_local_passes): Convert from a global struct to
2170         a subclass of simple_ipa_opt_pass along with...
2171         (pass_data_early_local_passes): ...new pass_data instance and...
2172         (make_pass_early_local_passes): ...new function.
2173         (pass_all_early_optimizations): Convert from a global struct to a
2174         subclass of gimple_opt_pass along with...
2175         (pass_data_all_early_optimizations): ...new pass_data instance and...
2176         (make_pass_all_early_optimizations): ...new function.
2177         (pass_all_optimizations): Convert from a global struct to a subclass
2178         of gimple_opt_pass along with...
2179         (pass_data_all_optimizations): ...new pass_data instance and...
2180         (make_pass_all_optimizations): ...new function.
2181         (pass_all_optimizations_g): Convert from a global struct to a subclass
2182         of gimple_opt_pass along with...
2183         (pass_data_all_optimizations_g): ...new pass_data instance and...
2184         (make_pass_all_optimizations_g): ...new function.
2185         (pass_rest_of_compilation): Convert from a global struct to a subclass
2186         of rtl_opt_pass along with...
2187         (pass_data_rest_of_compilation): ...new pass_data instance and...
2188         (make_pass_rest_of_compilation): ...new function.
2189         (pass_postreload): Convert from a global struct to a subclass of
2190         rtl_opt_pass along with...
2191         (pass_data_postreload): ...new pass_data instance and...
2192         (make_pass_postreload): ...new function.
2193         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
2194         subclass of rtl_opt_pass along with...
2195         (pass_data_gcse2): ...new pass_data instance and...
2196         (make_pass_gcse2): ...new function.
2197         * postreload.c (pass_postreload_cse): Convert from a global struct to
2198         a subclass of rtl_opt_pass along with...
2199         (pass_data_postreload_cse): ...new pass_data instance and...
2200         (make_pass_postreload_cse): ...new function.
2201         * predict.c (pass_profile): Convert from a global struct to a subclass
2202         of gimple_opt_pass along with...
2203         (pass_data_profile): ...new pass_data instance and...
2204         (make_pass_profile): ...new function.
2205         (pass_strip_predict_hints): Convert from a global struct to a subclass
2206         of gimple_opt_pass along with...
2207         (pass_data_strip_predict_hints): ...new pass_data instance and...
2208         (make_pass_strip_predict_hints): ...new function.
2209         * recog.c (pass_peephole2): Convert from a global struct to a subclass
2210         of rtl_opt_pass along with...
2211         (pass_data_peephole2): ...new pass_data instance and...
2212         (make_pass_peephole2): ...new function.
2213         (pass_split_all_insns): Convert from a global struct to a subclass of
2214         rtl_opt_pass along with...
2215         (pass_data_split_all_insns): ...new pass_data instance and...
2216         (make_pass_split_all_insns): ...new function.
2217         (pass_split_after_reload): Convert from a global struct to a subclass
2218         of rtl_opt_pass along with...
2219         (pass_data_split_after_reload): ...new pass_data instance and...
2220         (make_pass_split_after_reload): ...new function.
2221         (pass_split_before_regstack): Convert from a global struct to a
2222         subclass of rtl_opt_pass along with...
2223         (pass_data_split_before_regstack): ...new pass_data instance and...
2224         (make_pass_split_before_regstack): ...new function.
2225         (pass_split_before_sched2): Convert from a global struct to a subclass
2226         of rtl_opt_pass along with...
2227         (pass_data_split_before_sched2): ...new pass_data instance and...
2228         (make_pass_split_before_sched2): ...new function.
2229         (pass_split_for_shorten_branches): Convert from a global struct to a
2230         subclass of rtl_opt_pass along with...
2231         (pass_data_split_for_shorten_branches): ...new pass_data instance
2232         and...
2233         (make_pass_split_for_shorten_branches): ...new function.
2234         * ree.c (pass_ree): Convert from a global struct to a subclass of
2235         rtl_opt_pass along with...
2236         (pass_data_ree): ...new pass_data instance and...
2237         (make_pass_ree): ...new function.
2238         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
2239         subclass of rtl_opt_pass along with...
2240         (pass_data_stack_regs): ...new pass_data instance and...
2241         (make_pass_stack_regs): ...new function.
2242         (pass_stack_regs_run): Convert from a global struct to a subclass of
2243         rtl_opt_pass along with...
2244         (pass_data_stack_regs_run): ...new pass_data instance and...
2245         (make_pass_stack_regs_run): ...new function.
2246         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
2247         subclass of rtl_opt_pass along with...
2248         (pass_data_cprop_hardreg): ...new pass_data instance and...
2249         (make_pass_cprop_hardreg): ...new function.
2250         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
2251         subclass of rtl_opt_pass along with...
2252         (pass_data_reginfo_init): ...new pass_data instance and...
2253         (make_pass_reginfo_init): ...new function.
2254         * regmove.c (pass_regmove): Convert from a global struct to a subclass
2255         of rtl_opt_pass along with...
2256         (pass_data_regmove): ...new pass_data instance and...
2257         (make_pass_regmove): ...new function.
2258         * regrename.c (pass_regrename): Convert from a global struct to a
2259         subclass of rtl_opt_pass along with...
2260         (pass_data_regrename): ...new pass_data instance and...
2261         (make_pass_regrename): ...new function.
2262         * reorg.c (pass_delay_slots): Convert from a global struct to a
2263         subclass of rtl_opt_pass along with...
2264         (pass_data_delay_slots): ...new pass_data instance and...
2265         (make_pass_delay_slots): ...new function.
2266         (pass_machine_reorg): Convert from a global struct to a subclass of
2267         rtl_opt_pass along with...
2268         (pass_data_machine_reorg): ...new pass_data instance and...
2269         (make_pass_machine_reorg): ...new function.
2270         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
2271         of rtl_opt_pass along with...
2272         (pass_data_sched): ...new pass_data instance and...
2273         (make_pass_sched): ...new function.
2274         (pass_sched2): Convert from a global struct to a subclass of
2275         rtl_opt_pass along with...
2276         (pass_data_sched2): ...new pass_data instance and...
2277         (make_pass_sched2): ...new function.
2278         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
2279         to a subclass of rtl_opt_pass along with...
2280         (pass_data_stack_ptr_mod): ...new pass_data instance and...
2281         (make_pass_stack_ptr_mod): ...new function.
2282         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
2283         to a subclass of rtl_opt_pass along with...
2284         (pass_data_rtl_store_motion): ...new pass_data instance and...
2285         (make_pass_rtl_store_motion): ...new function.
2286         * tracer.c (pass_tracer): Convert from a global struct to a subclass
2287         of gimple_opt_pass along with...
2288         (pass_data_tracer): ...new pass_data instance and...
2289         (make_pass_tracer): ...new function.
2290         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
2291         to a subclass of gimple_opt_pass along with...
2292         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
2293         (make_pass_diagnose_tm_blocks): ...new function.
2294         (pass_lower_tm): Convert from a global struct to a subclass of
2295         gimple_opt_pass along with...
2296         (pass_data_lower_tm): ...new pass_data instance and...
2297         (make_pass_lower_tm): ...new function.
2298         (pass_tm_init): Convert from a global struct to a subclass of
2299         gimple_opt_pass along with...
2300         (pass_data_tm_init): ...new pass_data instance and...
2301         (make_pass_tm_init): ...new function.
2302         (pass_tm_mark): Convert from a global struct to a subclass of
2303         gimple_opt_pass along with...
2304         (pass_data_tm_mark): ...new pass_data instance and...
2305         (make_pass_tm_mark): ...new function.
2306         (pass_tm_edges): Convert from a global struct to a subclass of
2307         gimple_opt_pass along with...
2308         (pass_data_tm_edges): ...new pass_data instance and...
2309         (make_pass_tm_edges): ...new function.
2310         (pass_tm_memopt): Convert from a global struct to a subclass of
2311         gimple_opt_pass along with...
2312         (pass_data_tm_memopt): ...new pass_data instance and...
2313         (make_pass_tm_memopt): ...new function.
2314         (pass_ipa_tm): Convert from a global struct to a subclass of
2315         simple_ipa_opt_pass along with...
2316         (pass_data_ipa_tm): ...new pass_data instance and...
2317         (make_pass_ipa_tm): ...new function.
2318         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
2319         subclass of gimple_opt_pass along with...
2320         (pass_data_call_cdce): ...new pass_data instance and...
2321         (make_pass_call_cdce): ...new function.
2322         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
2323         subclass of gimple_opt_pass along with...
2324         (pass_data_build_cfg): ...new pass_data instance and...
2325         (make_pass_build_cfg): ...new function.
2326         (pass_split_crit_edges): Convert from a global struct to a subclass of
2327         gimple_opt_pass along with...
2328         (pass_data_split_crit_edges): ...new pass_data instance and...
2329         (make_pass_split_crit_edges): ...new function.
2330         (pass_warn_function_return): Convert from a global struct to a
2331         subclass of gimple_opt_pass along with...
2332         (pass_data_warn_function_return): ...new pass_data instance and...
2333         (make_pass_warn_function_return): ...new function.
2334         (pass_warn_function_noreturn): Convert from a global struct to a
2335         subclass of gimple_opt_pass along with...
2336         (pass_data_warn_function_noreturn): ...new pass_data instance and...
2337         (make_pass_warn_function_noreturn): ...new function.
2338         (pass_warn_unused_result): Convert from a global struct to a subclass
2339         of gimple_opt_pass along with...
2340         (pass_data_warn_unused_result): ...new pass_data instance and...
2341         (make_pass_warn_unused_result): ...new function.
2342         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
2343         a subclass of gimple_opt_pass along with...
2344         (pass_data_merge_phi): ...new pass_data instance and...
2345         (make_pass_merge_phi): ...new function.
2346         * tree-complex.c (pass_lower_complex): Convert from a global struct to
2347         a subclass of gimple_opt_pass along with...
2348         (pass_data_lower_complex): ...new pass_data instance and...
2349         (make_pass_lower_complex): ...new function.
2350         (pass_lower_complex_O0): Convert from a global struct to a subclass of
2351         gimple_opt_pass along with...
2352         (pass_data_lower_complex_O0): ...new pass_data instance and...
2353         (make_pass_lower_complex_O0): ...new function.
2354         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
2355         subclass of gimple_opt_pass along with...
2356         (pass_data_lower_eh): ...new pass_data instance and...
2357         (make_pass_lower_eh): ...new function.
2358         (pass_refactor_eh): Convert from a global struct to a subclass of
2359         gimple_opt_pass along with...
2360         (pass_data_refactor_eh): ...new pass_data instance and...
2361         (make_pass_refactor_eh): ...new function.
2362         (pass_lower_resx): Convert from a global struct to a subclass of
2363         gimple_opt_pass along with...
2364         (pass_data_lower_resx): ...new pass_data instance and...
2365         (make_pass_lower_resx): ...new function.
2366         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
2367         of gimple_opt_pass along with...
2368         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
2369         (make_pass_lower_eh_dispatch): ...new function.
2370         (pass_cleanup_eh): Convert from a global struct to a subclass of
2371         gimple_opt_pass along with...
2372         (pass_data_cleanup_eh): ...new pass_data instance and...
2373         (make_pass_cleanup_eh): ...new function.
2374         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
2375         to a subclass of simple_ipa_opt_pass along with...
2376         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
2377         (make_pass_ipa_lower_emutls): ...new function.
2378         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
2379         a subclass of gimple_opt_pass along with...
2380         (pass_data_if_conversion): ...new pass_data instance and...
2381         (make_pass_if_conversion): ...new function.
2382         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
2383         subclass of gimple_opt_pass along with...
2384         (pass_data_build_ssa): ...new pass_data instance and...
2385         (make_pass_build_ssa): ...new function.
2386         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
2387         global struct to a subclass of gimple_opt_pass along with...
2388         (pass_data_loop_distribution): ...new pass_data instance and...
2389         (make_pass_loop_distribution): ...new function.
2390         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
2391         subclass of gimple_opt_pass along with...
2392         (pass_data_mudflap_1): ...new pass_data instance and...
2393         (make_pass_mudflap_1): ...new function.
2394         (pass_mudflap_2): Convert from a global struct to a subclass of
2395         gimple_opt_pass along with...
2396         (pass_data_mudflap_2): ...new pass_data instance and...
2397         (make_pass_mudflap_2): ...new function.
2398         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
2399         subclass of gimple_opt_pass along with...
2400         (pass_data_mudflap_1): ...new pass_data instance and...
2401         (make_pass_mudflap_1): ...new function.
2402         (pass_mudflap_2): Convert from a global struct to a subclass of
2403         gimple_opt_pass along with...
2404         (pass_data_mudflap_2): ...new pass_data instance and...
2405         (make_pass_mudflap_2): ...new function.
2406         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
2407         gimple_opt_pass along with...
2408         (pass_data_nrv): ...new pass_data instance and...
2409         (make_pass_nrv): ...new function.
2410         (pass_return_slot): Convert from a global struct to a subclass of
2411         gimple_opt_pass along with...
2412         (pass_data_return_slot): ...new pass_data instance and...
2413         (make_pass_return_slot): ...new function.
2414         * tree-object-size.c (pass_object_sizes): Convert from a global struct
2415         to a subclass of gimple_opt_pass along with...
2416         (pass_data_object_sizes): ...new pass_data instance and...
2417         (make_pass_object_sizes): ...new function.
2418         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
2419         global struct to a subclass of gimple_opt_pass along with...
2420         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
2421         and...
2422         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2423         (pass_fixup_cfg): Convert from a global struct to a subclass of
2424         gimple_opt_pass along with...
2425         (pass_data_fixup_cfg): ...new pass_data instance and...
2426         (make_pass_fixup_cfg): ...new function.
2427         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
2428         (make_pass_mudflap_1): ...new function.
2429         (pass_mudflap_2): Replace declaration with that of...
2430         (make_pass_mudflap_2): ...new function.
2431         (pass_asan): Replace declaration with that of...
2432         (make_pass_asan): ...new function.
2433         (pass_asan_O0): Replace declaration with that of...
2434         (make_pass_asan_O0): ...new function.
2435         (pass_tsan): Replace declaration with that of...
2436         (make_pass_tsan): ...new function.
2437         (pass_tsan_O0): Replace declaration with that of...
2438         (make_pass_tsan_O0): ...new function.
2439         (pass_lower_cf): Replace declaration with that of...
2440         (make_pass_lower_cf): ...new function.
2441         (pass_refactor_eh): Replace declaration with that of...
2442         (make_pass_refactor_eh): ...new function.
2443         (pass_lower_eh): Replace declaration with that of...
2444         (make_pass_lower_eh): ...new function.
2445         (pass_lower_eh_dispatch): Replace declaration with that of...
2446         (make_pass_lower_eh_dispatch): ...new function.
2447         (pass_lower_resx): Replace declaration with that of...
2448         (make_pass_lower_resx): ...new function.
2449         (pass_build_cfg): Replace declaration with that of...
2450         (make_pass_build_cfg): ...new function.
2451         (pass_early_tree_profile): Replace declaration with that of...
2452         (make_pass_early_tree_profile): ...new function.
2453         (pass_cleanup_eh): Replace declaration with that of...
2454         (make_pass_cleanup_eh): ...new function.
2455         (pass_sra): Replace declaration with that of...
2456         (make_pass_sra): ...new function.
2457         (pass_sra_early): Replace declaration with that of...
2458         (make_pass_sra_early): ...new function.
2459         (pass_early_ipa_sra): Replace declaration with that of...
2460         (make_pass_early_ipa_sra): ...new function.
2461         (pass_tail_recursion): Replace declaration with that of...
2462         (make_pass_tail_recursion): ...new function.
2463         (pass_tail_calls): Replace declaration with that of...
2464         (make_pass_tail_calls): ...new function.
2465         (pass_tree_loop): Replace declaration with that of...
2466         (make_pass_tree_loop): ...new function.
2467         (pass_tree_loop_init): Replace declaration with that of...
2468         (make_pass_tree_loop_init): ...new function.
2469         (pass_lim): Replace declaration with that of...
2470         (make_pass_lim): ...new function.
2471         (pass_tree_unswitch): Replace declaration with that of...
2472         (make_pass_tree_unswitch): ...new function.
2473         (pass_predcom): Replace declaration with that of...
2474         (make_pass_predcom): ...new function.
2475         (pass_iv_canon): Replace declaration with that of...
2476         (make_pass_iv_canon): ...new function.
2477         (pass_scev_cprop): Replace declaration with that of...
2478         (make_pass_scev_cprop): ...new function.
2479         (pass_empty_loop): Replace declaration with that of...
2480         (make_pass_empty_loop): ...new function.
2481         (pass_record_bounds): Replace declaration with that of...
2482         (make_pass_record_bounds): ...new function.
2483         (pass_graphite): Replace declaration with that of...
2484         (make_pass_graphite): ...new function.
2485         (pass_graphite_transforms): Replace declaration with that of...
2486         (make_pass_graphite_transforms): ...new function.
2487         (pass_if_conversion): Replace declaration with that of...
2488         (make_pass_if_conversion): ...new function.
2489         (pass_loop_distribution): Replace declaration with that of...
2490         (make_pass_loop_distribution): ...new function.
2491         (pass_vectorize): Replace declaration with that of...
2492         (make_pass_vectorize): ...new function.
2493         (pass_slp_vectorize): Replace declaration with that of...
2494         (make_pass_slp_vectorize): ...new function.
2495         (pass_complete_unroll): Replace declaration with that of...
2496         (make_pass_complete_unroll): ...new function.
2497         (pass_complete_unrolli): Replace declaration with that of...
2498         (make_pass_complete_unrolli): ...new function.
2499         (pass_parallelize_loops): Replace declaration with that of...
2500         (make_pass_parallelize_loops): ...new function.
2501         (pass_loop_prefetch): Replace declaration with that of...
2502         (make_pass_loop_prefetch): ...new function.
2503         (pass_iv_optimize): Replace declaration with that of...
2504         (make_pass_iv_optimize): ...new function.
2505         (pass_tree_loop_done): Replace declaration with that of...
2506         (make_pass_tree_loop_done): ...new function.
2507         (pass_ch): Replace declaration with that of...
2508         (make_pass_ch): ...new function.
2509         (pass_ccp): Replace declaration with that of...
2510         (make_pass_ccp): ...new function.
2511         (pass_phi_only_cprop): Replace declaration with that of...
2512         (make_pass_phi_only_cprop): ...new function.
2513         (pass_build_ssa): Replace declaration with that of...
2514         (make_pass_build_ssa): ...new function.
2515         (pass_build_alias): Replace declaration with that of...
2516         (make_pass_build_alias): ...new function.
2517         (pass_build_ealias): Replace declaration with that of...
2518         (make_pass_build_ealias): ...new function.
2519         (pass_dominator): Replace declaration with that of...
2520         (make_pass_dominator): ...new function.
2521         (pass_dce): Replace declaration with that of...
2522         (make_pass_dce): ...new function.
2523         (pass_dce_loop): Replace declaration with that of...
2524         (make_pass_dce_loop): ...new function.
2525         (pass_cd_dce): Replace declaration with that of...
2526         (make_pass_cd_dce): ...new function.
2527         (pass_call_cdce): Replace declaration with that of...
2528         (make_pass_call_cdce): ...new function.
2529         (pass_merge_phi): Replace declaration with that of...
2530         (make_pass_merge_phi): ...new function.
2531         (pass_split_crit_edges): Replace declaration with that of...
2532         (make_pass_split_crit_edges): ...new function.
2533         (pass_pre): Replace declaration with that of...
2534         (make_pass_pre): ...new function.
2535         (pass_profile): Replace declaration with that of...
2536         (make_pass_profile): ...new function.
2537         (pass_strip_predict_hints): Replace declaration with that of...
2538         (make_pass_strip_predict_hints): ...new function.
2539         (pass_lower_complex_O0): Replace declaration with that of...
2540         (make_pass_lower_complex_O0): ...new function.
2541         (pass_lower_complex): Replace declaration with that of...
2542         (make_pass_lower_complex): ...new function.
2543         (pass_lower_vector): Replace declaration with that of...
2544         (make_pass_lower_vector): ...new function.
2545         (pass_lower_vector_ssa): Replace declaration with that of...
2546         (make_pass_lower_vector_ssa): ...new function.
2547         (pass_lower_omp): Replace declaration with that of...
2548         (make_pass_lower_omp): ...new function.
2549         (pass_diagnose_omp_blocks): Replace declaration with that of...
2550         (make_pass_diagnose_omp_blocks): ...new function.
2551         (pass_expand_omp): Replace declaration with that of...
2552         (make_pass_expand_omp): ...new function.
2553         (pass_expand_omp_ssa): Replace declaration with that of...
2554         (make_pass_expand_omp_ssa): ...new function.
2555         (pass_object_sizes): Replace declaration with that of...
2556         (make_pass_object_sizes): ...new function.
2557         (pass_strlen): Replace declaration with that of...
2558         (make_pass_strlen): ...new function.
2559         (pass_fold_builtins): Replace declaration with that of...
2560         (make_pass_fold_builtins): ...new function.
2561         (pass_stdarg): Replace declaration with that of...
2562         (make_pass_stdarg): ...new function.
2563         (pass_early_warn_uninitialized): Replace declaration with that of...
2564         (make_pass_early_warn_uninitialized): ...new function.
2565         (pass_late_warn_uninitialized): Replace declaration with that of...
2566         (make_pass_late_warn_uninitialized): ...new function.
2567         (pass_cse_reciprocals): Replace declaration with that of...
2568         (make_pass_cse_reciprocals): ...new function.
2569         (pass_cse_sincos): Replace declaration with that of...
2570         (make_pass_cse_sincos): ...new function.
2571         (pass_optimize_bswap): Replace declaration with that of...
2572         (make_pass_optimize_bswap): ...new function.
2573         (pass_optimize_widening_mul): Replace declaration with that of...
2574         (make_pass_optimize_widening_mul): ...new function.
2575         (pass_warn_function_return): Replace declaration with that of...
2576         (make_pass_warn_function_return): ...new function.
2577         (pass_warn_function_noreturn): Replace declaration with that of...
2578         (make_pass_warn_function_noreturn): ...new function.
2579         (pass_cselim): Replace declaration with that of...
2580         (make_pass_cselim): ...new function.
2581         (pass_phiopt): Replace declaration with that of...
2582         (make_pass_phiopt): ...new function.
2583         (pass_forwprop): Replace declaration with that of...
2584         (make_pass_forwprop): ...new function.
2585         (pass_phiprop): Replace declaration with that of...
2586         (make_pass_phiprop): ...new function.
2587         (pass_tree_ifcombine): Replace declaration with that of...
2588         (make_pass_tree_ifcombine): ...new function.
2589         (pass_dse): Replace declaration with that of...
2590         (make_pass_dse): ...new function.
2591         (pass_nrv): Replace declaration with that of...
2592         (make_pass_nrv): ...new function.
2593         (pass_rename_ssa_copies): Replace declaration with that of...
2594         (make_pass_rename_ssa_copies): ...new function.
2595         (pass_sink_code): Replace declaration with that of...
2596         (make_pass_sink_code): ...new function.
2597         (pass_fre): Replace declaration with that of...
2598         (make_pass_fre): ...new function.
2599         (pass_check_data_deps): Replace declaration with that of...
2600         (make_pass_check_data_deps): ...new function.
2601         (pass_copy_prop): Replace declaration with that of...
2602         (make_pass_copy_prop): ...new function.
2603         (pass_vrp): Replace declaration with that of...
2604         (make_pass_vrp): ...new function.
2605         (pass_uncprop): Replace declaration with that of...
2606         (make_pass_uncprop): ...new function.
2607         (pass_return_slot): Replace declaration with that of...
2608         (make_pass_return_slot): ...new function.
2609         (pass_reassoc): Replace declaration with that of...
2610         (make_pass_reassoc): ...new function.
2611         (pass_rebuild_cgraph_edges): Replace declaration with that of...
2612         (make_pass_rebuild_cgraph_edges): ...new function.
2613         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
2614         (make_pass_remove_cgraph_callee_edges): ...new function.
2615         (pass_build_cgraph_edges): Replace declaration with that of...
2616         (make_pass_build_cgraph_edges): ...new function.
2617         (pass_local_pure_const): Replace declaration with that of...
2618         (make_pass_local_pure_const): ...new function.
2619         (pass_tracer): Replace declaration with that of...
2620         (make_pass_tracer): ...new function.
2621         (pass_warn_unused_result): Replace declaration with that of...
2622         (make_pass_warn_unused_result): ...new function.
2623         (pass_diagnose_tm_blocks): Replace declaration with that of...
2624         (make_pass_diagnose_tm_blocks): ...new function.
2625         (pass_lower_tm): Replace declaration with that of...
2626         (make_pass_lower_tm): ...new function.
2627         (pass_tm_init): Replace declaration with that of...
2628         (make_pass_tm_init): ...new function.
2629         (pass_tm_mark): Replace declaration with that of...
2630         (make_pass_tm_mark): ...new function.
2631         (pass_tm_memopt): Replace declaration with that of...
2632         (make_pass_tm_memopt): ...new function.
2633         (pass_tm_edges): Replace declaration with that of...
2634         (make_pass_tm_edges): ...new function.
2635         (pass_split_functions): Replace declaration with that of...
2636         (make_pass_split_functions): ...new function.
2637         (pass_feedback_split_functions): Replace declaration with that of...
2638         (make_pass_feedback_split_functions): ...new function.
2639         (pass_strength_reduction): Replace declaration with that of...
2640         (make_pass_strength_reduction): ...new function.
2641         (pass_ipa_lower_emutls): Replace declaration with that of...
2642         (make_pass_ipa_lower_emutls): ...new function.
2643         (pass_ipa_function_and_variable_visibility): Replace declaration with
2644         that of...
2645         (make_pass_ipa_function_and_variable_visibility): ...new function.
2646         (pass_ipa_tree_profile): Replace declaration with that of...
2647         (make_pass_ipa_tree_profile): ...new function.
2648         (pass_early_local_passes): Replace declaration with that of...
2649         (make_pass_early_local_passes): ...new function.
2650         (pass_ipa_whole_program_visibility): Replace declaration with that
2651         of...
2652         (make_pass_ipa_whole_program_visibility): ...new function.
2653         (pass_ipa_lto_gimple_out): Replace declaration with that of...
2654         (make_pass_ipa_lto_gimple_out): ...new function.
2655         (pass_ipa_increase_alignment): Replace declaration with that of...
2656         (make_pass_ipa_increase_alignment): ...new function.
2657         (pass_ipa_inline): Replace declaration with that of...
2658         (make_pass_ipa_inline): ...new function.
2659         (pass_ipa_free_lang_data): Replace declaration with that of...
2660         (make_pass_ipa_free_lang_data): ...new function.
2661         (pass_ipa_free_inline_summary): Replace declaration with that of...
2662         (make_pass_ipa_free_inline_summary): ...new function.
2663         (pass_ipa_cp): Replace declaration with that of...
2664         (make_pass_ipa_cp): ...new function.
2665         (pass_ipa_reference): Replace declaration with that of...
2666         (make_pass_ipa_reference): ...new function.
2667         (pass_ipa_pure_const): Replace declaration with that of...
2668         (make_pass_ipa_pure_const): ...new function.
2669         (pass_ipa_pta): Replace declaration with that of...
2670         (make_pass_ipa_pta): ...new function.
2671         (pass_ipa_lto_finish_out): Replace declaration with that of...
2672         (make_pass_ipa_lto_finish_out): ...new function.
2673         (pass_ipa_tm): Replace declaration with that of...
2674         (make_pass_ipa_tm): ...new function.
2675         (pass_ipa_profile): Replace declaration with that of...
2676         (make_pass_ipa_profile): ...new function.
2677         (pass_ipa_cdtor_merge): Replace declaration with that of...
2678         (make_pass_ipa_cdtor_merge): ...new function.
2679         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
2680         of...
2681         (make_pass_cleanup_cfg_post_optimizing): ...new function.
2682         (pass_init_datastructures): Replace declaration with that of...
2683         (make_pass_init_datastructures): ...new function.
2684         (pass_fixup_cfg): Replace declaration with that of...
2685         (make_pass_fixup_cfg): ...new function.
2686         (pass_expand): Replace declaration with that of...
2687         (make_pass_expand): ...new function.
2688         (pass_instantiate_virtual_regs): Replace declaration with that of...
2689         (make_pass_instantiate_virtual_regs): ...new function.
2690         (pass_rtl_fwprop): Replace declaration with that of...
2691         (make_pass_rtl_fwprop): ...new function.
2692         (pass_rtl_fwprop_addr): Replace declaration with that of...
2693         (make_pass_rtl_fwprop_addr): ...new function.
2694         (pass_jump): Replace declaration with that of...
2695         (make_pass_jump): ...new function.
2696         (pass_jump2): Replace declaration with that of...
2697         (make_pass_jump2): ...new function.
2698         (pass_lower_subreg): Replace declaration with that of...
2699         (make_pass_lower_subreg): ...new function.
2700         (pass_cse): Replace declaration with that of...
2701         (make_pass_cse): ...new function.
2702         (pass_fast_rtl_dce): Replace declaration with that of...
2703         (make_pass_fast_rtl_dce): ...new function.
2704         (pass_ud_rtl_dce): Replace declaration with that of...
2705         (make_pass_ud_rtl_dce): ...new function.
2706         (pass_rtl_dce): Replace declaration with that of...
2707         (make_pass_rtl_dce): ...new function.
2708         (pass_rtl_dse1): Replace declaration with that of...
2709         (make_pass_rtl_dse1): ...new function.
2710         (pass_rtl_dse2): Replace declaration with that of...
2711         (make_pass_rtl_dse2): ...new function.
2712         (pass_rtl_dse3): Replace declaration with that of...
2713         (make_pass_rtl_dse3): ...new function.
2714         (pass_rtl_cprop): Replace declaration with that of...
2715         (make_pass_rtl_cprop): ...new function.
2716         (pass_rtl_pre): Replace declaration with that of...
2717         (make_pass_rtl_pre): ...new function.
2718         (pass_rtl_hoist): Replace declaration with that of...
2719         (make_pass_rtl_hoist): ...new function.
2720         (pass_rtl_store_motion): Replace declaration with that of...
2721         (make_pass_rtl_store_motion): ...new function.
2722         (pass_cse_after_global_opts): Replace declaration with that of...
2723         (make_pass_cse_after_global_opts): ...new function.
2724         (pass_rtl_ifcvt): Replace declaration with that of...
2725         (make_pass_rtl_ifcvt): ...new function.
2726         (pass_into_cfg_layout_mode): Replace declaration with that of...
2727         (make_pass_into_cfg_layout_mode): ...new function.
2728         (pass_outof_cfg_layout_mode): Replace declaration with that of...
2729         (make_pass_outof_cfg_layout_mode): ...new function.
2730         (pass_loop2): Replace declaration with that of...
2731         (make_pass_loop2): ...new function.
2732         (pass_rtl_loop_init): Replace declaration with that of...
2733         (make_pass_rtl_loop_init): ...new function.
2734         (pass_rtl_move_loop_invariants): Replace declaration with that of...
2735         (make_pass_rtl_move_loop_invariants): ...new function.
2736         (pass_rtl_unswitch): Replace declaration with that of...
2737         (make_pass_rtl_unswitch): ...new function.
2738         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
2739         (make_pass_rtl_unroll_and_peel_loops): ...new function.
2740         (pass_rtl_doloop): Replace declaration with that of...
2741         (make_pass_rtl_doloop): ...new function.
2742         (pass_rtl_loop_done): Replace declaration with that of...
2743         (make_pass_rtl_loop_done): ...new function.
2744         (pass_web): Replace declaration with that of...
2745         (make_pass_web): ...new function.
2746         (pass_cse2): Replace declaration with that of...
2747         (make_pass_cse2): ...new function.
2748         (pass_df_initialize_opt): Replace declaration with that of...
2749         (make_pass_df_initialize_opt): ...new function.
2750         (pass_df_initialize_no_opt): Replace declaration with that of...
2751         (make_pass_df_initialize_no_opt): ...new function.
2752         (pass_reginfo_init): Replace declaration with that of...
2753         (make_pass_reginfo_init): ...new function.
2754         (pass_inc_dec): Replace declaration with that of...
2755         (make_pass_inc_dec): ...new function.
2756         (pass_stack_ptr_mod): Replace declaration with that of...
2757         (make_pass_stack_ptr_mod): ...new function.
2758         (pass_initialize_regs): Replace declaration with that of...
2759         (make_pass_initialize_regs): ...new function.
2760         (pass_combine): Replace declaration with that of...
2761         (make_pass_combine): ...new function.
2762         (pass_if_after_combine): Replace declaration with that of...
2763         (make_pass_if_after_combine): ...new function.
2764         (pass_ree): Replace declaration with that of...
2765         (make_pass_ree): ...new function.
2766         (pass_partition_blocks): Replace declaration with that of...
2767         (make_pass_partition_blocks): ...new function.
2768         (pass_match_asm_constraints): Replace declaration with that of...
2769         (make_pass_match_asm_constraints): ...new function.
2770         (pass_regmove): Replace declaration with that of...
2771         (make_pass_regmove): ...new function.
2772         (pass_split_all_insns): Replace declaration with that of...
2773         (make_pass_split_all_insns): ...new function.
2774         (pass_fast_rtl_byte_dce): Replace declaration with that of...
2775         (make_pass_fast_rtl_byte_dce): ...new function.
2776         (pass_lower_subreg2): Replace declaration with that of...
2777         (make_pass_lower_subreg2): ...new function.
2778         (pass_mode_switching): Replace declaration with that of...
2779         (make_pass_mode_switching): ...new function.
2780         (pass_sms): Replace declaration with that of...
2781         (make_pass_sms): ...new function.
2782         (pass_sched): Replace declaration with that of...
2783         (make_pass_sched): ...new function.
2784         (pass_ira): Replace declaration with that of...
2785         (make_pass_ira): ...new function.
2786         (pass_reload): Replace declaration with that of...
2787         (make_pass_reload): ...new function.
2788         (pass_clean_state): Replace declaration with that of...
2789         (make_pass_clean_state): ...new function.
2790         (pass_branch_prob): Replace declaration with that of...
2791         (make_pass_branch_prob): ...new function.
2792         (pass_value_profile_transformations): Replace declaration with that
2793         of...
2794         (make_pass_value_profile_transformations): ...new function.
2795         (pass_postreload_cse): Replace declaration with that of...
2796         (make_pass_postreload_cse): ...new function.
2797         (pass_gcse2): Replace declaration with that of...
2798         (make_pass_gcse2): ...new function.
2799         (pass_split_after_reload): Replace declaration with that of...
2800         (make_pass_split_after_reload): ...new function.
2801         (pass_branch_target_load_optimize1): Replace declaration with that
2802         of...
2803         (make_pass_branch_target_load_optimize1): ...new function.
2804         (pass_thread_prologue_and_epilogue): Replace declaration with that
2805         of...
2806         (make_pass_thread_prologue_and_epilogue): ...new function.
2807         (pass_stack_adjustments): Replace declaration with that of...
2808         (make_pass_stack_adjustments): ...new function.
2809         (pass_peephole2): Replace declaration with that of...
2810         (make_pass_peephole2): ...new function.
2811         (pass_if_after_reload): Replace declaration with that of...
2812         (make_pass_if_after_reload): ...new function.
2813         (pass_regrename): Replace declaration with that of...
2814         (make_pass_regrename): ...new function.
2815         (pass_cprop_hardreg): Replace declaration with that of...
2816         (make_pass_cprop_hardreg): ...new function.
2817         (pass_reorder_blocks): Replace declaration with that of...
2818         (make_pass_reorder_blocks): ...new function.
2819         (pass_branch_target_load_optimize2): Replace declaration with that
2820         of...
2821         (make_pass_branch_target_load_optimize2): ...new function.
2822         (pass_leaf_regs): Replace declaration with that of...
2823         (make_pass_leaf_regs): ...new function.
2824         (pass_split_before_sched2): Replace declaration with that of...
2825         (make_pass_split_before_sched2): ...new function.
2826         (pass_compare_elim_after_reload): Replace declaration with that of...
2827         (make_pass_compare_elim_after_reload): ...new function.
2828         (pass_sched2): Replace declaration with that of...
2829         (make_pass_sched2): ...new function.
2830         (pass_stack_regs): Replace declaration with that of...
2831         (make_pass_stack_regs): ...new function.
2832         (pass_stack_regs_run): Replace declaration with that of...
2833         (make_pass_stack_regs_run): ...new function.
2834         (pass_df_finish): Replace declaration with that of...
2835         (make_pass_df_finish): ...new function.
2836         (pass_compute_alignments): Replace declaration with that of...
2837         (make_pass_compute_alignments): ...new function.
2838         (pass_duplicate_computed_gotos): Replace declaration with that of...
2839         (make_pass_duplicate_computed_gotos): ...new function.
2840         (pass_variable_tracking): Replace declaration with that of...
2841         (make_pass_variable_tracking): ...new function.
2842         (pass_free_cfg): Replace declaration with that of...
2843         (make_pass_free_cfg): ...new function.
2844         (pass_machine_reorg): Replace declaration with that of...
2845         (make_pass_machine_reorg): ...new function.
2846         (pass_cleanup_barriers): Replace declaration with that of...
2847         (make_pass_cleanup_barriers): ...new function.
2848         (pass_delay_slots): Replace declaration with that of...
2849         (make_pass_delay_slots): ...new function.
2850         (pass_split_for_shorten_branches): Replace declaration with that of...
2851         (make_pass_split_for_shorten_branches): ...new function.
2852         (pass_split_before_regstack): Replace declaration with that of...
2853         (make_pass_split_before_regstack): ...new function.
2854         (pass_convert_to_eh_region_ranges): Replace declaration with that
2855         of...
2856         (make_pass_convert_to_eh_region_ranges): ...new function.
2857         (pass_shorten_branches): Replace declaration with that of...
2858         (make_pass_shorten_branches): ...new function.
2859         (pass_set_nothrow_function_flags): Replace declaration with that of...
2860         (make_pass_set_nothrow_function_flags): ...new function.
2861         (pass_dwarf2_frame): Replace declaration with that of...
2862         (make_pass_dwarf2_frame): ...new function.
2863         (pass_final): Replace declaration with that of...
2864         (make_pass_final): ...new function.
2865         (pass_rtl_seqabstr): Replace declaration with that of...
2866         (make_pass_rtl_seqabstr): ...new function.
2867         (pass_release_ssa_names): Replace declaration with that of...
2868         (make_pass_release_ssa_names): ...new function.
2869         (pass_early_inline): Replace declaration with that of...
2870         (make_pass_early_inline): ...new function.
2871         (pass_inline_parameters): Replace declaration with that of...
2872         (make_pass_inline_parameters): ...new function.
2873         (pass_update_address_taken): Replace declaration with that of...
2874         (make_pass_update_address_taken): ...new function.
2875         (pass_convert_switch): Replace declaration with that of...
2876         (make_pass_convert_switch): ...new function.
2877         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
2878         to a subclass of simple_ipa_opt_pass along with...
2879         (pass_data_ipa_tree_profile): ...new pass_data instance and...
2880         (make_pass_ipa_tree_profile): ...new function.
2881         * tree-sra.c (pass_sra_early): Convert from a global struct to a
2882         subclass of gimple_opt_pass along with...
2883         (pass_data_sra_early): ...new pass_data instance and...
2884         (make_pass_sra_early): ...new function.
2885         (pass_sra): Convert from a global struct to a subclass of
2886         gimple_opt_pass along with...
2887         (pass_data_sra): ...new pass_data instance and...
2888         (make_pass_sra): ...new function.
2889         (pass_early_ipa_sra): Convert from a global struct to a subclass of
2890         gimple_opt_pass along with...
2891         (pass_data_early_ipa_sra): ...new pass_data instance and...
2892         (make_pass_early_ipa_sra): ...new function.
2893         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
2894         subclass of gimple_opt_pass along with...
2895         (pass_data_ccp): ...new pass_data instance and...
2896         (make_pass_ccp): ...new function.
2897         (pass_fold_builtins): Convert from a global struct to a subclass of
2898         gimple_opt_pass along with...
2899         (pass_data_fold_builtins): ...new pass_data instance and...
2900         (make_pass_fold_builtins): ...new function.
2901         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
2902         subclass of gimple_opt_pass along with...
2903         (pass_data_copy_prop): ...new pass_data instance and...
2904         (make_pass_copy_prop): ...new function.
2905         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
2906         global struct to a subclass of gimple_opt_pass along with...
2907         (pass_data_rename_ssa_copies): ...new pass_data instance and...
2908         (make_pass_rename_ssa_copies): ...new function.
2909         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
2910         subclass of gimple_opt_pass along with...
2911         (pass_data_dce): ...new pass_data instance and...
2912         (make_pass_dce): ...new function.
2913         (pass_dce_loop): Convert from a global struct to a subclass of
2914         gimple_opt_pass along with...
2915         (pass_data_dce_loop): ...new pass_data instance and...
2916         (make_pass_dce_loop): ...new function.
2917         (pass_cd_dce): Convert from a global struct to a subclass of
2918         gimple_opt_pass along with...
2919         (pass_data_cd_dce): ...new pass_data instance and...
2920         (make_pass_cd_dce): ...new function.
2921         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
2922         subclass of gimple_opt_pass along with...
2923         (pass_data_dominator): ...new pass_data instance and...
2924         (make_pass_dominator): ...new function.
2925         (pass_phi_only_cprop): Convert from a global struct to a subclass of
2926         gimple_opt_pass along with...
2927         (pass_data_phi_only_cprop): ...new pass_data instance and...
2928         (make_pass_phi_only_cprop): ...new function.
2929         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
2930         subclass of gimple_opt_pass along with...
2931         (pass_data_dse): ...new pass_data instance and...
2932         (make_pass_dse): ...new function.
2933         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
2934         a subclass of gimple_opt_pass along with...
2935         (pass_data_forwprop): ...new pass_data instance and...
2936         (make_pass_forwprop): ...new function.
2937         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
2938         struct to a subclass of gimple_opt_pass along with...
2939         (pass_data_tree_ifcombine): ...new pass_data instance and...
2940         (make_pass_tree_ifcombine): ...new function.
2941         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
2942         subclass of gimple_opt_pass along with...
2943         (pass_data_ch): ...new pass_data instance and...
2944         (make_pass_ch): ...new function.
2945         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
2946         subclass of gimple_opt_pass along with...
2947         (pass_data_tree_loop): ...new pass_data instance and...
2948         (make_pass_tree_loop): ...new function.
2949         (pass_tree_loop_init): Convert from a global struct to a subclass of
2950         gimple_opt_pass along with...
2951         (pass_data_tree_loop_init): ...new pass_data instance and...
2952         (make_pass_tree_loop_init): ...new function.
2953         (pass_lim): Convert from a global struct to a subclass of
2954         gimple_opt_pass along with...
2955         (pass_data_lim): ...new pass_data instance and...
2956         (make_pass_lim): ...new function.
2957         (pass_tree_unswitch): Convert from a global struct to a subclass of
2958         gimple_opt_pass along with...
2959         (pass_data_tree_unswitch): ...new pass_data instance and...
2960         (make_pass_tree_unswitch): ...new function.
2961         (pass_predcom): Convert from a global struct to a subclass of
2962         gimple_opt_pass along with...
2963         (pass_data_predcom): ...new pass_data instance and...
2964         (make_pass_predcom): ...new function.
2965         (pass_vectorize): Convert from a global struct to a subclass of
2966         gimple_opt_pass along with...
2967         (pass_data_vectorize): ...new pass_data instance and...
2968         (make_pass_vectorize): ...new function.
2969         (pass_graphite): Convert from a global struct to a subclass of
2970         gimple_opt_pass along with...
2971         (pass_data_graphite): ...new pass_data instance and...
2972         (make_pass_graphite): ...new function.
2973         (pass_graphite_transforms): Convert from a global struct to a subclass
2974         of gimple_opt_pass along with...
2975         (pass_data_graphite_transforms): ...new pass_data instance and...
2976         (make_pass_graphite_transforms): ...new function.
2977         (pass_check_data_deps): Convert from a global struct to a subclass of
2978         gimple_opt_pass along with...
2979         (pass_data_check_data_deps): ...new pass_data instance and...
2980         (make_pass_check_data_deps): ...new function.
2981         (pass_iv_canon): Convert from a global struct to a subclass of
2982         gimple_opt_pass along with...
2983         (pass_data_iv_canon): ...new pass_data instance and...
2984         (make_pass_iv_canon): ...new function.
2985         (pass_scev_cprop): Convert from a global struct to a subclass of
2986         gimple_opt_pass along with...
2987         (pass_data_scev_cprop): ...new pass_data instance and...
2988         (make_pass_scev_cprop): ...new function.
2989         (pass_record_bounds): Convert from a global struct to a subclass of
2990         gimple_opt_pass along with...
2991         (pass_data_record_bounds): ...new pass_data instance and...
2992         (make_pass_record_bounds): ...new function.
2993         (pass_complete_unroll): Convert from a global struct to a subclass of
2994         gimple_opt_pass along with...
2995         (pass_data_complete_unroll): ...new pass_data instance and...
2996         (make_pass_complete_unroll): ...new function.
2997         (pass_complete_unrolli): Convert from a global struct to a subclass of
2998         gimple_opt_pass along with...
2999         (pass_data_complete_unrolli): ...new pass_data instance and...
3000         (make_pass_complete_unrolli): ...new function.
3001         (pass_parallelize_loops): Convert from a global struct to a subclass
3002         of gimple_opt_pass along with...
3003         (pass_data_parallelize_loops): ...new pass_data instance and...
3004         (make_pass_parallelize_loops): ...new function.
3005         (pass_loop_prefetch): Convert from a global struct to a subclass of
3006         gimple_opt_pass along with...
3007         (pass_data_loop_prefetch): ...new pass_data instance and...
3008         (make_pass_loop_prefetch): ...new function.
3009         (pass_iv_optimize): Convert from a global struct to a subclass of
3010         gimple_opt_pass along with...
3011         (pass_data_iv_optimize): ...new pass_data instance and...
3012         (make_pass_iv_optimize): ...new function.
3013         (pass_tree_loop_done): Convert from a global struct to a subclass of
3014         gimple_opt_pass along with...
3015         (pass_data_tree_loop_done): ...new pass_data instance and...
3016         (make_pass_tree_loop_done): ...new function.
3017         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
3018         struct to a subclass of gimple_opt_pass along with...
3019         (pass_data_cse_reciprocals): ...new pass_data instance and...
3020         (make_pass_cse_reciprocals): ...new function.
3021         (pass_cse_sincos): Convert from a global struct to a subclass of
3022         gimple_opt_pass along with...
3023         (pass_data_cse_sincos): ...new pass_data instance and...
3024         (make_pass_cse_sincos): ...new function.
3025         (pass_optimize_bswap): Convert from a global struct to a subclass of
3026         gimple_opt_pass along with...
3027         (pass_data_optimize_bswap): ...new pass_data instance and...
3028         (make_pass_optimize_bswap): ...new function.
3029         (pass_optimize_widening_mul): Convert from a global struct to a
3030         subclass of gimple_opt_pass along with...
3031         (pass_data_optimize_widening_mul): ...new pass_data instance and...
3032         (make_pass_optimize_widening_mul): ...new function.
3033         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
3034         subclass of gimple_opt_pass along with...
3035         (pass_data_phiopt): ...new pass_data instance and...
3036         (make_pass_phiopt): ...new function.
3037         (pass_cselim): Convert from a global struct to a subclass of
3038         gimple_opt_pass along with...
3039         (pass_data_cselim): ...new pass_data instance and...
3040         (make_pass_cselim): ...new function.
3041         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
3042         subclass of gimple_opt_pass along with...
3043         (pass_data_phiprop): ...new pass_data instance and...
3044         (make_pass_phiprop): ...new function.
3045         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
3046         subclass of gimple_opt_pass along with...
3047         (pass_data_pre): ...new pass_data instance and...
3048         (make_pass_pre): ...new function.
3049         (pass_fre): Convert from a global struct to a subclass of
3050         gimple_opt_pass along with...
3051         (pass_data_fre): ...new pass_data instance and...
3052         (make_pass_fre): ...new function.
3053         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
3054         subclass of gimple_opt_pass along with...
3055         (pass_data_reassoc): ...new pass_data instance and...
3056         (make_pass_reassoc): ...new function.
3057         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
3058         subclass of gimple_opt_pass along with...
3059         (pass_data_sink_code): ...new pass_data instance and...
3060         (make_pass_sink_code): ...new function.
3061         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
3062         subclass of gimple_opt_pass along with...
3063         (pass_data_strlen): ...new pass_data instance and...
3064         (make_pass_strlen): ...new function.
3065         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
3066         struct to a subclass of gimple_opt_pass along with...
3067         (pass_data_build_alias): ...new pass_data instance and...
3068         (make_pass_build_alias): ...new function.
3069         (pass_build_ealias): Convert from a global struct to a subclass of
3070         gimple_opt_pass along with...
3071         (pass_data_build_ealias): ...new pass_data instance and...
3072         (make_pass_build_ealias): ...new function.
3073         (pass_ipa_pta): Convert from a global struct to a subclass of
3074         simple_ipa_opt_pass along with...
3075         (pass_data_ipa_pta): ...new pass_data instance and...
3076         (make_pass_ipa_pta): ...new function.
3077         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
3078         subclass of gimple_opt_pass along with...
3079         (pass_data_uncprop): ...new pass_data instance and...
3080         (make_pass_uncprop): ...new function.
3081         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
3082         global struct to a subclass of gimple_opt_pass along with...
3083         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
3084         (make_pass_late_warn_uninitialized): ...new function.
3085         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
3086         to a subclass of gimple_opt_pass along with...
3087         (pass_data_init_datastructures): ...new pass_data instance and...
3088         (make_pass_init_datastructures): ...new function.
3089         (pass_early_warn_uninitialized): Convert from a global struct to a
3090         subclass of gimple_opt_pass along with...
3091         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
3092         (make_pass_early_warn_uninitialized): ...new function.
3093         (pass_update_address_taken): Convert from a global struct to a
3094         subclass of gimple_opt_pass along with...
3095         (pass_data_update_address_taken): ...new pass_data instance and...
3096         (make_pass_update_address_taken): ...new function.
3097         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
3098         struct to a subclass of gimple_opt_pass along with...
3099         (pass_data_release_ssa_names): ...new pass_data instance and...
3100         (make_pass_release_ssa_names): ...new function.
3101         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
3102         subclass of gimple_opt_pass along with...
3103         (pass_data_stdarg): ...new pass_data instance and...
3104         (make_pass_stdarg): ...new function.
3105         * tree-switch-conversion.c (pass_convert_switch): Convert from a
3106         global struct to a subclass of gimple_opt_pass along with...
3107         (pass_data_convert_switch): ...new pass_data instance and...
3108         (make_pass_convert_switch): ...new function.
3109         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
3110         to a subclass of gimple_opt_pass along with...
3111         (pass_data_tail_recursion): ...new pass_data instance and...
3112         (make_pass_tail_recursion): ...new function.
3113         (pass_tail_calls): Convert from a global struct to a subclass of
3114         gimple_opt_pass along with...
3115         (pass_data_tail_calls): ...new pass_data instance and...
3116         (make_pass_tail_calls): ...new function.
3117         * tree-vect-generic.c (pass_lower_vector): Convert from a global
3118         struct to a subclass of gimple_opt_pass along with...
3119         (pass_data_lower_vector): ...new pass_data instance and...
3120         (make_pass_lower_vector): ...new function.
3121         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
3122         gimple_opt_pass along with...
3123         (pass_data_lower_vector_ssa): ...new pass_data instance and...
3124         (make_pass_lower_vector_ssa): ...new function.
3125         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
3126         to a subclass of gimple_opt_pass along with...
3127         (pass_data_slp_vectorize): ...new pass_data instance and...
3128         (make_pass_slp_vectorize): ...new function.
3129         (pass_ipa_increase_alignment): Convert from a global struct to a
3130         subclass of simple_ipa_opt_pass along with...
3131         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
3132         (make_pass_ipa_increase_alignment): ...new function.
3133         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
3134         gimple_opt_pass along with...
3135         (pass_data_vrp): ...new pass_data instance and...
3136         (make_pass_vrp): ...new function.
3137         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
3138         subclass of simple_ipa_opt_pass along with...
3139         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
3140         (make_pass_ipa_free_lang_data): ...new function.
3141         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
3142         gimple_opt_pass along with...
3143         (pass_data_tsan): ...new pass_data instance and...
3144         (make_pass_tsan): ...new function.
3145         (pass_tsan_O0): Convert from a global struct to a subclass of
3146         gimple_opt_pass along with...
3147         (pass_data_tsan_O0): ...new pass_data instance and...
3148         (make_pass_tsan_O0): ...new function.
3149         * var-tracking.c (pass_variable_tracking): Convert from a global
3150         struct to a subclass of rtl_opt_pass along with...
3151         (pass_data_variable_tracking): ...new pass_data instance and...
3152         (make_pass_variable_tracking): ...new function.
3153         * web.c (pass_web): Convert from a global struct to a subclass of
3154         rtl_opt_pass along with...
3155         (pass_data_web): ...new pass_data instance and...
3156         (make_pass_web): ...new function.
3157         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
3158         declaration with that of...
3159         (make_pass_mode_switch_use): ...new function.
3160         (pass_resolve_sw_modes): Replace declaration with that of...
3161         (make_pass_resolve_sw_modes): ...new function.
3162         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
3163         from a global struct to a subclass of rtl_opt_pass along with...
3164         (pass_data_mode_switch_use): ...new pass_data instance and...
3165         (make_pass_mode_switch_use): ...new function.
3166         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
3167         from a global struct to a subclass of rtl_opt_pass along with...
3168         (pass_data_resolve_sw_modes): ...new pass_data instance and...
3169         (make_pass_resolve_sw_modes): ...new function.
3170         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
3171         struct to a subclass of rtl_opt_pass along with...
3172         (pass_data_insert_vzeroupper): ...new pass_data instance and...
3173         (make_pass_insert_vzeroupper): ...new function.
3174         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
3175         global struct to a subclass of rtl_opt_pass along with...
3176         (pass_data_work_around_errata): ...new pass_data instance and...
3177         (make_pass_work_around_errata): ...new function.
3178         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
3179         struct to a subclass of rtl_opt_pass along with...
3180         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
3181         (make_pass_mips_machine_reorg2): ...new function.
3183 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3185         * passes.c (pass_manager::operator new): New.
3187 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
3189         Handwritten part of conversion of passes to C++ classes.
3191         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
3192         (toplev.o): Add dep on PASS_MANAGER_H.
3193         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
3194         of early local pases to reflect this moving from a global to a
3195         member of gcc::pass_manager.
3196         (cgraph_add_new_function): Likewise.
3197         * lto-cgraph.c (lto_output_node): Update for conversion of
3198         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
3199         * passes.c (opt_pass::clone): New.
3200         (opt_pass::gate): New.
3201         (opt_pass::execute): New.
3202         (opt_pass::opt_pass): New.
3203         (pass_manager::execute_early_local_passes): New.
3204         (pass_manager::execute_pass_mode_switching): new.
3205         (finish_optimization_passes): Convert to...
3206         (pass_manager::finish_optimization_passes): ...this.
3207         (finish_optimization_passes): Update for conversion of passes to
3208         C++ classes.
3209         (register_dump_files_1): Use has_gate since we cannot portably
3210         check a vtable entry against NULL.
3211         (dump_one_pass): Likewise.
3212         (ipa_write_summaries_2): Likewise.
3213         (ipa_write_optimization_summaries_1): Likewise.
3214         (ipa_read_summaries_1): Likewise.
3215         (ipa_read_optimization_summaries_1): Likewise.
3216         (execute_ipa_stmt_fixups): Likewise.
3217         (pass_manager::pass_manager): Rewrite pass-creation, invoking
3218         pass-creation functions rather than wiring up globals, and
3219         storing the results in fields of pass_manager generated using
3220         pass-instances.def.
3221         (pass_manager::dump_profile_report): Update for conversion of
3222         passes to C++ classes.
3223         (pass_manager::execute_ipa_summary_passes): Likewise.
3224         (execute_one_ipa_transform_pass): Likewise.
3225         (execute_one_pass): Use has_gate and has_execute since we cannot
3226         portably check a vtable entry against NULL.
3227         * pass_manager.h (pass_manager::finish_optimization_passes): New.
3228         (pass_manager): Use pass-instances.def to add fields for the
3229         various pass instances.
3230         * toplev.c (finalize): Update for move of
3231         finish_optimization_passes to a method of gcc::pass_manager.
3232         * toplev.h (finish_optimization_passes): Move to method of class
3233         pass_manager.
3234         * tree-pass.h (struct pass_data): New.
3235         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
3236         (opt_pass::gate): Convert to virtual function.
3237         (opt_pass::~opt_pass): New.
3238         (opt_pass::clone): New.
3239         (opt_pass::execute): Convert to virtual function.
3240         (opt_pass::opt_pass): New.
3241         (opt_pass::ctxt_): new.
3242         (gimple_opt_pass): Convert to subclass of opt_pass.
3243         (gimple_opt_pass::gimple_opt_pass): New.
3244         (rtl_opt_pass): Convert to subclass of opt_pass.
3245         (rtl_opt_pass::rtl_opt_pass): New.
3246         (ipa_opt_pass_d): Convert to subclass of opt_pass.
3247         (ipa_opt_pass_d::ipa_opt_pass_d): New.
3248         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
3249         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
3250         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
3251         invocation of pass_mode_switching to reflect this moving from a
3252         global to a member of gcc::pass_manager.
3253         (ix86_option_override): Rework how pass_insert_vzeroupper is
3254         added to the pass_manager to reflect autogenerated changes.
3255         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
3257 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
3259         PR rtl-optimization/57708
3260         * recog.c (peep2_find_free_register): Validate all regs in a
3261         multi-reg mode.
3263 2013-08-05  Jan Hubicka  <jh@suse.cz>
3265         PR lto/57602
3266         * cgraph.c (verify_cgraph_node): Accept local flags from other
3267         partitions.
3268         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
3269         (function_and_variable_visibility): Likewise.
3270         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
3272 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3274         * graph.c (init_graph_slim_pretty_print): Remove.
3275         (print_graph_cfg): Do not call it.  Use local pretty printer.
3276         (start_graph_dump): Likewise.
3278 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3280         * gimple-pretty-print.c (buffer): Remove.
3281         (initialized): Likewise.
3282         (maybe_init_pretty_print): Likewise.
3283         (print_gimple_stmt): Do not call it.  Use non-static local
3284         pretty_printer variable.
3285         (print_gimple_expr): Likewise.
3286         (print_gimple_seq): Likewise.
3287         (gimple_dump_bb): Likewise.
3289 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3291         * asan.c (asan_pp): Remove.
3292         (asan_pp_initialized): Likewise.
3293         (asan_pp_initialize): Likewise.
3294         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
3295         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
3296         (asan_add_global): Likewise.
3298 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3300         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
3301         * diagnostic.h (diagnostic_flush_buffer): Adjust.
3302         * pretty-print.c (pp_formatted_text_data): Likewise.
3303         (pp_indent): Rename from pp_base_indent.
3304         (pp_format): Rename from pp_base_format.
3305         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
3306         (pp_format_verbatim): Rename from pp_base_format_verbatim.
3307         (pp_flush): Rename from pp_base_flush.
3308         (pp_set_line_maximum_length): Rename from
3309         pp_base_set_line_maximum_length.
3310         (pp_clear_output_area): Rename from pp_base_clear_output_area.
3311         (pp_set_prefix): Rename from pp_base_set_prefix.
3312         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
3313         (pp_emit_prefix): Rename from pp_base_emit_prefix.
3314         (pp_append_text): Rename from pp_base_append_text.
3315         (pp_formatted_text): Rename from pp_base_formatted_text.
3316         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
3317         (pp_remaining_character_count_for_line): Rename from
3318         pp_base_remaining_character_count_for_line.
3319         (pp_newline): Rename from pp_base_newline.
3320         (pp_character): Rename from pp_base_character.
3321         (pp_string): Rename from pp_base_string.
3322         (pp_maybe_space): Rename from pp_base_maybe_space.
3323         * asan.c (asan_pp_string): Adjust.
3324         (asan_emit_stack_protection): Likewise.
3325         (asan_add_global): Likewise.
3326         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
3327         * tree-mudflap.c (mf_varname_tree): Likewise.
3328         * tree-pretty-print.c (pp_tree_identifier): Rename from
3329         pp_base_tree_identifier.
3330         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
3331         Declare as function.
3333 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3335         * pretty-print.h (pp_bar_bar): New.
3336         (pp_ampersand_ampersand): Likewise.
3337         (pp_less_equal): Likewise.
3338         (pp_greater_equal): Likewise.
3339         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
3340         printer functions instead of pp_string or operators and punctuators.
3341         (dump_gimple_call): Likewise.
3342         (dump_gimple_omp_for): Likewise.
3343         (dump_gimple_transaction): Likewise.
3344         (dump_gimple_phi): Likewise.
3345         (pp_gimple_stmt_1): Likewise.
3346         * sched-vis.c (print_insn): Likewise.
3347         * tree-mudflap.c (mf_varname_tree): Likewise.
3348         * tree-pretty-print.c (dump_block_node): Likewise.
3349         (dump_generic_node): Likewise.
3351 2013-08-02  Jan Hubicka  <jh@suse.cz>
3353         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
3354         boundaries.
3355         * lto-streamer-out.c (tree_is_indexable): Results decls and
3356         parm decls are not indexable.
3357         (DFS_write_tree_body): Do not follow args and results.
3358         (hash_tree): Likewise.
3359         (output_functions): Rearrange so struct function is needed
3360         only when real body is output; be able to also ouptut abstract
3361         functions; output DECL_ARGUMENTS and DECL_RESULT.
3362         (lto_output): When not in WPA, ale store abstract functions.
3363         (write_symbol): Do not care about RESULT_DECL.
3364         (output_symbol_p): Handle correctly sbtract decls.
3365         * lto-streamer-in.c (input_function): Rearrange so struct
3366         function can be NULL at entry; allow streaming of
3367         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
3368         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
3369         sanity check during LTO.
3370         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
3371         RESULT_DECl and DECL_ARGUMENTS.
3372         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
3373         Likewise.
3375 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
3377         * pretty-print.h (pp_underscore): New.
3378         (pp_comma): Tidy.
3379         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
3380         printer functions instead of pp_character.
3381         (dump_binary_rhs): Likewise.
3382         (dump_ternary_rhs): Likewise.
3383         (dump_gimple_call_args): Likewise.
3384         (pp_points_to_solution): Likewise.
3385         (dump_gimple_call): Likewise.
3386         (dump_gimple_switch): Likewise.
3387         (dump_gimple_cond): Likewise.
3388         (dump_gimple_bind): Likewise.
3389         (dump_gimple_try): Likewise.
3390         (dump_gimple_omp_for): Likewise.
3391         (dump_gimple_omp_continue): Likewise.
3392         (dump_gimple_omp_single): Likewise.
3393         (dump_gimple_omp_sections): Likewise.
3394         (dump_gimple_omp_block): Likewise.
3395         (dump_gimple_omp_critical): Likewise.
3396         (dump_gimple_transaction): Likewise.
3397         (dump_gimple_asm): Likewise.
3398         (dump_gimple_phi): Likewise.
3399         (dump_gimple_omp_parallel): Likewise.
3400         (dump_gimple_omp_task): Likewise.
3401         (dump_gimple_omp_atomic_load): Likewise.
3402         (dump_gimple_omp_atomic_store): Likewise.
3403         (dump_gimple_mem_ops): Likewise.
3404         (pp_gimple_stmt_1): Likewise.
3405         (pp_cfg_jump): Likewise.
3406         (dump_implicit_edges): Likewise.
3407         (gimple_dump_bb_for_graph): Likewise.
3408         * graph.c (draw_cfg_node): Likewise.
3409         * langhooks.c (lhd_print_error_function): Likewise.
3410         * sched-vis.c (print_exp): Likewise.
3411         (print_value): Likewise.
3412         (print_pattern): Likewise.
3413         (print_insn): Likewise.
3414         (rtl_dump_bb_for_graph): Likewise.
3415         * tree-pretty-print.c (dump_function_declaration): Likewise.
3416         (dump_array_domain): Likewise.
3417         (dump_omp_clause): Likewise.
3418         (dump_location): Likewise.
3419         (dump_generic_node): Likewise.
3420         (print_struct_decl): Likewise.
3421         * diagnostic.c (diagnostic_show_locus): Use pp_space.
3423 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3425         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
3426         candidate table when replacing a candidate statement.
3427         (replace_rhs_if_not_dup): Likewise.
3428         (replace_one_candidate): Likewise.
3430 2013-08-02  Jan Hubicka  <jh@suse.cz>
3431             Martin Liska  <marxin.liska@gmail.com>
3433         * cgraphunit.c (add_new_function): Fix logic when adding from
3434         late IPA pass.
3435         (assemble_thunk): Rename to ...
3436         (expand_thunk); .. this one; export; get it working with
3437         general functions; make produced gimple valid.
3438         * cgraph.h (expand_thunk): Declare.
3440 2013-08-02  Jan Hubicka  <jh@suse.cz>
3442         * ipa-cp.c (gather_context_independent_values): Use
3443         ipa_get_param_move_cost.
3444         (get_replacement_map): Remove PARAM; move parameter folding
3445         into tree-inline.c
3446         (create_specialized_node): Update.
3447         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
3448         assert that we have gimple body; update move_cost.
3449         (count_formal_params): Assert that we have gimple body.
3450         (ipa_dump_param): New function.
3451         (ipa_alloc_node_params): Break out from ...
3452         (ipa_initialize_node_params): ... here.
3453         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
3454         (ipa_write_node_info): Stream move costs.
3455         (ipa_read_node_info): Read move costs.
3456         (ipa_update_after_lto_read): Do not recompute node params.
3457         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
3458         (ipa_get_param): Check we are not in WPA.
3459         (ipa_get_param_move_cost): New.
3460         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
3461         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
3462         parm numbers to be present.
3464 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3466         PR rtl-optimization/58048
3467         * lra-constraints.c (process_alt_operands): Don't check asm
3468         operand on register.
3470 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
3472         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
3473         the implied StoreLoad barrier for atomic operations if before.
3475 2013-08-02  Jan Hubicka  <jh@suse.cz>
3476             Martin Liska  <marxin.liska@gmail.com>
3478         * cgraph.c (cgraph_function_body_availability): Do not check
3479         cgraph flags.
3480         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
3481         symtab_node_availability): Declare.
3482         * ipa.c (can_replace_by_local_alias): New.
3483         (function_and_variable_visibility): Use it.
3484         * symtab.c (symtab_for_node_and_aliases,
3485         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
3487 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
3489         PR rtl-optimization/57963
3490         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
3491         (lra_constraints): Use them.
3493 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
3495         * config/arm/types.md (define_attr "type"): Add "load_acq"
3496         and "store_rel".
3497         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
3498         changes.
3499         (cortex_a53_store1): Likewise.
3501 2013-08-01  Jan Hubicka  <jh@suse.cz>
3503         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
3504         partitions are not needed.
3506 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
3508         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
3509         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
3510         MAYBE_NON_Q_CLASS_P where appropriate.
3512 2013-08-01  Jan Hubicka  <jh@suse.cz>
3514         * cgraph.h (release_function_body): Declare.
3515         * tree.c (free_lang_data_in_decl): Free, parameters and return values
3516         of unused delcarations.
3518 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3520         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
3521         RTL form when subtracting a constant.
3523 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3525         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
3526         Generate canonical plus rtx with negated immediate instead of minus
3527         where appropriate.
3528         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
3530 2013-08-01  Jan Hubicka  <jh@suse.cz>
3532         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
3533         (cgraph_release_function_body): Likewise.
3534         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
3535         * cgraph.h (cgrpah_node): Rename abstract_and_needed
3536         to used_as_abstract_origin.
3537         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
3538         symbols used as abstract origins.
3539         * cgraphunit.c (analyze_functions): Update.
3540         * ipa.c (symtab_remove_unreachable_nodes): Recompute
3541         used_as_abstract_origin.
3542         * tree-inline.c (tree_function_versioning): Update
3543         used_as_abstract_origin; be ready for DECL_RESULT and
3544         DECL_ARGUMENTS to be NULL.
3546         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
3547         for abstract functions.
3548         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
3549         real symbols.
3551 2013-08-01  Jan Hubicka  <jh@suse.cz>
3553         * profile.c (compute_value_histograms): Fix thinko.
3555 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3557         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
3558         aarch-common-protos.h to extra_headers.
3559         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3560         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
3561         * config/aarch64/t-aarch64 (aarch-common.o): Define.
3563 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
3565         * config/aarch64/aarch64.md (define_attr "type"): Delete.
3566         Include "../arm/types.md".  Define "type" attribute for all patterns.
3567         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
3568         attribute changes.
3570 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
3572         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
3573         to support power8 load fusion.
3574         (fusion_gpr_mem_load): Likewise.
3576         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
3578         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
3579         declarations for power8 load fusion.
3580         (emit_fusion_gpr_load): Likewise.
3582         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
3583         tuning for power8, turn on fusion mode by default.  Turn on sign
3584         extending fusion mode if normal fusion mode is on, and we are at
3585         -O2 or -O3.
3586         (fusion_gpr_load_p): New function, return true if we can fuse an
3587         addis instruction with a dependent load to a GPR.
3588         (emit_fusion_gpr_load): Emit the instructions for power8 load
3589         fusion to GPRs.
3591         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
3592         (VSX load fusion peepholes): New peepholes to fuse together an
3593         addi instruction with a VSX load instruction.
3595         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
3596         peepholes to fuse an addis instruction with a load to a GPR base
3597         register.  If we are supporting sign extending fusions, convert
3598         sign extending loads to zero extending loads and add an explicit
3599         sign extension.
3601 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3603         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
3604         aarch-common-protos.h to extra_headers.
3605         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
3606         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
3607         (arm_early_store_addr_dep): Likewise.
3608         (arm_no_early_alu_shift_dep): Likewise.
3609         (arm_no_early_alu_shift_value_dep): Likewise.
3610         (arm_no_early_mul_dep): Likewise.
3611         (arm_no_early_store_addr_dep): Likewise.
3612         (arm_mac_accumulator_is_mul_result): Likewise.
3613         (arm_mac_accumulator_is_result): Likewise.
3614         * config/arm/aarch-common.c: ... here.  New file.
3615         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
3616         here to ...
3617         (arm_early_store_addr_dep): Likewise.
3618         (arm_no_early_alu_shift_dep): Likewise.
3619         (arm_no_early_alu_shift_value_dep): Likewise.
3620         (arm_no_early_mul_dep): Likewise.
3621         (arm_no_early_store_addr_dep): Likewise.
3622         (arm_mac_accumulator_is_mul_result): Likewise.
3623         (arm_mac_accumulator_is_result): Likewise.
3624         * config/arm/aarch-common-protos.h: ... here.  New file.
3625         * config/arm/t-arm (aarch-common.o): Define.
3627 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
3629         * config/arm/arm.md: Include new file "types.md".
3630         (define_attr "type"): Move from here to ...
3631         (define_attr "mul32"): Likewise.
3632         (define_attr "mul64"): Likewise.
3633         * config/arm/types.md: ... here.  New file.
3635 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
3637         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
3638         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
3640 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3642         * gen-pass-instances.awk: Fix offset of substr().
3644 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
3646         * Makefile.in (pass-instances.def): New.
3647         (passes.o): Replace dependency on passes.def with one on
3648         pass-instances.def
3650         * gen-pass-instances.awk: New.
3652         * passes.c (pass_manager::pass_manager): Use pass-instances.def
3653         rather than passes.def, updating local definition of NEXT_PASS
3654         macro to add an extra NUM parameter (currently unused).
3656 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
3658         * Makefile.in (PASS_MANAGER_H): New.
3659         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
3660         (passes.o): Likewise.
3661         (statistics.o): Likewise.
3662         (cgraphunit.o): Likewise.
3663         (context.o): Depend on PASS_MANAGER_H.
3665         * pass_manager.h: New.
3667         * cgraphunit.c (cgraph_add_new_function): Update for moves
3668         of globals to fields of pass_manager.
3669         (analyze_function): Likewise.
3670         (expand_function): Likewise.
3671         (ipa_passes): Likewise.
3672         (compile): Likewise.
3674         * context.c (context::context): New.
3675         * context.h  (context::context): New.
3676         (context::get_passes): New.
3677         (context::passes_): New.
3679         * lto-cgraph.c (input_node): Update for moves of globals to
3680         fields of pass_manager.
3682         * passes.c (all_passes): Remove, in favor of a field of the
3683         same name within the new class pass_manager.
3684         (all_small_ipa_passes): Likewise.
3685         (all_lowering_passes): Likewise.
3686         (all_regular_ipa_passes): Likewise.
3687         (all_late_ipa_passes): Likewise.
3688         (all_lto_gen_passes): Likewise.
3689         (passes_by_id): Likewise.
3690         (passes_by_id_size): Likewise.
3691         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3692         the new class pass_manager.
3693         (set_pass_for_id): Convert to...
3694         (pass_manager::set_pass_for_id): ...method.
3695         (get_pass_for_id): Convert to...
3696         (pass_manager::get_pass_for_id): ...method.
3697         (register_one_dump_file): Move body of implementation into...
3698         (pass_manager::register_one_dump_file): ...here.
3699         (register_dump_files_1): Convert to...
3700         (pass_manager::register_dump_files_1): ...method.
3701         (register_dump_files): Convert to...
3702         (pass_manager::register_dump_files): ...method.
3703         (create_pass_tab): Update for moves of globals to fields of
3704         pass_manager.
3705         (dump_passes): Move body of implementation into...
3706         (pass_manager::dump_passes): ...here.
3707         (register_pass): Move body of implementation into...
3708         (pass_manager::register_pass): ...here.
3709         (init_optimization_passes): Convert into...
3710         (pass_manager::pass_manager): ...constructor for new
3711         pass_manager class, and initialize the pass_lists array.
3712         (check_profile_consistency): Update for moves of globals to
3713         fields of pass_manager.
3714         (dump_profile_report): Move body of implementation into...
3715         (pass_manager::dump_profile_report): ...here.
3716         (ipa_write_summaries_1): Update for moves of pass lists from
3717         being globals to fields of pass_manager.
3718         (ipa_write_optimization_summaries): Likewise.
3719         (ipa_read_summaries):  Likewise.
3720         (ipa_read_optimization_summaries): Likewise.
3721         (execute_all_ipa_stmt_fixups): Likewise.
3723         * statistics.c (statistics_fini): Update for moves of globals to
3724         fields of pass_manager.
3726         * toplev.c (general_init): Replace call to
3727         init_optimization_passes with construction of the pass_manager
3728         instance.
3730         * tree-pass.h (all_passes): Remove, in favor of a field of the
3731         same name within the new class pass_manager.
3732         (all_small_ipa_passes): Likewise.
3733         (all_lowering_passes): Likewise.
3734         (all_regular_ipa_passes): Likewise.
3735         (all_lto_gen_passes): Likewise.
3736         (all_late_ipa_passes): Likewise.
3737         (passes_by_id): Likewise.
3738         (passes_by_id_size): Likewise.
3739         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
3740         the new class pass_manager.
3741         (get_pass_for_id): Remove.
3743 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3745         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
3746         configs.
3748 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
3750         * arm.md (mulhi3): New expand pattern.
3752 2013-07-30  Jan Hubicka  <jh@suse.cz>
3753             Martin Liska  <marxin.liska@gmail.com>
3755         * profile.c (compute_value_histograms): Do not ICE when
3756         there is mismatch only on some counters.
3758 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
3760         PR rtl-optimization/57637
3761         * function.c (move_insn_for_shrink_wrap): Also check the
3762         GEN set of the LIVE problem for the liveness analysis
3763         if it exists, otherwise give up.
3765 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3767         PR tree-optimization/57993
3768         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
3769         replaced statement in the candidate table.
3770         (phi_add_costs): Return infinite cost when the hidden basis does
3771         not dominate all phis on which the candidate is dependent.
3772         (replace_one_candidate): Record replaced statement in the
3773         candidate table.
3775 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
3777         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
3778         (ashlv2si3): New expander.
3779         (*ashlv2si3_i): New define_insn_and_split.
3780         * predicates.md (float_operation): Allow patterns with three
3781         basic sub-patterns.
3783         PR rtl-optimization/58021
3784         * mode-switching.c (create_pre_exit): Always split off preceding
3785         insns if we are not at the basic block head.
3787 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
3789         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
3790         (UCLIBC_DYNAMIC_LINKER): New macro.
3791         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
3792         `-mnan=2008'.
3793         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
3794         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
3795         `-mnan=2008'.
3796         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
3797         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
3798         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
3799         for SF and DF modes.  Use ieee_quad_format for TF mode.
3800         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
3801         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
3802         (mips_option_override): Handle `-mnan=legacy'.
3803         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
3804         `-mabs=2008' and `-mnan=2008'.
3805         (OPTION_DEFAULT_SPECS): Add "nan" default.
3806         (ASM_SPEC): Handle `-mnan='.
3807         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
3808         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
3809         comment accordingly.
3810         (neg<mode>2): Likewise.
3811         * config/mips/mips.opt (mabs, mnan): New options.
3812         * doc/install.texi (Configuration): Document `--with-nan=' option.
3813         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
3814         `-mnan=' options.
3815         (MIPS Options): Document them.
3816         * config.gcc <mips*-*-*>: Handle `--with-nan='.
3817         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
3818         * configure: Regenerate.
3819         * config.in: Regenerate.
3821 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3823         * config/i386/i386.md (float post-reload splitters): Do not check
3824         for subregs of SSE registers.
3826 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
3827             H.J. Lu  <hongjiu.lu@intel.com>
3829         PR target/57954
3830         PR target/57988
3831         * config/i386/i386.md (post-reload splitter
3832         to avoid partial SSE reg dependency stalls): New pattern.
3834 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
3836         * config/s390/s390.md ("movcc"): Swap load and store instructions.
3838 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
3840         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
3841         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
3843 2013-07-26  Cary Coutant  <ccoutant@google.com>
3845         * dwarf2out.c (die_checksum_ordered): Don't include template
3846         instantiations in signature.
3847         (is_template_parameter): New function.
3848         (is_template_instantiation): New function.
3849         (generate_skeleton_bottom_up): Don't include template instantiations
3850         in type unit DIE.
3851         (generate_skeleton): Likewise.
3852         (break_out_comdat_types): Move recursive call to break out nested
3853         types earlier.
3854         (prune_unused_types_mark_generic_parms_dies): Call
3855         is_template_parameter.
3857 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
3859         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
3860         uses vector registers.
3861         * config/aarch64/iterators.md: Add attributes rtn and vas.
3863 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3864             Richard Earnshaw  <richard.earnshaw@arm.com>
3866         * combine.c (simplify_comparison): Re-canonicalize operands
3867         where appropriate.
3868         * config/arm/arm.md (movcond_addsi): New splitter.
3870 2013-07-25  Sterling Augustine  <saugustine@google.com>
3872         * dwarf2out.c (size_of_pubnames): Move code to...
3873         (include_pubname_in_output): ...here.  New.
3874         (want_pubnames): Rearrange.
3875         (output_pubnames): Call include_pubname_in_output.  Move assertion.
3877 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3879         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
3881 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
3883         PR target/38836
3884         * doc/extend.texi: Remove obsolete builtins. Fix
3885         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
3887 2013-07-25  Jan Hubicka  <jh@suse.cz>
3889         * cgraph.c (release_function_body): Break out from ...
3890         (cgraph_release_function_body): ... this one; also release DECL_RESULT
3891         and DECL_ARGUMENTS.
3892         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
3893         old_tree in the map.
3894         (create_specialized_node): Update.
3895         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
3896         into index.
3897         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
3898         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
3899         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
3900         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
3901         DECL_RESULT.
3903 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3905         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
3906         addsi3_carryin_alt2_<optab>): Correct output template.
3908 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3910         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
3911         Adjust for arm_restrict_it.
3912         Remove trailing whitespace.
3914 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
3916         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
3917         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
3919         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
3921 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
3923         PR rtl-optimization/57960
3924         * lra-constraints.c (process_alt_operands): Use the right mode
3925         when checking strict_low.
3927 2013-07-25  Jan Hubicka  <jh@suse.cz>
3929         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
3930         * cgraph.c (cgraph_remove_node): Do not release function body
3931         when in cgraph streaming.
3932         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
3933         in other partitions are not considered reachable; fix handling of
3934         clones.
3936 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3938         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
3940 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
3942         PR target/19599
3943         PR target/57731
3944         PR target/57837
3945         * config/arm/arm.md ("*sibcall_insn): Replace use of
3946         Ss with US. Adjust output for v5 and v4t.
3947         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
3949         * config/arm/constraints.md ("Ss"): Rename to US.
3951 2013-07-25  Terry Guo  <terry.guo@arm.com>
3953         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
3954         shift_add/shift_sub0/shift_sub1 RTXs.
3956 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
3957             Anton Blanchard  <anton@au1.ibm.com>
3959         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
3960         (altivec_vpks<VI_char>ss): Likewise.
3961         (altivec_vpks<VI_char>us): Likewise.
3962         (altivec_vpku<VI_char>us): Likewise.
3963         (altivec_vpku<VI_char>um): Likewise.
3965 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
3967         Introduce context class.
3969         * Makefile.in (CONTEXT_H): New.
3970         (OBJS): Add context.o.
3971         (toplev.o): Add CONTEXT_H to dependencies.
3972         (context.o): New.
3974         * toplev.c (general_init): Create the singleton gcc::context instance.
3976         * context.c: New.
3978         * context.h: New.
3980 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
3982         PR rtl-optimization/57968
3983         * mode-switching.c (create_pre_exit): Allow instructions that
3984         don't set a return register to need a non-exit mode.
3986 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
3987             Anton Blanchard  <anton@au1.ibm.com>
3989         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
3990         operands to vperm for little endian.
3991         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
3992         of lvsl to create the control mask for a vperm for little endian.
3994 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3995             Anton Blanchard  <anton@au1.ibm.com>
3997         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
3998         two operands for little-endian.
4000 2013-07-23  Steve Ellcey  <sellcey@mips.com>
4002         * config/mips/mips.c (mips_case_values_threshold): New.
4003         (TARGET_CASE_VALUES_THRESHOLD): Define.
4005 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4006             Anton Blanchard  <anton@au1.ibm.com>
4008         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
4009         selection of field for vector splat in little endian mode.
4011 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4013         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
4014         expanders to rs6000.md.
4015         (ior<mode>3): Likewise.
4016         (and<mode>3): Likewise.
4017         (one_cmpl<mode>2): Likewise.
4018         (nor<mode>3): Likewise.
4019         (andc<mode>3): Likewise.
4020         (eqv<mode>3): Likewise.
4021         (nand<mode>3): Likewise.
4022         (orc<mode>3): Likewise.
4024         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
4025         declaration.
4027         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
4028         to split multi-word logical operations.
4029         (rs6000_split_logical_di): Likewise.
4030         (rs6000_split_logical): Likewise.
4032         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
4033         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
4034         and allow TImode operations in 32-bit.
4035         (vsx_and<mode>3_64bit): Likewise.
4036         (vsx_ior<mode>3_32bit): Likewise.
4037         (vsx_ior<mode>3_64bit): Likewise.
4038         (vsx_xor<mode>3_32bit): Likewise.
4039         (vsx_xor<mode>3_64bit): Likewise.
4040         (vsx_one_cmpl<mode>2_32bit): Likewise.
4041         (vsx_one_cmpl<mode>2_64bit): Likewise.
4042         (vsx_nor<mode>3_32bit): Likewise.
4043         (vsx_nor<mode>3_64bit): Likewise.
4044         (vsx_andc<mode>3_32bit): Likewise.
4045         (vsx_andc<mode>3_64bit): Likewise.
4046         (vsx_eqv<mode>3_32bit): Likewise.
4047         (vsx_eqv<mode>3_64bit): Likewise.
4048         (vsx_nand<mode>3_32bit): Likewise.
4049         (vsx_nand<mode>3_64bit): Likewise.
4050         (vsx_orc<mode>3_32bit): Likewise.
4051         (vsx_orc<mode>3_64bit): Likewise.
4053         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
4054         logical types in GPRs.
4056         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
4057         logical insns to rs6000.md, and allow TImode operations in
4058         32-bit.
4059         (altivec_ior<mode>3): Likewise.
4060         (altivec_xor<mode>3): Likewise.
4061         (altivec_one_cmpl<mode>2): Likewise.
4062         (altivec_nor<mode>3): Likewise.
4063         (altivec_andc<mode>3): Likewise.
4065         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
4066         attributes for moving the 128-bit logical operations into
4067         rs6000.md.
4068         (BOOL_REGS_OUTPUT): Likewise.
4069         (BOOL_REGS_OP1): Likewise.
4070         (BOOL_REGS_OP2): Likewise.
4071         (BOOL_REGS_UNARY): Likewise.
4072         (BOOL_REGS_AND_CR0): Likewise.
4073         (one_cmpl<mode>2): Add support for DI logical operations on
4074         32-bit, splitting the operations to 32-bit.
4075         (anddi3): Likewise.
4076         (iordi3): Likewise.
4077         (xordi3): Likewise.
4078         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
4079         changes to combine the 32/64-bit code, allow logical operations on
4080         TI mode in 32-bit, and to use similar match_operator patterns like
4081         scalar mode uses.  Combine the Altivec and VSX code for logical
4082         operations, and move it here.
4083         (ior<mode>3, 128-bit types): Likewise.
4084         (xor<mode>3, 128-bit types): Likewise.
4085         (one_cmpl<mode>3, 128-bit types): Likewise.
4086         (nor<mode>3, 128-bit types): Likewise.
4087         (andc<mode>3, 128-bit types): Likewise.
4088         (eqv<mode>3, 128-bit types): Likewise.
4089         (nand<mode>3, 128-bit types): Likewise.
4090         (orc<mode>3, 128-bit types): Likewise.
4091         (and<mode>3_internal): Likewise.
4092         (bool<mode>3_internal): Likewise.
4093         (boolc<mode>3_internal1): Likewise.
4094         (boolc<mode>3_internal2): Likewise.
4095         (boolcc<mode>3_internal1): Likewise.
4096         (boolcc<mode>3_internal2): Likewise.
4097         (eqv<mode>3_internal1): Likewise.
4098         (eqv<mode>3_internal2): Likewise.
4099         (one_cmpl1<mode>3_internal): Likewise.
4101 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4103         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4104         Rename flag_stack_usage to flag_stack_usage_info.
4106 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
4108         * config/microblaze/sync.md: New file.
4109         * config/microblaze/microblaze.md: Include sync.md
4110         * config/microblaze/microblaze.c: Add print_operand 'y'.
4111         * config/microblaze/constraints.md: Add memory_contraint
4112         'Q' which is a single register.
4114 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
4116         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
4118 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4120         * reload.c (find_reloads): Exit loop once we find this operand
4121         cannot be reloaded somehow for this alternative.
4123         * reload.c (find_reloads): Exit loop once we find a hard register.
4125         * rtlanal.c (computed_jump_p): Exit loop once we find label
4126         reference is used.
4128         * i386.c (ix86_pad_returns): Exit loop after setting replace.
4130         * cfgloopmanip.c (remove_path): Exit loop after setting
4131         irred_invalidated.
4133         * gensupport.c (subst_dup): Avoid loop if code is not
4134         MATCH_DUP nor MATCH_OP_DUP.
4136 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
4138         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
4140 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4142         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
4143         true for SP_REGNUM if mode == ptr_mode.
4144         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
4145         with value R0_REGNUM + 31.
4147 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4149         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
4150         pad pointer-typed argument downward.
4152 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4154         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
4155         and __ILP32__ when the ILP32 model is in use.
4157 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4159         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
4160         (aarch64_load_symref_appropriately): In the case of
4161         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
4162         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
4163         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
4164         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
4165         if 'mode' doesn't equal to 'ptr_mode'.
4166         (aarch64_output_mi_thunk): Add an assertion on the alignment of
4167         'vcall_offset'; change to call aarch64_emit_move differently depending
4168         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
4169         to calculate the upper bound of 'vcall_offset'.
4170         (aarch64_cannot_force_const_mem): Change to also return true if
4171         mode != ptr_mode.
4172         (aarch64_legitimize_reload_address): In the case of large
4173         displacements, add new local variable 'xmode' and an assertion
4174         based on it; change to use 'xmode' to generate the new rtx and
4175         reload.
4176         (aarch64_asm_trampoline_template): Change to generate the template
4177         differently depending on TARGET_ILP32 or not; change to use
4178         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
4179         (aarch64_trampoline_size): Removed.
4180         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
4181         and replace immediate literals with it.  Change to use 'ptr_mode'
4182         instead of 'DImode' and call convert_memory_address if the mode
4183         of 'fnaddr' doesn't equal to 'ptr_mode'.
4184         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
4185         to output symbol.
4186         (aarch64_elf_asm_destructor): Likewise.
4187         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
4188         on TARGET_ILP32 instead of aarch64_trampoline_size.
4189         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
4190         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
4191         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
4192         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
4193         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
4194         (storewb_pair<GPI:mode>_<P:mode>): ... this.
4195         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
4196         depending on the value of 'mode'.
4197         (add_losym_<mode>): New.
4198         (ldr_got_small_<mode>): New, based on ldr_got_small.
4199         (ldr_got_small): Remove.
4200         (ldr_got_small_sidi): New.
4201         * config/aarch64/iterators.md (P): New.
4202         (PTR): Change to 'ptr_mode' in the condition.
4204 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
4206         * config.gcc (aarch64*-*-*): Support --with-abi.
4207         (aarch64*-*-elf): Support --with-multilib-list.
4208         (aarch64*-*-linux*): Likewise.
4209         (supported_defaults): Add abi to aarch64*-*-*.
4210         * configure.ac: Mention AArch64 for --with-multilib-list.
4211         * configure: Re-generated.
4212         * config/aarch64/biarchilp32.h: New file.
4213         * config/aarch64/biarchlp64.h: New file.
4214         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
4215         (ABI_SPEC): Ditto.
4216         (MULTILIB_DEFAULTS): Ditto.
4217         (DRIVER_SELF_SPECS): Ditto.
4218         (ASM_SPEC): Update to also substitute -mabi.
4219         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
4220         file whose name depends on -mabi= and -mbig-endian.
4221         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
4222         TARGET_ILP32.
4223         (POINTER_SIZE): New define.
4224         (POINTERS_EXTEND_UNSIGNED): Ditto.
4225         (enum aarch64_abi_type): New enumeration tag.
4226         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
4227         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
4228         (TARGET_ILP32): New define.
4229         * config/aarch64/aarch64.opt (mabi): New.
4230         (aarch64_abi): New.
4231         (ilp32, lp64): New values for -mabi.
4232         * config/aarch64/t-aarch64 (comma): New define.
4233         (MULTILIB_OPTIONS): Ditto.
4234         (MULTILIB_DIRNAMES): Ditto.
4235         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
4236         * doc/invoke.texi: Document -mabi for AArch64.
4238 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
4240         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
4242 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
4243             Anton Blanchard  <anton@au1.ibm.com>
4245         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
4246         endianness when selecting field to splat.
4248 2013-07-22  Eric Christopher  <echristo@gmail.com>
4250         * dwarf2out.c (die_odr_checksum): New function to use
4251         CHECKSUM_ macros and ULEB128 for DIE tag.
4252         (generate_type_signature): Use.
4254 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
4256         * config.gcc (sparc*-*-*): Accept leon3 processor.
4257         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
4258         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
4259         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
4260         * config/sparc/sparc.opt (enum processor_type): Add leon3.
4261         (mfix-ut699): Adjust comment.
4262         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
4263         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
4264         (CPP_CPU_SPEC): Likewise.
4265         (ASM_CPU_SPEC): Likewise.
4266         * config/sparc/sparc.c (leon3_cost): New constant.
4267         (sparc_option_override): Add leon3 support.
4268         (mem_ref): New function.
4269         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
4270         (sparc_do_work_around_errata): Look into the instruction in the delay
4271         slot and adjust accordingly.  Add fix for the data cache nullify issues
4272         of the UT699.  Change insertion position for the NOP.
4273         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
4274         (leon3_load): New reservation.
4275         (leon_store): Bump latency to 2.
4276         (grfpu): New automaton.
4277         (grfpu_alu): New unit.
4278         (grfpu_ds): Likewise.
4279         (leon_fp_alu): Adjust.
4280         (leon_fp_mult): Delete.
4281         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
4282         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
4283         * config/sparc/sparc.md (cpu): Add leon3.
4284         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
4285         (swapsi): Likewise.
4286         (atomic_test_and_set): Likewise.
4287         (ldstub): Likewise.
4289 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
4291         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
4292         default for R5900 targets.
4293         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
4294         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
4295         * config/mips/mips.c (mips_option_override): Report an error for
4296         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
4297         for -march=r5900 -mhard-float.
4299 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
4301         * df-problems.c (can_move_insns_across): Exit loop once we
4302         find a non-fixed, non-global register.
4304         * ipa-pure-const.c (propagate_nothrow): Exit loop after
4305         setting can_throw.
4307         * omega.c (omega_eliminate_red): Break after setting red_found.
4308         (omega_problem_has_red_equations): Similarly after setting found.
4309         (omega_query_variable): Similarly after setting coupled.
4311 2013-07-22  Marek Polacek  <polacek@redhat.com>
4313         * gimplify.c: Don't include gimple.h twice.
4315 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4317         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
4318         instead of TARGET_THUMB1.
4319         (Pz): New constraint.
4320         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
4321         encodings.
4322         (compare_negsi_si): Likewise.
4323         (compare_addsi2_op0): Likewise.
4324         (compare_addsi2_op1): Likewise.
4325         (addsi3_carryin_<optab>): Likewise.
4326         (addsi3_carryin_alt2_<optab>): Likewise.
4327         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
4328         for arm_restrict_it.
4329         (subsi3_carryin): Likewise.
4330         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
4331         (minmax_arithsi): Disable for arm_restrict_it.
4332         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
4333         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
4334         (satsi_<SAT:code>_shift): Likewise.
4335         (arm_shiftsi3): Add alternative for 16-bit encoding.
4336         (arm32_movhf): Disable for arm_restrict_it.
4337         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
4338         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
4340 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
4342         * config/arm/arm.md (attribute "insn"): Delete.
4343         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
4344         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
4345         (not_shiftsi): Update for attribute change.
4346         (not_shiftsi_compare0): Likewise.
4347         (not_shiftsi_compare0_scratch): Likewise.
4348         (arm_one_cmplsi2): Likewise.
4349         (thumb1_one_cmplsi2): Likewise.
4350         (notsi_compare0): Likewise.
4351         (notsi_compare0_scratch): Likewise.
4352         (thumb1_movdi_insn): Likewise.
4353         (arm_movsi_insn): Likewise.
4354         (movhi_insn_arch4): Likewise.
4355         (movhi_bytes): Likewise.
4356         (arm_movqi_insn): Likewise.
4357         (thumb1_movqi_insn): Likewise.
4358         (arm32_movhf): Likewise.
4359         (thumb1_movhf): Likewise.
4360         (arm_movsf_soft_insn): Likewise.
4361         (thumb1_movsf_insn): Likewise.
4362         (thumb_movdf_insn): Likewise.
4363         (movsicc_insn): Likewise.
4364         (movsfcc_soft_insn): Likewise.
4365         (and_scc): Likewise.
4366         (cond_move): Likewise.
4367         (if_move_not): Likewise.
4368         (if_not_move): Likewise.
4369         (if_shift_move): Likewise.
4370         (if_move_shift): Likewise.
4371         (if_shift_shift): Likewise.
4372         (if_not_arith): Likewise.
4373         (if_arith_not): Likewise.
4374         (cond_move_not): Likewise.
4375         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
4376         (neon_mov<mode>): Likewise.
4377         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
4378         (thumb2_movsi_vfp): Likewise.
4379         (movsf_vfp): Likewise.
4380         (thumb2_movsf_vfp): Likewise.
4381         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4382         change.
4383         (cortexa7_older_only): Likewise.
4384         (cortexa7_younger): Likewise.
4385         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
4386         (1020alu_shift_op): Likewise.
4387         (1020alu_shift_reg_op): Likewise.
4388         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
4389         (alu_shift_op): Likewise.
4390         (alu_shift_reg_op): Likewise.
4391         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
4392         (11_alu_shift_op): Likewise.
4393         (11_alu_shift_reg_op): Likewise.
4394         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
4395         (9_alu_shift_reg_op): Likewise.
4396         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4397         change.
4398         (cortex_a15_alu_shift): Likewise.
4399         (cortex_a15_alu_shift_reg): Likewise.
4400         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
4401         (cortex_a5_alu_shift): Likewise.
4402         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
4403         change.
4404         (cortex_a53_alu_shift): Likewise.
4405         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4406         change.
4407         (cortex_a7_alu_reg): Likewise.
4408         (cortex_a7_alu_shift): Likewise.
4409         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4410         (cortex_a8_alu_shift): Likewise.
4411         (cortex_a8_alu_shift_reg): Likewise.
4412         (cortex_a8_mov): Likewise.
4413         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
4414         (cortex_a9_dp_shift): Likewise.
4415         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
4416         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
4417         (cortex_r4_mov): Likewise.
4418         (cortex_r4_alu_shift): Likewise.
4419         (cortex_r4_alu_shift_reg): Likewise.
4420         * config/arm/fa526.md (526_alu_op): Update for attribute change.
4421         (526_alu_shift_op): Likewise.
4422         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
4423         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
4424         (626te_alu_shift_op): Likewise.
4425         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
4426         (726te_alu_op): Likewise.
4427         (726te_alu_shift_op): Likewise.
4428         (726te_alu_shift_reg_op): Likewise.
4429         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
4430         (mp626_alu_shift_op): Likewise.
4431         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
4432         (pj4_alu_e1_conds): Likewise.
4433         (pj4_alu): Likewise.
4434         (pj4_alu_conds): Likewise.
4435         (pj4_shift): Likewise.
4436         (pj4_shift_conds): Likewise.
4437         (pj4_alu_shift): Likewise.
4438         (pj4_alu_shift_conds): Likewise.
4440 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4442         * config/arm/predicates.md (shiftable_operator_strict_it):
4443         New predicate.
4444         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
4445         Disable cond_exec version for arm_restrict_it.
4446         (thumb2_smaxsi3): Convert to generate cond_exec.
4447         (thumb2_sminsi3): Likewise.
4448         (thumb32_umaxsi3): Likewise.
4449         (thumb2_uminsi3): Likewise.
4450         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
4451         (thumb2_neg_abssi2): Likewise.
4452         (thumb2_mov_scc): Add alternative for 16-bit encoding.
4453         (thumb2_movsicc_insn): Adjust alternatives.
4454         (thumb2_mov_negscc): Disable for arm_restrict_it.
4455         (thumb2_mov_negscc_strict_it): New pattern.
4456         (thumb2_mov_notscc_strict_it): New pattern.
4457         (thumb2_mov_notscc): Disable for arm_restrict_it.
4458         (thumb2_ior_scc): Likewise.
4459         (thumb2_ior_scc_strict_it): New pattern.
4460         (thumb2_cond_move): Adjust for arm_restrict_it.
4461         (thumb2_cond_arith): Disable for arm_restrict_it.
4462         (thumb2_cond_arith_strict_it): New pattern.
4463         (thumb2_cond_sub): Adjust for arm_restrict_it.
4464         (thumb2_movcond): Likewise.
4465         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
4466         (thumb2_zero_extendhisi2_v6): Likewise.
4467         (thumb2_zero_extendqisi2_v6): Likewise.
4468         (orsi_notsi_si): Likewise.
4469         (orsi_not_shiftsi_si): Likewise.
4471 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
4473         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
4474         instruction sequence is 1 byte shorter.
4476 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
4478         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
4479         it is not needed after split.
4481 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
4483         PR target/51784
4484         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
4485         second label for nonlocal goto receivers. Don't output pic base labels
4486         unless we're producing PIC; mark that action unreachable().
4487         (ix86_save_reg): If the function contains a nonlocal label, save the
4488         PIC base reg.
4489         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
4490         * gcc/config/darwin.c (emitted_pic_label_num): New GTY.
4491         (update_pic_label_number_if_needed): New.
4492         (machopic_output_function_base_name): Adjust for nonlocal receiver
4493         case.
4494         (machopic_should_output_picbase_label): New.
4495         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
4496         (nonlocal_goto_receiver): New insn and split.
4498 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
4500         * config/aarch64/aarch64-builtins.c
4501         (aarch64_fold_builtin): Fold abs in all modes.
4502         * config/aarch64/aarch64-simd-builtins.def
4503         (abs): Enable for all modes.
4504         * config/aarch64/arm_neon.h
4505         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
4506         (vabs_f64): Add missing intrinsic.
4508 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
4510         * config/aarch64/arm_neon.h (vabs_s64): New function
4512 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
4514         PR target/57516
4515         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
4516         * config/avr/avr.md (adjust_len): Add `round'.
4517         * config/avr/avr-protos.h (avr_out_round): New prototype.
4518         (avr_out_plus): Add `out_label' argument.
4519         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
4520         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
4521         Handle the case where `insn' is just a pattern.
4522         (avr_out_bitop): Handle the case where `insn' is just a pattern.
4523         (avr_out_round): New function.
4524         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
4526 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
4528         * config/microblaze/microblaze.c (microblaze_expand_prologue):
4529         Add check for flag_stack_usage to handle -fstack-usage support
4531 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
4533         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
4534         interaction for new Power8 flags and VSX.
4536 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
4538         PR middle-end/57698
4539         * tree-inline.c (expand_call_inline): Emit errors during
4540         early_inlining only if optimization is not turned on.
4542 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4544         * passes.def: New.
4546         * passes.c (init_optimization_passes): Move the construction of
4547         the pass hierarchy into a new passes.def file.
4549         * Makefile.in (passes.o): Add dependency on passes.def.
4551 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4553         * passes.c (init_optimization_passes): Introduce macros for
4554         constructing the tree of passes (INSERT_PASSES_AFTER,
4555         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
4556         TERMINATE_PASS_LIST).
4558 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
4559             Wei Mi  <wmi@google.com>
4561         PR rtl-optimization/57878
4562         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
4563         top.
4564         (reload_pseudo_compare_func): Check nregs first for reload
4565         pseudos.
4567 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4569         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
4571 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
4573         * read-rtl.c (validate_const_int): Once an invalid character is
4574         seen, quit the loop.
4576         * gengtype.c (write_roots): Similarly once we find the "deletable"
4577         or "if_marked" option.
4579 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4581         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
4582         "xtab" and "sat".  Move value "clz" from here to ...
4583         (attriubte "type"): ... here.
4584         (satsi_<SAT:code>): Delete "insn" attribute.
4585         (satsi_<SAT:code>_shift): Likewise.
4586         (arm_zero_extendqisi2addsi): Likewise.
4587         (arm_extendqisi2addsi): Likewise.
4588         (clzsi2): Update for attribute changes.
4589         (rbitsi2): Likewise.
4590         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
4591         attribute.
4592         (arm_usatsihi): Likewise.
4593         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
4595 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
4597         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
4598         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
4599         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
4600         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
4601         in alphabetical order.
4602         (attribute "core_cycles"): Update for attribute changes.
4603         (arm_addsi3): Likewise.
4604         (addsi3_compare0): Likewise.
4605         (addsi3_compare0_scratch): Likewise.
4606         (addsi3_compare_op1): Likewise.
4607         (addsi3_compare_op2): Likewise.
4608         (compare_addsi2_op0): Likewise.
4609         (compare_addsi2_op1): Likewise.
4610         (addsi3_carryin_shift_<optab>): Likewise.
4611         (subsi3_carryin_shift): Likewise.
4612         (rsbsi3_carryin_shift): Likewise.
4613         (arm_subsi3_insn): Likewise.
4614         (subsi3_compare0): Likewise.
4615         (subsi3_compare): Likewise.
4616         (arm_andsi3_insn): Likewise.
4617         (thumb1_andsi3_insn): Likewise.
4618         (andsi3_compare0): Likewise.
4619         (andsi3_compare0_scratch): Likewise.
4620         (zeroextractsi_compare0_scratch
4621         (andsi_not_shiftsi_si): Likewise.
4622         (iorsi3_insn): Likewise.
4623         (iorsi3_compare0): Likewise.
4624         (iorsi3_compare0_scratch): Likewise.
4625         (arm_xorsi3): Likewise.
4626         (thumb1_xorsi3_insn): Likewise.
4627         (xorsi3_compare0): Likewise.
4628         (xorsi3_compare0_scratch): Likewise.
4629         (satsi_<SAT:code>_shift): Likewise.
4630         (rrx): Likewise.
4631         (arm_shiftsi3): Likewise.
4632         (shiftsi3_compare0): Likewise.
4633         (not_shiftsi): Likewise.
4634         (not_shiftsi_compare0): Likewise.
4635         (not_shiftsi_compare0_scratch): Likewise.
4636         (arm_one_cmplsi2): Likewise.
4637         (thumb_one_complsi2): Likewise.
4638         (notsi_compare0): Likewise.
4639         (notsi_compare0_scratch): Likewise.
4640         (thumb1_zero_extendhisi2): Likewise.
4641         (arm_zero_extendhisi2): Likewise.
4642         (arm_zero_extendhisi2_v6): Likewise.
4643         (arm_zero_extendhisi2addsi): Likewise.
4644         (thumb1_zero_extendqisi2): Likewise.
4645         (thumb1_zero_extendqisi2_v6): Likewise.
4646         (arm_zero_extendqisi2): Likewise.
4647         (arm_zero_extendqisi2_v6): Likewise.
4648         (arm_zero_extendqisi2addsi): Likewise.
4649         (thumb1_extendhisi2): Likewise.
4650         (arm_extendhisi2): Likewise.
4651         (arm_extendhisi2_v6): Likewise.
4652         (arm_extendqisi): Likewise.
4653         (arm_extendqisi_v6): Likewise.
4654         (arm_extendqisi2addsi): Likewise.
4655         (thumb1_extendqisi2): Likewise.
4656         (thumb1_movdi_insn): Likewise.
4657         (arm_movsi_insn): Likewise.
4658         (movsi_compare0): Likewise.
4659         (movhi_insn_arch4): Likewise.
4660         (movhi_bytes): Likewise.
4661         (arm_movqi_insn): Likewise.
4662         (thumb1_movqi_insn): Likewise.
4663         (arm32_movhf): Likewise.
4664         (thumb1_movhf): Likewise.
4665         (arm_movsf_soft_insn): Likewise.
4666         (thumb1_movsf_insn): Likewise.
4667         (movdf_soft_insn): Likewise.
4668         (thumb_movdf_insn): Likewise.
4669         (arm_cmpsi_insn): Likewise.
4670         (cmpsi_shiftsi): Likewise.
4671         (cmpsi_shiftsi_swp): Likewise.
4672         (arm_cmpsi_negshiftsi_si): Likewise.
4673         (movsicc_insn): Likewise.
4674         (movsfcc_soft_insn): Likewise.
4675         (arith_shiftsi): Likewise.
4676         (arith_shiftsi_compare0
4677         (arith_shiftsi_compare0_scratch
4678         (sub_shiftsi): Likewise.
4679         (sub_shiftsi_compare0
4680         (sub_shiftsi_compare0_scratch
4681         (and_scc): Likewise.
4682         (cond_move): Likewise.
4683         (if_plus_move): Likewise.
4684         (if_move_plus): Likewise.
4685         (if_move_not): Likewise.
4686         (if_not_move): Likewise.
4687         (if_shift_move): Likewise.
4688         (if_move_shift): Likewise.
4689         (if_shift_shift): Likewise.
4690         (if_not_arith): Likewise.
4691         (if_arith_not): Likewise.
4692         (cond_move_not): Likewise.
4693         (thumb1_ashlsi3): Set type attribute.
4694         (thumb1_ashrsi3): Likewise.
4695         (thumb1_lshrsi3): Likewise.
4696         (thumb1_rotrsi3): Likewise.
4697         (shiftsi3_compare0_scratch): Likewise.
4698         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
4699         (neon_mov<mode>): Likewise.
4700         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
4701         attribute changes.
4702         (thumb2_movsi_insn): Likewise.
4703         (thumb2_cmpsi_neg_shiftsi): Likewise.
4704         (thumb2_extendqisi_v6): Likewise.
4705         (thumb2_zero_extendhisi2_v6): Likewise.
4706         (thumb2_zero_extendqisi2_v6): Likewise.
4707         (thumb2_shiftsi3_short): Likewise.
4708         (thumb2_addsi3_compare0_scratch): Likewise.
4709         (orsi_not_shiftsi_si): Likewise.
4710         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
4711         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
4712         changes.
4713         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
4714         (1020alu_shift_op): Likewise.
4715         (1020alu_shift_reg_op): Likewise.
4716         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
4717         (alu_shift_op): Likewise.
4718         (alu_shift_reg_op): Likewise.
4719         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
4720         (11_alu_shift_op): Likewise.
4721         (11_alu_shift_reg_op): Likewise.
4722         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
4723         (9_alu_shift_reg_op): Likewise.
4724         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
4725         changes.
4726         (cortex_a15_alu_shift): Likewise.
4727         (cortex_a15_alu_shift_reg): Likewise.
4728         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
4729         changes.
4730         (cortex_a5_alu_shift): Likewise.
4731         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4732         changes.
4733         (cortex_a53_alu_shift): Likewise.
4734         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4735         changes.
4736         (cortex_a7_alu_reg): Likewise.
4737         (cortex_a7_alu_shift): Likewise.
4738         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
4739         changes.
4740         (cortex_a8_alu_shift): Likewise.
4741         (cortex_a8_alu_shift_reg): Likewise.
4742         (cortex_a8_mov): Likewise.
4743         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
4744         (cortex_a9_dp_shift): Likewise.
4745         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
4746         changes.
4747         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
4748         changes.
4749         (cortex_r4_mov): Likewise.
4750         (cortex_r4_alu_shift): Likewise.
4751         (cortex_r4_alu_shift_reg): Likewise.
4752         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
4753         (526_alu_shift_op): Likewise.
4754         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
4755         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
4756         (626te_alu_shift_op): Likewise.
4757         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
4758         (726te_alu_op): Likewise.
4759         (726te_alu_shift_op): Likewise.
4760         (726te_alu_shift_reg_op): Likewise.
4761         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
4762         (mp626_alu_shift_op): Likewise.
4763         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
4764         (pj4_alu_e1_conds): Likewise.
4765         (pj4_alu): Likewise.
4766         (pj4_alu_conds): Likewise.
4767         (pj4_shift): Likewise.
4768         (pj4_shift_conds): Likewise.
4769         (pj4_alu_shift): Likewise.
4770         (pj4_alu_shift_conds): Likewise.
4771         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
4772         changes.
4773         (cortexa7_older_only): Likewise.
4774         (cortexa7_younger): Likewise.
4776 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
4778         * ipa-pure-const.c (generate_summary): Rename to...
4779         (pure_const_generate_summary): ... this.
4781 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
4783         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
4785 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
4787         PR target/57909
4788         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
4789         usage in HI mode.
4791 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
4793         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
4794         enabled without -march=zEC12.
4795         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
4796         flags to be set.
4798 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4800         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
4801         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
4802         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
4803         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
4804         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
4805         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
4806         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
4807         ISA_HAS_FP4.
4808         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
4809         and ISA_HAS_NMADD3_NMSUB3.
4810         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
4811         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
4812         (nmsub4<mode>, nmsub3<mode>): Likewise.
4813         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
4815 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4817         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
4818         TARGET_MIPS5400 checking.
4820 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
4821             Peter Bergner  <bergner@vnet.ibm.com>
4823         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
4824         registers in the comment.
4825         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
4826         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
4827         rather than FIRST_PSEUDO_REGISTERS.
4829 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
4831         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
4832         enable extra ISA flags with TARGET_HTM.
4834 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
4836         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
4837         Fix comment typos.
4839 2013-07-15  Cong Hou  <congh@google.com>
4841         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
4842         in compare function for sorting.
4844 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
4846         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
4847         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
4848         * config/rs6000/rs6000.opt: Add -mhtm option.
4849         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
4850         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
4851         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
4852         __HTM__ if the HTM instructions are available.
4853         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
4854         htm_spr_reg_operand): New define_predicates.
4855         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
4856         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
4857         Include htm.md.
4858         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
4859         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
4860         HTM builtin functions.
4861         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
4862         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
4863         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
4864         (rs6000_builtin_mask_calculate): Likewise.
4865         (rs6000_option_override_internal): Likewise.
4866         (bdesc_htm): Add new HTM builtin support.
4867         (htm_spr_num): New function.
4868         (htm_spr_regno): Likewise.
4869         (rs6000_htm_spr_icode): Likewise.
4870         (htm_expand_builtin): Likewise.
4871         (htm_init_builtins): Likewise.
4872         (rs6000_expand_builtin): Add support for HTM builtin functions.
4873         (rs6000_init_builtins): Likewise.
4874         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
4875         option.
4876         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
4877         (TARGET_HTM, MASK_HTM): Define macros.
4878         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
4879         (FIXED_REGISTERS): Likewise.
4880         (CALL_USED_REGISTERS): Likewise.
4881         (CALL_REALLY_USED_REGISTERS): Likewise.
4882         (REG_ALLOC_ORDER): Likewise.
4883         (enum reg_class): Likewise.
4884         (REG_CLASS_NAMES): Likewise.
4885         (REG_CLASS_CONTENTS): Likewise.
4886         (REGISTER_NAMES): Likewise.
4887         (ADDITIONAL_REGISTER_NAMES): Likewise.
4888         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
4889         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
4890         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
4891         * config/rs6000/htm.md: New file.
4892         * config/rs6000/htmintrin.h: New file.
4893         * config/rs6000/htmxlintrin.h: New file.
4895 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
4897         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
4898         Define SYMBOL_TINY_GOT, update comment.
4899         * config/aarch64/aarch64.c
4900         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
4901         (aarch64_expand_mov_immediate): Likewise.
4902         (aarch64_print_operand): Likewise.
4903         (aarch64_classify_symbol): Likewise.
4904         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
4905         (ldr_got_tiny): Define.
4907 2013-07-13  Tobias Grosser  <tobias@grosser.es>
4909         PR tree-optimization/54094
4910         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
4911           scheduling dimension for the parallelism check from the polyhedral
4912           information in the AST.
4913         * graphite-dependences.c (carries_deps): Do not assume the schedule is
4914           in 2D + 1 form.
4916 2013-07-13  Jason Merrill  <jason@redhat.com>
4918         * print-tree.c (debug_vec_tree): Use debug_raw.
4919         (debug_raw (vec<tree, va_gc> &)): New.
4920         (debug_raw (vec<tree, va_gc> *)): New.
4921         * tree.h: Declare them.
4923 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
4925         * ifcvt.c (ifcvt_after_combine): New static variable.
4926         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
4927         for size.
4928         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
4929         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
4930         rest_of_handle_if_after_reload): Pass new argument for if_convert.
4932 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
4934         * config/mips/mips.c (mips_expand_call): Remove empty statement.
4936 2013-07-12  Michael Matz  <matz@suse.de>
4938         PR middle-end/55771
4939         * convert.c (convert_to_real): Reject non-float inner types.
4941 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
4943         * config/aarch64/aarch64-protos.h
4944         (aarch64_simd_immediate_valid_for_move): Remove.
4945         * config/aarch64/aarch64.c (simd_immediate_info): New member.
4946         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
4947         cases.
4948         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
4950 2013-07-11  Steve Ellcey  <sellcey@mips.com>
4952         * config/mips/mips.c (mips_conditional_register_usage): Do not
4953         use t[0-7] registers in MIPS16 mode when optimizing for size.
4955 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
4957         * config/i386/i386.c (dispatch_function_versions): Fix array
4958         indexing of function_version_info to match actual_versions.
4960 2013-07-11  Teresa Johnson  <tejohnson@google.com>
4962         * vec.h (struct va_gc): Move release out-of-line.
4963         (va_gc::release): Call ggc_free on released vec.
4965 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
4967         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
4968         Require GOT register as additional operand in UNSPEC.
4969         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
4970         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4971         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
4972         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
4973         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
4974         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
4975         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
4977 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
4979         PR target/57631
4980         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
4981         name seen by assembler/linker rather if available.
4983 2013-07-11  Andreas Schwab  <schwab@suse.de>
4985         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
4987 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
4989         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
4991 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
4993         * doc/tm.texi.in: Move hook documentation to ....
4994         * target.def: ... here.
4996         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
4997         text on @hook line.
4998         * doc/tm.texi: Regenerate.
5000 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
5002         PR c++/57869
5003         * doc/invoke.texi: Document -Wconditionally-supported.
5005 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5007         PR target/57844
5008         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5009         of my_fp.
5011 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5013         PR target/57506
5014         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
5015         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
5016         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
5017         Remove duplicate devices.
5018         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
5019         * config/avr/t-multilib: Regenerate.
5020         * config/avr/avr-tables.opt: Regenerate.
5021         * doc/avr-mmcu.texi: Regenerate.
5023 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
5025         PR target/56987
5026         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
5028 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
5030         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
5031         the cost of MULT when optimizing for size.
5033 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5035         * config/cr16/cr16-protos.h: Don't include target.h.
5037 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
5039         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
5040         adjust register size for TDmode and TFmode for VSX registers.
5042 2013-07-08  Kai Tietz  <ktietz@redhat.com>
5044         PR target/56892
5045         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
5046         hook_bool_const_tree_true.
5048 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5050         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
5051         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
5052         * config/s390/s390.md: Define FPR*_REGNUM constants.
5053         Fix FPR2_REGNUM constant (18 -> 17).
5054         ("*trunc<BFP:mode><DFP_ALL:mode>2")
5055         ("*trunc<DFP_ALL:mode><BFP:mode>2")
5056         ("trunc<BFP:mode><DFP_ALL:mode>2")
5057         ("trunc<DFP_ALL:mode><BFP:mode>2")
5058         ("*extend<BFP:mode><DFP_ALL:mode>2")
5059         ("*extend<DFP_ALL:mode><BFP:mode>2")
5060         ("extend<BFP:mode><DFP_ALL:mode>2")
5061         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
5062         FPR4_REGNUM.
5064 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
5066         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
5068 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5070         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
5071         and cfun_fpr_bit_p to cfun_fpr_save_p.
5072         (s390_frame_area, s390_register_info, s390_frame_info)
5073         (s390_emit_prologue, s390_emit_epilogue)
5074         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
5075         register numbers.
5076         * config/s390/s390.h: Define *_REGNUM macros for floating point
5077         register numbers.
5079 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
5081         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
5083 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
5085         PR rtl-optimization/57786
5086         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
5087         and break out of the loop when it is set to false.
5089 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
5091         PR target/57819
5092         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
5093         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
5094         (const_int 63)) 0)).
5095         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
5096         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
5097         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
5099         PR rtl-optimization/57829
5100         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
5101         mask bits outside of mode are just sign-extension from mode to HWI.
5103 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5105         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
5106         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
5107         adjust_address instead of change_address to keep info about alignment.
5108         (emit_strmov): Remove.
5109         (emit_memmov): New function.
5110         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
5111         (expand_movmem_epilogue): Likewise and return updated rtx for
5112         destination.
5113         (expand_constant_movmem_prologue): Likewise and return updated rtx for
5114         destination and source.
5115         (decide_alignment): Refactor, handle vector_loop.
5116         (ix86_expand_movmem): Likewise.
5117         (ix86_expand_setmem): Likewise.
5118         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
5120 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
5122         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
5123         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
5125 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5127         * config/i386/sse.md (sse_movlhps): Change alternative 3
5128         of operand 2 to "m".
5130 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
5132         PR target/57807
5133         * config/i386/sse.md (iptr): New mode attribute.
5134         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
5135         (<sse>_vm<plusminus_insn><mode>3): Ditto.
5136         (<sse>_vmmul<mode>3): Ditto.
5137         (<sse>_vmdiv<mode>3): Ditto.
5138         (sse_vmrcpv4sf2): Ditto.
5139         (<sse>_vmsqrt<mode>2): Ditto.
5140         (sse_vmrsqrtv4sf2): Ditto.
5141         (<sse>_vm<code><mode>3): Ditto.
5142         (avx_vmcmp<mode>3): Ditto.
5143         (<sse>_vmmaskcmp<mode>3): Ditto.
5144         (<sse>_comi): Ditto.
5145         (<sse>_ucomi): Ditto.
5146         (*xop_vmfrcz_<mode>): Ditto.
5147         (*fmai_fmadd_<mode>): Ditto.
5148         (*fmai_fmsub_<mode>): Ditto.
5149         (*fmai_fnmadd_<mode>): Ditto.
5150         (*fmai_fnmsub_<mode>): Ditto.
5151         (*fma4i_vmfmadd_<mode>): Ditto.
5152         (*fma4i_vmfmsub_<mode>): Ditto.
5153         (*fma4i_vmfnmadd_<mode>): Ditto.
5154         (*fma4i_vmfnmsub_<mode>): Ditto.
5155         (*xop_vmfrcz_<mode>): Ditto.
5156         (sse_cvtps2pi): Ditto.
5157         (sse_cvttps2pi): Ditto.
5158         (sse_cvtss2si): Ditto.
5159         (sse_cvtss2si_2): Ditto.
5160         (sse_cvtss2siq_2): Ditto.
5161         (sse_cvttss2si): Ditto.
5162         (sse_cvttss2siq): Ditto.
5163         (sse_cvtsd2si): Ditto.
5164         (sse_cvtsd2si_2): Ditto.
5165         (sse_cvtsd2siq_2): Ditto.
5166         (sse_cvttsd2si): Ditto.
5167         (sse_cvttsd2siq): Ditto.
5168         (sse_cvtsd2ss): Ditto.
5169         (sse_cvtss2sd): Ditto.
5170         (avx2_pbroadcast<mode>): Ditto.
5171         (avx2_pbroadcast<mode>_1): Ditto.
5172         (*avx_vperm_broadcast_v4sf): Ditto.
5174         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
5175         (sse_movlhps): Ditto.
5176         (sse_storehps): Ditto.
5177         (sse_loadhps): Ditto.
5178         (sse_storelps): Ditto.
5179         (sse_loadlps): Ditto.
5180         (*vec_concatv4sf): Ditto.
5181         (*vec_interleave_highv2df): Ditto.
5182         (*vec_interleave_lowv2df): Ditto.
5183         (*vec_extractv2df_1_sse): Ditto.
5184         (*vec_extractv2df_0_sse): Ditto.
5185         (sse2_storelpd): Ditto.
5186         (sse2_loadlpd): Ditto.
5187         (sse2_movsd): Ditto.
5188         (*vec_concatv4si): Ditto.
5189         (vec_concatv2di): Ditto.
5191         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
5192         for Intel asm dialect.
5193         (mmx_punpcklwd): Ditto.
5194         (mmx_punpckldq): Ditto.
5196         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
5197         for intel assembler dialect.
5199 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
5201         PR target/29776
5202         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
5203         for BUILT_IN_C{LZ,LRSB}*.
5204         * tree.h (CASE_INT_FN): Add FN##IMAX case.
5205         * tree-vrp.c (extract_range_basic): Handle
5206         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
5207         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
5208         fall thru to code calling set_value*.
5209         * builtins.c (expand_builtin): Remove *IMAX cases.
5210         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
5211         if width is bigger than 2*HWI.
5213 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
5215         PR rtl-optimization/55342
5216         * lra-int.h (lra_subreg_reload_pseudos): New.
5217         * lra.c: Add undoing optional reloads to the block diagram.
5218         (lra_subreg_reload_pseudos): New.
5219         (lra_optional_reload_pseudos): Change comments.
5220         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
5221         lra_optional_reload_pseudos after undo transformations.
5222         * lra-assigns.c (pseudo_prefix_title): New.
5223         (lra_setup_reg_renumber): Use it.
5224         (spill_for): Ditto.  Check subreg reload pseudos too.
5225         (assign_by_spills): Consider subreg reload pseudos too.
5226         * lra-constraints.c (simplify_operand_subreg): Use
5227         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
5228         (curr_insn_transform): Recognize and do optional reloads.
5229         (undo_optional_reloads): New.
5230         (lra_undo_inheritance): Call undo_optional_reloads.
5232 2013-07-05  Thomas Quinot  <quinot@adacore.com>
5234         * tree-complex.c (expand_complex_operations_1): Fix typo.
5236 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
5238         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
5239         (tune_params): New member 'const vec_costs'.
5240         * config/aarch64/aarch64.c (generic_vector_cost): New.
5241         (generic_tunings): New member 'generic_vector_cost'.
5242         (aarch64_builtin_vectorization_cost): New.
5243         (aarch64_add_stmt_cost): New.
5244         (TARGET_VECTORIZE_ADD_STMT_COST): New.
5245         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
5247 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
5249         PR target/57777
5250         * config/i386/predicates.md (vsib_address_operand): Disallow
5251         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
5253 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
5255         PR middle-end/55030
5256         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
5257         expand_builtin_setjmp_receiver.
5258         (expand_label): Adjust, call expand_builtin_setjmp_receiver
5259         with NULL for the label parameter.
5260         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
5261         the frame-pointer.  Adjust comments.
5262         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
5263         only if LABEL is non-NULL.
5265 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
5267         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
5268         (ARM_ABI_AAPCS64): Ditto.
5269         (arm_abi): Ditto.
5270         (ARM_DEFAULT_ABI): Ditto.
5272 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
5274         * config/aarch64/aarch64-builtins.c
5275         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
5276         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
5277         (st1): Likewise.
5278         * config/aarch64/aarch64-simd.md
5279         (aarch64_ld1<VALL:mode>): New.
5280         (aarch64_st1<VALL:mode>): Likewise.
5281         * config/aarch64/arm_neon.h
5282         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
5284 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
5286         * config/i386/i386.c (gate_insert_vzeroupper): Check if
5287         target ISA is AVX.
5288         (ix86_option_override_internal):Turn on all -mavx target flags by
5289         default as they are dependent on AVX anyway.
5291 2013-07-02  Cary Coutant  <ccoutant@google.com>
5293         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
5294         deterministic hash.
5295         (loc_checksum_ordered): Likewise.
5296         (hash_loc_operands): Remove inline keyword.
5298 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
5300         PR tree-optimization/57741
5301         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
5302         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
5303         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
5304         Allow REAL_CST step_exprs if flag_associative_math.
5305         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
5307 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5309         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
5311 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
5313         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
5315 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5317         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
5318         encoding.
5319         (iorsi3_insn): Likewise.
5320         (arm_xorsi3): Likewise.
5322 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
5324         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
5325         to ...
5326         (attribute "type"): ... here, and prefix with "wmmx_".
5327         (attribute "core_cycles"): Update for attribute changes.
5328         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
5329         (tbcstv4hi): Likewise.
5330         (tbcstv2si): Likewise.
5331         (iwmmxt_iordi3): Likewise.
5332         (iwmmxt_xordi3): Likewise.
5333         (iwmmxt_anddi3): Likewise.
5334         (iwmmxt_nanddi3): Likewise.
5335         (iwmmxt_arm_movdi): Likewise.
5336         (iwmmxt_movsi_insn): Likewise.
5337         (mov<mode>_internal): Likewise.
5338         (and<mode>3_iwmmxt): Likewise.
5339         (ior<mode>3_iwmmxt): Likewise.
5340         (xor<mode>3_iwmmxt): Likewise.
5341         (add<mode>3_iwmmxt): Likewise.
5342         (ssaddv8qi3): Likewise.
5343         (ssaddv4hi3): Likewise.
5344         (ssaddv2si3): Likewise.
5345         (usaddv8qi3): Likewise.
5346         (usaddv4hi3): Likewise.
5347         (usaddv2si3): Likewise.
5348         (sub<mode>3_iwmmxt): Likewise.
5349         (sssubv8qi3): Likewise.
5350         (sssubv4hi3): Likewise.
5351         (sssubv2si3): Likewise.
5352         (ussubv8qi3): Likewise.
5353         (ussubv4hi3): Likewise.
5354         (ussubv2si3): Likewise.
5355         (mulv4hi3_iwmmxt): Likewise.
5356         (smulv4hi3_highpart): Likewise.
5357         (umulv4hi3_highpart): Likewise.
5358         (iwmmxt_wmacs): Likewise.
5359         (iwmmxt_wmacsz): Likewise.
5360         (iwmmxt_wmacu): Likewise.
5361         (iwmmxt_wmacuz): Likewise.
5362         (iwmmxt_clrdi): Likewise.
5363         (iwmmxt_clrv8qi): Likewise.
5364         (iwmmxt_clr4hi): Likewise.
5365         (iwmmxt_clr2si): Likewise.
5366         (iwmmxt_uavgrndv8qi3): Likewise.
5367         (iwmmxt_uavgrndv4hi3): Likewise.
5368         (iwmmxt_uavgv8qi3): Likewise.
5369         (iwmmxt_uavgv4hi3): Likewise.
5370         (iwmmxt_tinsrb): Likewise.
5371         (iwmmxt_tinsrh): Likewise.
5372         (iwmmxt_tinsrw): Likewise.
5373         (iwmmxt_textrmub): Likewise.
5374         (iwmmxt_textrmsb): Likewise.
5375         (iwmmxt_textrmuh): Likewise.
5376         (iwmmxt_textrmsh): Likewise.
5377         (iwmmxt_textrmw): Likewise.
5378         (iwmxxt_wshufh): Likewise.
5379         (eqv8qi3): Likewise.
5380         (eqv4hi3): Likewise.
5381         (eqv2si3): Likewise.
5382         (gtuv8qi3): Likewise.
5383         (gtuv4hi3): Likewise.
5384         (gtuv2si3): Likewise.
5385         (gtv8qi3): Likewise.
5386         (gtv4hi3): Likewise.
5387         (gtv2si3): Likewise.
5388         (smax<mode>3_iwmmxt): Likewise.
5389         (umax<mode>3_iwmmxt): Likewise.
5390         (smin<mode>3_iwmmxt): Likewise.
5391         (umin<mode>3_iwmmxt): Likewise.
5392         (iwmmxt_wpackhss): Likewise.
5393         (iwmmxt_wpackwss): Likewise.
5394         (iwmmxt_wpackdss): Likewise.
5395         (iwmmxt_wpackhus): Likewise.
5396         (iwmmxt_wpackwus): Likewise.
5397         (iwmmxt_wpackdus): Likewise.
5398         (iwmmxt_wunpckihb): Likewise.
5399         (iwmmxt_wunpckihh): Likewise.
5400         (iwmmxt_wunpckihw): Likewise.
5401         (iwmmxt_wunpckilb): Likewise.
5402         (iwmmxt_wunpckilh): Likewise.
5403         (iwmmxt_wunpckilw): Likewise.
5404         (iwmmxt_wunpckehub): Likewise.
5405         (iwmmxt_wunpckehuh): Likewise.
5406         (iwmmxt_wunpckehuw): Likewise.
5407         (iwmmxt_wunpckehsb): Likewise.
5408         (iwmmxt_wunpckehsh): Likewise.
5409         (iwmmxt_wunpckehsw): Likewise.
5410         (iwmmxt_wunpckelub): Likewise.
5411         (iwmmxt_wunpckeluh): Likewise.
5412         (iwmmxt_wunpckeluw): Likewise.
5413         (iwmmxt_wunpckelsb): Likewise.
5414         (iwmmxt_wunpckelsh): Likewise.
5415         (iwmmxt_wunpckelsw): Likewise.
5416         (ror<mode>3): Likewise.
5417         (ashr<mode>3_iwmmxt): Likewise.
5418         (lshr<mode>3_iwmmxt): Likewise.
5419         (ashl<mode>3_iwmmxt): Likewise.
5420         (ror<mode>3_di): Likewise.
5421         (ashr<mode>3_di): Likewise.
5422         (lshr<mode>3_di): Likewise.
5423         (ashl<mode>3_di): Likewise.
5424         (iwmmxt_wmadds): Likewise.
5425         (iwmmxt_wmaddu): Likewise.
5426         (iwmmxt_tmia): Likewise.
5427         (iwmmxt_tmiaph): Likewise.
5428         (iwmmxt_tmiabb): Likewise.
5429         (iwmmxt_tmiatb): Likewise.
5430         (iwmmxt_tmiabt): Likewise.
5431         (iwmmxt_tmiatt): Likewise.
5432         (iwmmxt_tmovmskb): Likewise.
5433         (iwmmxt_tmovmskh): Likewise.
5434         (iwmmxt_tmovmskw): Likewise.
5435         (iwmmxt_waccb): Likewise.
5436         (iwmmxt_wacch): Likewise.
5437         (iwmmxt_waccw): Likewise.
5438         (iwmmxt_waligni): Likewise.
5439         (iwmmxt_walignr): Likewise.
5440         (iwmmxt_walignr0): Likewise.
5441         (iwmmxt_walignr1): Likewise.
5442         (iwmmxt_walignr2): Likewise.
5443         (iwmmxt_walignr3): Likewise.
5444         (iwmmxt_wsadb): Likewise.
5445         (iwmmxt_wsadh): Likewise.
5446         (iwmmxt_wsadbz): Likewise.
5447         (iwmmxt_wsadhz): Likewise.
5448         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
5449         (iwmmxt_wabsdiffb): Likewise.
5450         (iwmmxt_wabsdiffh): Likewise.
5451         (iwmmxt_wabsdiffw): Likewise.
5452         (iwmmxt_waddsubhx): Likewise
5453         (iwmmxt_wsubaddhx): Likewise.
5454         (addc<mode>3): Likewise.
5455         (iwmmxt_avg4): Likewise.
5456         (iwmmxt_avg4r): Likewise.
5457         (iwmmxt_wmaddsx): Likewise.
5458         (iwmmxt_wmaddux): Likewise.
5459         (iwmmxt_wmaddsn): Likewise.
5460         (iwmmxt_wmaddun): Likewise.
5461         (iwmmxt_wmulwsm): Likewise.
5462         (iwmmxt_wmulwum): Likewise.
5463         (iwmmxt_wmulsmr): Likewise.
5464         (iwmmxt_wmulumr): Likewise.
5465         (iwmmxt_wmulwsmr): Likewise.
5466         (iwmmxt_wmulwumr): Likewise.
5467         (iwmmxt_wmulwl): Likewise.
5468         (iwmmxt_wqmulm): Likewise.
5469         (iwmmxt_wqmulwm): Likewise.
5470         (iwmmxt_wqmulmr): Likewise.
5471         (iwmmxt_wqmulwmr): Likewise.
5472         (iwmmxt_waddbhusm): Likewise.
5473         (iwmmxt_waddbhusl): Likewise.
5474         (iwmmxt_wqmiabb): Likewise.
5475         (iwmmxt_wqmiabt): Likewise.
5476         (iwmmxt_wqmiatb): Likewise.
5477         (iwmmxt_wqmiatt): Likewise.
5478         (iwmmxt_wqmiabbn): Likewise.
5479         (iwmmxt_wqmiabtn): Likewise.
5480         (iwmmxt_wqmiatbn): Likewise.
5481         (iwmmxt_wqmiattn): Likewise.
5482         (iwmmxt_wmiabb): Likewise.
5483         (iwmmxt_wmiabt): Likewise.
5484         (iwmmxt_wmiatb): Likewise.
5485         (iwmmxt_wmiatt): Likewise.
5486         (iwmmxt_wmiabbn): Likewise.
5487         (iwmmxt_wmiabtn): Likewise.
5488         (iwmmxt_wmiatbn): Likewise.
5489         (iwmmxt_wmiattn): Likewise.
5490         (iwmmxt_wmiawbb): Likewise.
5491         (iwmmxt_wmiawbt): Likewise.
5492         (iwmmxt_wmiawtb): Likewise.
5493         (iwmmxt_wmiawtt): Likewise.
5494         (iwmmxt_wmiawbbn): Likewise.
5495         (iwmmxt_wmiawbtn): Likewise.
5496         (iwmmxt_wmiawtbn): Likewise.
5497         (iwmmxt_wmiawttn): Likewise.
5498         (iwmmxt_wmerge): Likewise.
5499         (iwmmxt_tandc<mode>3): Likewise.
5500         (iwmmxt_torc<mode>3): Likewise.
5501         (iwmmxt_torvsc<mode>3): Likewise.
5502         (iwmmxt_textrc<mode>3): Likewise.
5503         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
5504         (wmmxt_pack): Likewise.
5505         (wmmxt_mult_c1): Likewise.
5506         (wmmxt_mult_c2): Likewise.
5507         (wmmxt_alu_c1): Likewise.
5508         (wmmxt_alu_c2): Likewise.
5509         (wmmxt_alu_c3): Likewise.
5510         (wmmxt_transfer_c1): Likewise.
5511         (wmmxt_transfer_c2): Likewise.
5512         (wmmxt_transfer_c3): Likewise.
5513         (marvell_f_iwmmxt_wstr): Likewise.
5514         (marvell_f_iwmmxt_wldr): Likewise.
5516 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
5518         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
5520 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5522         Revert:
5523         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5524         * lra-constraints.c (need_for_split_p): Check call used hard regs
5525         living through calls.
5527         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
5528         call used regs for call insn.
5530 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
5532         PR target/57736
5533         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
5534         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
5536 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
5538         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
5540 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
5542         * lra-constraints.c (need_for_split_p): Check call used hard regs
5543         living through calls.
5545 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
5547         PR target/57744
5548         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
5549         to tie with any other modes.  Eliminate Altivec vector mode tests,
5550         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
5551         code, to return 0 if testing MODE2 for a condition, if we've
5552         already tested MODE1 for the same condition.
5554 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5556         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
5557         layout.
5559 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5561         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
5562         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
5564 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5566         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
5567         Define.
5568         (aarch64_symbolic_constant_p): Remove.
5569         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
5570         static.  Fix line length and white space.
5571         (aarch64_symbolic_constant_p): Remove.
5572         * config/aarch64/predicates.md (aarch64_valid_symref):
5573         Use aarch64_classify_symbol_expression.
5575 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5577         * config/arm/constraints.md (Ts): New constraint.
5578         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
5579         16-bit encodings.
5580         (compare_scc): Use "Ts" constraint for operand 0.
5581         (ior_scc_scc): Likewise.
5582         (and_scc_scc): Likewise.
5583         (and_scc_scc_nodom): Likewise.
5584         (ior_scc_scc_cmp): Likewise for operand 7.
5585         (and_scc_scc_cmp): Likewise.
5586         * config/arm/thumb2.md (thumb2_movsi_insn):
5587         Add alternatives for 16-bit encodings.
5588         (thumb2_movhi_insn): Likewise.
5589         (thumb2_movsicc_insn): Likewise.
5590         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
5591         (thumb2_negscc): Use "Ts" constraint.
5592         Move mvn instruction outside cond_exec block.
5593         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
5594         for 16-bit encodings.
5596 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5598         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
5599         encoding.
5600         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
5601         (mulsi3subsi): Likewise.
5602         (mulsidi3adddi): Likewise.
5603         (mulsidi3_v6): Likewise.
5604         (umulsidi3_v6): Likewise.
5605         (umulsidi3adddi_v6): Likewise.
5606         (smulsi3_highpart_v6): Likewise.
5607         (umulsi3_highpart_v6): Likewise.
5608         (mulhisi3tb): Likewise.
5609         (mulhisi3bt): Likewise.
5610         (mulhisi3tt): Likewise.
5611         (maddhisi4): Likewise.
5612         (maddhisi4tb): Likewise.
5613         (maddhisi4tt): Likewise.
5614         (maddhidi4): Likewise.
5615         (maddhidi4tb): Likewise.
5616         (maddhidi4tt): Likewise.
5617         (zeroextractsi_compare0_scratch): Likewise.
5618         (insv_zero): Likewise.
5619         (insv_t2): Likewise.
5620         (anddi_notzesidi_di): Likewise.
5621         (anddi_notsesidi_di): Likewise.
5622         (andsi_notsi_si): Likewise.
5623         (iordi_zesidi_di): Likewise.
5624         (xordi_zesidi_di): Likewise.
5625         (andsi_iorsi3_notsi): Likewise.
5626         (smax_0): Likewise.
5627         (smax_m1): Likewise.
5628         (smin_0): Likewise.
5629         (not_shiftsi): Likewise.
5630         (unaligned_loadsi): Likewise.
5631         (unaligned_loadhis): Likewise.
5632         (unaligned_loadhiu): Likewise.
5633         (unaligned_storesi): Likewise.
5634         (unaligned_storehi): Likewise.
5635         (extv_reg): Likewise.
5636         (extzv_t2): Likewise.
5637         (divsi3): Likewise.
5638         (udivsi3): Likewise.
5639         (arm_zero_extendhisi2addsi): Likewise.
5640         (arm_zero_extendqisi2addsi): Likewise.
5641         (compareqi_eq0): Likewise.
5642         (arm_extendhisi2_v6): Likewise.
5643         (arm_extendqisi2addsi): Likewise.
5644         (arm_movt): Likewise.
5645         (thumb2_ldrd): Likewise.
5646         (thumb2_ldrd_base): Likewise.
5647         (thumb2_ldrd_base_neg): Likewise.
5648         (thumb2_strd): Likewise.
5649         (thumb2_strd_base): Likewise.
5650         (thumb2_strd_base_neg): Likewise.
5651         (arm_negsi2): Add alternative for 16-bit encoding.
5652         (arm_one_cmplsi2): Likewise.
5654 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5656         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
5657         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
5658         (movdfcc): Likewise.
5659         * config/arm/vfp.md (*thumb2_movsf_vfp):
5660         Disable predication for arm_restrict_it.
5661         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
5662         (*thumb2_movdfcc_vfp): Likewise.
5663         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
5664         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
5665         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
5666         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
5667         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
5668         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
5669         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
5670         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
5671         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
5672         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
5673         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
5674         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
5675         Disable predication for arm_restrict_it.
5677 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
5679         * config/i386/bmiintrin.h (_bextr_u32): New.
5680         (_bextr_u64): Ditto.
5682 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
5684         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
5685         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
5686         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
5687         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
5688         * config/mips/n32-elf.h: ...this new file.
5690 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
5692         PR target/57224
5693         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
5694         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
5696 2013-06-27  Catherine Moore  <clm@codesourcery.com>
5698         * config/mips/mips-tables.opt: Regenerate.
5699         * config/mips/mips-cpus.def: Add m14ke and m14kec.
5700         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
5701         * doc/invoke.texi: Add -m14kc.
5703 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5705         PR target/57623
5706         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5707         constraints of operand 1 and 2.
5709         PR target/57623
5710         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
5711         to match RTL canonicalization.  Swap predicates and
5712         constraints of operand 1 and 2.
5714 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
5716         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
5717         Process OP_INOUT regs for splitting too.
5719 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
5721         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
5722         decl before the loop, initialize to NULL.
5723         (vectorizable_load): Initialize ptr_incr to NULL.
5725 2013-06-27  Martin Jambor  <mjambor@suse.cz>
5727         PR lto/57208
5728         * ipa-ref.h (ipa_maybe_record_reference): Declare.
5729         * ipa-ref.c (ipa_maybe_record_reference): New function.
5730         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
5731         * ipa-cp.c (create_specialized_node): Record potential references from
5732         aggvals.
5733         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
5735 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5737         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
5738         parameter 'mode' of type 'enum machine_mode mode'; change to pass
5739         'mode' to force_reg.
5740         (aarch64_add_offset): Update calls to aarch64_force_temporary.
5741         (aarch64_expand_mov_immediate): Likewise.
5743 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
5745         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
5746         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
5748 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5750         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
5751         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5752         (struct machine_function): Add tbegin_p.
5753         (s390_canonicalize_comparison): Fold CC mode compares to
5754         conditional jump if possible.
5755         (s390_emit_jump): Return the emitted jump.
5756         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
5757         Handle CCRAWmode compares.
5758         (s390_option_override): Default to -mhtm if available.
5759         (s390_reg_clobbered_rtx): Handle floating point regs as well.
5760         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
5761         FPRs instead of df_regs_ever_live_p.
5762         (s390_optimize_nonescaping_tx): New function.
5763         (s390_init_frame_layout): Extend clobbered_regs array to cover
5764         FPRs as well.
5765         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
5766         (s390_expand_tbegin): New function.
5767         (enum s390_builtin): New enum definition.
5768         (code_for_builtin): New array definition.
5769         (s390_init_builtins): New function.
5770         (s390_expand_builtin): New function.
5771         (TARGET_INIT_BUILTINS): Define.
5772         (TARGET_EXPAND_BUILTIN): Define.
5773         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
5774         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
5775         (s390_alc_comparison): Likewise.
5776         * config/s390/s390-modes.def: Add CCRAWmode.
5777         * config/s390/s390.h (processor_flags): Add PF_TX.
5778         (TARGET_CPU_HTM): Define macro.
5779         (TARGET_HTM): Define macro.
5780         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
5781         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
5782         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
5783         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
5784         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
5785         (TBEGIN_MASK, TBEGINC_MASK): New constants.
5786         ("*cc_to_int"): Move up.
5787         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
5788         constants other than 0.
5789         ("*ccraw_to_int"): New insn and splitter definition.
5790         ("tbegin", "tbegin_nofloat", "tbegin_retry")
5791         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
5792         ("tx_assist"): New expander.
5793         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
5794         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
5795         * config/s390/s390.opt: Add -mhtm option.
5796         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
5797         * config/s390/htmxlintrin.h: New file.
5798         * config/s390/htmintrin.h: New file.
5799         * config/s390/s390intrin.h: New file.
5800         * doc/extend.texi: Document htm builtins.
5801         * config.gcc: Add the new header files to extra_headers.
5803 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
5805         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
5806         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
5808 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
5809             Pat Haugen  <pthaugen@us.ibm.com>
5810             Peter Bergner  <bergner@vnet.ibm.com>
5812         * config/rs6000/power8.md: New.
5813         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
5814         setting for power8 entry.
5815         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
5816         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
5817         test for Power4/Power5 only.
5818         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
5819         support.
5820         (force_new_group): Adjust comment.
5821         * config/rs6000/rs6000.md: Include power8.md.
5823 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
5825         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
5826         * config/arm/arm-protos.h (arm_max_conditional_execute): New
5827         declaration.
5828         (tune_params): Update comment.
5829         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
5830         (arm_max_conditional_execute): New function.
5831         (thumb2_final_prescan_insn): Use max_insn_skipped and
5832         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
5834 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
5836         PR tree-optimization/57705
5837         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
5838         SSA_NAME step, provided that it is not defined inside the loop.
5839         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
5840         (get_initial_def_for_induction): Handle SSA_NAME IV step.
5842 2013-06-25  Martin Jambor  <mjambor@suse.cz>
5844         PR middle-end/57670
5845         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
5846         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
5847         calls in the dump.
5848         (ipa_note_param_call): Initialize member_ptr flag.
5849         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
5850         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
5851         (ipa_write_indirect_edge_info): Stream member_ptr flag.
5852         (ipa_read_indirect_edge_info): Likewise.
5854 2013-06-25  Richard Biener  <rguenther@suse.de>
5856         PR middle-end/56977
5857         * passes.c (init_optimization_passes): Move pass_fold_builtins
5858         and pass_dce earlier with -Og.
5860 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
5862         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
5863         <BIT_FIELD_REF>: Remove trailing TAB.
5864         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
5865         remove blank line.
5867 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5869         PR tree-optimization/57358
5870         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
5871         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
5872         (ipa_analyze_params_uses): Generate pessimistic info when true.
5874 2013-06-24  Martin Jambor  <mjambor@suse.cz>
5876         PR tree-optimization/57539
5877         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
5878         global.inlined_to of the new node to it.  All callers changed.
5879         * ipa-inline-transform.c (clone_inlined_nodes): New variable
5880         inlining_into, pass it to cgraph_clone_node.
5881         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
5882         ipa_free_edge_args_substructures.
5883         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
5884         rdesc linked list.  Do not assert rdesc edges have inlined caller.
5885         Assert we have found an rdesc in the rdesc list.
5887 2013-06-24  Richard Biener  <rguenther@suse.de>
5889         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
5890         (pointer_set_lookup): Declare.
5891         (class pointer_map): New template class implementing a
5892         generic pointer to T map.
5893         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
5894         pointer_map<T>::contains, pointer_map<T>::insert,
5895         pointer_map<T>::traverse): New functions.
5896         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
5897         (pointer_set_lookup): New function.
5898         (pointer_set_contains): Use pointer_set_lookup.
5899         (pointer_set_insert): Likewise.
5900         (insert_aux): Remove.
5901         (struct pointer_map_t): Embed a pointer_set_t.
5902         (pointer_map_create): Adjust.
5903         (pointer_map_destroy): Likewise.
5904         (pointer_map_contains): Likewise.
5905         (pointer_map_insert): Likewise.
5906         (pointer_map_traverse): Likewise.
5907         * tree-streamer.h (struct streamer_tree_cache_d): Use a
5908         pointer_map<unsigned> instead of a pointer_map_t.
5909         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
5910         (streamer_tree_cache_lookup): Likewise.
5911         (streamer_tree_cache_create): Likewise.
5912         (streamer_tree_cache_delete): Likewise.
5913         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
5914         pointer_map<unsigned> instead of a pointer_map_t.
5915         (lto_init_tree_ref_encoder): Adjust.
5916         (lto_destroy_tree_ref_encoder): Likewise.
5917         * lto-section-out.c (lto_output_decl_index): Likewise.
5918         (lto_record_function_out_decl_state): Likewise.
5919         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
5921 2013-06-24  Richard Biener  <rguenther@suse.de>
5923         PR tree-optimization/57488
5924         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
5926 2013-06-24  Alan Modra  <amodra@gmail.com>
5928         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
5929         (gen_easy_altivec_constant): Likewise.
5930         * config/rs6000/predicates.md (easy_vector_constant_add_self,
5931         easy_vector_constant_msb): Likewise.
5933 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
5935         PR target/57688
5936         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
5937         add missing return true.
5939 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
5941         PR target/52483
5942         * config/sh/predicates.md (general_extend_operand): Invoke
5943         general_movsrc_operand for memory operands.
5944         (general_movsrc_operand): Allow reg+reg addressing, do not use
5945         general_operand for memory operands.
5947 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
5949         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
5950         when current target options does not apply.
5951         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
5952         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
5953         * config/i386/bmiintrin.h: Pass appropriate target
5954         attributes to header.
5955         * config/i386/mmintrin.h: Ditto.
5956         * config/i386/nmmintrin.h: Ditto.
5957         * config/i386/avx2intrin.h: Ditto.
5958         * config/i386/fxsrintrin.h: Ditto.
5959         * config/i386/tbmintrin.h: Ditto.
5960         * config/i386/xsaveintrin.h: Ditto.
5961         * config/i386/f16cintrin.h: Ditto.
5962         * config/i386/xtestintrin.h: Ditto.
5963         * config/i386/xsaveoptintrin.h: Ditto.
5964         * config/i386/bmi2intrin.h: Ditto.
5965         * config/i386/lzcntintrin.h: Ditto.
5966         * config/i386/smmintrin.h: Ditto.
5967         * config/i386/wmmintrin.h: Ditto.
5968         * config/i386/x86intrin.h: Remove all header include guards.
5969         * config/i386/prfchwintrin.h: Ditto.
5970         * config/i386/pmmintrin.h: Ditto.
5971         * config/i386/tmmintrin.h: Ditto.
5972         * config/i386/xmmintrin.h: Ditto.
5973         * config/i386/popcntintrin.h: Ditto.
5974         * config/i386/rdseedintrin.h: Ditto.
5975         * config/i386/ammintrin.h: Ditto.
5976         * config/i386/emmintrin.h: Ditto.
5977         * config/i386/immintrin.h: Remove all header include guards.
5978         * config/i386/fma4intrin.h: Ditto.
5979         * config/i386/lwpintrin.h: Ditto.
5980         * config/i386/xopintrin.h: Ditto.
5981         * config/i386/ia32intrin.h: Ditto.
5982         * config/i386/avxintrin.h: Ditto.
5983         * config/i386/rtmintrin.h: Ditto.
5984         * config/i386/fmaintrin.h: Ditto.
5985         * config/i386/mm3dnow.h: Ditto.
5987 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
5989         * common/config/i386/i386-common.c: Handle LZCNT.
5991 2013-06-22  Andi Kleen  <ak@linux.intel.com>
5993         * doc/extend.texi: Use __atomic_store_n instead of
5994         __atomic_store in HLE example.
5996 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
5998         * config/sh/sh.c: Remove <cstdlib> workaround.
6000 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6002         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
6004 2013-06-21  Andi Kleen  <ak@linux.intel.com>
6006         * doc/extend.texi: Document that __atomic_clear and
6007           __atomic_test_and_set should only be used with bool.
6009 2013-06-20  Jan Hubicka  <jh@suse.cz>
6011         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
6012         types_same_for_odr.
6013         * tree.c (decls_same_for_odr): New function.
6014         (same_for_edr): New function.
6015         (types_same_for_odr): New function.
6016         (get_binfo_at_offset): Use it.
6017         * tree.h (types_same_for_odr): Declare.
6019 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
6020             Jason Merrill  <jason@redhat.com>
6022         * system.h: Include <cstdlib> as well as <stdlib.h>.
6024 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
6026         PR target/57655
6027         * config/i386/i386.c (construct_container): Report error if
6028         long double is used with disabled x87 float returns.
6030 2013-06-20  Jan Hubicka  <jh@suse.cz>
6032         * lto-cgraph.c (input_symtab): Do not set cgraph state.
6034 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
6036         PR rtl-optimization/57425
6037         PR rtl-optimization/57569
6038         * alias.c (write_dependence_p): Remove parameters mem_mode and
6039         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
6040         Changed all callers.
6041         (canon_anti_dependence): Get comments and semantics in sync.
6042         Add parameter mem_canonicalized.  Changed all callers.
6043         * rtl.h (canon_anti_dependence): Update prototype.
6045 2013-06-20  Richard Biener  <rguenther@suse.de>
6047         * data-streamer-in.c (streamer_read_uhwi): Optimize single
6048         byte case, inline streamer_read_uchar and defer section
6049         overrun check.
6051 2013-06-20  Richard Biener  <rguenther@suse.de>
6053         PR tree-optimization/57584
6054         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
6055         SSA names into the expanded expression that take part in
6056         abnormal coalescing.
6058 2013-06-19  Sharad Singhai  <singhai@google.com>
6060         * gcov.c (print_usage): Handle new option.
6061         (process_args): Ditto.
6062         (get_gcov_intermediate_filename): New function.
6063         (output_intermediate_file): New function.
6064         (output_gcov_file): New function
6065         (generate_results): Handle new option.
6066         (release_function): Relase demangled name.
6067         (read_graph_file): Handle demangled name.
6068         (output_lines): Ditto.
6069         * doc/gcov.texi: Document gcov intermediate format.
6071 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
6073         PR bootstrap/57604
6074         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6075         (lra_emit_add): Use the functions.  Add comment about Y as an
6076         address segment.
6078 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
6080         PR driver/57652
6081         * collect2.c (collect_atexit): New.
6082         (collect_exit): Delete.
6083         (main): Register collect_atexit with atexit.
6084         (collect_wait): Change collect_exit to exit.
6085         (do_wait): Same.
6086         * collect2.h (collect_exit): Delete.
6087         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
6089 2013-06-19  Wei Mi  <wmi@google.com>
6091         PR rtl-optimization/57518
6092         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
6093         if regno is used in paradoxical subreg.
6094         (update_equiv_regs): Check pdx_subregs[regno] before
6095         set a reg to be equivalent with a mem.
6097 2013-06-19  Matthias Klose  <doko@ubuntu.com>
6099         PR driver/57651
6100         * file-find.h (find_a_file): Add a mode parameter.
6101         * file-find.c (find_a_file): Likewise.
6102         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
6103         with X_OK for the executables.
6104         * collect2.c (main): Call find_a_file with X_OK.
6106 2013-06-19  Steve Ellcey  <sellcey@mips.com>
6108         PR target/56942
6109         * config/mips/mips.md (casesi_internal_mips16_<mode>):
6110         Use NEXT_INSN instead of next_real_insn.
6112 2013-06-19  Jan Hubicka  <jh@suse.cz>
6114         * cgraph.h (const_value_known_p): Replace by ...
6115         (ctor_for_folding): .. this one.
6116         * cgraphunit.c (process_function_and_variable_attributes): Use it.
6117         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
6118         * expr.c (expand_expr_real_1): Likewise.
6119         (string_constant): Likewise.
6120         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
6121         * ipa.c (process_references): Likewise.
6122         (symtab_remove_unreachable_nodes): Likewise.
6123         * ipa-inline-analysis.c (param_change_prob): Likewise.
6124         * gimple-fold.c (canonicalize_constructor_val): Likewise.
6125         (get_base_constructor): Likwise.
6126         * varpool.c (varpool_remove_node): Likewise.
6127         (varpool_remove_initializer): LIkewise.
6128         (dump_varpool_node): LIkwise.
6129         (const_value_known_p): Rewrite to ...
6130         (ctor_for_folding): ... this one.
6132 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
6134         PR driver/57651
6135         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
6136         PERSONALITY in $PATH derived prefixes.
6138 2013-06-19  Jeff Law  <law@redhat.com>
6140         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
6141         in comment.
6143         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6144         (simplify_bitwise_binary): Use it to simpify certain binary ops on
6145         booleans.
6147 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
6149         * config/arm/vfp.md: Move VFP instruction classification documentation
6150         to ...
6151         * config/arm/arm.md: ... here.  Update instruction classification
6152         documentation.
6154 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
6156         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
6157         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
6158         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
6160 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
6162         PR target/57609
6163         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6164         with NEXT_INSN.  Use tablejump_p to check for jump table data
6165         insns.
6167 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
6169         PR c++/56544
6170         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
6171         that now in C++ the value is correct per the C++ standards.
6173 2013-06-19  Richard Biener  <rguenther@suse.de>
6175         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
6176         for global context.
6178 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6180         Revert:
6181         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6183         PR target/57609
6184         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6185         with next_active_insn.
6187 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
6189         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
6190         functions are inlined during failures to flag an error.
6191         * tree-inline.c (expand_call_inline): Allow the error to be flagged
6192         in early inline pass.
6194 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
6196         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
6197         in comments.
6199 2013-06-18  Julian Brown  <julian@codesourcery.com>
6201         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
6202         Permit virtual register pre-reload if !strict.
6203         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
6204         change.
6205         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
6206         prototype.
6207         * config/arm/neon.md (movmisalign<mode>): Use
6208         neon_perm_struct_or_reg_operand instead of
6209         neon_struct_or_register_operand.
6210         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
6211         neon_permissive_struct_operand instead of neon_struct_operand.
6212         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
6213         neon_vector_mem_operand.
6214         * config/arm/predicates.md (neon_struct_operand): Adjust call to
6215         neon_vector_mem_operand.
6216         (neon_permissive_struct_operand): New.
6217         (neon_struct_or_register_operand): Rename to...
6218         (neon_perm_struct_or_reg_operand): This. Adjust call to
6219         neon_vector_mem_operand.
6221 2013-06-18  Richard Biener  <rguenther@suse.de>
6223         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
6224         * lto-streamer.h: Include pointer-set.h.
6225         (struct lto_decl_slot): Remove.
6226         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
6227         Remove next_index entry.
6228         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6229         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6230         (lto_init_tree_ref_encoder): Adjust.
6231         (lto_destroy_tree_ref_encoder): Likewise.
6232         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
6233         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
6234         (lto_output_decl_index): Adjust.
6235         (lto_new_out_decl_state): Likewise.
6236         (lto_record_function_out_decl_state): Likewise.
6237         * lto-streamer-out.c (copy_function): Likewise.
6239 2013-06-18  Richard Biener  <rguenther@suse.de>
6241         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
6242         * cgraphunit.c: Include cfgloop.h.
6243         (init_lowered_empty_function): Initialize the loop tree.
6244         (assemble_thunk): Insert new BBs into loops.
6246 2013-06-18  Richard Biener  <rguenther@suse.de>
6248         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
6249         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
6250         the map from cache entry to cache index optional.
6251         (streamer_tree_cache_replace_tree): Adjust accordingly.
6252         (streamer_tree_cache_append): Likewise.
6253         (streamer_tree_cache_delete): Likewise.
6254         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
6255         streamer cache map from cache entry to cache index.
6256         * lto-streamer-out.c (create_output_block): Adjust.
6258 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
6260         * config/arm/arm.md (attribute "insn"): Move multiplication and
6261         division attributes to...
6262         (attribute "type"): ... here.  Remove mult.
6263         (attribute "mul32"): New attribute.
6264         (attribute "mul64"): Add umaal.
6265         (*arm_mulsi3): Update attributes.
6266         (*arm_mulsi3_v6): Likewise.
6267         (*thumb_mulsi3): Likewise.
6268         (*thumb_mulsi3_v6): Likewise.
6269         (*mulsi3_compare0): Likewise.
6270         (*mulsi3_compare0_v6): Likewise.
6271         (*mulsi_compare0_scratch): Likewise.
6272         (*mulsi_compare0_scratch_v6): Likewise.
6273         (*mulsi3addsi): Likewise.
6274         (*mulsi3addsi_v6): Likewise.
6275         (*mulsi3addsi_compare0): Likewise.
6276         (*mulsi3addsi_compare0_v6): Likewise.
6277         (*mulsi3addsi_compare0_scratch): Likewise.
6278         (*mulsi3addsi_compare0_scratch_v6): Likewise.
6279         (*mulsi3subsi): Likewise.
6280         (*mulsidi3adddi): Likewise.
6281         (*mulsi3addsi_v6): Likewise.
6282         (*mulsidi3adddi_v6): Likewise.
6283         (*mulsidi3_nov6): Likewise.
6284         (*mulsidi3_v6): Likewise.
6285         (*umulsidi3_nov6): Likewise.
6286         (*umulsidi3_v6): Likewise.
6287         (*umulsidi3adddi): Likewise.
6288         (*umulsidi3adddi_v6): Likewise.
6289         (*smulsi3_highpart_nov6): Likewise.
6290         (*smulsi3_highpart_v6): Likewise.
6291         (*umulsi3_highpart_nov6): Likewise.
6292         (*umulsi3_highpart_v6): Likewise.
6293         (mulhisi3): Likewise.
6294         (*mulhisi3tb): Likewise.
6295         (*mulhisi3bt): Likewise.
6296         (*mulhisi3tt): Likewise.
6297         (maddhisi4): Likewise.
6298         (*maddhisi4tb): Likewise.
6299         (*maddhisi4tt): Likewise.
6300         (maddhidi4): Likewise.
6301         (*maddhidi4tb): Likewise.
6302         (*maddhidi4tt): Likewise.
6303         (divsi3): Likewise.
6304         (udivsi3): Likewise.
6305         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
6306         (thumb2_mulsi_short_compare0): Likewise.
6307         (thumb2_mulsi_short_compare0_scratch): Likewise.
6308         * config/arm/arm1020e.md (1020mult1): Update attribute change.
6309         (1020mult2): Likewise.
6310         (1020mult3): Likewise.
6311         (1020mult4): Likewise.
6312         (1020mult5): Likewise.
6313         (1020mult6): Likewise.
6314         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
6315         change.
6316         (cortex_a15_mult64): Likewise.
6317         (cortex_a15_sdiv): Likewise.
6318         (cortex_a15_udiv): Likewise.
6319         * config/arm/arm1026ejs.md (mult1): Update attribute change.
6320         (mult2): Likewise.
6321         (mult3): Likewise.
6322         (mult4): Likewise.
6323         (mult5): Likewise.
6324         (mult6): Likewise.
6325         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
6326         (pj4_ir_div): Likewise.
6327         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
6328         (11_mult2): Likewise.
6329         (11_mult3): Likewise.
6330         (11_mult4): Likewise.
6331         (11_mult5): Likewise.
6332         (11_mult6): Likewise.
6333         (11_mult7): Likewise.
6334         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
6335         (cortex_a8_mla): Likewise.
6336         (cortex_a8_mull): Likewise.
6337         (cortex_a8_smulwy): Likewise.
6338         (cortex_a8_smlald): Likewise.
6339         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
6340         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
6341         (cortex_r4_mul_3): Likewise.
6342         (cortex_r4_mla_4): Likewise.
6343         (cortex_r4_mla_3): Likewise.
6344         (cortex_r4_smlald): Likewise.
6345         (cortex_r4_mull): Likewise.
6346         (cortex_r4_sdiv): Likewise.
6347         (cortex_r4_udiv): Likewise.
6348         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
6349         (cortex_a7_idiv): Likewise.
6350         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
6351         (9_mult2): Likewise.
6352         (9_mult3): Likewise.
6353         (9_mult4): Likewise.
6354         (9_mult5): Likewise.
6355         (9_mult6): Likewise.
6356         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
6357         (cortex_a53_sdiv): Likewise.
6358         (cortex_a53_udiv): Likewise.
6359         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
6360         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
6361         (mp626_mult2): Likewise.
6362         (mp626_mult3): Likewise.
6363         (mp626_mult4): Likewise.
6364         * config/arm/fa526.md (526_mult1): Update attribute change.
6365         (526_mult2): Likewise.
6366         * config/arm/arm-generic.md (mult): Update attribute change.
6367         (mult_ldsched_strongarm): Likewise.
6368         (mult_ldsched): Likewise.
6369         (multi_cycle): Likewise.
6370         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
6371         * config/arm/fa606te.md (606te_mult1): Update attribute change.
6372         (606te_mult2): Likewise.
6373         (606te_mult3): Likewise.
6374         (606te_mult4): Likewise.
6375         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
6376         (cortex_a9_mac16): Likewise.
6377         (cortex_a9_multiply): Likewise.
6378         (cortex_a9_mac): Likewise.
6379         (cortex_a9_multiply_long): Likewise.
6380         * config/arm/fa626te.md (626te_mult1): Update attribute change.
6381         (626te_mult2): Likewise.
6382         (626te_mult3): Likewise.
6383         (626te_mult4): Likewise.
6385 2013-06-18  Richard Biener  <rguenther@suse.de>
6387         PR lto/57334
6388         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6390 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6392         PR target/57609
6393         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
6394         with next_active_insn.
6396 2013-06-18  Alan Modra  <amodra@gmail.com>
6398         * config/rs6000/rs6000.h (enum data_align): New.
6399         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
6400         (DATA_ABI_ALIGNMENT): Define.
6401         (CONSTANT_ALIGNMENT): Correct comment.
6402         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
6403         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
6405 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
6407         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
6408         ATTRIBUTE_UNUSED marking.
6410 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
6412         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
6413         alternative and update.
6414         (aarch64_dup_lanedi): Delete.
6415         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
6416         * config/aarch64/aarch64-simd-builtins.def: Update.
6418 2013-06-17  Richard Biener  <rguenther@suse.de>
6420         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
6421         (lto_input_scc): Declare.
6422         (lto_input_tree_1): Likewise.
6423         (struct lto_stats_d): Add num_tree_bodies_output and
6424         num_pickle_refs_output.
6425         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
6426         (lto_read_tree_1): Split out from ...
6427         (lto_read_tree): ... this.
6428         (lto_input_scc): New function.
6429         (lto_input_tree_1): Split out from ...
6430         (lto_input_tree): ... this.  Handle LTO_tree_scc.
6431         (lto_data_in_create): Create the streamer cache without hashes.
6432         * lto-streamer-out.c (create_output_block): Create the streamer
6433         cache with hashes when not doing WPA.
6434         (lto_write_tree_1): Split out from ...
6435         (lto_write_tree): ... this.
6436         (get_symbol_initial_value): New function.
6437         (lto_output_tree_1): Split out from ...
6438         (lto_output_tree): ... this.  Write trees as series of SCCs
6439         using a DFS walk via DFS_write_tree.
6440         (struct sccs, struct scc_entry): New types.
6441         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
6442         (DFS_write_tree_body): New function.
6443         (DFS_write_tree): Likewise.
6444         (hash_tree): Likewise.
6445         (scc_entry_compare): Likewise.
6446         (hash_scc): Likewise.
6447         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
6448         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
6449         TREE_CHAIN as regular reference.
6450         (streamer_read_integer_cst): Remove.
6451         (streamer_get_pickled_tree): Adjust.
6452         * tree-streamer-out.c (streamer_write_chain): Disable streaming
6453         of DECL_EXTERNALs in BLOCK_VARS for now.
6454         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
6455         reference.
6456         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6457         Add hash value argument and record that if hashes are recorded
6458         in the cache.
6459         (streamer_tree_cache_insert_1): Adjust.
6460         (streamer_tree_cache_insert): Likewise.
6461         (streamer_tree_cache_insert_at): Rename to ...
6462         (streamer_tree_cache_replace_tree): ... this and adjust.
6463         (streamer_tree_cache_append): Adjust.
6464         (record_common_node): Likewise.
6465         (streamer_tree_cache_create): Add argument whether to
6466         record hash values together with trees.
6467         (streamer_tree_cache_delete): Adjust.
6468         * tree-streamer.h (struct streamer_tree_cache_d): Add
6469         vector of hashes.
6470         (streamer_read_integer_cst): Remove.
6471         (streamer_tree_cache_insert): Adjust.
6472         (streamer_tree_cache_append): Likewise.
6473         (streamer_tree_cache_insert_at): Rename to ...
6474         (streamer_tree_cache_replace_tree): ... this and adjust.
6475         (streamer_tree_cache_create): Add argument whether to record hashes.
6476         (streamer_tree_cache_get): Rename to ...
6477         (streamer_tree_cache_get_tree): ... this.
6478         (streamer_tree_cache_get_hash): New function.
6479         * tree.c (cache_integer_cst): New function.
6480         * tree.h (cache_integer_cst): Declare.
6481         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
6482         * lto-symtab.c (lto_varpool_replace_node): Only release
6483         DECL_INITIAL of non-prevailing decls.
6484         * varpool.c (varpool_remove_initializer): Do not release
6485         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
6487 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6489         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
6490         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
6491         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
6492         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
6493         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
6494         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
6495         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
6496         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
6497         instead of TARGET_64BIT.
6498         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
6499         Require ISA_HAS_<D>DIV.
6501 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
6503         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
6504         (mips*-*-linux*): Move default with_llsc setting to where other
6505         defaults are set.
6506         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
6507         with_arch block.
6508         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
6509         Likewise.  Remove default with_tune setting.  Move default float
6510         setting to its own block.  Handle with_llsc in the same block as above.
6512 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
6514         PR rtl-optimization/57425
6515         PR rtl-optimization/57569
6516         * alias.c (write_dependence_p): Add new parameters mem_mode,
6517         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
6518         Changed all callers.
6519         (canon_anti_dependence): New function.
6520         * cse.c (check_dependence): Use canon_anti_dependence.
6521         * cselib.c (cselib_invalidate_mem): Likewise.
6522         * rtl.h (canon_anti_dependence): Declare.
6524 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
6526         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
6527         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
6528         ".set mips3" for 64-bit targets.
6530 2013-06-15  Dehao Chen  <dehao@google.com>
6532         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
6533         * gimple-low.c (gimple_check_call_matching_types): Likewise.
6534         (gimple_check_call_args): Likewise.
6535         * value-prof.c (check_ic_target): Likewise.
6536         * ipa-inline.c (early_inliner): Likewise.
6537         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
6538         * cgraph.c (cgraph_create_edge_1): Likewise.
6539         (cgraph_make_edge_direct): Likewise.
6541 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
6543         PR target/57615
6544         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
6545         rs6000_output_move_128bit to handle emitting quad memory
6546         operations.  Set attribute length to 8 bytes.
6548 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
6550         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
6551         New pattern.
6552         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
6553         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
6554         (aarch64_<su>mlsl<mode>): Likewise.
6556 2013-06-14  Mike Stump  <mikestump@comcast.net>
6558         * Makefile.in (TARGET_H): Add insn-codes.h.
6560 2013-06-14  Alan Modra  <amodra@gmail.com>
6562         PR middle-end/57134
6563         PR middle-end/57586
6564         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
6565         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
6566         bitfield expansion when EXPAND_MEMORY.
6567         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
6569 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
6571         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
6572         test for clearing quad memory on 32-bit later.
6574 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6576         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
6577         (fold_negate_expr): Likewise.
6578         (fold_real_zero_addition_p): Handle vectors.
6579         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
6581 2013-06-14  Alan Modra  <amodra@gmail.com>
6583         * varasm.c (force_const_mem): Revert 2013-06-07 change.
6585 2013-06-13  Jan Hubicka  <jh@suse.cz>
6587         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
6588         Local comdats are not externally visible.
6589         * symtab.c (dump_symtab_base): Dump externally visible.
6590         (verify_symtab_base): Verify back links in the symtab hash.
6592 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6594         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
6595         CONVERT_EXPR as equal nodes.
6597 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
6599         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
6601 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
6603         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
6604         Generalize to complex and vector.
6605         * tree.c (build_all_ones_cst): New function.
6606         * tree.h (build_all_ones_cst): Declare it.
6608 2013-06-13  Alan Modra  <amodra@gmail.com>
6610         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
6611         * config/rs6000/rs6000.md (signbittf2): New insn.
6612         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
6613         (abstf2_internal, cmptf_internal2): Likewise.
6614         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
6616 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
6617             Pat Haugen  <pthaugen@us.ibm.com>
6618             Peter Bergner  <bergner@vnet.ibm.com>
6620         * config/rs6000/rs6000.c (emit_load_locked): Add support for
6621         power8 byte, half-word, and quad-word atomic instructions.
6622         (emit_store_conditional): Likewise.
6623         (rs6000_expand_atomic_compare_and_swap): Likewise.
6624         (rs6000_expand_atomic_op): Likewise.
6626         * config/rs6000/sync.md (larx): Add new modes for power8.
6627         (stcx): Likewise.
6628         (AINT): New mode iterator to include TImode as well as normal
6629         integer modes on power8.
6630         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
6631         that VSX registers are not considered.  Use AINT mode iterator
6632         instead of INT1 to allow inclusion of quad word atomic operations
6633         on power8.
6634         (load_locked<mode>): Likewise.
6635         (store_conditional<mode>): Likewise.
6636         (atomic_compare_and_swap<mode>): Likewise.
6637         (atomic_exchange<mode>): Likewise.
6638         (atomic_nand<mode>): Likewise.
6639         (atomic_fetch_<fetchop_name><mode>): Likewise.
6640         (atomic_nand_fetch<mode>): Likewise.
6641         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
6642         each type.
6643         (ATOMIC): On power8, add QImode, HImode modes.
6644         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
6645         modes that promote to SImode.
6646         (load_lockedti): Convert TImode arguments to PTImode, so that we
6647         get a guaranteed even/odd register pair.
6648         (load_lockedpti): Likewise.
6649         (store_conditionalti): Likewise.
6650         (store_conditionalpti): Likewise.
6652         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
6653         atomic load/store instructions.
6654         (HSI): Likewise.
6656 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
6658         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
6659         loads.
6660         (insn_count): New attribute, with most cases extracted from...
6661         (length): ...here.  Redefine most cases in terms of insn_count.
6662         (single_insn): Delete.
6663         (can_delay): Use insn_count to check for single instructions.
6664         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
6665         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
6666         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
6667         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
6668         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
6669         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
6670         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
6671         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
6672         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
6673         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
6674         rather than "length".
6675         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
6676         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
6677         Use "insn_count" rather than "length".
6678         * config/mips/mips-dsp.md
6679         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
6680         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
6681         length attributes.
6683 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
6685         PR tree-optimization/57361
6686         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
6688 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
6690         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
6691         to split.
6692         (aarch64_simd_combine<mode>): New instruction expansion.
6693         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
6694         function prototype.
6695         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
6696         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
6698 2013-06-12  Jan Hubicka  <jh@suse.cz>
6700         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
6701         decl has when in streaming stage.
6702         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
6703         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
6705 2013-06-12  Roland Stigge  <stigge@antcom.de>
6707         PR target/57578
6708         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
6710 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6712         PR tree-optimization/57537
6713         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
6714         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
6716 2013-06-12  Richard Biener  <rguenther@suse.de>
6718         * data-streamer.h (streamer_write_char_stream): CSE
6719         obs->current_pointer.
6720         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
6721         streamer_write_char_stream manually and optimize the resulting loop.
6722         (streamer_write_hwi_stream): Likewise.
6724 2013-06-12  Jan Hubicka  <jh@suse.cz>
6726         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
6727         * cgraph.h (varpool_create_empty_node): Declare.
6728         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
6729         duplicated nodes.
6730         * symtab.c (symtab_unregister_node): Be lax about missin entries
6731         in node hash.
6732         (symtab_get_node): Update comment.
6733         * varpool.c (varpool_create_empty_node): Break out from ...
6734         (varpool_node_for_decl): ... here.
6735         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
6737 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
6739         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
6740         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
6741         part.  Use straight-line flow at the end.
6742         <COMPONENT_REF>: Remove superfluous else.
6743         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
6745 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
6747         PR target/56564
6748         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
6749         target hook even for !TREE_PUBLIC decls.  If no resolution info
6750         is available, return false for common and external decls.
6752 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
6754         * config/rl78/constraints.md (U): New constraint.
6755         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
6756         valloc attribute.
6758 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
6760         PR target/57589
6761         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
6762         to allow returning address to AT_PLATFORM name.
6764 2013-06-11  Jan Hubicka  <jh@suse.cz>
6766         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
6767         * cgraph.h (symtab_node_base): Add weakref flag.
6768         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
6769         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
6770         (output_weakrefs): Use weakref flag.
6771         * fold-const.c (simple_operand_p): Handle WEAK.
6772         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
6773         * ipa.c (varpool_externally_visible_p): Drop weakref.
6774         (function_and_variable_visibility): Update comment; fix weakref
6775         sanity checks; do not clear DECL_WEAK on them.
6776         * lto-cgraph.c (lto_output_node): update.
6777         (lto_output_varpool_node): Update.
6778         (input_overwrite_node): Update.
6779         (input_node): Update.
6780         (input_varpool_node): Update.
6781         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
6782         (lto_symtab_merge_symbols): Add sanity check.
6783         (lto_symtab_prevailing_decl): Do not special case weakrefs.
6784         * passes.c (rest_of_decl_compilation): Set static flag, too.
6785         * symtab.c (dump_symtab_base): Dump weakref.
6786         (verify_symtab_base): Sanity check weakrefs.
6787         (symtab_make_decl_local): Remove duplicated code.
6788         (symtab_alias_ultimate_target): Simplify.
6789         * varpool.c (varpool_create_variable_alias): Set weakref flag.
6791 2013-06-11  DJ Delorie  <dj@redhat.com>
6793         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
6794         (rl78_unwind_word_mode): New.
6796 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6798         * final.c (debug_prefix_maps): Make static.
6800 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
6802         * function.c (initial_trampoline): Remove stray copy.
6804 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
6806         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
6808 2013-06-11  Martin Jambor  <mjambor@suse.cz>
6810         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
6811         within bounds at the beginning of the function.
6813 2013-06-11  Alan Modra  <amodra@gmail.com>
6815         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
6816         reporting.
6817         (get_named_section): Don't NULL !DECL_P decl.
6819 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
6821         * doc/invoke.texi (core-avx2): Document.
6822         (slm): Likewise.
6823         (atom): Updated with MOVBE.
6825 2013-06-11  Richard Biener  <rguenther@suse.de>
6827         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
6829 2013-06-11  Anton Blanchard  <anton@samba.org>
6831         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
6832         correct shift value in little-endian mode.
6834 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
6836         PR target/56564
6837         * varasm.c (get_variable_align): Move #endif to the right place.
6839 2013-06-10  Cary Coutant  <ccoutant@google.com>
6841         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
6842         for hash so that hash table traversal order is deterministic.
6844 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
6845             Pat Haugen  <pthaugen@us.ibm.com>
6846             Peter Bergner  <bergner@vnet.ibm.com>
6848         * config/rs6000/vector.md (GPR move splitter): Do not split moves
6849         of vectors in GPRS if they are direct moves or quad word load or
6850         store moves.
6852         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
6853         declaration.
6854         (direct_move_p): Likewise.
6855         (quad_load_store_p): Likewise.
6857         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
6858         classes into bins based on the physical register type.
6859         (reg_class_to_reg_type): Likewise.
6860         (IS_STD_REG_TYPE): Likewise.
6861         (IS_FP_VECT_REG_TYPE): Likewise.
6862         (reload_fpr_gpr): Arrays to determine what insn to use if we can
6863         use direct move instructions.
6864         (reload_gpr_vsx): Likewise.
6865         (reload_vsx_gpr): Likewise.
6866         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
6867         information that is a simplification of register classes.  Also
6868         precalculate direct move reload helpers.
6869         (direct_move_p): New function to return true if the operation can
6870         be done as a direct move instruciton.
6871         (quad_load_store_p): New function to return true if the operation
6872         is a quad memory operation.
6873         (rs6000_legitimize_address): If quad memory, only allow register
6874         indirect for TImode addresses.
6875         (rs6000_legitimate_address_p): Likewise.
6876         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
6877         (rs6000_reload_register_type): Likewise.
6878         (register_to_reg_type): Return register type.
6879         (rs6000_secondary_reload_simple_move): New helper function for
6880         secondary reload and secondary memory needed to identify anything
6881         that is a simple move, and does not need reloading.
6882         (rs6000_secondary_reload_direct_move): New helper function for
6883         secondary reload to identify cases that can be done with several
6884         instructions via the direct move instructions.
6885         (rs6000_secondary_reload_move): New helper function for secondary
6886         reload to identify moves between register types that can be done.
6887         (rs6000_secondary_reload): Add support for quad memory operations
6888         and for direct move.
6889         (rs6000_secondary_memory_needed): Likewise.
6890         (rs6000_debug_secondary_memory_needed): Change argument names.
6891         (rs6000_output_move_128bit): New function to return the move to
6892         use for 128-bit moves, including knowing about the various
6893         limitations of quad memory operations.
6895         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
6896         memory operations.  call rs6000_output_move_128bit for the actual
6897         instruciton(s) to generate.
6898         (vsx_movti_64bit): Likewise.
6900         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
6901         (UNSPEC_P8V_MTVSRWZ): Likewise.
6902         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
6903         (UNSPEC_P8V_MTVSRD): Likewise.
6904         (UNSPEC_P8V_XXPERMDI): Likewise.
6905         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
6906         (UNSPEC_FUSION_GPR): Likewise.
6907         (FMOVE128_GPR): New iterator for direct move.
6908         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
6909         (f32_sv): Likewise.
6910         (f32_dm): Likewise.
6911         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
6912         loads and direct move instructions.
6913         (zero_extendsidi2_lfiwzx): Likewise.
6914         (extendsidi2_lfiwax): Likewise.
6915         (extendsidi2_nocell): Likewise.
6916         (floatsi<mode>2_lfiwax): Likewise.
6917         (lfiwax): Likewise.
6918         (floatunssi<mode>2_lfiwzx): Likewise.
6919         (lfiwzx): Likewise.
6920         (fix_trunc<mode>_stfiwx): Likewise.
6921         (fixuns_trunc<mode>_stfiwx): Likewise.
6922         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
6923         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
6924         (parity<mode>2_cmpb): Set length/type attr.
6925         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
6926         for 'mr.' to fast_compare.
6927         (bpermd_<mode>): Change type attr to popcnt.
6928         (p8_fmrgow_<mode>): New insns for power8 direct move support.
6929         (p8_mtvsrwz_1): Likewise.
6930         (p8_mtvsrwz_2): Likewise.
6931         (reload_fpr_from_gpr<mode>): Likewise.
6932         (p8_mtvsrd_1): Likewise.
6933         (p8_mtvsrd_2): Likewise.
6934         (p8_xxpermdi_<mode>): Likewise.
6935         (reload_vsx_from_gpr<mode>): Likewise.
6936         (reload_vsx_from_gprsf): Likewise.
6937         (p8_mfvsrd_3_<mode>): LIkewise.
6938         (reload_gpr_from_vsx<mode>): Likewise.
6939         (reload_gpr_from_vsxsf): Likewise.
6940         (p8_mfvsrd_4_disf): Likewise.
6941         (multi-word GPR splits): Do not split direct moves or quad memory
6942         operations.
6944 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
6946         * tree-into-ssa.c (interesting_blocks): Make static.
6948 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
6950         PR target/56564
6951         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
6952         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
6953         Use DATA_ABI_ALIGNMENT for that case instead if defined.
6954         (get_variable_align): New function.
6955         (get_variable_section, emit_bss, emit_common,
6956         assemble_variable_contents, place_block_symbol): Use
6957         get_variable_align instead of DECL_ALIGN.
6958         (assemble_noswitch_variable): Add align argument, use it
6959         instead of DECL_ALIGN.
6960         (assemble_variable): Adjust caller.  Use get_variable_align
6961         instead of DECL_ALIGN.
6962         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
6963         caller.
6964         (DATA_ABI_ALIGNMENT): Define.
6965         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
6966         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
6967         opt is false, only return the psABI mandated alignment increase.
6968         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
6969         (DATA_ABI_ALIGNMENT): ... this.
6970         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
6971         (DATA_ABI_ALIGNMENT): ... this.
6972         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
6973         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
6974         (DATA_ABI_ALIGNMENT): ... this.
6975         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
6976         * doc/tm.texi: Regenerated.
6978 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
6980         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
6981         cmp_code to construct REG_EQUAL note.
6983 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
6985         PR target/57568
6986         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
6987         that operands[2] doesn't overlap with operands[0].
6989 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
6990             Jan Hubicka  <jh@suse.cz>
6992         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
6993         hack to mark symbols as used.
6995 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
6997         PR rtl-optimization/57559
6998         * lra-constraints.c (process_alt_operands): Don't discourage
6999         memory with known offset for offsetable memory constraint.
7000         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
7002 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
7004         * varasm.c (struct oc_local_state): Reorder fields.
7005         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
7006         and adjust accordingly.
7007         (output_constructor): Reorder initialization code and adjust call to
7008         output_constructor_bitfield.
7010 2013-06-07  Jan Hubicka  <jh@suse.cz>
7012         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
7014 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
7016         * tree-object-size.c (unknown): Make const.
7018 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
7020         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
7021         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
7022         for last alternative in the cpu_facility attribute.
7024 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7026         PR target/56315
7027         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
7028         (xordi3): Change operand 2 constraint to arm_xordi_operand.
7029         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
7030         * config/arm/constraints.md (Dg): New constraint.
7031         * config/arm/neon.md (xordi3_neon): Remove.
7032         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
7033         * config/arm/predicates.md (arm_xordi_operand): New predicate.
7035 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7037         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
7038         Clean up alternatives.
7040 2013-06-07  Alan Modra  <amodra@gmail.com>
7042         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
7043         va_list_gpr_size.
7045 2013-06-07  Alan Modra  <amodra@gmail.com>
7047         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
7049 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7051         * config/arm/constraints.md (Df): New constraint.
7052         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
7053         Correct length attribute for last two alternatives.
7055 2013-06-07  Alan Modra  <amodra@gmail.com>
7057         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
7058         override user -mfp-in-toc.
7059         (offsettable_ok_by_alignment): Consider just the current access
7060         rather than the whole object, unless BLKmode.  Handle
7061         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
7062         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
7063         for -mcmodel=medium.
7064         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
7065         override user -mfp-in-toc or -msum-in-toc.  Default to
7066         -mno-fp-in-toc for -mcmodel=medium.
7068 2013-06-06  DJ Delorie  <dj@redhat.com>
7070         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
7071         TARGET_VALID_POINTER_MODE.
7073 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
7074             Pat Haugen  <pthaugen@us.ibm.com>
7075             Peter Bergner  <bergner@vnet.ibm.com>
7077         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
7078         Document new power8 builtins.
7080         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
7081         condition code register, to allow 128-bit logical operations to be
7082         done in the VSX or GPR registers.
7083         (nor<mode>3): Use the canonical form for nor.
7084         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
7085         vclz*, and vpopcnt* vector instructions.
7086         (nand<mode>3): Likewise.
7087         (orc<mode>3): Likewise.
7088         (clz<mode>2): LIkewise.
7089         (popcount<mode>2): Likewise.
7091         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
7092         that only the GPRs are recognized.
7094         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7095         support for new power8 builtins.
7097         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
7098         builtin functions.
7099         (xscvdpspn): Likewise.
7100         (vclz): Likewise.
7101         (vclzb): Likewise.
7102         (vclzh): Likewise.
7103         (vclzw): Likewise.
7104         (vclzd): Likewise.
7105         (vpopcnt): Likewise.
7106         (vpopcntb): Likewise.
7107         (vpopcnth): Likewise.
7108         (vpopcntw): Likewise.
7109         (vpopcntd): Likewise.
7110         (vgbbd): Likewise.
7111         (vmrgew): Likewise.
7112         (vmrgow): Likewise.
7113         (eqv): Likewise.
7114         (eqv_v16qi3): Likewise.
7115         (eqv_v8hi3): Likewise.
7116         (eqv_v4si3): Likewise.
7117         (eqv_v2di3): Likewise.
7118         (eqv_v4sf3): Likewise.
7119         (eqv_v2df3): Likewise.
7120         (nand): Likewise.
7121         (nand_v16qi3): Likewise.
7122         (nand_v8hi3): Likewise.
7123         (nand_v4si3): Likewise.
7124         (nand_v2di3): Likewise.
7125         (nand_v4sf3): Likewise.
7126         (nand_v2df3): Likewise.
7127         (orc): Likewise.
7128         (orc_v16qi3): Likewise.
7129         (orc_v8hi3): Likewise.
7130         (orc_v4si3): Likewise.
7131         (orc_v2di3): Likewise.
7132         (orc_v4sf3): Likewise.
7133         (orc_v2df3): Likewise.
7135         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
7136         allow power8 quad mode in 64-bit.
7137         (rs6000_builtin_vectorized_function): Add support to vectorize
7138         ISA 2.07 count leading zeros, population count builtins.
7139         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
7140         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
7141         (builtin_function_type): Add vgbbd builtin function which takes an
7142         unsigned argument.
7143         (altivec_expand_vec_perm_const): Add support for new power8 merge
7144         instructions.
7146         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
7147         that does not include TImdoe for use with 32-bit.
7148         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
7149         instructions.
7150         (UNSPEC_VSX_CVDPSPN): Likewise.
7151         (vsx_xscvdpspn): Likewise.
7152         (vsx_xscvspdpn): Likewise.
7153         (vsx_xscvdpspn_scalar): Likewise.
7154         (vsx_xscvspdpn_directmove): Likewise.
7155         (vsx_and<mode>3): Split logical operations into 32-bit and
7156         64-bit. Add support to do logical operations on TImode as well as
7157         VSX vector types.  Allow logical operations to be done in either
7158         VSX registers or in general purpose registers in 64-bit mode.  Add
7159         splitters if GPRs were used. For AND, add clobber of CCmode to
7160         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
7161         encoding.
7162         (vsx_and<mode>3_32bit): Likewise.
7163         (vsx_and<mode>3_64bit): Likewise.
7164         (vsx_ior<mode>3): Likewise.
7165         (vsx_ior<mode>3_32bit): Likewise.
7166         (vsx_ior<mode>3_64bit): Likewise.
7167         (vsx_xor<mode>3): Likewise.
7168         (vsx_xor<mode>3_32bit): Likewise.
7169         (vsx_xor<mode>3_64bit): Likewise.
7170         (vsx_one_cmpl<mode>2): Likewise.
7171         (vsx_one_cmpl<mode>2_32bit): Likewise.
7172         (vsx_one_cmpl<mode>2_64bit): Likewise.
7173         (vsx_nor<mode>3): Likewise.
7174         (vsx_nor<mode>3_32bit): Likewise.
7175         (vsx_nor<mode>3_64bit): Likewise.
7176         (vsx_andc<mode>3): Likewise.
7177         (vsx_andc<mode>3_32bit): Likewise.
7178         (vsx_andc<mode>3_64bit): Likewise.
7179         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
7180         and xxlorc instructions.
7181         (vsx_eqv<mode>3_64bit): Likewise.
7182         (vsx_nand<mode>3_32bit): Likewise.
7183         (vsx_nand<mode>3_64bit): Likewise.
7184         (vsx_orc<mode>3_32bit): Likewise.
7185         (vsx_orc<mode>3_64bit): Likewise.
7187         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
7189         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
7190         instruction.
7191         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
7192         (p8_vmrgow): Likewise.
7193         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
7194         GPRs to be split under VSX.
7195         (p8v_clz<mode>2): Add power8 count leading zero support.
7196         (p8v_popcount<mode>2): Add power8 population count support.
7197         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
7198         support.
7200         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7201         instruction.
7203         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
7204         builtin functions.
7205         (vec_nand): Likewise.
7206         (vec_vclz): Likewise.
7207         (vec_vclzb): Likewise.
7208         (vec_vclzd): Likewise.
7209         (vec_vclzh): Likewise.
7210         (vec_vclzw): Likewise.
7211         (vec_vgbbd): Likewise.
7212         (vec_vmrgew): Likewise.
7213         (vec_vmrgow): Likewise.
7214         (vec_vpopcnt): Likewise.
7215         (vec_vpopcntb): Likewise.
7216         (vec_vpopcntd): Likewise.
7217         (vec_vpopcnth): Likewise.
7218         (vec_vpopcntw): Likewise.
7220 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7222         PR rtl-optimization/57468
7223         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
7224         spilled pseudos.
7226 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7228         PR rtl-optimization/57459
7229         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
7230         type when setting live regs.
7232 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
7234         * config/s390/s390.opt (mlra): New option.
7235         * config/s390/s390.c (s390_decompose_address): Check displacement
7236         for all registers for LRA.
7237         (s390_secondary_reload): Don't used secondary reloads for LRA.
7238         (s390_lra_p): New function.
7239         (TARGET_LRA_P): Define.
7240         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
7241         of attribute cpu_facility to zarch for the last alternative.
7242         (*cmpmem_short): Ditto.
7244 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
7246         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
7247         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
7248         (arm_expand_prologue): Likewise.
7250 2013-06-06  Teresa Johnson  <tejohnson@google.com>
7252         PR c++/53743
7253         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
7254         as this is now done by redirect_edge_and_branch_force.
7255         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
7256         barriers, and fix interaction with splitting.
7257         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
7258         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
7259         reflect changes made in the routine.
7260         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
7261         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
7262         since this is called in cfglayout mode, and replace partition fixup
7263         with assert as that is now done by force_nonfallthru_and_redirect.
7264         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
7265         already be marked with region crossing note.
7266         (insert_section_boundary_note): Make non-static, gate on flag
7267         has_bb_partition, rewrite to also check for multiple partitions.
7268         (rest_of_handle_reorder_blocks): Remove call to
7269         insert_section_boundary_note, now done later during free_cfg.
7270         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
7271         * bb-reorder.h (insert_section_boundary_note): Declare.
7272         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
7273         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
7274         invoke insert_section_boundary_note.
7275         (try_redirect_by_replacing_jump): Remove unnecessary
7276         check for region crossing note.
7277         (fixup_partition_crossing): New function.
7278         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
7279         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
7280         in non-cfglayout mode.
7281         (force_nonfallthru_and_redirect): Fixup partition boundaries,
7282         remove old code that tried to do this. Emit barrier correctly
7283         when we are in cfglayout mode.
7284         (last_bb_in_partition): New function.
7285         (rtl_split_edge): Correctly fixup partition boundaries.
7286         (commit_one_edge_insertion): Remove old code that tried to
7287         fixup region crossing edge since this is now handled in
7288         split_block, and set up insertion point correctly since
7289         block may now end in a jump.
7290         (verify_hot_cold_block_grouping): Guard against checking when not in
7291         linearized RTL mode.
7292         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
7293         notes.
7294         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
7295         rtl_verify_flow_info, so not called in cfglayout mode.
7296         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
7297         (fixup_reorder_chain): Remove old code that attempted to fixup region
7298         crossing note as this is now handled in force_nonfallthru_and_redirect.
7299         (duplicate_insn_chain): Don't duplicate switch section notes.
7300         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
7301         note.
7302         * basic-block.h (emit_barrier_after_bb): Declare.
7304 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7306         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
7307         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
7308         arm_usatsihi): Adjust alternatives for arm_restrict_it.
7310 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7312         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
7313         where appropriate.
7314         * config/arm/ldmstm.md: Regenerate.
7316 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7318         * config/arm/sync.md (atomic_loaddi_1):
7319         Disable predication for arm_restrict_it.
7320         (arm_load_exclusive<mode>): Likewise.
7321         (arm_load_exclusivesi): Likewise.
7322         (arm_load_exclusivedi): Likewise.
7323         (arm_load_acquire_exclusive<mode>): Likewise.
7324         (arm_load_acquire_exclusivesi): Likewise.
7325         (arm_load_acquire_exclusivedi): Likewise.
7326         (arm_store_exclusive<mode>): Likewise.
7327         (arm_store_exclusive<mode>): Likewise.
7328         (arm_store_release_exclusivedi): Likewise.
7329         (arm_store_release_exclusive<mode>): Likewise.
7331 2013-06-06  Richard Biener  <rguenther@suse.de>
7333         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
7334         after LTO_null.
7335         (lto_tag_is_tree_code_p): Adjust.
7336         (lto_tag_is_gimple_code_p): Likewise.
7337         (lto_gimple_code_to_tag): Likewise.
7338         (lto_tag_to_gimple_code): Likewise.
7339         (lto_tree_code_to_tag): Likewise.
7340         (lto_tag_to_tree_code): Likewise.
7341         * data-streamer.h (streamer_write_hwi_in_range): Use
7342         uhwi streaming to stream the normalized range.
7343         (streamer_read_hwi_in_range): Likewise.
7345 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7347         * config/arm/arm.md (enabled_for_depr_it): New attribute.
7348         (predicable_short_it): Likewise.
7349         (predicated): Likewise.
7350         (enabled): Handle above.
7351         (define_cond_exec): Set predicated attribute to yes.
7353 2013-06-05  Mike Stump  <mikestump@comcast.net>
7355         * gdbinit.in (__FUNCTION__): Add.
7357 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
7359         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
7360         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
7362 2013-06-05  Jan Hubicka  <jh@suse.cz>
7364         * varasm.c (mark_decl_referenced): Revert the removal until targets
7365         are fixed.
7367 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
7369         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
7370         instead of mark_decl_referenced.
7372 2013-06-05  Jan Hubicka  <jh@suse.cz>
7374         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
7375         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
7376         and symtab_used_from_object_file_p.
7377         (cgraph_make_node_local_1): Clear forced_by_abi.
7378         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
7379         * cgraph.h (symtab_node_base): Add forced_by_abi.
7380         (decide_is_variable_needed): Remove.
7381         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
7382         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
7383         (decide_is_symbol_needed): ... this one; handle symbols in general;
7384         always analyze virtuals; honnor forced_by_abi.
7385         (cgraph_finalize_function): Update.
7386         (varpool_finalize_decl): Update.
7387         (symbol_defined_and_needed): Remove.
7388         (analyze_functions): Update.
7389         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
7390         output_refs, input_overwrite_node): Handle forced_by_abi.
7391         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
7392         (address_taken_from_non_vtable_p): ... this one.
7393         (comdat_can_be_unshared_p_1): New function.
7394         (cgraph_comdat_can_be_unshared_p): Rename to ...
7395         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
7396         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
7397         (function_and_variable_visibility): Clear forced_by_abi as needed.
7398         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
7399         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
7400         * symtab.c (dump_symtab_base): Dump forced_by_abi.
7401         * varpool.c (decide_is_variable_needed): Remove.
7403 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7405         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
7406         (arm_option_override): Override arm_restrict_it where appropriate.
7407         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
7408         * config/arm/arm.opt (mrestrict-it): New command-line option.
7409         * doc/invoke.texi: Document -mrestrict-it.
7411 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7413         * tsan.c (tsan_atomic_table): Make const.
7415 2013-06-05  Richard Biener  <rguenther@suse.de>
7417         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
7418         index associated with the tree we are supposed to replace.
7419         * tree-streamer-out.c (pack_ts_base_value_fields): Output
7420         TREE_ASM_WRITTEN as zero for everything but SSA names.
7422 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
7424         * tree-ssa-structalias.c (call_stmt_vars): Make static.
7426 2013-06-04  Jan Hubicka  <jh@suse.cz>
7428         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
7429         (input_node, input_varpool_node): Handle correctly external same
7430         body aliases.
7431         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
7432         nodes at ltrans stage.
7434 2013-06-04  Jan Hubicka  <jh@suse.cz>
7436         * ipa-inline.c (update_caller_keys): Fix availability test.
7437         (update_callee_keys): Likewise.
7438         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
7439         to follow ELF standard.
7441 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
7443         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
7444         (mips64r5900el-*-elf*): New configurations.
7445         * config/mips/mips-cpus.def (r5900): New processor.
7446         * config/mips/mips-tables.opt: Regenerate.
7447         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
7448         (mips_issue_rate): Handle PROCESSOR_R5900.
7449         (mips_reorg_process_insns): Force reorder mode for the R5900.
7450         * config/mips/mips.h (TARGET_MIPS5900): Define.
7451         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
7452         TARGET_MIPS5900.
7453         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
7454         TARGET_MIPS5900.
7455         * config/mips/mips.md (processor): Add r5900.
7456         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
7458 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7460         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
7461         into function to generate MOVI instruction.
7462         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
7463         (aarch64_preferred_simd_mode): Turn into wrapper.
7464         (aarch64_output_scalar_simd_mov_immediate): New function.
7465         * config/aarch64/aarch64-protos.h: Add prototype for above.
7467 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7469         * config/aarch64/aarch64.c (simd_immediate_info): Remove
7470         element_char member.
7471         (sizetochar): Return signed char.
7472         (aarch64_simd_valid_immediate): Remove elchar and other
7473         unnecessary variables.
7474         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
7475         Calculate element_char as required.
7476         * config/aarch64/aarch64-protos.h: Update and move prototype
7477         for aarch64_output_simd_mov_immediate.
7478         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7479         Update arguments.
7481 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7483         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
7484         information completed by aarch64_simd_valid_immediate.
7485         (aarch64_legitimate_constant_p): Update arguments.
7486         (aarch64_simd_valid_immediate): Work with struct rather than many
7487         pointers.
7488         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
7489         (aarch64_simd_make_constant): Update arguments.
7490         (aarch64_output_simd_mov_immediate): Work with struct rather than
7491         many pointers.  Output immediate directly rather than as operand.
7492         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7493         Update prototype.
7494         * config/aarch64/constraints.md (Dn): Update arguments.
7496 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7498         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
7499         longer static.
7500         (aarch64_simd_immediate_valid_for_move): Remove.
7501         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
7502         (aarch64_simd_make_constant): Update call.
7503         (aarch64_output_simd_mov_immediate): Update call.
7504         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
7505         Add prototype.
7506         * config/aarch64/constraints.md (Dn): Update call.
7508 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
7510         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
7511         return type to bool for prototype.
7512         (aarch64_legitimate_constant_p): Check for true instead of not -1.
7513         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
7514         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
7516 2013-06-04  Catherine Moore  <clm@codesourcery.com>
7518         * config/mips/mips.opt (meva): New.
7519         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
7520         (ASM_SPEC): Handle -meva.
7521         * doc/invoke.texi (meva):  Document.
7523 2013-06-04  Alan Modra  <amodra@gmail.com>
7525         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
7526         constant output.
7528 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7530         * rtl.def: Add extra fourth optional field to define_cond_exec.
7531         * gensupport.c (process_one_cond_exec): Process attributes from
7532         define_cond_exec.
7533         * doc/md.texi: Document fourth field in define_cond_exec.
7535 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
7537         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
7538         out the processing order as in store_bit_field_1.
7540 2013-06-04  Jan Hubicka  <jh@suse.cz>
7542         PR middle-end/57500
7543         * cgraphunit.c (cgraph_process_same_body_aliases): Create
7544         non-VAR_DECL node if it does not exist yet.
7546 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
7548         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
7549         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
7550         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
7551         target_cpu_default setting.
7553 2013-06-03  Teresa Johnson  <tejohnson@google.com>
7555         * dumpfile.c (opt_info_switch_p): Change -fopt-info
7556         default to -fopt-info=optimized instead of all.
7557         * doc/invoke.texi: Ditto.
7558         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
7559         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
7560         (execute_vect_slp): Emit BB vectorization success under
7561         MSG_OPTIMIZED_LOCATIONS.
7562         * tree-vect-slp.c (vect_slp_transform_bb): Change
7563         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
7564         * tree-vect-loop.c (vect_transform_loop): Ditto.
7566 2013-06-03  Jason Merrill  <jason@redhat.com>
7568         PR c++/57415
7569         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7570         Use TARGET_EXPR for C++.
7572 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
7574         PR rtl-optimization/57268
7575         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
7576         if DEBUG_INSN_P (insn).
7578         Reapply
7579         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7581         PR rtl-optimization/57268
7582         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7583         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7585 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7587         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
7588         (ix86_avoid_lea_for_addr): Likewise.
7589         (exact_dependency_1): Likewise.
7590         (ix86_adjust_cost): Likewise.
7591         (swap_top_of_ready_list): Fix formatting and !reload_completed check
7592         removed.
7593         (do_reorder_for_imul): Fix typo, formatting and
7594         !reload_completed check removed.
7595         (ix86_sched_reorder): Fix typo and formatting.
7596         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
7597         list.
7599 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
7601         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
7603 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
7605         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
7606         <STRING_CST>: Likewise.
7607         <VECTOR_CST>: Likewise.
7609 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
7610             Mikael Morin  <mikael@gcc.gnu.org>
7612         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
7613         * config.in: Regenerated.
7614         * configure: Regenerated.
7616 2013-06-01  Jan Hubicka  <jh@suse.cz>
7618         PR middle-end/57366
7619         * cgraphunit.c (compile): When weakref is not supported,
7620         set up transparent aliases before final output pass.
7621         * varasm.c (assemble_alias): Do not try to do it here.
7623 2013-06-01  Jan Hubicka  <jh@suse.cz>
7625         PR middle-end/57467
7626         * passes.c (for_per_function): Skip unanalyzed functions.
7628 2013-06-01  Jan Hubicka  <jh@suse.cz>
7630         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
7631         (lto_symtab_merge_symbols_1): ... this one.
7632         (lto_symtab_merge_cgraph_nodes): Rename to ...
7633         (lto_symtab_merge_symbols): ... this one; simplify.
7634         * cgraph.c (same_body_aliases_done): Rename to ...
7635         (cpp_implicit_aliases_done): ... this one.
7636         (cgraph_create_function_alias): Update.
7637         (cgraph_same_body_alias): Update.
7638         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
7639         (verify_edge_corresponds_to_fndecl): Simplify.
7640         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
7641         (cgraph_node): Remove same_body_alias.
7642         (varpool_node): Remove alias_of and extra_name_alias.
7643         (same_body_aliases_done): Rename to ..
7644         (cpp_implicit_aliases_done): ... this one.
7645         (symtab_alias_ultimate_target): Add default parameter.
7646         (symtab_resolve_alias): New function.
7647         (fixup_same_cpp_alias_visibility): Declare.
7648         (cgraph_function_node): Add default parameter.
7649         (cgraph_node_asm_name): Likewise.
7650         (cgraph_function_or_thunk_node): Add default parameter; do
7651         not ICE when it is NULL.
7652         (varpool_variable_node): Likewise.
7653         * tree-emutls.c (create_emultls_var): Update.
7654         (ipa_lower_emutls): Update.
7655         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
7656         (cgraph_reset_node): Reset alias info.
7657         (cgraph_finalize_function): Update.
7658         (fixup_same_cpp_alias_visibility): Move to symtab.c.
7659         (analyze_function): Simplify.
7660         (cgraph_process_same_body_aliases): Simplify.
7661         (analyze_functions): Fixup same body aliases.
7662         (handle_alias_pairs): Simplify.
7663         (assemble_thunk): Update.
7664         (assemble_thunks_and_aliases): Update.
7665         (output_weakrefs): Rewrite.
7666         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
7667         (lto_output_varpool_node): Likewise.
7668         (compute_ltrans_boundary): Remve assert.
7669         (get_alias_symbol): New functoin.
7670         (input_node): Rewrite alias handling.
7671         (input_varpool_node): Likewise.
7672         * ipa-pure-const.c (propagate_pure_const): Fix formating.
7673         * ipa.c (process_references): Handle weakrefs correctly.
7674         (symtab_remove_unreachable_nodes): Likewise.
7675         * trans-mem.c (get_cg_data): Update.
7676         (ipa_tm_create_version_alias): Update.
7677         (ipa_tm_execute): Update.
7678         * symtab.c (dump_symtab_base): Dump aliases.
7679         (verify_symtab_base): Verify aliases.
7680         (symtab_node_availability): New function.
7681         (symtab_alias_ultimate_target): Simplify.
7682         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
7683         handle all the fixup cases.
7684         (symtab_resolve_alias): New function.
7685         * passes.c (ipa_write_summaries): Handle weakrefs.
7686         * varpool.c (varpool_analyze_node): Simplify.
7687         (assemble_aliases): Update.
7688         (varpool_create_variable_alias): Simplify.
7689         (varpool_extra_name_alias): Simplify.
7690         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
7691         (lto_symtab_merge_symbols): ... this one.
7693 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
7695         Revert
7696         PR rtl-optimization/57268
7697         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7698         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7700 2013-06-01  Tobias Burnus  <burnus@net-b.de>
7702         Partially reverted:
7703         2013-05-31  Tobias Burnus  <burnus@net-b.de>
7705         PR middle-end/57073
7706         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
7707         further up.
7709 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
7711         PR rtl-optimization/57268
7712         * sched-deps.c (sched_analyze_2): Flush dependence lists if
7713         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
7715 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7717         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
7718         unordered comparison operators when -fno-trapping-math is in effect
7719         on the e500.
7720         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
7721         and implement unordered comparison operators properly on the e500.
7723 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
7725         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
7726         for constant scalar integers.
7727         (simplify_relational_operation_1): Likewise.
7729 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
7731         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
7732         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
7733         Fix comment.
7735 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7736             Igor Zamyatin  <igor.zamyatin@intel.com>
7738         Silvermont (SLM) architecture performance tuning.
7739         * config/i386/i386.h (enum ix86_tune_indices): Add
7740         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
7741         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
7743         * config/i386/i386.c (initial_ix86_tune_features)
7744         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
7745         (ix86_lea_outperforms): Handle Silvermont tuning.
7746         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
7747         call.
7748         (ix86_use_lea_for_mov): Likewise.
7749         (ix86_avoid_lea_for_addr): Likewise.
7750         (ix86_lea_for_add_ok): Likewise.
7751         (exact_dependency_1): New function.
7752         (exact_store_load_dependency): Likewise.
7753         (ix86_adjust_cost): Handle Silvermont tuning.
7754         (do_reoder_for_imul): Likewise.
7755         (swap_top_of_ready_list): New function.
7756         (ix86_sched_reorder): Changed to handle Silvermont tuning.
7758         * config/i386/i386.md (peepholes that split memory operand in fp
7759         converts): New.
7761 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7763         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
7764         Remove un-necessary braces.
7766 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
7768         * config/aarch64/aarch64.c (aarch64_classify_symbol):
7769         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
7771 2013-05-31  Tobias Burnus  <burnus@net-b.de>
7773         PR middle-end/57073
7774         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
7776 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7778         PR target/56315
7779         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
7780         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
7781         * config/arm/neon.md (iordi3_neon): Remove.
7782         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
7783         * config/arm/predicates.md (imm_for_neon_logic_operand):
7784         Move to earlier in the file.
7785         (neon_logic_op2): Likewise.
7786         (arm_iordi_operand_neon): New predicate.
7788 2013-05-31  Richard Biener  <rguenther@suse.de>
7790         PR tree-optimization/57478
7791         PR tree-optimization/57453
7792         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
7793         are life as well.
7795 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
7797         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
7798         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
7800 2013-05-30  Tobias Burnus  <burnus@net-b.de>
7801             Thomas Koenig  <tkoenig@gcc.gnu.org>
7803         PR middle-end/57073
7804         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
7805         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
7807 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
7809         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
7811 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
7813         * target.def (register_usage_leveling_p): New hook.
7814         * targhooks.c (default_register_usage_leveling_p): New.
7815         * targhooks.h (default_register_usage_leveling_p): New prototype.
7816         * lra-assigns.c (register_usage_leveling_p): Use the hook.
7817         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
7818         * doc/tm.texi: Update.
7819         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
7821 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
7823         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
7824         (*insv_reg<mode>): New define_insn.
7826 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7828         PR rtl-optimization/57439
7829         * postreload.c (move2add_valid_value_p): Check that we have
7830         a zero subreg_regno_offset when accessing the register in
7831         the requested mode.
7833 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
7834             Igor Zamyatin  <igor.zamyatin@intel.com>
7836         Silvermont (SLM) architecture pipeline model, tuning and
7837         insn selection.
7838         * config.gcc: Add slm config options and target.
7840         * config/i386/slm.md: New.
7842         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
7844         * gcc/config/i386/i386-c.c (ix86_target_macros_internal):  New case
7845         PROCESSOR_SLM.
7846         (ix86_target_macros_internal): Likewise.
7848         * gcc/config/i386/i386.c (slm_cost): New cost.
7849         (m_SLM): New macro flag.
7850         (initial_ix86_tune_features): Set m_SLM.
7851         (x86_accumulate_outgoing_args): Likewise.
7852         (x86_arch_always_fancy_math_387): Likewise.
7853         (processor_target_table): Add slm cost.
7854         (cpu_names): Add slm cpu name.
7855         (x86_option_override_internal): Set SLM ISA.
7856         (ix86_issue_rate): New case PROCESSOR_SLM.
7857         (ia32_multipass_dfa_lookahead): Likewise.
7858         (fold_builtin_cpu): Add slm.
7860         * config/i386/i386.h (TARGET_SLM): New target macro.
7861         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
7862         (processor_type): Add PROCESSOR_SLM.
7864         * config/i386/i386.md (cpu): Add new value "slm".
7865         (slm.md): Include slm.md.
7867 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
7868             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7870         * config/arm/arm-protos.h: Add and update function protos.
7871         * config/arm/arm.c (use_simple_return_p): New added.
7872         (thumb2_expand_return): Check simple_return flag.
7873         * config/arm/arm.md: Add simple_return and conditional simple_return.
7874         * config/arm/iterators.md: Add iterator for return and simple_return.
7876 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
7878         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
7879         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
7880         (arm_emit_vfp_multi_reg_pop): Likewise.
7881         (thumb2_emit_ldrd_pop): Likewise.
7882         (arm_expand_epilogue): Add misc REG_CFA notes.
7883         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
7885 2013-05-29  Lawrence Crowl  <crowl@google.com>
7887         * config/arm/t-arm: Update for below.
7889         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
7890         Change type to hash_table.  Update dependent calls and types.
7892         * config/i386/t-cygming: Update for below.
7894         * config/i386/t-interix: Update for below.
7896         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
7897         Change type to hash_table.  Update dependent calls and types.
7898         (i386_find_on_wrapper_list::wrappers): Likewise.
7900         * config/ia64/t-ia64: Update for below.
7902         * config/ia64/ia64.c (bundle_state_table):
7903         Change type to hash_table.  Update dependent calls and types.
7905         * config/mips/mips.c (mips_reorg_process_insns::htab):
7906         Change type to hash_table.  Update dependent calls and types.
7908         * config/sol2.c (solaris_comdat_htab):
7909         Change type to hash_table.  Update dependent calls and types.
7911         * config/t-sol2: Update for above.
7913 2013-05-29  Teresa Johnson  <tejohnson@google.com>
7915         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
7916         functions are not yet marked as defined.
7918 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
7919             Pat Haugen  <pthaugen@us.ibm.com>
7920             Peter Bergner  <bergner@vnet.ibm.com>
7922         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
7923         instructions.
7924         (VEC_A): Likewise.
7925         (VEC_C): Likewise.
7926         (vrotl<mode>3): Likewise.
7927         (vashl<mode>3): Likewise.
7928         (vlshr<mode>3): Likewise.
7929         (vashr<mode>3): Likewise.
7931         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7932         support for power8 V2DI builtins.
7934         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
7935         power8 V2DI builtins.
7936         (vupkhsw): Likewise.
7937         (vupklsw): Likewise.
7938         (vaddudm): Likewise.
7939         (vminsd): Likewise.
7940         (vmaxsd): Likewise.
7941         (vminud): Likewise.
7942         (vmaxud): Likewise.
7943         (vpkudum): Likewise.
7944         (vpksdss): Likewise.
7945         (vpkudus): Likewise.
7946         (vpksdus): Likewise.
7947         (vrld): Likewise.
7948         (vsld): Likewise.
7949         (vsrd): Likewise.
7950         (vsrad): Likewise.
7951         (vsubudm): Likewise.
7952         (vcmpequd): Likewise.
7953         (vcmpgtsd): Likewise.
7954         (vcmpgtud): Likewise.
7955         (vcmpequd_p): Likewise.
7956         (vcmpgtsd_p): Likewise.
7957         (vcmpgtud_p): Likewise.
7958         (vupkhsw): Likewise.
7959         (vupklsw): Likewise.
7960         (vaddudm): Likewise.
7961         (vmaxsd): Likewise.
7962         (vmaxud): Likewise.
7963         (vminsd): Likewise.
7964         (vminud): Likewise.
7965         (vpksdss): Likewise.
7966         (vpksdus): Likewise.
7967         (vpkudum): Likewise.
7968         (vpkudus): Likewise.
7969         (vrld): Likewise.
7970         (vsld): Likewise.
7971         (vsrad): Likewise.
7972         (vsrd): Likewise.
7973         (vsubudm): Likewise.
7975         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
7976         support for power8 V2DI instructions.
7978         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
7979         power8 V2DI instructions.  Combine pack and unpack insns to use an
7980         iterator for each mode.  Check whether a particular mode supports
7981         Altivec instructions instead of just checking TARGET_ALTIVEC.
7982         (UNSPEC_VPKUWUM): Likewise.
7983         (UNSPEC_VPKSHSS): Likewise.
7984         (UNSPEC_VPKSWSS): Likewise.
7985         (UNSPEC_VPKUHUS): Likewise.
7986         (UNSPEC_VPKSHUS): Likewise.
7987         (UNSPEC_VPKUWUS): Likewise.
7988         (UNSPEC_VPKSWUS): Likewise.
7989         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
7990         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
7991         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
7992         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
7993         (UNSPEC_VUPKHSB): Likewise.
7994         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
7995         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
7996         (UNSPEC_VUPKHSH): Likewise.
7997         (UNSPEC_VUPKLSB): Likewise.
7998         (UNSPEC_VUPKLSH): Likewise.
7999         (VI2): Likewise.
8000         (VI_char): Likewise.
8001         (VI_scalar): Likewise.
8002         (VI_unit): Likewise.
8003         (VP): Likewise.
8004         (VP_small): Likewise.
8005         (VP_small_lc): Likewise.
8006         (VU_char): Likewise.
8007         (add<mode>3): Likewise.
8008         (altivec_vaddcuw): Likewise.
8009         (altivec_vaddu<VI_char>s): Likewise.
8010         (altivec_vadds<VI_char>s): Likewise.
8011         (sub<mode>3): Likewise.
8012         (altivec_vsubcuw): Likewise.
8013         (altivec_vsubu<VI_char>s): Likewise.
8014         (altivec_vsubs<VI_char>s): Likewise.
8015         (altivec_vavgs<VI_char>): Likewise.
8016         (altivec_vcmpbfp): Likewise.
8017         (altivec_eq<mode>): Likewise.
8018         (altivec_gt<mode>): Likewise.
8019         (altivec_gtu<mode>): Likewise.
8020         (umax<mode>3): Likewise.
8021         (smax<mode>3): Likewise.
8022         (umin<mode>3): Likewise.
8023         (smin<mode>3): Likewise.
8024         (altivec_vpkuhum): Likewise.
8025         (altivec_vpkuwum): Likewise.
8026         (altivec_vpkshss): Likewise.
8027         (altivec_vpkswss): Likewise.
8028         (altivec_vpkuhus): Likewise.
8029         (altivec_vpkshus): Likewise.
8030         (altivec_vpkuwus): Likewise.
8031         (altivec_vpkswus): Likewise.
8032         (altivec_vpks<VI_char>ss): Likewise.
8033         (altivec_vpks<VI_char>us): Likewise.
8034         (altivec_vpku<VI_char>us): Likewise.
8035         (altivec_vpku<VI_char>um): Likewise.
8036         (altivec_vrl<VI_char>): Likewise.
8037         (altivec_vsl<VI_char>): Likewise.
8038         (altivec_vsr<VI_char>): Likewise.
8039         (altivec_vsra<VI_char>): Likewise.
8040         (altivec_vsldoi_<mode>): Likewise.
8041         (altivec_vupkhsb): Likewise.
8042         (altivec_vupkhs<VU_char>): Likewise.
8043         (altivec_vupkls<VU_char>): Likewise.
8044         (altivec_vupkhsh): Likewise.
8045         (altivec_vupklsb): Likewise.
8046         (altivec_vupklsh): Likewise.
8047         (altivec_vcmpequ<VI_char>_p): Likewise.
8048         (altivec_vcmpgts<VI_char>_p): Likewise.
8049         (altivec_vcmpgtu<VI_char>_p): Likewise.
8050         (abs<mode>2): Likewise.
8051         (vec_unpacks_hi_v16qi): Likewise.
8052         (vec_unpacks_hi_v8hi): Likewise.
8053         (vec_unpacks_lo_v16qi): Likewise.
8054         (vec_unpacks_hi_<VP_small_lc>): Likewise.
8055         (vec_unpacks_lo_v8hi): Likewise.
8056         (vec_unpacks_lo_<VP_small_lc>): Likewise.
8057         (vec_pack_trunc_v8h): Likewise.
8058         (vec_pack_trunc_v4si): Likewise.
8059         (vec_pack_trunc_<mode>): Likewise.
8061         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
8062         V2DI builtins.
8063         (vec_vmaxsd): Likewise.
8064         (vec_vmaxud): Likewise.
8065         (vec_vminsd): Likewise.
8066         (vec_vminud): Likewise.
8067         (vec_vpksdss): Likewise.
8068         (vec_vpksdus): Likewise.
8069         (vec_vpkudum): Likewise.
8070         (vec_vpkudus): Likewise.
8071         (vec_vrld): Likewise.
8072         (vec_vsld): Likewise.
8073         (vec_vsrad): Likewise.
8074         (vec_vsrd): Likewise.
8075         (vec_vsubudm): Likewise.
8076         (vec_vupkhsw): Likewise.
8077         (vec_vupklsw): Likewise.
8079 2013-05-29  Jan Hubicka  <jh@suse.cz>
8081         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
8082         flags; reorder rest of fields in more consistent way.
8083         (varpool_node): Remove analyzed, finalized and alias.
8084         (cgraph_ndoe): Likewise.
8085         (symtab_alias_ultimate_target): New function.
8086         (cgraph_function_node): Move offline.
8087         (cgraph_reset_node): Declare.
8088         (cgraph_comdat_can_be_unshared_p): Remove.
8089         (varpool_remove_initializer): Declare.
8090         (varpool_first_defined_variable, varpool_next_defined_variable
8091         cgraph_first_defined_function, cgraph_next_defined_function): Update.
8092         (cgraph_function_with_gimple_body_p): Update.
8093         (varpool_all_refs_explicit_p): Update.
8094         (symtab_alias_target): New function.
8095         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
8096         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
8097         (cgraph_function_or_thunk_node): Simplify using
8098         symtab_alias_ultimate_target.
8099         (varpool_variable_node): Likewise.
8100         * cgraph.c (cgraph_create_function_alias): Update.
8101         (cgraph_add_thunk): Update.
8102         (cgraph_remove_node): Update.
8103         (dump_cgraph_node): Do not dump removed flags.
8104         (cgraph_function_body_availability): Update.
8105         (cgraph_propagate_frequency): Update.
8106         (verify_cgraph_node): Check sanity of local flag.
8107         (cgraph_function_node): Move here from cgraph.h; revamp for
8108         cgraph_function_or_thunk_node.
8109         * lto-symtab.c (lto_varpool_replace_node): Update.
8110         (lto_symtab_resolve_can_prevail_p): Update.
8111         (lto_symtab_merge_cgraph_nodes): Update.
8112         * ipa-cp.c (determine_versionability, initialize_node_lattices,
8113         propagate_constants_accross_call, devirtualization_time_bonus,
8114         ipcp_propagate_stage): Update.
8115         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
8116         * ipa-inline-transform.c (clone_inlined_nodes,
8117         preserve_function_body_p): Update.
8118         * ipa-reference.c (propagate): Update.
8119         (write_node_summary_p): Update.
8120         * toplev.c (wrapup_global_declaration_2): Update.
8121         * cgraphunit.c (cgraph_analyze_function): Rename to ...
8122         (analyze_function) ... this one.
8123         (cgraph_process_new_functions): Update.
8124         (cgraph_reset_node): Export.
8125         (cgraph_finalize_function): Update.
8126         (cgraph_add_new_function): Update.
8127         (process_function_and_variable_attributes): Update.
8128         (varpool_finalize_decl): Update.
8129         (symbol_finalized): Remove.
8130         (symbol_finalized_and_needed): Rename to ...
8131         (symbol_defined_and_needed): ... update.
8132         (cgraph_analyze_functions): Update.
8133         (handle_alias_pairs): Update.
8134         (mark_functions_to_output): Update.
8135         (assemble_thunk): Update.
8136         (output_in_order): Update.
8137         (output_weakrefs): Update.
8138         (finalize_compilation_unit): Update.
8139         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
8140         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
8141         input_node, input_varpool_node): Update.
8142         * dbxout.c (dbxout_expand_expr): Update.
8143         * cgraphclones.c (cgraph_clone_node): Update.
8144         (cgraph_copy_node_for_versioning): Update.
8145         (cgraph_materialize_clone): Update.
8146         (cgraph_materialize_all_clones): Update.
8147         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
8148         propagate_pure_const, propagate_nothrow): Update.
8149         * lto-streamer-out.c (lto_output, write_symbol): Update.
8150         * ipa-utils.c (ipa_reverse_postorder): Update.
8151         * ipa-inline.c (can_inline_edge_p): Update.
8152         (update_caller_keys, ipa_inline): Update.
8153         * dwarf2out.c (reference_to_unused,
8154         premark_types_used_by_global_vars_helper): Update.
8155         * tree-eh.c (tree_could_trap_p): Update.
8156         * ipa-split.c (consider_split, execute_split_functions): Update.
8157         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
8158         has_addr_references_p): Update; move ahead in file for better
8159         readability.
8160         (process_references): Simplify.
8161         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
8162         bodies are removed.
8163         (cgraph_comdat_can_be_unshared_p): Make static.
8164         (cgraph_externally_visible_p): Update.
8165         (varpool_externally_visible_p): Update.
8166         (function_and_variable_visibility): Update.
8167         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
8168         ipa_tm_mark_force_output_node): Update.
8169         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
8170         estimate_edge_devirt_benefit, inline_generate_summary,
8171         inline_write_summary): Update.
8172         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
8173         * ipa-prop.c (ipa_compute_jump_functions): Update.
8174         (ipa_print_node_params, ipa_prop_read_section,
8175         ipa_update_after_lto_read, read_replacements_section): Update.
8176         * varasm.c (mark_decl_referenced): Update.
8177         (assemble_alias, dump_tm_clone_pairs): Update.
8178         * tree-inline.c (copy_bb): Update.
8179         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
8180         Update.
8181         * symtab.c (dump_symtab_base): Print new flags.
8182         (verify_symtab_base): Verify new flags.
8183         (symtab_alias_ultimate_target): New function.
8184         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
8185         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
8186         Update.
8187         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
8188         Update.
8189         * i386.c (ix86_get_function_versions_dispatcher,
8190         ix86_generate_version_dispatcher_body): Update.
8191         (fold_builtin_cpu): Use varpool_add_new_variable.
8192         * varpool.c (varpool_remove_initializer): Break out from ...
8193         (varpool_remove_node): ... this one.
8194         (dump_varpool_node, varpool_node_for_asm,
8195         cgraph_variable_initializer_availability, varpool_analyze_node,
8196         varpool_assemble_decl, varpool_remove_unreferenced_decls,
8197         varpool_finalize_named_section_flags, varpool_create_variable_alias):
8198         Update.
8200 2013-05-29  Jan Hubicka  <jh@suse.cz>
8202         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
8204 2013-05-29  Easwaran Raman  <eraman@google.com>
8206         PR tree-optimization/57442
8207         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
8208         when control exits the main loop.
8210 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
8212         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
8213         and RX600.
8214         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
8215         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
8216         * rx/t-rx: Add rx100 under multi library matches option for nofpu
8217         option.
8219 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8221         PR tree-optimization/57441
8222         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
8223         Don't limit size of incr_vec to number of candidates.
8225 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
8227         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
8228         and mips16 directories.
8229         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
8230         (MULTILIB_DIRNAMES): Ditto.
8231         (MULTILIB_EXCEPTIONS): Add new exceptions.
8232         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
8233         (MULTILIB_DIRNAMES): Ditto.
8234         (MULTILIB_EXCEPTIONS): Add new exceptions.
8236 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8237             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8239         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
8240         SYMBOL_TINY_ABSOLUTE.
8241         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
8242         SYMBOL_TINY_ABSOLUTE.
8243         (aarch64_expand_mov_immediate): Likewise.
8244         (aarch64_classify_symbol): Likewise.
8245         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
8246         Permit SYMBOL_TINY_ABSOLUTE.
8247         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
8249 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8250             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8252         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
8253         Refactor if/switch.  Replace gcc_assert with if.
8255 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
8257         * config/i386/i386.c (initial_ix86_tune_features): Enable
8258         FP Reassociation for AMD bdver1 and bdver2.
8260 2013-05-29  Martin Jambor  <mjambor@suse.cz>
8262         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
8263         and IMAGPART_EXPR do not occur within other handled_components.
8265 2013-05-29  Richard Biener  <rguenther@suse.de>
8267         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
8268         access on whether the use is in the BB we currently try to
8269         vectorize.
8270         (vect_bb_vectorization_profitable_p): Pass the BB we currently
8271         vectorize to vect_bb_slp_scalar_cost.
8273 2013-05-29  Richard Biener  <rguenther@suse.de>
8275         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
8276         computing scalar cost offsetted by stmts that are kept live
8277         by scalar uses.
8278         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
8279         for computation of scalar cost.
8281 2013-05-28  Steve Ellcey  <sellcey@mips.com>
8283         * config/mips/mips-cpus.def (mips32r2): Change processor type.
8285 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
8287         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
8288         array notation built-in reduction functions.
8289         * doc/passes.texi (Passes): Added documentation about changes done
8290         for Cilk Plus.
8291         * doc/invoke.texi (C Dialect Options): Added documentation about
8292         the -fcilkplus flag.
8293         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
8294         (BUILTINS_DEF): Depend on cilkplus.def.
8295         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
8296         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
8297         * cilkplus.def: New file.
8299 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
8301         PR rtl-optimization/57439
8302         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
8304 2013-05-28  Easwaran Raman  <eraman@google.com>
8306         PR tree-optimization/57337
8307         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
8308         (find_insert_point): Correctly identify the insertion point
8309         when two statements with the same UID is compared.
8311 2013-05-28  Richard Biener  <rguenther@suse.de>
8313         PR tree-optimization/56787
8314         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
8315         from the list of data references.
8316         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
8317         clobbers.
8318         (vect_analyze_loop_operations): Likewise.
8319         (vect_transform_loop): Remove clobbers.
8321 2013-05-28  Martin Jambor  <mjambor@suse.cz>
8323         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
8324         and REALPART_EXPRs have scalar type.
8326 2013-05-28  Richard Biener  <rguenther@suse.de>
8328         PR tree-optimization/57411
8329         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
8330         virtual operands.
8331         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
8332         virtual operand propagation.
8334 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8336         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
8337         destination register for bmasksi_vis.
8338         (vector_init_bshuffle): Likewise.
8339         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
8341 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
8343         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
8344         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
8345         mode if the instruction isn't available in the original mode.
8346         * config/sparc/sparc.opt (mfix-ut699): New option.
8347         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
8348         (divdf3): Turn into expander.
8349         (divdf3_nofix): New insn.
8350         (divdf3_fix): Likewise.
8351         (divsf3): Disable if -mfix-ut699.
8352         (sqrtdf2): Turn into expander.
8353         (sqrtdf2_nofix): New insn.
8354         (sqrtdf2_fix): Likewise.
8355         (sqrtsf2): Disable if -mfix-ut699.
8357 2013-05-27  Richard Biener  <rguenther@suse.de>
8359         PR middle-end/57412
8360         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
8361         block for the new loop.
8363 2013-05-27  Richard Biener  <rguenther@suse.de>
8365         PR tree-optimization/57343
8366         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
8367         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
8368         (number_of_iterations_cond): Do not build the folded tree.
8370 2013-05-27  Richard Biener  <rguenther@suse.de>
8372         Revert
8373         PR middle-end/57381
8374         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8375         OEP_CONSTANT_ADDRESS_OF retained.
8377         PR tree-optimization/57417
8378         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
8379         for unchanged base.
8380         (set_ssa_val_to): Compare addresses using
8381         get_addr_base_and_unit_offset.
8383 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
8385         PR rtl-optimization/56833
8386         * postreload.c (move2add_record_mode): New function.
8387         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
8388         (move2add_use_add2_insn): Use move2add_record_sym_value.
8389         (move2add_use_add3_insn): Likewise.
8390         (reload_cse_move2add): Use move2add_valid_value_p and
8391         move2add_record_mode.  Invalidate call-clobbered and REG_INC
8392         affected regs by setting reg_mode to VOIDmode.
8393         (move2add_note_store): Don't pretend the inside of a SUBREG is
8394         the actual destination.  Invalidate single/leading registers by
8395         setting reg_mode to VOIDmode.
8396         Use move2add_record_sym_value, move2add_valid_value_p and
8397         move2add_record_mode.
8399 2013-05-27  Richard Biener  <rguenther@suse.de>
8401         PR tree-optimization/57396
8402         * tree-affine.c (double_int_constant_multiple_p): Properly
8403         return false for val == 0 and div != 0.
8405 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8407         * config/mips/mips.h: Use #elif in preprocessor conditions.
8409 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8411         PR target/53916
8412         * config/mips/constraints.md (kl): New constraint.
8413         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
8414         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
8415         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
8416         from LO for MIPS16.
8417         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8419 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
8421         PR target/55777
8422         * config/mips/mips.c (mips_can_inline_p): New function.
8423         (TARGET_CAN_INLINE_P): Define.
8425 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
8427         * sched-int.h (ds_t, dw_t): Make unsigned int.
8428         Fix documentation that describes how all the ds_t bits are used.
8429         Reserve the last bit for delayed-branch scheduling.
8430         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
8431         (BITS_PER_DEP_WEAK): Fix definition and documentation.
8432         (gen_dep_weak_1): Remove prototype.
8433         * sched-deps.c (get_dep_weak_1): Make static.
8434         * target.def (speculate_insn, needs_block_p, gen_spec_check,
8435         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
8436         * doc/tm.texi: Regenerate.
8437         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
8439 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
8441         PR debug/56950
8442         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
8444 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
8445             Sandra Loosemore  <sandra@codesourcery.com>
8447         * config.gcc (powerpc-*): Allow native for with-cpu.
8449 2013-05-24  Jeff Law  <law@redhat.com>
8451         PR tree-optimization/57124
8452         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
8453         conversion feeding a condition if the range has an overflow
8454         if -fstrict-overflow.  Add warnings for when we do make the
8455         transformation.
8457 2013-05-24  Dehao Chen  <dehao@google.com>
8459         * tree-cfg.c (locus_discrim_map): Fix the typo.
8460         (locus_discrim_hasher): Likewise.
8461         (locus_discrim_hasher::hash): Likewise.
8462         (locus_discrim_hasher::equal): Likewise.
8464 2013-05-24  Martin Jambor  <mjambor@suse.cz>
8466         PR tree-optimization/57294
8467         * cgraph.h (ipa_record_stmt_references): Declare.
8468         * cgraphbuild.c (ipa_record_stmt_references): New function.
8469         (build_cgraph_edges): Use ipa_record_stmt_references.
8470         (rebuild_cgraph_edges): Likewise.
8471         (cgraph_rebuild_references): Likewise.
8472         * ipa-prop.c (ipa_modify_call_arguments): Discard references
8473         associated with the old statement and build references from the
8474         newly built statements.
8475         * ipa-ref.c (ipa_remove_stmt_references): New function.
8476         * ipa-ref.h (ipa_remove_stmt_references): Declare.
8478 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
8480         * lra-constraints.c (emit_spill_move): Use smaller mode for
8481         mem-mem moves.
8482         (check_and_process_move): Consider mem-reg moves for secondary
8483         too.
8484         (curr_insn_transform): Don't lose insns emitted before for
8485         secondary memory moves.
8486         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
8487         reg set up in the current insn.
8489 2013-05-24  Dehao Chen  <dehao@google.com>
8491         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
8492         hash function.
8493         (locus_descrim_hasher::equal): Likewise.
8494         (build_gimple_cfg): New discriminator assignment algorithm.
8495         (make_edges): Likewise.
8496         (next_discriminator_for_locus): Likewise.
8497         (same_line_p): Likewise.
8498         (assign_discriminators): Likewise.
8499         (make_cond_expr_edges): Likewise.
8500         (make_gimple_switch_edges): Likewise.
8501         (make_goto_expr_edges): Likewise.
8502         (make_gimple_asm_edges): Likewise.
8504 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
8506         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
8507         X format specifier to only display bottom 16 bits.
8508         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
8509         immediate to match for operand 2, since it will be masked.
8511 2013-05-24  Richard Biener  <rguenther@suse.de>
8513         PR tree-optimization/57287
8514         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
8515         all SSA names that occur in abnormal PHIs.
8517 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8519         PR tree-ssa/57385
8520         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
8521         that index is not negative.
8523 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
8525         PR rtl-optimization/55177
8526         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
8527         (simplify_byte_swapping_operation): New.
8528         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
8529         (simplify_relational_operation_1): Deal with BSWAP.
8531 2013-05-23  Richard Henderson  <rth@redhat.com>
8533         PR target/56742
8534         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
8535         (ix86_reorg): Call it.
8537 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
8539         PR target/57379
8540         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
8541         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
8542         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
8544 2013-05-23  Christian Bruel  <christian.bruel@st.com>
8546         PR debug/57351
8547         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
8549 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8550             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8552         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
8553         * config/aarch64/constraints.md (Usa): Remove.
8554         * doc/md.texi (AArch64 Usa): Remove.
8556 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
8557             Marcus Shawcroft  <marcus.shawcroft@arm.com>
8559         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
8560         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
8561         * config/aarch64/predicates.md (aarch64_const_address): Remove.
8562         (aarch64_mov_operand): Use aarch64_mov_operand_p.
8564 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
8566         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
8567         instruction (AdvSIMD).
8568         * config/aarch64/aarch64-builtins.c
8569         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
8570         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
8572 2013-05-23  Martin Jambor  <mjambor@suse.cz>
8574         PR middle-end/57347
8575         * tree.h (contains_bitfld_component_ref_p): Declare.
8576         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
8577         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
8578         caller.
8579         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
8580         not access a bit-field.  Assert all final offsets are byte-aligned.
8582 2013-05-23  Richard Biener  <rguenther@suse.de>
8584         PR tree-optimization/57380
8585         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
8586         least one invariant or re-used load.
8587         * passes.c (init_optimization_passes): Move pass_phiprop before
8588         pass_forwprop.
8590 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
8592         * config/aarch64/aarch64-simd.md
8593         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
8595 2013-05-23  Richard Biener  <rguenther@suse.de>
8597         PR middle-end/57381
8598         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
8599         OEP_CONSTANT_ADDRESS_OF retained.
8601 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
8603         PR middle-end/57344
8604         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
8605         don't lower unit.  Handle unit not being always BITS_PER_WORD.
8607 2013-05-23  Richard Biener  <rguenther@suse.de>
8609         PR rtl-optimization/57341
8610         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
8611         instead of true_dependence.
8613 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
8615         * bb-reorder.c (branch_threshold): Make const.
8616         (exec_threshold): Ditto.
8618 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8619             Pat Haugen  <pthaugen@us.ibm.com>
8620             Peter Bergner  <bergner@vnet.ibm.com>
8622         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
8623         documentation for the power8 crypto builtins.
8625         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
8627         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
8628         macros for defining power8 builtin functions.
8629         (BU_P8V_AV_2): Likewise.
8630         (BU_P8V_AV_P): Likewise.
8631         (BU_P8V_VSX_1): Likewise.
8632         (BU_P8V_OVERLOAD_1): Likewise.
8633         (BU_P8V_OVERLOAD_2): Likewise.
8634         (BU_CRYPTO_1): Likewise.
8635         (BU_CRYPTO_2): Likewise.
8636         (BU_CRYPTO_3): Likewise.
8637         (BU_CRYPTO_OVERLOAD_1): Likewise.
8638         (BU_CRYPTO_OVERLOAD_2): Likewise.
8639         (XSCVSPDP): Fix typo, point to the correct instruction.
8640         (VCIPHER): Add power8 crypto builtins.
8641         (VCIPHERLAST): Likewise.
8642         (VNCIPHER): Likewise.
8643         (VNCIPHERLAST): Likewise.
8644         (VPMSUMB): Likewise.
8645         (VPMSUMH): Likewise.
8646         (VPMSUMW): Likewise.
8647         (VPERMXOR_V2DI): Likewise.
8648         (VPERMXOR_V4SI: Likewise.
8649         (VPERMXOR_V8HI: Likewise.
8650         (VPERMXOR_V16QI: Likewise.
8651         (VSHASIGMAW): Likewise.
8652         (VSHASIGMAD): Likewise.
8653         (VPMSUM): Likewise.
8654         (VPERMXOR): Likewise.
8655         (VSHASIGMA): Likewise.
8657         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
8658         __CRYPTO__ if the crypto instructions are available.
8659         (altivec_overloaded_builtins): Add support for overloaded power8
8660         builtins.
8662         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
8663         support for power8 crypto builtins.
8664         (builtin_function_type): Likewise.
8665         (altivec_init_builtins): Add support for builtins that take vector
8666         long long (V2DI) arguments.
8668         * config/rs6000/crypto.md: New file, define power8 crypto
8669         instructions.
8671 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
8672             Pat Haugen  <pthaugen@us.ibm.com>
8673             Peter Bergner  <bergner@vnet.ibm.com>
8675         * doc/invoke.texi (Option Summary): Add power8 options.
8676         (RS/6000 and PowerPC Options): Likewise.
8678         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
8679         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
8680         wm, wn, wr documentation.
8682         * gcc/config/rs6000/constraints.md (wm): New constraint for VSX
8683         registers if direct move instructions are enabled.
8684         (wn): New constraint for no registers.
8685         (wq): New constraint for quad word even GPR registers.
8686         (wr): New constraint if 64-bit instructions are enabled.
8687         (wv): New constraint if power8 vector instructions are enabled.
8688         (wQ): New constraint for quad word memory locations.
8690         * gcc/config/rs6000/predicates.md (const_0_to_15_operand): New
8691         constraint for 0..15 for crypto instructions.
8692         (gpc_reg_operand): If VSX allow registers in VSX registers as well
8693         as GPR and floating point registers.
8694         (int_reg_operand): New predicate to match only GPR registers.
8695         (base_reg_operand): New predicate to match base registers.
8696         (quad_int_reg_operand): New predicate to match even GPR registers
8697         for quad memory operations.
8698         (vsx_reg_or_cint_operand): New predicate to allow vector logical
8699         operations in both GPR and VSX registers.
8700         (quad_memory_operand): New predicate for quad memory operations.
8701         (reg_or_indexed_operand): New predicate for direct move support.
8703         * gcc/config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
8704         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
8705         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
8706         (POWERPC_MASKS): Add power8 options.
8707         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
8708         various options.
8710         * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
8711         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
8713         * gcc/config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
8714         (-mpower8-fusion): New power8 options.
8715         (-mpower8-fusion-sign): Likewise.
8716         (-mpower8-vector): Likewise.
8717         (-mcrypto): Likewise.
8718         (-mdirect-move): Likewise.
8719         (-mquad-memory): Likewise.
8721         * gcc/config/rs6000/rs6000.c (power8_cost): Initial definition for
8722         power8.
8723         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
8724         registers.
8725         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
8726         (rs6000_debug_vector_unit): Add p8_vector.
8727         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
8728         definitions.  Also print fusion state.
8729         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
8730         (rs6000_builtin_mask_calculate): Add power8 builtin support.
8731         (rs6000_option_override_internal): Add support for power8.
8732         (rs6000_common_init_builtins): Add debugging for skipped builtins
8733         if -mdebug=builtin.
8734         (rs6000_adjust_cost): Add power8 support.
8735         (rs6000_issue_rate): Likewise.
8736         (insn_must_be_first_in_group): Likewise.
8737         (insn_must_be_last_in_group): Likewise.
8738         (force_new_group): Likewise.
8739         (rs6000_register_move_cost): Likewise.
8740         (rs6000_opt_masks): Likewise.
8742         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
8743         power8 capable assembler, default to power7 options.
8744         (TARGET_DIRECT_MOVE): Likewise.
8745         (TARGET_CRYPTO): Likewise.
8746         (TARGET_P8_VECTOR): Likewise.
8747         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
8748         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
8749         (VECTOR_MEM_P8_VECTOR_P): Likewise.
8750         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
8751         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
8752         (TARGET_XSCVDPSPN): Likewise.
8753         (TARGET_XSCVSPDPN): Likewsie.
8754         (TARGET_SYNC_HI_QI): Likewise.
8755         (TARGET_SYNC_TI): Likewise.
8756         (MASK_CRYPTO): Likewise.
8757         (MASK_DIRECT_MOVE): Likewise.
8758         (MASK_P8_FUSION): Likewise.
8759         (MASK_P8_VECTOR): Likewise.
8760         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
8761         temporary used by some of the direct move instructions to get two FP
8762         temporary registers does not force creation of a stack frame.
8763         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
8764         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
8765         that any VSX registers are tieable, even if they are also an
8766         Altivec vector mode.
8767         (r6000_reg_class_enum): Add wm, wr, wv constraints.
8768         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
8769         (RS6000_BTM_CRYPTO): Likewise.
8770         (RS6000_BTM_COMMON): Likewise.
8772         * config/rs6000/rs6000.md (cpu attribute): Add power8.
8773         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
8774         (enum rs6000_vector): Add power8 vector support.
8776 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
8778         PR target/19599
8779         PR target/57340
8780         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
8781         (any_sibcall_could_use_r3): this and handle indirect calls.
8782         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
8784 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
8786         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
8788 2013-05-22  Richard Biener  <rguenther@suse.de>
8790         PR middle-end/57349
8791         * profile.c (branch_prob): Do not split blocks that are
8792         abnormally receiving from ECF_RETURNS_TWICE functions.
8794 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
8796         * recog.c (offsettable_address_addr_space_p): Fix calculation of
8797         address mode.  Move pointer mode initialization to the same place.
8799 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8801         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
8802         while it has any effect.
8804 2013-05-21  Easwaran Raman  <eraman@google.com>
8806         PR tree-optimization/57322
8807         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
8808         UID of the statement added to the BB to be 1.
8810 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
8812         PR tree-optimization/57331
8813         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
8814         of conversion from pointer type to integral type with integer.
8816 2013-05-21  Martin Jambor  <mjambor@suse.cz>
8818         PR lto/57289
8819         * ipa-prop.c (ipa_read_node_info): Process param_used and
8820         controlled_uses in the same order as when writing.
8822 2013-05-21  Magnus Granberg  <baldrick@free.fr>
8824         PR plugins/56754
8825         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
8827 2013-05-21  Richard Biener  <rguenther@suse.de>
8829         PR tree-optimization/57318
8830         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
8831         estimate stmts with side-effects as likely eliminated.
8833 2013-05-21  Richard Biener  <rguenther@suse.de>
8835         PR tree-optimization/57330
8836         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
8837         preserve the call stmts fntype.
8839 2013-05-21  Richard Biener  <rguenther@suse.de>
8841         PR tree-optimization/57303
8842         * tree-ssa-sink.c (statement_sink_location): Improve killing
8843         stmt detection and properly handle self-assignments.
8845 2013-05-21  Christian Bruel  <christian.bruel@st.com>
8847         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
8848         spanning registers. LEAF_REG_REMAP is supported only for contiguous
8849         registers. Set register size out of the PARALLEL loop.
8851 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
8853         PR target/56547
8854         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
8855         (*fmasf4, *fmasf4_media): New insns.
8857 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8859         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
8860         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
8861         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
8862         (mips_idiv_insns): Update the comments to say that the returned
8863         instruction counts are in units of BASE_INSN_LENGTH.
8864         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
8865         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
8866         using 2 rather than 4 as the length of indirect MIPS16 and
8867         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
8868         length of a NOP.  Don't divide MIPS16 lengths by 2.
8869         (mips16_split_long_branches): Assume a branch is long if the
8870         length is greater than 4 rather than 8.
8871         * config/mips/mips.md (length): Give MIPS16 lengths directly,
8872         rather than multiplying them by 2.  Multiply instruction counts
8873         by BASE_INSN_LENGTH rather than 4.
8874         (*jump_mips16, tls_get_tp_mips16_<mode>)
8875         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
8877 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8879         * config/mips/mips.md (extended_mips16): Remove branch case.
8880         (length): Remove duplicated extended_mips16 test.
8882 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
8884         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
8886 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
8888         * recog.h (Recog_data): Rename to...
8889         (recog_data_d): ...this.
8890         (recog_data): Update accordingly.
8891         * recog.c (recog_data): Likewise.
8892         * reload.c (save_recog_data): Likewise.
8893         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
8894         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
8896 2013-05-17  Julian Brown  <julian@codesourcery.com>
8898         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
8899         found in a REG_EQUAL note, invalidate it.
8901 2013-05-17  Easwaran Raman  <eraman@google.com>
8903         * tree-ssa-reassoc.c (find_insert_point): New function.
8904         (insert_stmt_after): Likewise.
8905         (get_def_stmt): Likewise.
8906         (ensure_ops_are_available): Likewise.
8907         (not_dominated_by): Likewise.
8908         (rewrite_expr_tree): Do not move statements beyond what is
8909         necessary. Remove call to swap_ops_for_binary_stmt...
8910         (reassociate_bb): ... and move it here.
8911         (build_and_add_sum): Assign UIDs for new statements.
8912         (linearize_expr): Likewise.
8913         (do_reassoc): Renumber gimple statement UIDs.
8915 2013-05-17  Jan Hubicka  <jh@suse.cz>
8917         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
8918         weakrefs.
8919         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8920         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8921         weakrefs.
8922         (output_weakrefs): Update.
8924 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
8925             Martin Jambor  <mjambor@suse.cz>
8927         PR middle-end/57276
8928         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
8929         value that corresponds to the given aggval is found in values vector.
8931 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
8933         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
8934         sse, sse2, sse3, ssse3 and sse4a flags to options.
8936 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
8938         * gengtype-state.c: (s_expr_writer): New class, to handle
8939         prettifying of output layout of s-expressions.
8940         (state_writer): New class, to write out gtype.state.
8941         (state_written_type_count): Move this variable into member data of
8942         state_writer.
8943         (s_expr_writer::s_expr_writer): New code: constructor for new class
8944         (state_writer::state_writer(): ditto
8945         (s_expr_writer::write_new_line): New function
8946         (s_expr_writer::write_any_indent): ditto
8947         (s_expr_writer::begin_s_expr): ditto
8948         (s_expr_writer::end_s_expr): ditto
8949         (write_state_fileloc): convert to method of state_writer...
8950         (state_writer:: write_state_fileloc): ...and use methods of
8951         s_expr_writer to write indentation into the gtype.state output file
8952         to visually represent the hierarchical structure of the list
8953         structures
8954         (write_state_fields): ditto, renaming to...
8955         (state_writer::write_state_fields)
8956         (write_state_a_string): ditto, renaming to...
8957         (state_writer::write_state_a_string)
8958         (write_state_string_option): ditto, renaming to...
8959         (state_writer::write_state_string_option)
8960         (write_state_type_option): ditto, renaming to...
8961         (state_writer::write_state_type_option)
8962         (write_state_nested_option): ditto, renaming to...
8963         (state_writer::write_state_nested_option)
8964         (write_state_option): ditto, renaming to...
8965         (state_writer::write_state_option)
8966         (write_state_options): ditto, renaming to...
8967         (state_writer::write_state_options)
8968         (write_state_lang_bitmap): ditto, renaming to...
8969         (state_writer::write_state_lang_bitmap)
8970         (write_state_version): ditto, renaming to...
8971         (state_writer::write_state_version)
8972         (write_state_scalar_type): ditto, renaming to...
8973         (state_writer::write_state_scalar_type)
8974         (write_state_string_type): ditto, renaming to...
8975         (state_writer::write_state_string_type)
8976         (write_state_undefined_type): ditto, renaming to...
8977         (state_writer::write_state_undefined_type)
8978         (write_state_struct_union_type): ditto, renaming to...
8979         (state_writer::write_state_struct_union_type)
8980         (write_state_struct_type): ditto, renaming to...
8981         (state_writer::write_state_struct_type)
8982         (write_state_user_struct_type): ditto, renaming to...
8983         (state_writer::write_state_user_struct_type)
8984         (write_state_lang_struct_type): ditto, renaming to...
8985         (state_writer::write_state_lang_struct_type)
8986         (write_state_param_struct_type): ditto, renaming to...
8987         (state_writer::write_state_param_struct_type)
8988         (write_state_pointer_type): ditto, renaming to...
8989         (state_writer::write_state_pointer_type)
8990         (write_state_array_type): ditto, renaming to...
8991         (state_writer::write_state_array_type)
8992         (write_state_gc_used): ditto, renaming to...
8993         (state_writer::write_state_gc_used)
8994         (write_state_common_type_content): ditto, renaming to...
8995         (state_writer::write_state_common_type_content)
8996         (write_state_type): ditto, renaming to...
8997         (state_writer::write_state_type)
8998         (write_state_pair_list): ditto, renaming to...
8999         (state_writer::write_state_pair_list)
9000         (write_state_pair): ditto, renaming to...
9001         (state_writer::write_state_pair)
9002         (write_state_typedefs): ditto, renaming to...
9003         (state_writer::write_state_typedefs)
9004         (write_state_structures): ditto, renaming to...
9005         (state_writer::write_state_structures)
9006         (write_state_param_structs): ditto, renaming to...
9007         (state_writer::write_state_param_structs)
9008         (write_state_variables): ditto, renaming to...
9009         (state_writer::write_state_variables)
9010         (write_state_srcdir): ditto, renaming to...
9011         (state_writer::write_state_srcdir)
9012         (write_state_files_list): ditto, renaming to...
9013         (state_writer::write_state_files_list)
9014         (write_state_languages): ditto, renaming to...
9015         (state_writer::write_state_languages)
9016         (write_state): create a state_writer instance and use it when
9017         writing out the state file
9019 2013-05-17  Mike Stump  <mikestump@comcast.net>
9021         PR rtl-optimization/57304
9022         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
9023         accessing DF_REF_REAL_LOC.
9025 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9027         PR rtl-optimization/57281
9028         PR rtl-optimization/57300
9029         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
9030         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
9031         what the other splitter did if the registers are dead.
9033 2013-05-17  Richard Biener  <rguenther@suse.de>
9035         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
9036         MEM_REF offsets.
9038 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9040         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
9041         linking.
9043 2013-05-17  Marek Polacek  <polacek@redhat.com>
9045         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
9046         length when doing non-zero store of storing '\0' to '\0'.
9048 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
9050         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
9051         vect_external_def oprnd1 with loop_vinfo, try to emit
9052         optional cast, negation and and stmts on the loop preheader
9053         edge instead of into the pattern def seq.
9055         PR tree-optimization/57051
9056         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
9057         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
9059 2013-05-16  Nick Clifton  <nickc@redhat.com>
9061         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
9062         (rl78_is_naked_func): New function.
9063         (rl78_expand_prologue): Skip prologue generation for naked functions.
9064         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
9065         * doc/extend.texi (naked): Add RL78 to the list of processors
9066         that supports this attribute.
9068 2013-05-16  Jeff Law  <law@redhat.com>
9070         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
9072 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9074         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
9075         cache parameters using detect_caches_amd also for CYRIX,
9076         NSC and TM2 signatures.
9078 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
9079             Dzianis Kahanovich  <mahatma@eu.by>
9081         PR target/45359
9082         PR target/46396
9083         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
9084         VIA/Centaur processors and determine their cache parameters
9085         using detect_caches_amd.
9087 2013-05-16  Teresa Johnson  <tejohnson@google.com>
9089         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
9090         (rtl_verify_edges): New function.
9091         (rtl_verify_bb_insns): Ditto.
9092         (rtl_verify_bb_pointers): Ditto.
9093         (rtl_verify_bb_insn_chain): Ditto.
9094         (rtl_verify_fallthru): Ditto.
9095         (rtl_verify_bb_layout): Ditto.
9096         (rtl_verify_flow_info_1): Outline checks into new functions.
9097         (rtl_verify_flow_info): Ditto.
9099 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
9101         * cfghooks.c (copy_bbs): Add update_dominance argument.
9102         * cfghooks.h (copy_bbs): Update prototype.
9103         * tree-cfg.c (gimple_duplicate_sese_region):
9104         Add update_dominance argument.
9105         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
9106         * tree-ssa-loop-ch.c (copy_loop_headers): Update
9107         gimple_duplicate_sese_region call.
9108         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
9109         Update copy_bbs call.
9110         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
9111         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
9113 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9115         * tree-vectorizer.h (NUM_PATTERNS): Increment.
9116         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
9117         vect_recog_rotate_pattern.
9118         (vect_recog_rotate_pattern): New function.
9120 2013-05-16  Jason Merrill  <jason@redhat.com>
9122         * Makefile.in (LLINKER): New variable.
9123         (mostlyclean): Remove link mutex.
9124         * configure.ac: Handle --enable-link-mutex.
9125         * lock-and-run.sh: New script.
9127 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9129         PR target/19599
9130         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
9131         for NULL decl.
9133 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9135         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
9137 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
9139         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
9140         * config/arm/arm.c (next_consecutive_mem): New function.
9141         (gen_movmem_ldrd_strd): Likewise.
9142         * config/arm/arm.md (movmemqi): Update condition and code.
9143         (unaligned_loaddi, unaligned_storedi): New patterns.
9145 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9147         * config.gcc: Obsolete *-*-solaris2.9*.
9148         * doc/install.texi (Specific, *-*-solaris2*): Document it.
9150 2013-05-16  Richard Biener  <rguenther@suse.de>
9152         * passes.c (init_optimization_passes): Move pass_parallelize_loops
9153         earlier, after GRAPHITE transforms and IV canonicalization.
9155 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
9157         * omp-low.c (extract_omp_for_data): For collapsed loops,
9158         if at least one of the loops is known at compile time to
9159         iterate zero times, set count to 0.
9160         (expand_omp_regimplify_p): New function.
9161         (expand_omp_for_generic): For collapsed loops, if at least
9162         one of the loops isn't known to iterate at least once,
9163         add runtime check with setting count to 0.
9164         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
9165         For unsigned types if it isn't known at compile time that
9166         the loop will iterate at least once, add runtime check to bypass
9167         the whole loop if initial condition isn't true.
9169 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
9171         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
9173 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
9175         PR middle-end/57286
9176         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
9177         transformations to avoid an infinite loop.
9179 2013-05-16  Marek Polacek  <polacek@redhat.com>
9181         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
9183 2013-05-15  Leif Ekblad  <leif@rdos.net>
9185         * config/i386/i386.c (ix86_decompose_address): Use
9186         DEFAULT_TLS_SEG_REG to access TLS segment register.
9187         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
9188         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
9189         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
9191 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
9193         PR target/57260
9194         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
9195         sibling calls to functions that would normally be lazily bound,
9196         unless $gp is call-clobbered.
9198 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
9200         * config/i386/i386.c (ix86_option_override_internal): Update
9201         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
9202         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
9203         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
9204         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
9205         of TARGET_3DNOW.
9206         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
9208 2013-05-15  Andreas Schwab  <schwab@suse.de>
9210         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
9211         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
9212         third operand.
9214 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9216         * loop-unroll.c (report_unroll_peel): Check decision before
9217         emitting unroll/peel message.
9219 2013-05-15  Teresa Johnson  <tejohnson@google.com>
9221         * function.h (has_bb_partition): New rtl_data flag.
9222         (bb_reorder_complete): Ditto.
9223         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
9224         instead of flag_reorder_blocks_and_partition.
9225         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
9226         with some enhancements.
9227         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
9228         * bb-reorder.c (connect_traces): Check for has_bb_partition
9229         instead of flag_reorder_blocks_and_partition.
9230         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
9231         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
9232         verify_hot_cold_block_grouping.
9233         (partition_hot_cold_basic_blocks): Set has_bb_partition.
9235 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
9237         PR target/19599
9238         * config/arm/predicates.md (call_insn_operand): New predicate.
9239         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
9240         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
9241         if insn is not a tail call.
9242         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
9243         registers.
9244         * config/arm/arm.h (enum reg_class): New caller save register class.
9245         (REG_CLASS_NAMES): Likewise.
9246         (REG_CLASS_CONTENTS): Likewise.
9247         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
9248         without decls.
9250 2013-05-15  Richard Biener  <rguenther@suse.de>
9252         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
9253         of MSG_OPTIMIZED_LOCATIONS.
9254         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
9255         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
9256         message.
9257         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
9258         of MSG_OPTIMIZED_LOCATIONS.
9259         (execute_vect_slp): Likewise.
9260         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
9261         (vect_create_cond_for_alias_checks): Likewise.
9262         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9263         (vect_recog_widen_mult_pattern): Likewise.
9264         (vect_recog_widen_sum_pattern): Likewise.
9265         (vect_recog_over_widening_pattern): Likewise.
9266         (vect_recog_widen_shift_pattern): Likewise.
9267         (vect_recog_vector_vector_shift_pattern): Likewise.
9268         (vect_recog_divmod_pattern): Likewise.
9269         (vect_recog_mixed_size_cond_pattern): Likewise.
9270         (vect_recog_bool_pattern): Likewise.
9271         (vect_pattern_recog_1): Likewise.
9273 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9275         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
9276         non-functions to builtin_unreachable.
9277         * ipa-inline-transform.c (inline_call): Do not assert estimates were
9278         correct when new direct edges were discovered.
9280 2013-05-15  Martin Jambor  <mjambor@suse.cz>
9282         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
9283         header, print symbol order instead of node uid, print more information
9284         about indirect edge targets.
9285         (ipa_make_edge_direct_to_target): Print symbol order instead of node
9286         uids.
9287         (ipa_make_edge_direct_to_target): Likewise.
9288         (remove_described_reference): Likewise.
9289         (propagate_controlled_uses): Likewise.
9290         (ipa_print_node_params): Also print symbol order.
9291         (ipcp_transform_function): Print symbol order instead of node uids.
9292         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
9293         (cgraph_get_create_real_symbol_node): Likewise.
9294         * ipa-cp.c (print_lattice): Likewise.
9295         (print_all_lattices): Likewise.
9296         (determine_versionability): Likewise.
9297         (initialize_node_lattices): Likewise.
9298         (estimate_local_effects): Likewise.
9299         (update_profiling_info): Likewise.
9300         (create_specialized_node): Likewise.
9301         (perhaps_add_new_callers): Likewise.
9302         (decide_about_value): Likewise.
9303         (decide_whether_version_node): Likewise.
9304         (identify_dead_nodes): Likewise.
9305         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
9306         (dump_inline_summary): Likewise.
9307         (estimate_node_size_and_time): Likewise.
9308         (inline_analyze_function): Likewise.
9309         * ipa-inline.c (report_inline_failed_reason): Likewise.
9310         (want_early_inline_function_p): Likewise.
9311         (edge_badness): Likewise.
9312         (update_edge_key): Likewise.
9313         (inline_small_functions): Likewise.  Add dumping of order to two other
9314         dumps.
9315         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
9316         instead of node uids.
9317         (propagate_pure_const): Likewise.
9318         (propagate_pure_const): Likewise.
9319         * ipa-utils.c (dump_cgraph_node_set): Likewise.
9320         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
9321         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
9322         of node uids.
9323         * tree-pretty-print.c (dump_function_header): Likewise.
9324         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
9325         Print symbol order instead of node uids.
9327 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9329         * config/s390/s390.c (s390_register_move_cost): Don't impose the
9330         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
9332 2013-05-15  Richard Biener  <rguenther@suse.de>
9334         PR tree-optimization/57275
9335         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
9336         return value for fail to do runtime alias checks for gather loads.
9338 2013-05-15  Jan Hubicka  <jh@suse.cz>
9340         PR lto/57038
9341         PR lto/47375
9342         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
9343         weakrefs are not external.
9344         (lto_symtab_merge_decls): Fix thinko when dealing with
9345         non-lto_symtab decls.
9346         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
9347         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
9348         * varpool.c (dump_varpool_node): Dump more flags.
9350 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9352         * config/i386/i386.c (processor_alias_table): Add instruction
9353         FSGSBASE for AMD bdver3 architecture.
9355 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9357         * tree.c (warn_deprecated_use): Print file:line using locus color.
9358         * diagnostic.c (diagnostic_report_current_module): Print file:line
9359         and file:line:column using locus color.
9361 2013-05-14  Mike Stump  <mikestump@comcast.net>
9363         * gdbinit.in: Add __null.
9365 2013-05-14  Mike Stump  <mikestump@comcast.net>
9367         * recog.h: Rename struct recog_data to Recog_data.
9368         * recog.c: Likewise.
9369         * reload.c (can_reload_into): Likewise.
9370         * config/picochip/picochip.c: Likewise.
9372 2013-05-14  Mike Stump  <mikestump@comcast.net>
9374         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
9376 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
9378         * resource.h (struct resources): Remove unch_memory member.
9379         (CLEAR_RESOURCE): Don't clear unch_memory.
9380         * resource.c (mark_referenced_resources): Don't set it.
9381         (mark_set_resources): Likewise.
9382         (mark_target_live_regs): Don't clear it.
9383         (init_resource_info): Likewise.
9384         * reorg.c (resource_conflicts_p): Don't compare it.
9385         (redundant_insn): Don't set it.
9387         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
9388         Remove prototypes.
9389         * emit-rtl.c (next_label): Remove unused function.
9390         (skip_consecutive_labels, link_cc0_insns): Move to ...
9391         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
9392         only place where these functions are used, and make them static.
9394 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9396         * fold-const.c (fold_negate_expr): Handle vectors.
9397         (fold_truth_not_expr): Make it static.
9398         (fold_invert_truthvalue): New static function.
9399         (invert_truthvalue_loc): Handle vectors. Do not call
9400         fold_truth_not_expr directly.
9401         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
9402         <TRUTH_NOT_EXPR>: Do not cast to boolean.
9403         (fold_comparison): Handle vector constants.
9404         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
9405         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
9406         * tree.h (fold_truth_not_expr): Remove declaration.
9408 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
9410         * config/aarch64/aarch64-simd.md
9411         (aarch64_vcond_internal<mode>): Rename to...
9412         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
9413         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
9414         float modes. Clarify all iterator modes.
9415         (vcond<mode><mode>): Use new name for vcond expanders.
9416         (vcond<v_cmp_result><mode>): Likewise.
9417         (vcondu<mode><mode>: Likewise.
9418         * config/aarch64/iterators.md (VDQF_COND): New.
9420 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
9422         PR bootstrap/57266
9423         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9424         variable for the shift amount. Check that we shift by non-negative
9425         amounts.
9427 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
9429         PR target/42017
9430         * config/arm/arm.h (EPILOGUE_USES): Only return true
9431         for LR_REGNUM after epilogue_completed.
9433 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
9435         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
9436         is error_mark_node.
9438 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
9440         PR target/57261
9441         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
9442         and Solaris 11+/x86 with gld.
9443         * configure: Regenerate.
9445 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9447         * expmed.c (expand_shift_1): Canonicalize rotates by
9448         constant bitsize / 2 to bitsize - 1.
9449         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
9450         case ROTATERT>: Likewise.
9452         Revert:
9453         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9455         * config/i386/i386.md (rotateinv): New code attr.
9456         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9457         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9458         roll $31, %eax, etc.
9460 2013-05-14  Richard Biener  <rguenther@suse.de>
9462         PR middle-end/57235
9463         * tree-eh.c (sink_clobbers): Give up for successors with
9464         multiple predecessors and no virtual uses.
9466 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
9468         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
9469         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
9471 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
9473         PR middle-end/57251
9474         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
9475         the case when both op0 and op1 have VOIDmode.
9477 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
9479         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
9480         in multiply-accumulate mode.
9482 2013-05-13  Guozhi Wei  <carrot@google.com>
9484         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
9486 2013-05-13  Kai Tietz  <ktietz@redhat.com>
9488         PR target/56975
9489         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
9490         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
9491         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
9492         * config/i386/i386.c (ix86_option_override_internal): Likewise.
9493         (ix86_expand_prologue): Likewise.
9494         (ix86_expand_split_stack_prologue): Likewise.
9495         (legitimate_pic_address_disp_p): Likewise.
9496         (legitimize_pic_address): Likewise.
9497         (legitimize_tls_address): Likewise.
9498         (legitimize_pe_coff_symbol): Likewise.
9499         (output_pic_addr_const): Likewise.
9500         (construct_plt_address): Likewise.
9501         (ix86_expand_call): Likewise.
9502         (x86_output_mi_thunk): Likewise.
9503         (x86_function_profiler): Likewise.
9505 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
9507         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
9508         similar switch cases.
9509         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
9510         (aarch64_simd_mov_to_<mode>low): Delete.
9511         (aarch64_simd_mov_to_<mode>high): Delete.
9512         (move_lo_quad_<mode>): Add w<-r alternative.
9513         (aarch64_simd_move_hi_quad_<mode>): Likewise.
9514         (aarch64_simd_mov_from_*): Update type attribute.
9515         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
9516         statement.
9518 2013-05-13  Jan Hubicka  <jh@suse.cz>
9520         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
9521         * config/i386/i386.c (ix86_compute_frame_layout,
9522         ix86_expand_epilogue, emit_i387_cw_initialization,
9523         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
9524         ix86_local_alignment): Fix use of size/speed predicates.
9526 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9528         PR tree-optimization/45216
9529         PR tree-optimization/57157
9530         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
9531         the (-Y) & (B - 1) variant if OP is |.
9532         * expmed.c (expand_shift_1): For rotations by const0_rtx just
9533         return shifted.  Use (-op1) & (prec - 1) as other_amount
9534         instead of prec - op1.
9536 2013-05-13  Martin Jambor  <mjambor@suse.cz>
9538         PR middle-end/42371
9539         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
9540         (ipa_constant_data): New type.
9541         (ipa_jump_func): Use ipa_constant_data to hold information about
9542         constant jump functions.
9543         (ipa_get_jf_constant): Adjust to jump function type changes.
9544         (ipa_get_jf_constant_rdesc): New function.
9545         (ipa_param_descriptor): New field controlled_uses.
9546         (ipa_get_controlled_uses): New function.
9547         (ipa_set_controlled_uses): Likewise.
9548         * ipa-ref.h (ipa_find_reference): Declare.
9549         * ipa-prop.c (ipa_cst_ref_desc): New type.
9550         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
9551         changes.
9552         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
9553         New parameter cs.  Adjust all callers.
9554         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
9555         (remove_described_reference): New function.
9556         (jfunc_rdesc_usable): Likewise.
9557         (try_make_edge_direct_simple_call): Decrement controlled use count,
9558         attempt to remove reference if it hits zero.
9559         (combine_controlled_uses_counters): New function.
9560         (propagate_controlled_uses): Likewise.
9561         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
9562         (ipa_edge_duplication_hook): Duplicate reference descriptions.
9563         (ipa_print_node_params): Print described use counter.
9564         (ipa_write_jump_function): Adjust to jump function type changes.
9565         (ipa_read_jump_function): New parameter CS, pass it to
9566         ipa_set_jf_constant.  Adjust caller.
9567         (ipa_write_node_info): Stream controlled use count
9568         (ipa_read_node_info): Likewise.
9569         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
9570         asserting.
9571         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
9572         count.  Remove cloning-added reference if it reaches zero.
9573         * ipa-ref.c (ipa_find_reference): New function.
9575 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
9577         * config/i386/i386.c (processor_target_table): Modified default
9578         alignment values for AMD BD and BT architectures.
9580 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
9582         * tree-vect-generic.c (uniform_vector_p): Move ...
9583         * tree.c (uniform_vector_p): ... here.
9584         * tree.h (uniform_vector_p): Declare it.
9585         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
9586         into a scalar.
9588 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
9590         PR tree-optimization/57230
9591         * tree-ssa-strlen.c (handle_char_store): Record length for
9592         array store from STRING_CST.
9594         PR tree-optimization/57230
9595         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
9596         check.
9598 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
9600         * config/epiphany/epiphany.c (epiphany_init): Check size of
9601         NUM_MODES_FOR_MODE_SWITCHING.
9602         (epiphany_expand_prologue):
9603         Remove CONFIG_REGNUM initial value handling code.
9604         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
9605         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
9606         (emit_set_fp_mode, epiphany_mode_after): Likewise.
9607         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
9608         Don't return 1 for FP_MODE_NONE.
9609         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
9610         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
9611         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
9612         * config/epiphany/epiphany.md (save_config): New pattern.
9614 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
9616         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
9618 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
9620         * config/i386/i386.md (memory): Handle sseishft1.
9621         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
9622         (*vec_extractv2di_1): Ditto.
9624 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
9626         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
9627         saved registers.
9629 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
9631         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
9632         Add mthumb/march=armv7-a multilib.
9633         Add mthumb/march=armv7-r multilib.
9634         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
9636 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
9638         * config/v850/t-rtems: Add more multilibs.
9640 2013-05-10  Richard Biener  <rguenther@suse.de>
9642         PR tree-optimization/57214
9643         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
9644         not propagate from SSA names that occur in abnormal PHI nodes.
9646 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
9648         * stor-layout.c (element_precision): New function.
9649         * machmode.h (element_precision): Declare it.
9650         * tree.c (build_minus_one_cst): New function.
9651         (element_precision): Likewise.
9652         * tree.h (build_minus_one_cst): Declare new function.
9653         (element_precision): Likewise.
9654         * fold-const.c (operand_equal_p): Use element_precision.
9655         (fold_binary_loc): Handle vector types.
9656         * convert.c (convert_to_integer): Use element_precision.
9657         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
9658         separately.
9660 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
9662         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
9663         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
9664         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
9665         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9666         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
9667         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
9668         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
9669         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
9670         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
9671         (Uuw8): New constraints.
9672         (Usb4): Move into alphabetical order.
9673         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
9674         (sd8_operand, ub8_operand, uw8_operand): New predicates.
9675         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
9676         previously unnamed patterns.
9677         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
9678         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
9679         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
9680         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
9681         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
9682         of specific lengths.
9684 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
9686         * config/i386/i386.md (rotateinv): New code attr.
9687         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
9688         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
9689         roll $31, %eax, etc.
9691         PR tree-optimization/45216
9692         PR tree-optimization/57157
9693         * tree-ssa-forwprop.c (simplify_rotate): New function.
9694         (ssa_forward_propagate_and_combine): Call it.
9696 2013-05-10  Richard Biener  <rguenther@suse.de>
9698         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
9699         disable peeling when we version for aliasing.
9700         (vector_alignment_reachable_p): Honor explicit user alignment.
9701         (vect_supportable_dr_alignment): Likewise.
9702         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
9703         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
9704         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
9705         then peeling to arrange for the cost-model check to come first.
9707 2013-05-10  Alan Modra  <amodra@gmail.com>
9709         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
9710         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
9711         * configure: Regenerate.
9713 2013-05-10  Alan Modra  <amodra@gmail.com>
9715         PR target/55033
9716         * varasm.c (default_elf_select_section): Move !DECL_P check..
9717         (get_named_section): ..to here before calling get_section_name.
9718         Adjust assertion.
9719         (default_section_type_flags): Add DECL_P check.
9720         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
9721         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
9723 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
9725         * config/epiphany/epiphany.c (epiphany_expand_prologue):
9726         When using gen_stack_adjust_str with a register offset, add a
9727         REG_FRAME_RELATED_EXPR note.
9729 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
9731         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
9732         (*vec_extractv4si_zext_mem): Ditto.
9733         (*vec_extractv2di): Add 0->x and x->x alternatives.
9734         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
9735         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
9737 2013-05-09  Jason Merrill  <jason@redhat.com>
9739         N3639 C++1y VLA support
9740         * gimplify.c (gimplify_vla_decl): Don't touch an existing
9741         DECL_VALUE_EXPR.
9743         * tree.c (build_constructor_va): New.
9744         * tree.h: Declare it.
9746 2013-05-09  Martin Jambor  <mjambor@suse.cz>
9748         PR lto/57084
9749         * gimple-fold.c (canonicalize_constructor_val): Call
9750         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
9752 2013-05-09  Jan Hubicka  <jh@suse.cz>
9753             Richard Biener  <rguenther@suse.de>
9755         PR lto/54095
9756         * symtab.c (symtab_make_decl_local): Do not add private names.
9758 2013-05-09  Jan Hubicka  <jh@suse.cz>
9760         PR lto/54095
9761         * symtab.c (insert_to_assembler_name_hash): Handle clones.
9762         (unlink_from_assembler_name_hash): Likewise.
9763         (symtab_prevail_in_asm_name_hash, symtab_register_node,
9764         symtab_unregister_node, symtab_initialize_asm_name_hash,
9765         change_decl_assembler_name): Update.
9767 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
9769         * config/aarch64/aarch64.md: New movtf split.
9770         (*movtf_aarch64): Update.
9771         (aarch64_movdi_tilow): Handle TF modes and rename to
9772         aarch64_movdi_<mode>low.
9773         (aarch64_movdi_tihigh): Handle TF modes and rename to
9774         aarch64_movdi_<mode>high
9775         (aarch64_movtihigh_di): Handle TF modes and rename to
9776         aarch64_mov<mode>high_di
9777         (aarch64_movtilow_di): Handle TF modes and rename to
9778         aarch64_mov<mode>low_di
9779         (aarch64_movtilow_tilow): Remove spurious whitespace.
9780         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
9781         splits.
9782         (aarch64_print_operand): Update.
9784 2013-05-09  Alan Modra  <amodra@gmail.com>
9786         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
9787         powerpc64le.
9788         * configure: Regenerate.
9790 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9792         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
9793         splitter preparation statements.
9794         * config/i386/sse.md (*vec_extract* splitters): Ditto.
9795         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
9796         adjust_address_nv.
9798 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9800         * gimple-ssa-strength-reduction.c (count_candidates): Change
9801         return value to int.
9802         (analyze_candidates_and_replace): Change type of length to int.
9804 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
9806         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
9807         (*vec_extract<mode>): Use VI12_128 mode iterator.
9808         (*vec_extract<mode>_mem): Ditto.
9809         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
9810         attribute.
9812 2013-05-08  Diego Novillo  <dnovillo@google.com>
9814         PR bootstrap/54659
9816         Revert:
9817         2012-08-17  Diego Novillo  <dnovillo@google.com>
9819         PR bootstrap/54281
9820         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
9821         * config.in: Regenerate.
9822         * configure: Regenerate.
9823         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
9825 2013-05-08  Jan Hubicka  <jh@suse.cz>
9827         PR lto/54095
9828         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
9829         * cgraph.h (symtab_node_base): Add unique_name.
9830         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
9831         input_overwrite_node, input_varpool_node): Stream unique_name.
9832         * cgraphclones.c (cgraph_create_virtual_clone,
9833         cgraph_function_versioning): Set unique_name.
9834         * ipa.c (function_and_variable_visibility): Set unique_name.
9836 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9838         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
9839         (alloc_cand_and_find_basis): Restrict conditional candidate
9840         processing to CAND_MULTs.
9842 2013-05-08  Jan Hubicka  <jh@suse.cz>
9844         PR lto/54095
9845         lto-symtab.c (lto_symtab_symbol_p): New function.
9846         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
9847         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
9848         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
9849         Skip static symbols.
9851 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
9853         PR tree-optimization/57200
9854         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
9855         Only call inform if the preceding warning_at returns true.
9857 2013-05-07  Han Shen  <shenhan@google.com>
9859         * cfgexpand.c (record_or_union_type_has_array_p): New function.
9860         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
9861         * common.opt (fstack-protector-strong): New option.
9862         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
9863         * doc/invoke.texi (Optimization Options): Document
9864         "-fstack-protector-strong".
9865         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
9867 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
9869         * config/mips/mips.c (mips_machine_reorg2): Return 0.
9871 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
9873         * ira.c (update_equiv_regs): Add insn having equiv memory even if
9874         it is not lhs of the insn.
9875         (setup_reg_equiv): Remove insn having equiv memory which it is not
9876         lhs of the insn.
9877         * lra-constraints.c (process_address): Try to improve generation
9878         code for address base + disp.
9879         (lra_constraints): Make correct the code for checking insn setting
9880         up backward equivalence.  Remove insn only if it is in the init
9881         insn list.
9882         * lra-eliminations.c (update_reg_eliminate): Change return value.
9883         (lra_eliminate): Use the result.
9885 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
9887         * config/i386/sse.md (ssescalarnummask): New mode attribute.
9888         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
9889         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
9890         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
9891         register target operands.
9892         (*vec_extractv8hi_sse2): New pattern.
9893         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
9894         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
9895         (*vec_extract<mode>_mem): New insn and split pattern.
9897 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
9899         * config/arm/arm.c (arm_asan_shadow_offset): New function.
9900         (TARGET_ASAN_SHADOW_OFFSET): Define.
9901         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
9902         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
9904 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9906         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
9907         (incr_vec_index): Return -1 if increment not found.
9908         (create_add_on_incoming_edge): Assert if increment not found.
9909         (record_increment): Limit number of increments recorded.
9910         (all_phi_incrs_profitable): Return false if an increment not found.
9911         (replace_profitable_candidates): Don't process increments that were
9912         not recorded.
9913         (analyze_candidates_and_replace): Limit size of incr_vec.
9915 2013-05-07  Richard Biener  <rguenther@suse.de>
9917         * calls.c (special_function_p): setjmp-like functions are leaf.
9918         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
9919         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
9921 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
9923         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
9924         (aarch64_simd_mov<mode>): New expander.
9925         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
9926         (aarch64_simd_mov_to_<mode>high): Likewise.
9927         (aarch64_simd_mov_from_<mode>low): Likewise.
9928         (aarch64_simd_mov_from_<mode>high): Likewise.
9929         (aarch64_dup_lane<mode>): Update.
9930         (aarch64_dup_lanedi): New instruction pattern.
9931         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
9932         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
9934 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9936         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
9937         (replace_mult_candidate): Remove unnecessary argument; remove
9938         unnecessary parameter from call to introduce_cast_before_cand.
9939         (replace_unconditional_candidate): Remove unnecessary parameter
9940         from call to replace_mult_candidate.
9941         (replace_conditional_candidate): Likewise.
9942         (insert_initializers): Use make_temp_ssa_name.
9943         (introduce_cast_before_cand): Remove unnecessary argument; use
9944         make_temp_ssa_name.
9945         (replace_one_candidate): Remove unnecessary argument; remove
9946         unnecessary parameter from calls to introduce_cast_before_cand.
9947         (replace_profitable_candidates): Remove unnecessary parameters
9948         from calls to replace_one_candidate.
9950 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9952         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9953         phi def as possibly hiding a basis for a CAND_ADD whose operands
9954         have been commuted in the analysis.
9955         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
9957 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
9959         * config/aarch64/aarch64.md
9960         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
9961         shift value between 0-4.
9963 2013-05-07  Richard Biener  <rguenther@suse.de>
9965         * double-int.h (rshift): New overload.
9966         * double-int.c (rshift): New function.
9967         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
9968         (create_reference_ops_from_ref): Remove.
9969         (vn_reference_insert): Use shared ops for constructing the
9970         reference and copy it.
9972 2013-05-07  Richard Biener  <rguenther@suse.de>
9974         PR middle-end/57190
9975         * tree-eh.c (sink_clobbers): Properly propagate
9976         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
9978 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
9980         PR tree-optimization/57149
9981         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
9982         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
9983         collect_phi_def_edges, execute_late_warn_uninitialized): Use
9984         uninit_undefined_value_p instead of ssa_undefined_value_p.
9986         PR debug/57184
9987         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
9988         for modifier == EXPAND_INITIALIZER.
9990 2013-05-07  Anton Blanchard  <anton@samba.org>
9992         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
9993         for powerpc64 little endian.
9994         * configure: Regenerate.
9996 2013-05-06  Graham Stott  <grahams@btinternet.com>
9998         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
9999         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
10000         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
10001         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
10003 2013-05-06  Graham Stott  <grahams@btinternet.com>
10005         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
10006         codes which allow non-lvalues.
10008 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
10010         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
10011         components are all 1s.
10012         (integer_minus_onep): New function.
10013         * tree.h (integer_minus_onep): Declare it.
10014         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
10015         integer_minus_onep instead of integer_all_onesp.
10017 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10019         PR target/52933
10020         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
10021         variations of these patterns.
10023 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10025         * config/i386/i386.md (isa): Add x64_sse4 member.
10026         (enabled): Handle x64_sse4.
10027         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
10028         instruction for 64bit SSE4_1 targets.  Update insn attributes.
10029         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
10030         instruction for SSE4_1 targets.  Update insn attributes.
10031         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10032         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
10033         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
10034         const_1 selector.
10035         (*vec_extractv4si): Rename from *sse4_1_pextrd.
10036         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
10037         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
10039 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10041         PR target/57108
10042         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
10044 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
10046         * final.c (do_assembler_dialects): Don't handle curly braces and
10047         vertical bar escaped by % as dialect delimiters.
10048         (output_asm_insn): Print curly braces and vertical bar if escaped
10049         by % and ASSEMBLER_DIALECT defined.
10050         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
10051         * doc/tm.texi: Regenerated.
10053 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
10055         * config/mips/mips.c: Include tree-pass.h.
10056         (mips_reorg): Split in pre- and post-dbr_schedule parts.
10057         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
10058         (pass_mips_machine_reorg2): New machine specific pass.
10059         (insert_pass_mips_machine_reorg2): New pass plugin definition.
10060         (mips_option_override): Register the new pass.
10061         * rtl.h (cleanup_barriers): Remove prototype.
10062         (dbr_schedule): Likewise.
10063         * jump.c (cleanup_barriers): Make static.
10064         * reorg.c (dbr_schedule): Likewise.
10066 2013-05-06  Richard Biener  <rguenther@suse.de>
10068         PR tree-optimization/57185
10069         * tree-parloops.c (add_field_for_reduction): Handle anonymous
10070         SSA names properly.
10072 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
10074         PR target/57106
10075         * config/i386/i386.c (add_parameter_dependencies): Add dependence
10076         between "first_arg" and "insn", not "last" and "insn".
10078 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10080         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
10081         (find_candidates_in_block): Re-enable slsr_process_phi.
10082         (create_phi_basis): Fix double counting of candidate adjustment.
10084 2013-05-06  Richard Biener  <rguenther@suse.de>
10086         PR middle-end/57147
10087         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
10088         the edge is also fallthru, preserve it and just clear the
10089         abnormal flag.
10090         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
10091         also complex, preserve that and just clear the fallthru flag.
10092         * tree-inline.c (update_ssa_across_abnormal_edges): Also
10093         update virtual operands.
10095 2013-05-06  Alan Modra  <amodra@gmail.com>
10097         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
10098         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
10099         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
10100         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
10101         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10102         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
10104 2013-05-06  Alan Modra  <amodra@gmail.com>
10106         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
10107         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
10108         (DEFAULT_ASM_ENDIAN): Define.
10109         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
10110         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
10111         Update -K PIC clause from sysv4.h.
10112         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
10113         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
10115 2013-05-06  Alan Modra  <amodra@gmail.com>
10117         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
10118         twice for little-endian.
10119         (ashrdi3_no_power, ashrdi3): Support little-endian.
10121 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
10123         PR target/55303
10124         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
10125         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
10126         related expanders.
10127         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
10128         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
10129         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
10130         New predicates.
10132 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
10133             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
10135         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
10136         * config/pa/pa.opt: Make mbig-switch a no-op.
10137         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
10138         (CASE_VECTOR_MODE): Always return SImode.
10139         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
10140         for the !TARGET_BIG_SWITCH case.
10141         * config/pa/pa-linux.h: Likewise.
10142         * config/pa/pa-openbsd.h: Likewise.
10143         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
10144         * config/pa/pa.md (short_jump): Remove define_insn.
10145         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
10146         (casesi0): Remove define_insn.
10147         (type): Remove btable_branch.
10148         (pa_combine_type): Likewise.
10149         (in_nullified_branch_delay): Likewise.
10150         (in_call_delay): Likewise.
10151         (define_delay): Likewise.
10152         (define_insn_reservation "Z3"): Likewise.
10153         (define_insn_reservation "Z4"): Likewise.
10154         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
10155         (pa_adjust_insn_length): Remove adjustment for btable branches.
10156         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
10157         and mno-big-switch
10159 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
10161         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
10162         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
10163         Add m->r,x alternatives.
10164         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
10165         splitters using SWI48x mode iterator.
10166         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
10167         TARGET_64BIT.  Add m->x alternative.
10168         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
10169         Add o->x alternative.  Enable for TARGET_SSE.
10170         (sse_storeq): Remove expander.
10171         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
10172         with memory input operand.
10173         (*vec_extractv2di_1 splitter): New.
10174         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
10175         * config/i386/i386.md (ssevecmodelower): New mode attribute.
10177 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10179         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
10180         (INT_LOWPART): Delete.
10181         (extract_MB): Adjust.
10182         (extract_ME): Adjust.
10183         (print_operand): Adjust.
10185 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
10187         * config/rs6000/predicates.md (reg_or_add_cint_operand,
10188         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10189         (reg_or_logical_cint_operand, easy_fp_constant,
10190         logical_const_operand): Delete "CONST_DOUBLE" case.
10191         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
10192         "HOST_BITS_PER_WIDE_INT == 64" test.
10193         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
10194         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
10195         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
10196         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
10197         test.
10198         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
10199         CONST_DOUBLE DImode/VOIDmode case.
10200         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
10201         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10202         CONST_DOUBLE VOIDmode case.
10203         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
10204         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
10205         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
10206         Delete CONST_DOUBLE case.
10207         (splitters for mov FMOVE64 const_double): Delete
10208         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
10209         "HOST_BITS_PER_WIDE_INT >= 64" test.
10210         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
10211         case.
10212         (mov DI const_double): Delete.
10214 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
10216         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
10217         on op shows all bits zero in mode of a lowpart subreg, return zero.
10219 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
10221         PR target/57150
10222         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
10223         to save TFmode registers and DImode to save TImode registers for
10224         caller save operations.
10225         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
10226         mark being partially clobbered since they only use the first
10227         double word.
10229         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
10230         and TDmode only use the upper 64-bits of each VSX register.
10232 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10234         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
10235         (find_candidates_in_block): Disable slsr_process_phi.
10237 2013-05-03  Guozhi Wei  <carrot@google.com>
10239         * coverage.c (coverage_obj_init): Move the construction of gcov
10240         constructor to ...
10241         (build_init_ctor): ... here.
10243 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
10245         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
10246         (slsr_cand_d): Redefine def_phi.
10247         (stride_status, phi_adjust_status, count_phis_status): New enums.
10248         (find_phi_def): New.
10249         (find_basis_for_base_expr): New.
10250         (find_basis_for_candidate): Handle hidden bases.
10251         (alloc_cand_and_find_basis): Handle phi candidates.
10252         (slsr_process_phi): New.
10253         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
10254         (create_mul_imm_cand): Likewise.
10255         (create_add_ssa_cand): Exclude phi base candidates.
10256         (create_add_imm_cand): Likewise.
10257         (slsr_process_cast): Likewise.
10258         (slsr_process_copy): Likewise.
10259         (find_candidates_in_block): Handle phi candidates.
10260         (dump_candidate): Likewise.
10261         (unconditional_cands): Delete.
10262         (unconditional_cands_with_known_stride_p): Delete.
10263         (phi_dependent_cand_p): New.
10264         (cand_increment): Handle phi-dependent candidates.
10265         (replace_dependent): Delete.
10266         (replace_mult_candidate): New.
10267         (replace_unconditional_candidate): New.
10268         (incr_vec_index): Move to avoid forward reference.
10269         (create_add_on_incoming_edge): New.
10270         (create_phi_basis): New.
10271         (replace_dependents): Delete.
10272         (replace_conditional_candidate): New.
10273         (phi_add_costs): New.
10274         (replace_uncond_cands_and_profitable_phis): New.
10275         (record_increment): Handle phi adjustments.
10276         (record_phi_increments): New.
10277         (record_increments): Handle phi adjustments.
10278         (phi_incr_cost): New.
10279         (lowest_cost_path): Handle phis.
10280         (total_savings): Likewise.
10281         (analyze_increments): Likewise.
10282         (ncd_with_phi): New.
10283         (ncd_of_cand_and_phis): New.
10284         (nearest_common_dominator_for_cands): Handle phi increments.
10285         (all_phi_incrs_profitable): New.
10286         (replace_profitable_candidates): Handle phi-dependent candidates.
10287         (analyze_candidates_and_replace): Likewise.
10289 2013-05-03  Teresa Johnson  <tejohnson@google.com>
10291         PR bootstrap/57154
10292         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
10293         do not exceed REG_BR_PROB_BASE.
10295 2013-05-03  Jeff Law  <law@redhat.com>
10297         PR tree-optimization/57144
10298         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
10299         operand of the condition will bit into the new type when eliminating
10300         a cast feeding a condition.
10302 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
10304         PR rtl-optimization/57130
10305         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
10306         of COMPARE as in_code to the recursive call if needed.
10308 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
10310         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
10311         (enabled): Handle new members.
10312         * config/i386/sse.md (*vec_concatv2si): Merge from
10313         *vec_concatv2si_sse2 and vec_concatv2si_sse.
10314         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
10316 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
10318         PR tree-optimization/57027
10319         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
10320         for fnms opportunity, check we got the prerequisite kind
10321         of tree / gimple before using accessor functions.
10323 2013-05-03  Richard Biener  <rguenther@suse.de>
10325         * double-int.h (lshift): New overload without precision
10326         and arith argument.
10327         (operator *=, operator +=, operator -=): Move ...
10328         * double-int.c (operator *=, operator +=, operator -=): ... here
10329         and implement more efficiently.
10330         (mul_double_with_sign): Remove.
10331         (lshift_double): Adjust to take unsinged shift argument, push
10332         dispatching code to callers.
10333         (mul_double_wide_with_sign): Add early out for callers that
10334         are not interested in high parts or overflow.
10335         (lshift): New function.
10336         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
10337         dispatch code here.
10338         (lrotate, rrotate): Use logical shifts.
10339         * expr.c (get_inner_reference): Use lshift.
10340         * fixed-value.c (do_fixed_divide): Likewise.
10341         * tree-dfa.c (get_ref_base_and_extent): Likewise.
10342         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
10343         (indirect_refs_may_alias_p): Likewise.
10344         (stmt_kills_ref_p_1): Likewise.
10346 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10348         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
10350 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
10352         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
10353         scalar form of FABD instruction.
10355 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
10357         * lra-constraints.c (process_alt_operands): Add checking alt
10358         number to choose the best alternative.
10360 2013-05-02  Richard Biener  <rguenther@suse.de>
10362         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
10363         bitmap and its handling.
10364         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
10366 2013-05-02  Richard Biener  <rguenther@suse.de>
10368         PR middle-end/57140
10369         * tree-inline.c (copy_loops): Properly handle removed loops.
10370         (copy_cfg_body): Mark destination loops for fixup if source
10371         loops needed fixup.
10373 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
10375         PR target/56732
10376         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
10377         generating simple_return for naked functions.
10379 2013-05-02  Martin Jambor  <mjambor@suse.cz>
10381         PR middle-end/56988
10382         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
10383         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
10384         flags match.
10385         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
10386         ipa_agg_replacement_value structures.
10387         (known_aggs_to_agg_replacement_list): Likewise.
10388         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
10389         (read_agg_replacement_chain): Likewise.
10390         (ipcp_transform_function): Also check that by_ref flags match.
10392 2013-05-02  Richard Biener  <rguenther@suse.de>
10394         * graphds.h (struct graph): Add obstack member.
10395         * graphds.c (new_graph): Initialize obstack and allocate
10396         vertices from it.
10397         (add_edge): Allocate edge from the obstack.
10398         (free_graph): Free the obstack instead of all edges and vertices.
10400 2013-05-02  Teresa Johnson  <tejohnson@google.com>
10402         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
10403         divides.
10404         * cfg.c (update_bb_profile_for_threading): Ditto.
10405         * tree-inline.c (copy_bb): Ditto.
10406         (copy_edges_for_bb): Ditto.
10407         (initialize_cfun): Ditto.
10408         (copy_cfg_body): Ditto.
10409         (expand_call_inline): Ditto.
10410         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
10411         (estimate_node_size_and_time): Ditto.
10412         (inline_merge_summary): Ditto.
10413         * cgraphclones.c (cgraph_clone_edge): Ditto.
10414         (cgraph_clone_node): Ditto.
10415         * sched-rgn.c (compute_dom_prob_ps): Ditto.
10416         (compute_trg_info): Ditto.
10418 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10420         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
10421         S reg when fp attribute set.
10422         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
10424 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
10426         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
10427         New pattern.
10428         (*and_one_cmplsi3_compare0_uxtw): Likewise.
10429         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
10430         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
10432 2013-05-02  Richard Biener  <rguenther@suse.de>
10434         * tree-scalar-evolution.c (scev_info_hasher): Remove.
10435         (struct instantiate_cache_entry): New type.
10436         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
10437         (struct instantiate_cache_type): New type.
10438         (set_instantiated_value, get_instantiated_value): Remove.
10439         (get_instantiated_value_entry): New function.
10440         (instantiate_scev_name): Use the new cache and adjust.
10441         (instantiate_scev_poly): Adjust.
10442         (instantiate_scev_binary): Likewise.
10443         (instantiate_array_ref): Likewise.
10444         (instantiate_scev_convert): Likewise.
10445         (instantiate_scev_not): Likewise.
10446         (instantiate_scev_3): Likewise.
10447         (instantiate_scev_2): Likewise.
10448         (instantiate_scev_r): Likewise.
10449         (instantiate_scev): Likewise.
10450         (resolve_mixers): Likewise.
10452 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
10454         PR target/57091
10455         * lra-constraints.c (best_small_class_operands_num): Remove.
10456         (process_alt_operands): Remove small_class_operands_num.  Take
10457         small classes operands into losers and only if the operand is not
10458         matched.  Modify debugging output.
10459         (curr_insn_transform): Remove best_small_class_operands_num.
10460         Print insn name.
10462 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10464         * config/aarch64/aarch64-builtins.c
10465         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
10466         * config/aarch64/aarch64-simd-builtins.def
10467         (reduc_splus_): Add new modes.
10468         (reduc_uplus_): New.
10469         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
10470         (reduc_uplus_v4sf): Likewise.
10471         (reduc_splus_v4sf): Likewise.
10472         (aarch64_addv<mode>): Likewise.
10473         (reduc_uplus_<mode>): Likewise.
10474         (reduc_splus_<mode>): Likewise.
10475         (aarch64_addvv2di): Likewise.
10476         (reduc_uplus_v2di): Likewise.
10477         (reduc_splus_v2di): Likewise.
10478         (aarch64_addvv2si): Likewise.
10479         (reduc_uplus_v2si): Likewise.
10480         (reduc_splus_v2si): Likewise.
10481         (reduc_<sur>plus_<mode>): New.
10482         (reduc_<sur>plus_v2di): Likewise.
10483         (reduc_<sur>plus_v2si): Likewise.
10484         (reduc_<sur>plus_v4sf): Likewise.
10485         (aarch64_addpv4sf): Likewise.
10486         * config/aarch64/arm_neon.h
10487         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
10488         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
10489         add UNSPEC_SADDV, UNSPEC_UADDV.
10490         (SUADDV): New.
10491         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
10493 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10495         * config/aarch64/arm_neon.h
10496         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
10498 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10500         * config/aarch64/aarch64-builtins
10501         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
10503 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10505         * config/aarch64/aarch64-simd-builtins.def
10506         (reduc_smax_): New.
10507         (reduc_smin_): Likewise.
10508         (reduc_umax_): Likewise.
10509         (reduc_umin_): Likewise.
10510         (reduc_smax_nan_): Likewise.
10511         (reduc_smin_nan_): Likewise.
10512         (fmax): Remove.
10513         (fmin): Likewise.
10514         (smax): Update for V2SF, V4SF and V2DF modes.
10515         (smin): Likewise.
10516         (smax_nan): New.
10517         (smin_nan): Likewise.
10518         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
10519         (<su><maxmin><mode>3): ...This, refactor.
10520         (s<maxmin><mode>3): New.
10521         (<maxmin_uns><mode>3): Likewise.
10522         (reduc_<maxmin_uns>_<mode>): Refactor.
10523         (reduc_<maxmin_uns>_v4sf): Likewise.
10524         (reduc_<maxmin_uns>_v2si): Likewise.
10525         (aarch64_<fmaxmin><mode>: Remove.
10526         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
10527         new builtin names.
10528         (vmin<q>_f<32,64>): Likewise.
10529         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10530         (FMAXMIN): New.
10531         (su): Add mappings for smax, smin, umax, umin.
10532         (maxmin): New.
10533         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10534         (FMAXMIN): Rename as...
10535         (FMAXMIN_UNS): ...This.
10536         (maxminv): Remove.
10537         (fmaxminv): Likewise.
10538         (fmaxmin): Likewise.
10539         (maxmin_uns): New.
10540         (maxmin_uns_op): Likewise.
10542 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10544         * config/aarch64/arm_neon.h
10545         (vac<ge, gt><sd>_f<32, 64>): Rename to...
10546         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
10547         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
10549 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10551         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
10552         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
10554 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10556         * config/aarch64/aarch64-simd.md
10557         (vcond<mode>_internal): Handle special cases for constant masks.
10558         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
10559         (vcondu<mode><mode>): Likewise.
10560         (vcond<v_cmp_result><mode>): New.
10562 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10564         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
10565         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
10566         * config/aarch64/aarch64-simd-builtins.def
10567         (cmeq): Update to BUILTIN_VALLDI.
10568         (cmgt): Likewise.
10569         (cmge): Likewise.
10570         (cmle): Likewise.
10571         (cmlt): Likewise.
10572         * config/aarch64/arm_neon.h
10573         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
10574         to builtins or C as appropriate.
10576 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
10578         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
10579         (cmgeu): ...This.
10580         (cmhi): Rename to...
10581         (cmgtu): ...This.
10582         * config/aarch64/aarch64-simd.md
10583         (simd_mode): Add SF.
10584         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
10585         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
10586         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
10587         (cstore<mode>_neg): ...This.
10588         * config/aarch64/iterators.md
10589         (VALLF): new.
10590         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10591         (COMPARISONS): New.
10592         (UCOMPARISONS): Likewise.
10593         (optab): Add missing comparisons.
10594         (n_optab): New.
10595         (cmp_1): Likewise.
10596         (cmp_2): Likewise.
10597         (CMP): Likewise.
10598         (cmp): Remove.
10599         (VCMP_S): Likewise.
10600         (VCMP_U): Likewise.
10601         (V_cmp_result): Add DF, SF modes.
10602         (v_cmp_result): Likewise.
10603         (v): Likewise.
10604         (vmtype): Likewise.
10605         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
10607 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
10609         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
10610         define_insn to define_insn_and_split.
10611         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
10612         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
10613         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
10614         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
10615         (thumb2_negscc): Likewise.
10617 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10619         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
10621 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
10623         * config/arm/thumb2.md: Remove trailing whitespaces.
10625 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
10627         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
10628         Use gen_int_mode rather than GEN_INT.
10630 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
10632         * value-prof.c (stream_in_histogram_value): Remove the strayed
10633         debug_gimple_stmt.
10635 2013-04-30  Richard Biener  <rguenther@suse.de>
10637         PR middle-end/57122
10638         * cfghooks.c (split_edge): Properly check for the loop latch edge.
10640 2013-04-30  Richard Biener  <rguenther@suse.de>
10642         PR middle-end/57107
10643         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
10645 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
10647         PR rtl-optimization/56957
10648         PR rtl-optimization/57105
10649         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
10650         variable.  Use just INSN_UID for determining whether an insn
10651         should be only disconnected from the insn stream.
10652         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
10654 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
10656         PR tree-optimization/57104
10657         * tsan.c (instrument_expr): Don't instrument accesses to
10658         DECL_HARD_REGISTER VAR_DECLs.
10660 2013-04-30  Richard Biener  <rguenther@suse.de>
10662         * function.h (loops_for_fn): New inline function.
10663         (set_loops_for_fn): Likewise.
10664         * cfgloop.h (place_new_loop): Add struct function parameter.
10665         (get_loop): Likewise.
10666         (get_loops): Likewise.
10667         (number_of_loops): Likewise.
10668         (fel_next): Adjust.
10669         (fel_init): Likewise.
10670         * cfg.c (get_loop_copy): Adjust.
10671         * cfgloop.c (flow_loops_dump): Likewise.
10672         (record_loop_exits): Likewise.
10673         (verify_loop_structure): Likewise.
10674         * cfgloopanal.c (mark_irreducible_loops): Likewise.
10675         (estimate_reg_pressure_cost): Likewise.
10676         (mark_loop_exit_edges): Likewise.
10677         * cfgloopmanip.c (place_new_loop): Likewise.
10678         (add_loop): Likewise.
10679         (duplicate_loop): Likewise.
10680         * graph.c (draw_cfg_nodes): Likewise.
10681         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
10682         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
10683         (extract_affine_chrec): Likewise.
10684         (build_scop_iteration_domain): Likewise.
10685         * graphite.c (graphite_initialize): Likewise.
10686         * ira-build.c (create_loop_tree_nodes): Likewise.
10687         (more_one_region_p): Likewise.
10688         (rebuild_regno_allocno_maps): Likewise.
10689         (mark_loops_for_removal): Likewise.
10690         (mark_all_loops_for_removal): Likewise.
10691         (remove_unnecessary_regions): Likewise.
10692         (ira_build): Likewise.
10693         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
10694         * loop-init.c (fix_loop_structure): Likewise.
10695         (gate_rtl_move_loop_invariants): Likewise.
10696         (gate_rtl_unswitch): Likewise.
10697         (gate_rtl_unroll_and_peel_loops): Likewise.
10698         (rtl_doloop): Likewise.
10699         * lto-streamer-in.c (input_cfg): Likewise.
10700         * lto-streamer-out.c (output_cfg): Likewise.
10701         * modulo-sched.c (sms_schedule): Likewise.
10702         * predict.c (tree_estimate_probability): Likewise.
10703         (tree_estimate_probability_driver): Likewise.
10704         (estimate_loops): Likewise.
10705         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
10706         (move_sese_region_to_fn): Likewise.
10707         (debug_loop_num): Likewise.
10708         * tree-chrec.c (chrec_evaluate): Likewise.
10709         (hide_evolution_in_other_loops_than_loop): Likewise.
10710         (chrec_component_in_loop_num): Likewise.
10711         (reset_evolution_in_loop): Likewise.
10712         (evolution_function_is_invariant_rec_p): Likewise.
10713         * tree-if-conv.c (main_tree_if_conversion): Likewise.
10714         * tree-inline.c (copy_loops): Likewise.
10715         (copy_cfg_body): Likewise.
10716         (tree_function_versioning): Likewise.
10717         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
10718         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
10719         Likewise.
10720         (add_to_evolution_1): Likewise.
10721         (scev_const_prop): Likewise.
10722         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
10723         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
10724         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
10725         (tree_ssa_lim_initialize): Likewise.
10726         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
10727         (verify_loop_closed_ssa): Likewise.
10728         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
10729         (tree_ssa_loop_im): Likewise.
10730         (tree_ssa_loop_unswitch): Likewise.
10731         (tree_vectorize): Likewise.
10732         (check_data_deps): Likewise.
10733         (tree_ssa_loop_ivcanon): Likewise.
10734         (tree_ssa_loop_bounds): Likewise.
10735         (tree_complete_unroll): Likewise.
10736         (tree_complete_unroll_inner): Likewise.
10737         (tree_parallelize_loops): Likewise.
10738         (tree_ssa_loop_prefetch): Likewise.
10739         (tree_ssa_loop_ivopts): Likewise.
10740         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
10741         * tree-vectorizer.c (vectorize_loops): Likewise.
10743 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
10745         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
10746         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
10747         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
10748         with EABI_LINK_SPEC.
10750 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10752         PR target/44578
10753         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
10754         alternative.
10756 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
10758         PR target/57097
10759         * lra-constraints.c (process_alt_operands): Discourage a bit more
10760         using memory for pseudos.  Print cost dump for alternatives.
10761         Modify cost values for conflicts with early clobbers.
10762         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
10764 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10766         PR target/57098
10767         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
10769 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10771         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
10772         from/to S register.
10773         (movdi_aarch64): Support LDR/STR from/to D register.
10775 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
10777         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
10778         or higher by default.
10780 2013-04-29  Richard Biener  <rguenther@suse.de>
10782         PR middle-end/57075
10783         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
10784         even if not adding abnormal edges for calls that can make
10785         abnormal gotos.
10787 2013-04-29  Richard Biener  <rguenther@suse.de>
10789         PR middle-end/57103
10790         * tree-cfg.c (move_stmt_op): Fix condition under which to update
10791         TREE_BLOCK.
10792         (move_stmt_r): Remove redundant checking.
10794 2013-04-29  Teresa Johnson  <tejohnson@google.com>
10796         PR bootstrap/57077
10797         * basic-block.h (apply_scale): New function.
10798         (apply_probability): Use apply_scale.
10799         * gimple-streamer-in.c (input_bb): Ditto.
10800         * lto-streamer-in.c (input_cfg): Ditto.
10801         * lto-cgraph.c (merge_profile_summaries): Ditto.
10802         * tree-optimize.c (execute_fixup_cfg): Ditto.
10803         * tree-inline.c (copy_bb): Update comment to use apply_scale.
10804         (copy_edges_for_bb): Ditto.
10805         (copy_cfg_body): Ditto.
10807 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10809         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
10810         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
10811         (tail_merge_optimize): Handle current_loops == NULL.
10813 2013-04-26  Jeff Law  <law@redhat.com>
10815         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
10816         (simplify_cond_using_ranges): Generalize code to simplify
10817         COND_EXPRs where one argument is a constant and the other
10818         is an SSA_NAME created by an integral type conversion.
10820 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10822         * config/arm/arm.md (store_minmaxsi): Use only when
10823         optimize_insn_for_size_p.
10825 2013-04-29  Christian Bruel  <christian.bruel@st.com>
10827         PR target/57108
10828         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
10830 2013-04-29  Richard Biener  <rguenther@suse.de>
10832         PR middle-end/57089
10833         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
10834         loop tree make sure to schedule a fixup for the child as well.
10835         (expand_omp_for_generic): Properly add loops.
10836         (expand_omp_for_static_nochunk): Likewise.
10837         (expand_omp_for_static_chunk): Likewise.
10838         (expand_omp_for): For the degenerate case fixup loops.
10839         (expand_omp_sections): Fix default bb placement in loops.
10840         (expand_omp_atomic_pipeline): Properly add loops.
10842 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
10844         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
10846 2013-04-29  Tom de Vries  <tom@codesourcery.com>
10848         * tree-ssa-tail-merge.c: Update header comment.
10850 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10852         * config/aarch64/arm_neon.h
10853         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
10854         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
10855         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
10856         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
10857         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
10858         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
10859         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
10860         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
10862 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10864         * config/aarch64/aarch64-simd.md
10865         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
10866         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
10867         fix_trunc, fixuns_trunc.
10868         (ftrunc<VDQF:mode>2): New.
10869         * config/aarch64/iterators.md (optab): Add fix, fixuns.
10870         (fix_trunc_optab): New.
10872 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10874         * config/aarch64/aarch64-builtins.c
10875         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
10876         iceilf, lround, iroundf.
10878 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
10880         PR target/54349
10881         * config/i386/i386.h (enum ix86_tune_indices)
10882         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
10883         New, split from X86_TUNE_INTER_UNIT_MOVES.
10884         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
10885         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
10886         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
10887         (TARGET_INTER_UNIT_MOVES): Remove.
10888         * config/i386/i386.c (initial_ix86_tune_features): Update.
10889         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
10890         (ix86_expand_convert_uns_didf_sse): Use
10891         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10892         (ix86_expand_vector_init_one_nonzero): Ditto.
10893         (ix86_expand_vector_init_interleave): Ditto.
10894         (inline_secondary_memory_needed): Return true for moves from SSE class
10895         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
10896         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
10897         * config/i386/constraints.md (Yi, Ym): Depend on
10898         TARGET_INTER_UNIT_MOVES_TO_VEC.
10899         (Yj, Yn): New constraints.
10900         * config/i386/i386.md (*movdi_internal): Change constraints of
10901         operand 1 from Yi to Yj and from Ym to Yn.
10902         (*movsi_internal): Ditto.
10903         (*movdf_internal): Ditto.
10904         (*movsf_internal): Ditto.
10905         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
10906         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
10907         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
10908         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
10909         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
10910         * config/i386/sse.md (movdi_to_sse): Ditto.
10911         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
10912         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
10913         TARGET_INTER_UNIT_MOVES.
10914         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
10915         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
10916         instead of TARGET_INTER_UNIT_MOVES.
10917         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
10918         operand 1 from Yi to Yj and from Ym to Yn.
10920 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10922         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
10923         (float_truncate_hi_): Likewise.
10924         (float_extend_lo_): Likewise.
10925         (float_truncate_lo_): Likewise.
10926         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
10927         (aarch64_float_extend_lo_v2df): Likewise.
10928         (vec_unpacks_hi_v4sf): Likewise.
10929         (aarch64_float_truncate_lo_v2sf): Likewise.
10930         (aarch64_float_truncate_hi_v4sf): Likewise.
10931         (vec_pack_trunc_v2df): Likewise.
10932         (vec_pack_trunc_df): Likewise.
10934 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10936         * config/aarch64/aarch64-builtins.c
10937         (aarch64_fold_builtin): Fold float conversions.
10938         * config/aarch64/aarch64-simd-builtins.def
10939         (floatv2si, floatv4si, floatv2di): New.
10940         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
10941         * config/aarch64/aarch64-simd.md
10942         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
10943         * config/aarch64/iterators.md (FLOATUORS): New.
10944         (optab): Add float, floatuns.
10945         (su_optab): Likewise.
10947 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10949         * config/aarch64/aarch64-builtins.c
10950         (aarch64_builtin_vectorized_function): Use new names for
10951         fcvt builtins.
10952         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
10953         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
10954         (fcvtzu): Split as...
10955         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
10956         (fcvtas): Split as...
10957         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
10958         (fcvtau): Split as...
10959         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
10960         (fcvtps): Split as...
10961         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
10962         (fcvtpu): Split as...
10963         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
10964         (fcvtms): Split as...
10965         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
10966         (fcvtmu): Split as...
10967         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
10968         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
10969         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
10970         (lfrintnusf, lfrintnudf): Likewise.
10971         * config/aarch64/aarch64-simd.md
10972         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
10973         define_insn.
10974         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
10975         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
10976         (fcvt_pattern): Likewise.
10978 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10980         * config/aarch64/aarch64-simd.md
10981         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
10982         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
10984 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10986         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
10987         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
10988         (vrnd<a,m,n,p>_f32): Implement using builtins.
10989         (vrnd<i,x><q>_f<32, 64>): New.
10991 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
10993         * config/aarch64/aarch64-builtins.c
10994         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
10995         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
10996         (frintz): Rename to...
10997         (btrunc): ...this.
10998         (frintp): Rename to...
10999         (ceil): ...this.
11000         (frintm): Rename to...
11001         (floor): ...this.
11002         (frinti): Rename to...
11003         (nearbyint): ...this.
11004         (frintx): Rename to...
11005         (rint): ...this.
11006         (frinta): Rename to...
11007         (round): ...this.
11008         * config/aarch64/aarch64-simd.md
11009         (aarch64_frint<frint_suffix><mode>): Delete.
11010         (<frint_pattern><mode>2): Convert to insn.
11011         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
11012         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
11013         (frint_pattern): Likewise.
11014         (frint_suffix): Likewise.
11016 2013-04-29  Richard Biener  <rguenther@suse.de>
11018         PR tree-optimization/57081
11019         * loop-init.c: Include tree-flow.h.
11020         (loop_optimizer_finalize): Free number of iteration estimates.
11021         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
11023 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
11025         PR tree-optimization/57083
11026         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
11027         non-singleton shift count range, zero extend low_bound for uns case.
11029         * config/i386/predicates.md (general_vector_operand): New predicate.
11030         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
11031         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
11032         if they aren't nonimmediate operands.  If their original values
11033         satisfy const_vector_equal_evenodd_p, don't shift them.
11034         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
11035         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
11036         if not nonimmediate_operand.
11037         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
11038         instead of register_operand.
11039         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
11041 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
11043         * stor-layout.c (finalize_size_functions): Allocate a structure and
11044         reset cfun before dumping the functions.
11046 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
11048         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
11050         PR target/56866
11051         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
11052         use xop_pmacsdqh if uns_p.
11053         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
11054         the immediate rotate count.
11056 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
11058         * rtl.h (struct rtx_def): Add comment for field jump.
11059         (LRA_SUBREG_P): New macro.
11060         * recog.c (register_operand): Check LRA_SUBREG_P.
11061         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11062         stack frame.
11063         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11064         (lra_final_code_change): Skip subreg change for operators.
11065         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11066         if there are no operand changes.
11067         * lra-constraints.c (curr_insn_set): New.
11068         (match_reload): Set LRA_SUBREG_P.
11069         (emit_spill_move): Ditto.
11070         (check_and_process_move): Use curr_insn_set. Process only single
11071         set insns.  Don't initialize sec_mem_p and change_p.
11072         (simplify_operand_subreg): Use LRA_SUBREG_P.
11073         (reg_in_class_p): New function.
11074         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11075         of #ifdef.  Add code to remove cycling.
11076         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11077         non-null disp.  Reload inner instead of disp when base and index
11078         are null.  Try to put lo_sum into register.
11079         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11080         (check_and_process_move): Move code for move cost check to
11081         simple_move_p.  Remove equiv_substitution.
11082         (simple_move_p): New function.
11083         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11084         curr_insn_set.  Call check_and_process_move only for single set
11085         insns.  Use the new function.  Move call of check_and_process_move
11086         after operand equiv substitution and address process.
11088 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11090         PR go/57045
11091         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
11092         with nonlocal goto receivers or returns twice calls, ignore
11093         unininitialized values from abnormal edges to nl goto receiver
11094         or returns twice call.
11096 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11098         PR tree-optimization/57051
11099         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
11100         and VEC_RSHIFT_EXPR if shift count is a multiple of element
11101         bitsize.
11103 2013-04-26  Richard Biener  <rguenther@suse.de>
11105         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
11106         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
11107         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
11108         (fixup_loop_arrays_after_move): New function.
11109         (move_sese_region_to_fn): Properly outline the loop tree parts
11110         of the SESE region.
11112 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
11114         * config/i386/i386.md (type, unit): Fix long lines.
11116 2013-04-26  Richard Biener  <rguenther@suse.de>
11118         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
11119         (lto-streamer-out.o): Likewise.
11120         * cfgloop.c (init_loops_structure): Export, add struct function
11121         argument and adjust.
11122         (flow_loops_find): Adjust.
11123         * cfgloop.h (enum loop_estimation): Add EST_LAST.
11124         (init_loops_structure): Declare.
11125         * lto-streamer-in.c: Include cfgloop.h.
11126         (input_cfg): Input the loop tree.
11127         * lto-streamer-out.c: Include cfgloop.h.
11128         (output_cfg): Output the loop tree.
11129         (output_struct_function_base): Do not drop PROP_loops.
11131 2013-03-26  Richard Biener  <rguenther@suse.de>
11133         * tree-cfg.c (execute_build_cfg): Build the loop tree.
11134         (pass_build_cfg): Provide PROP_loops.
11135         (move_sese_region_to_fn): Remove loops that are outlined into fn
11136         for now.
11137         * tree-inline.c: Include cfgloop.h.
11138         (initialize_cfun): Do not drop PROP_loops.
11139         (copy_loops): New function.
11140         (copy_cfg_body): Copy loop structure.
11141         (tree_function_versioning): Initialize destination loop tree.
11142         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
11143         (pass_parallelize_loops): Do IL verification.
11144         * loop-init.c (loop_optimizer_init): Fixup loops if required.
11145         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
11146         the CFG make sure we fixup loops as well.
11147         * tree-ssa-tail-merge.c: Include cfgloop.h.
11148         (replace_block_by): When merging loop latches mark loops for fixup.
11149         * lto-streamer-out.c (output_struct_function_base): Drop
11150         PROP_loops for now.
11151         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
11152         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
11153         * ipa-split.c: Include cfgloop.h.
11154         (split_function): Add the new return block to the loop tree root.
11155         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
11156         whether we have removed the forwarder block.
11157         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
11158         * cfgloop.h (place_new_loop): Declare.
11159         * cfgloopmanip.c (place_new_loop): Export.
11160         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
11161         (tree-switch-conversion.o): Likewise.
11162         (tree-complex.o): Likewise.
11163         (tree-inline.o): Likewise.
11164         (tree-ssa-tailmerge.o): Likewise.
11165         (ipa-split.o): Likewise.
11166         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
11167         (tree-ssa-copy.o): Likewise.
11168         * tree-switch-conversion.c: Include cfgloop.h
11169         (process_switch): If we emit a bit-test cascade, schedule loops
11170         for fixup.
11171         * tree-complex.c: Include cfgloop.h.
11172         (expand_complex_div_wide): Properly add new basic-blocks to loops.
11173         * asan.c: Include cfgloop.h.
11174         (create_cond_insert_point): Properly add new basic-blocks to
11175         loops, schedule loop fixup.
11176         * cfgloop.c (verify_loop_structure): Check that looks are not
11177         marked for fixup.
11178         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
11179         to loops.
11180         (expand_omp_for_generic): Likewise.
11181         (expand_omp_sections): Likewise.
11182         (expand_omp_atomic_pipeline): Schedule loops for fixup.
11183         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
11184         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
11185         is initialized, not when loops are present.
11186         * tree-parloops.c (parallelize_loops): Remove checking here.
11187         * passes.c (init_optimization_passes): Schedule a copy-propagation
11188         pass before complete unrolling of inner loops.
11190 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
11192         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
11193         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
11194         (colorize_init): Add argument to _WIN32 version.
11195         * toplev.c: Include diagnostic-color.h.
11196         (process_options): Default to -fdiagnostics-color=auto if
11197         GCC_COLORS env var is in the environment.
11198         * common.opt (fdiagnostics-color=): Add Var and Init.
11199         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
11200         env var is in the environment, the default is auto rather than never.
11202         * diagnostic.h (file_name_as_prefix): Add context argument.
11203         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
11204         the string as locus.
11205         * langhooks.c (lhd_print_error_function): Adjust caller.
11207 2013-04-25  Lawrence Crowl  <crowl@google.com>
11209         * var-tracking.c (shared_hash_def::htab):
11210         Change type to hash_table.  Update dependent calls and types.
11212 2013-04-25  Lawrence Crowl  <crowl@google.com>
11214         * Makefile.in: Update as needed below.
11216         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
11217         Move declaration to after the type's method definitons.
11219         * attribs.c (htab_t scoped_attributes::attribute_hash):
11220         Change type to hash_table.  Update dependent calls and types.
11222         * bitmap.c (htab_t bitmap_desc_hash):
11223         Change type to hash_table.  Update dependent calls and types.
11225         * cselib.c (htab_t cselib_hash_table):
11226         Change type to hash_table.  Update dependent calls and types.
11228         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
11229         (hash_string_slot_node): Move implementation into lto-streamer.h
11230         struct string_slot_hasher.
11231         (eq_string_slot_node): Likewise.
11233         * data-streamer-out.c: Update output_block::string_hash_table
11234         dependent calls and types.
11236         * dwarf2cfi.c (htab_t trace_index):
11237         Change type to hash_table.  Update dependent calls and types.
11239         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
11240         Change type to hash_table.  Update dependent calls and types.
11241         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
11242         (htab_t optimize_external_refs::map): Likewise.
11243         (htab_t output_comp_unit::extern_map): Likewise.
11244         (htab_t output_comdat_type_unit::extern_map): Likewise.
11245         (htab_t output_macinfo::macinfo_htab): Likewise.
11246         (htab_t optimize_location_lists::htab): Likewise.
11247         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
11249         * except.c (htab_t ehspec_hash_type):
11250         Change type to hash_table.  Update dependent calls and types.
11251         (assign_filter_values::ttypes): Likewise.
11252         (assign_filter_values::ehspec): Likewise.
11253         (sjlj_assign_call_site_values::ar_hash): Likewise.
11254         (convert_to_eh_region_ranges::ar_hash): Likewise.
11256         * gcse.c (htab_t pre_ldst_table):
11257         Change type to hash_table.  Update dependent calls and types.
11259         * ggc-common.c (htab_t saving_htab):
11260         Change type to hash_table.  Update dependent calls and types.
11261         (htab_t loc_hash): Likewise.
11262         (htab_t ptr_hash): Likewise.
11263         (call_count): Rename ggc_call_count.
11264         (call_alloc): Rename ggc_call_alloc.
11265         (loc_descriptor): Rename make_loc_descriptor.
11266         (add_statistics): Rename ggc_add_statistics.
11268         * ggc-common.c (saving_htab):
11269         Change type to hash_table.  Update dependent calls and types.
11271         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
11272         (push_gimplify_context): Likewise.
11273         (pop_gimplify_context): Likewise.
11274         (struct gimple_temp_hash_elt): Added.
11275         (struct gimplify_hasher): Likewise.
11276         (struct gimplify_ctx.temp_htab):
11277         Change type to hash_table.  Update dependent calls and types.
11279         * gimple-fold.c: Include gimplify-ctx.h.
11281         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
11282         Change type to hash_table.  Update dependent calls and types.
11283         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
11284         avoid potential global name collision.
11286         * gimplify.c: Include gimplify-ctx.h.
11287         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
11288         (htab_t gimplify_ctx::temp_htab):
11289         Update dependent calls and types for new type hash_table.
11290         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
11291         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
11293         * gimplify-ctx.h: New.
11294         (struct gimple_temp_hash_elt): Move from gimplify.c.
11295         (class gimplify_hasher): New.
11296         (struct gimplify_ctx): Move from gimple.h.
11297         (htab_t gimplify_ctx::temp_htab):
11298         Change type to hash_table.  Update dependent calls and types.
11300         * graphite-clast-to-gimple.c: Include graphite-htab.h.
11301         (htab_t ivs_params::newivs_index):
11302         Change type to hash_table.  Update dependent calls and types.
11303         (htab_t ivs_params::params_index): Likewise.
11304         (htab_t print_generated_program::params_index): Likewise.
11305         (htab_t gloog::newivs_index): Likewise.
11306         (htab_t gloog::params_index): Likewise.
11308         * graphite.c: Include graphite-htab.h.
11309         4htab_t graphite_transform_loops::bb_pbb_mapping):
11310         Change type to hash_table.  Update dependent calls and types.
11312         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
11313         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
11314         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
11316         * graphite-dependences.c: Include graphite-htab.h.
11317         (loop_is_parallel_p): Change hash table type of parameter.
11319         * graphite-htab.h: New.
11320         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
11321         (extern find_pbb_via_hash): Move from graphite-poly.h.
11322         (extern loop_is_parallel_p): Move from graphite-poly.h.
11323         (extern get_loop_body_pbbs): Move from graphite-poly.h.
11325         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
11326         (extern loop_is_parallel_p): Move to graphite-htab.h.
11327         (extern get_loop_body_pbbs): Move to graphite-htab.h.
11329         * haifa-sched.c (htab_t delay_htab):
11330         Change type to hash_table.  Update dependent calls and types.
11331         (htab_t delay_htab_i2): Likewise.
11333         * ira-color.c (htab_t allocno_hard_regs_htab):
11334         Change type to hash_table.  Update dependent calls and types.
11336         * ira-costs.c (htab_t cost_classes_htab):
11337         Change type to hash_table.  Update dependent calls and types.
11339         * loop-invariant.c (htab_t merge_identical_invariants::eq):
11340         Change type to hash_table.  Update dependent calls and types.
11342         * loop-iv.c (htab_t bivs):
11343         Change type to hash_table.  Update dependent calls and types.
11345         * loop-unroll.c (htab_t opt_info::insns_to_split):
11346         Change type to hash_table.  Update dependent calls and types.
11347         (htab_t opt_info::insns_with_var_to_expand): Likewise.
11349         * lto-streamer.h (struct string_slot): Move from data-streamer.h
11350         (struct string_slot_hasher): New.
11351         (htab_t output_block::string_hash_table):
11352         Change type to hash_table.  Update dependent calls and types.
11354         * lto-streamer-in.c (freeing_string_slot_hasher): New.
11355         (htab_t file_name_hash_table):
11356         Change type to hash_table.  Update dependent calls and types.
11358         * lto-streamer-out.c: Update output_block::string_hash_table dependent
11359         calls and types.
11361         * lto-streamer.c (htab_t tree_htab):
11362         Change type to hash_table.  Update dependent calls and types.
11364         * omp-low.c: Include gimplify-ctx.h.
11366         * passes.c (htab_t name_to_pass_map):
11367         Change type to hash_table.  Update dependent calls and types.
11368         (pass_traverse): Rename to passes_pass_traverse.
11370         * plugin.c (htab_t event_tab):
11371         Change type to hash_table.  Update dependent calls and types.
11373         * postreload-gcse.c (htab_t expr_table):
11374         Change type to hash_table.  Update dependent calls and types.
11375         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
11377         * sese.c (debug_rename_map_1): Make extern.
11378         (htab_t copy_bb_and_scalar_dependences::rename_map):
11379         Change type to hash_table.  Update dependent calls and types.
11381         * sese.h (extern debug_rename_map): Move to .c file.
11383         * store-motion.c (htab_t store_motion_mems_table):
11384         Change type to hash_table.  Update dependent calls and types.
11386         * trans-mem.c (htab_t tm_new_mem_hash):
11387         Change type to hash_table.  Update dependent calls and types.
11389         * tree-browser.c (htab_t TB_up_ht):
11390         Change type to hash_table.  Update dependent calls and types.
11392         * tree-cfg.c (htab_t discriminator_per_locus):
11393         Change type to hash_table.  Update dependent calls and types.
11395         * tree-complex.c: Include tree-hasher.h
11396         (htab_t complex_variable_components):
11397         Change type to hash_table.  Update dependent calls and types.
11399         * tree-eh.c (htab_t finally_tree):
11400         Change type to hash_table.  Update dependent calls and types.
11402         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
11403         struct int_tree_hasher.
11404         (extern int_tree_map_eq): Likewise.
11405         (uid_decl_map_hash): Removed.
11406         (extern decl_tree_map_eq): Likewise.
11408         * tree-hasher.h: New.
11409         (struct int_tree_hasher): New.
11410         (typedef int_tree_htab_type): New.
11412         * tree-inline.c: Include gimplify-ctx.h.
11414         * tree-mudflap.c: Include gimplify-ctx.h.
11416         * tree-parloops.c: Include tree-hasher.h.
11417         (htab_t eliminate_local_variables_stmt::decl_address):
11418         Change type to hash_table.  Update dependent calls and types.
11419         (htab_t separate_decls_in_region::decl_copies): Likewise.
11421         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
11422         Change type to hash_table.  Update dependent calls and types.
11424         * tree-sra.c (candidates):
11425         Change type to hash_table.  Update dependent calls and types.
11427         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
11428         in tree-flow.h.
11429         (int_tree_map_hash): Likewise.
11431         * tree-ssa-dom.c (htab_t avail_exprs):
11432         Change type to hash_table.  Update dependent calls and types.
11434         * tree-ssa-live.c (var_map_base_init::tree_to_index):
11435         Change type to hash_table.  Update dependent calls and types.
11437         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
11438         Change type to hash_table.  Update dependent calls and types.
11440         * tree-ssa-phiopt.c (seen_ssa_names):
11441         Change type to hash_table.  Update dependent calls and types.
11443         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
11444         Change type to hash_table.  Update dependent calls and types.
11446         * tree-ssa-uncprop.c (equiv):
11447         Change type to hash_table.  Update dependent calls and types.
11449 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
11451         PR rtl-optimization/57003
11452         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
11453         call note_stores with kill_clobbered_value callback again after
11454         killing regs_invalidated_by_call.
11456 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11458         * config/aarch64/aarch64-simd.md
11459         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
11460         (aarch64_simd_bsl<mode>): Likewise.
11461         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
11463 2013-04-25  Marek Polacek  <polacek@redhat.com>
11465         PR tree-optimization/57066
11466         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
11468 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
11470         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
11472 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11474         * config/aarch64/aarch64-builtins.c
11475         (aarch64_fold_builtin): New.
11476         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
11477         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
11478         * config/aarch64/aarch64-simd-builtins.def (abs): New.
11479         * config/aarch64/arm_neon.h
11480         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
11482 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11483             Tejas Belagod  <tejas.belagod@arm.com>
11485         * config/aarch64/aarch64-builtins.c
11486         (aarch64_gimple_fold_builtin): New.
11487         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
11488         * config/aarch64/aarch64-simd-builtins.def (addv): New.
11489         * config/aarch64/aarch64-simd.md (addpv4sf): New.
11490         (addvv4sf): Update.
11491         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
11493 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11495         * config/aarch64/aarch64.md
11496         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
11498 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
11500         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
11501         (*ngcsi_uxtw): New pattern.
11503 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11504             Julian Brown  <julian@codesourcery.com>
11506         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
11507         (TB_DREG): Add T_V4HF.
11508         (v4hf_UP): New macro.
11509         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11510         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11511         Handle initialisation of V4HF. Adjust initialisation of reinterpret
11512         built-ins.
11513         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
11514         (arm_vector_mode_supported_p): Handle V4HF.
11515         (arm_mangle_map): Handle V4HFmode.
11516         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
11517         * config/arm/arm_neon_builtins.def: Add entries for
11518         vcvtv4hfv4sf, vcvtv4sfv4hf.
11519         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
11520         (neon_vcvtv4hfv4sf): Likewise.
11521         * config/arm/neon-gen.ml: Handle half-precision floating point
11522         features.
11523         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
11524         * config/arm/arm_neon.h: Regenerate.
11525         * config/arm/neon.ml (type elts): Add F16.
11526         (type vectype): Add T_float16x4, T_floatHF.
11527         (type vecmode): Add V4HF.
11528         (type features): Add Requires_FP_bit feature.
11529         (elt_width): Handle F16.
11530         (elt_class): Likewise.
11531         (elt_of_class_width): Likewise.
11532         (mode_of_elt): Refactor.
11533         (type_for_elt): Handle F16, fix error messages.
11534         (vectype_size): Handle T_float16x4.
11535         (vcvt_sh): New function.
11536         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
11537         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
11538         (string_of_mode): Handle V4HF.
11539         * doc/arm-neon-intrinsics.texi: Regenerate.
11541 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
11543         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
11544         format specifier in 'X' case.
11546 2013-04-25  Alan Modra  <amodra@gmail.com>
11548         PR target/57052
11549         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
11550         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
11551         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
11552         Repeat for many other rotate/shift and mask patterns using subregs.
11553         Name lshiftrt insns.
11554         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
11555         on WORDS_BIG_ENDIAN.
11557 2013-04-25  Alan Modra  <amodra@gmail.com>
11559         * config.gcc: Support little-endian powerpc-linux targets.
11560         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
11561         (LINK_OS_LINUX_SPEC): Define.
11562         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
11563         Preserve MASK_LITTLE_ENDIAN.
11564         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
11565         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
11566         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
11567         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
11568         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
11569         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
11570         Correct fp word order for little-endian.  Don't shift toc entries
11571         smaller than a word for little-endian.
11572         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
11573         (bswapdi2 splits): Correct low-part subreg for little-endian.
11574         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
11575         low/high where such is correct only for be.
11576         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
11577         little-endian for -mcall-aixdesc.
11579 2013-04-25  Alan Modra  <amodra@gmail.com>
11581         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
11582         replace_equiv_address_nv.
11584 2013-04-25  Alan Modra  <amodra@gmail.com>
11586         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
11588 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11590         Revert:
11591         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11592         * rtl.h (struct rtx_def): ...
11594 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11596         PR rtl-optimizations/57046
11597         * lra-constraints (split_reg): Set up lra_risky_transformations_p
11598         for multi-reg splits.
11600 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
11602         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
11604 2013-04-24  Sterling Augustine  <saugustine@google.com>
11606         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
11607         (comp_dir_string, debug_str_dwo_section): New.
11608         (DEBUG_STR_DWO_SECTION): Rename to ...
11609         (DEBUG_DWO_STR_SECTION): ... this.
11610         (DEBUG_NORM_STR_SECTION): Delete.
11611         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
11612         (DEBUG_STR_DWO_SECTION_FLAGS): New.
11613         (find_AT_string): Move most logic to ...
11614         (find_AT_string_in_table): ... here.  New.
11615         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
11616         add_skeleton_AT_string.  Delete logic.
11617         (output_skeleton_debug_sections): Remove call to
11618         add_top_level_skeleton_die_attrs.
11619         (add_comp_dir_attribute): Move logic to comp_dir_string.
11620         (dwarf2out_init): Initialize debug_str_dwo_section.
11621         (output_indirect_string): Call find_string_form.
11622         (output_indirect_strings): Rewrite.
11623         (prune_unused_types): Empty skeleton_debug_str_hash.
11624         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
11625         (dwarf2out_finish):  Call output_indirect_strings.
11627 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11629         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
11631 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
11633         * rtl.h (struct rtx_def): Add comment for field jump.
11634         (LRA_SUBREG_P): New macro.
11635         * recog.c (register_operand): Check LRA_SUBREG_P.
11636         * lra.c (lra): Add note at the end of RTL code. Align non-empty
11637         stack frame.
11638         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
11639         (lra_final_code_change): Skip subreg change for operators.
11640         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
11641         if there are no operand changes.
11642         * lra-constraints.c (curr_insn_set): New.
11643         (match_reload): Set LRA_SUBREG_P.
11644         (emit_spill_move): Ditto.
11645         (check_and_process_move): Use curr_insn_set. Process only single
11646         set insns.  Don't initialize sec_mem_p and change_p.
11647         (simplify_operand_subreg): Use LRA_SUBREG_P.
11648         (reg_in_class_p): New function.
11649         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
11650         of #ifdef.  Add code to remove cycling.
11651         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
11652         non-null disp.  Reload inner instead of disp when base and index
11653         are null.  Try to put lo_sum into register.
11654         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
11655         (check_and_process_move): Move code for move cost check to
11656         simple_move_p.  Remove equiv_substitution.
11657         (simple_move_p): New function.
11658         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
11659         curr_insn_set.  Call check_and_process_move only for single set
11660         insns.  Use the new function.  Move call of check_and_process_move
11661         after operand equiv substitution and address process.
11663 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
11665         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
11666         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
11667         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
11669 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
11671         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
11673 2013-04-24  Marek Polacek  <polacek@redhat.com>
11675         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
11676         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
11677         (select_loops_exit_conditions): Likewise.
11678         (number_of_iterations_for_all_loops): Likewise.
11679         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
11680         (scev_analysis): Likewise.
11682 2013-04-02  Catherine Moore  <clm@codesourcery.com>
11683             Chao-ying Fu  <fu@mips.com>
11685         * config/mips/micromips.md (jraddiusp): New pattern.
11686         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
11687         instruction if possible.
11689 2013-04-24  Alan Modra  <amodra@gmail.com>
11691         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
11693 2013-04-24  Julian Brown  <julian@codesourcery.com>
11694             Chung-Lin Tang  <cltang@codesourcery.com>
11696         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
11697         dependency behavior in enumeration type DIE generation. Add TODO note
11698         to comments about future DW_FORM_sdata/udata re-work of related code.
11700 2013-04-23  Lawrence Crowl  <crowl@google.com>
11702         * Makefile.in: Update as needed below.
11704         * hash-table.h (class hash_table):
11705         Correct many methods with parameter types compare_type to the correct
11706         value_type.  (Correct code was unlikely to notice the change.)
11707         (hash_table::elements_with_deleted) New.
11708         (class hashtable::iterator): New.
11709         (hashtable::begin()): New.
11710         (hashtable::end()): New.
11711         (FOR_EACH_HASH_TABLE_ELEMENT): New.
11713         * statistics.c (statistics_hashes):
11714         Change type to hash_table.  Update dependent calls and types.
11716         * tree-into-ssa.c (var_infos):
11717         Change type to hash_table.  Update dependent calls and types.
11719         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
11720         Change type to hash_table.  Update dependent calls and types.
11722         * tree-ssa-loop-im.c (struct mem_ref.refs):
11723         Change type to hash_table.  Update dependent calls and types.
11725         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
11726         Change type to hash_table.  Update dependent calls and types.
11728         * tree-ssa-sccvn.c (vn_tables_s::nary):
11729         Change type to hash_table.  Update dependent calls and types.
11730         (vn_tables_s::phis): Likewise.
11731         (vn_tables_s::references): Likewise.
11733         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
11734         (vn_reference_eq): Update parameter and return types.
11736         * tree-ssa-structalias.c (pointer_equiv_class_table):
11737         Change type to hash_table.  Update dependent calls and types.
11738         (location_equiv_class_table): Likewise.
11740         * tree-vect-data-refs.c: Consequential changes for making
11741         peeling a hash_table.
11743         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
11744         (destroy_loop_vec_info): Dependent hash_table update.
11746         * tree-vectorizer.h (peeling_htab):
11747         Change type to hash_table.  Update dependent calls and types.
11749 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
11751         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
11752         to check the register content is equal or not.
11753         * lra-constraints.c (match_reload): Use lra_assign_reg_val
11754         to assign register content record.
11755         * lra-eliminations.c (update_reg_eliminate): Use
11756         lra_update_reg_val_offset to update register content offset.
11757         * lra-int.h (struct lra_reg): Add offset member.
11758         (lra_reg_val_equal_p): New static inline function.
11759         (lra_update_reg_val_offset): New static inline function.
11760         (lra_assign_reg_val): New static inline function.
11761         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
11762         to assign register content record.
11763         (initialize_lra_reg_info_element): Initial offset to zero.
11765 2013-04-23  Catherine Moore  <clm@codesourcery.com>
11767         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
11768         operands.  Record compression.
11770 2013-04-23  Xinliang David Li  <davidxl@google.com>
11772         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
11774 2013-04-23  Richard Biener  <rguenther@suse.de>
11776         PR middle-end/57036
11777         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
11778         parameter, only add abnormal goto edges from the copied body
11779         if the call could perform abnormal gotos.
11780         (copy_cfg_body): Adjust.
11782 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
11784         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
11786 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
11788         * coretypes.h (gimple_stmt_iterator): Add struct to make
11789         compatible with C.
11791 2013-04-23  Richard Biener  <rguenther@suse.de>
11793         PR tree-optimization/57026
11794         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
11795         from SSA names occuring in abnormal PHI nodes.
11797 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11799         * lto/lto.c (print_lto_report_1): Fix LTO report names.
11801 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11803         * lto/lto.c (print_lto_report_1): Declare early.
11804         (read_cgraph_and_symbols): Call print_lto_report_1 early.
11806 2013-04-22  Andi Kleen  <ak@linux.intel.com>
11808         * common.opt (-flto-report-wpa): Add.
11809         * doc/invoke.texi (-flto-report-wpa): Add.
11810         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
11811         (lto_main): dito.
11813 2013-04-22  Xinliang David Li  <davidxl@google.com>
11815         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
11816         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
11817         * Makefile.in: New dependency
11819             David Daney <ddaney.cavm@gmail.com>
11821         * configure.ac (gcc_cv_as_micromips_support):  Use the
11822         --fatal-warnings option.
11823         * configure: Regenerate.
11825 2013-04-22  Marek Polacek  <polacek@redhat.com>
11827         PR sanitizer/56990
11828         * tsan.c (instrument_expr): Don't instrument expression
11829         in case its size is zero.
11831 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
11833         PR target/57032
11834         Revert:
11835         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
11837         * config/alpha/alpha.c (TARGET_LRA_P): New define.
11839 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11841         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
11842         (gimple_stmt_iterator): New typedef.
11843         * gimple.h (gimple_stmt_iterator): Rename to...
11844         (gimple_stmt_iterator_d): ... This.
11845         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
11846         trees be valid for GIMPLE and GENERIC.
11847         (TARGET_GIMPLE_FOLD_BUILTIN): New.
11848         * gimple-fold.c (gimple_fold_call): Call target hook
11849         gimple_fold_builtin.
11850         * hooks.c (hook_bool_gsiptr_false): New.
11851         * hooks.h (hook_bool_gsiptr_false): New.
11852         * target.def (fold_stmt): New.
11853         * doc/tm.texi: Regenerate.
11855 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
11857         PR target/57018
11858         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
11859         a set sp if no stack realignment.
11861 2013-04-22  Nick Clifton  <nickc@redhat.com>
11863         * config.gcc (tilegx-linux): Extend extra_objs rather than
11864         overwriting it.
11865         (tilepro-linux): Likewise.
11867 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11869         * config/aarch64/aarch64-builtins.c
11870         (CF): Remove.
11871         (CF0, CF1, CF2, CF3, CF4, CF10): New.
11872         (VAR<1-12>): Add MAP parameter.
11873         (BUILTIN_*): Likewise.
11874         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
11875         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
11876         (aarch64_ushl_n<mode>): Likewise.
11877         (aarch64_sshr_n<mode>): Likewise.
11878         (aarch64_ushr_n<mode>): Likewise.
11879         (aarch64_<maxmin><mode>): Likewise.
11880         (aarch64_sqrt<mode>): Likewise.
11881         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
11882         (vshr<q>_n_*): Likewise.
11884 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
11886         * config/aarch64/aarch64-builtins.c
11887         (aarch64_simd_builtin_type_mode): Handle SF types.
11888         (sf_UP): Define.
11889         (BUILTIN_GPF): Define.
11890         (aarch64_init_simd_builtins): Handle SF types.
11891         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
11892         (frecps): Likewise.
11893         (frecpx): Likewise.
11894         * config/aarch64/aarch64-simd.md
11895         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
11896         (aarch64_frecpe<mode>): New.
11897         (aarch64_frecps<mode>): Likewise.
11898         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
11899         (v8type): Add frecp<esx>.
11900         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
11901         (aarch64_frecps<mode>): Likewise.
11902         * config/aarch64/iterators.md (FRECP): New.
11903         (frecp_suffix): Likewise.
11904         * config/aarch64/arm_neon.h
11905         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
11907 2013-04-22  Christian Bruel  <christian.bruel@st.com>
11909         PR target/56995
11910         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
11911         (REG_CLASS_NAMES): Idem.
11912         (REG_CLASS_CONTENTS): Idem.
11913         (REGCLASS_HAS_FP_REG): Idem.
11914         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
11915         (sh_conditional_register_usage): Idem.
11917 2013-04-21  Jeff Law  <law@redhat.com>
11919         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
11920         (ssa_forward_propagate_and_combine): Use it.
11922 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11924         * lra.c: Update the flow chart diagram.
11926 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
11928         PR rtl-optimization/56847
11929         * lra-constraints.c (process_alt_operands): Discourage alternative
11930         with non-matche doffsettable memory constraint fro memory with
11931         known offset.
11933 2013-04-19  Richard Biener  <rguenther@suse.de>
11935         PR tree-optimization/56982
11936         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
11937         function.
11938         * gimplify.c (gimplify_call_expr): Notice special calls.
11939         (gimplify_modify_expr): Likewise.
11940         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
11941         abnormal control flow receivers.
11942         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
11943         in the same way as cfun->has_nonlocal_labels.
11944         (gimple_purge_dead_abnormal_call_edges): Likewise.
11945         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
11946         receivers start a basic-block.
11948 2013-04-19  Richard Biener  <rguenther@suse.de>
11950         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
11951         member ...
11952         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
11953         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
11954         (SLP_TREE_LOAD_PERMUTATION): Add.
11955         (vect_transform_slp_perm_load): Adjust prototype.
11956         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
11957         (vect_free_slp_instance): Likewise.
11958         (vect_create_new_slp_node): Likewise.
11959         (vect_supported_slp_permutation_p): Remove.
11960         (vect_slp_rearrange_stmts): Adjust.
11961         (vect_supported_load_permutation_p): Likewise.  Inline
11962         vect_supported_slp_permutation_p here.
11963         (vect_analyze_slp_instance): Compute load permutations per
11964         slp node instead of per instance.
11965         (vect_get_slp_defs): Adjust.
11966         (vect_transform_slp_perm_load): Likewise.
11967         (vect_schedule_slp_instance): Remove redundant code.
11968         (vect_schedule_slp): Remove hack for PR56270, add it ...
11969         * tree-vect-stmts.c (vectorizable_load): ... here, do not
11970         CSE loads for SLP.  Adjust.
11972 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11974         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
11975         spelling in two comments.
11977 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
11979         PR target/56797
11980         * config/arm/arm.c (load_multiple_sequence): Require SP
11981         as base register for loads if SP is in the register list.
11983 2013-04-19  Martin Jambor  <mjambor@suse.cz>
11985         PR tree-optimization/56718
11986         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
11987         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
11988         and made public.  Adjusted all callers.
11989         (ipa_intraprocedural_devirtualization): New function.
11990         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
11991         (ipa_intraprocedural_devirtualization): Likewise.
11992         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
11994 2013-04-19  Richard Biener  <rguenther@suse.de>
11996         PR tree-optimization/57000
11997         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
11999 2013-04-19  Terry Guo  <terry.guo@arm.com>
12001         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
12002         Replace with ...
12003         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
12004         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
12005         (cortex_m4_fmacs): Use new reservations.
12006         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
12008 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
12010         PR rtl-optimization/56999
12011         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
12012         related code.
12013         (lra_coalesce): Remove split_origin_bitmap and related code.
12014         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
12015         ranges if necessary.
12017 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
12019         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
12020         New array.
12021         (ix86_expand_call): Remove clobbered_registers array and use
12022         x86_64_ms_sysv_extra_clobbered_registers instead.
12023         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
12024         Declare here.
12025         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
12026         predicate.
12027         * config/i386/i386.md (*call_rex64_ms_sysv): Use
12028         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
12029         (*call_value_rex64_ms_sysv): Ditto.
12031 2013-04-18  Cary Coutant  <ccoutant@google.com>
12033         * dwarf2out.c (output_pubnames): Check die_perennial_p of
12034         parent instead of die_mark.
12036 2013-04-18  Diego Novillo  <dnovillo@google.com>
12038         * gimple.c (create_gimple_tmp): New.
12039         (get_expr_type): New.
12040         (build_assign): New.
12041         (build_type_cast): New.
12042         * gimple.h (enum ssa_mode): Define.
12043         (gimple_seq_set_location): New.
12044         * asan.c (build_check_stmt): Change some gimple_build_* calls
12045         to use build_assign and build_type_cast.
12047 2013-04-18  Richard Biener  <rguenther@suse.de>
12049         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
12050         handle negative step.  Remove redundant checks.
12051         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
12052         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
12053         for negative step and grouped loads fail to vectorize.
12055 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12057         * emit-rtl.c (reset_insn_used_flags): New function.
12058         (reset_all_used_flags): Use it.
12059         (verify_insn_sharing): New function.
12060         (verify_rtl_sharing): Fix verification for SEQUENCEs.
12062 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
12064         PR tree-optimization/56984
12065         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
12066         and (x >> M) >= N don't register any assertion if N << M is the
12067         minimum value.
12069 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
12071         * lower-subreg.c (resolve_simple_move): If called self-recursive,
12072         do not delete_insn insns that have not yet been emitted, only
12073         unlink them with remove_insn.
12074         * df-scan.c (df_insn_delete): Revert r197492.
12076 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
12078         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
12079         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
12081 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12083         * config/arm/arm.md (movsicc_insn): Convert define_insn into
12084         define_insn_and_split.
12085         (and_scc,ior_scc,negscc): Likewise.
12086         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
12088 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
12090         * config/arm/arm.c (use_return_insn): Return 0 for targets that
12091         can benefit from using a sequence of LDRD instructions in epilogue
12092         instead of a single LDM instruction.
12094 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12096         PR 45688
12097         * doc/extend.texi: Fix typo.
12099 2013-04-17  Richard Biener  <rguenther@suse.de>
12101         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
12102         (vect_build_slp_tree): ... here.
12103         (vect_build_slp_tree_1): Compute which stmts of the SLP group
12104         match.  Remove special-casing of mismatched complex loads.
12105         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
12106         re-try the match with swapped commutative operands.
12107         (vect_supported_load_permutation_p): Remove special-casing of
12108         mismatched complex loads.
12109         (vect_analyze_slp_instance): Adjust.
12111 2013-04-17  Richard Biener  <rguenther@suse.de>
12113         PR rtl-optimization/56921
12114         * cfgloop.h (struct loop): Add simple_loop_desc member.
12115         (struct niter_desc): Mark with GTY(()).
12116         (simple_loop_desc): Do not use aux field but simple_loop_desc.
12117         * loop-iv.c (get_simple_loop_desc): Likewise.
12118         (free_simple_loop_desc): Likewise.
12120         Revert
12121         2013-04-16  Richard Biener  <rguenther@suse.de>
12123         PR rtl-optimization/56921
12124         * loop-init.c (pass_rtl_move_loop_invariants): Add
12125         TODO_do_not_ggc_collect to todo_flags_finish.
12126         (pass_rtl_unswitch): Same.
12127         (pass_rtl_unroll_and_peel_loops): Same.
12128         (pass_rtl_doloop): Same.
12130 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
12132         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
12133         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
12134         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
12135         references.
12136         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
12137         * tree-streamer.c (record_common_node): Adjust reference in comment.
12139 2013-04-17  Terry Guo  <terry.guo@arm.com>
12141         * config/arm/cortex-m4.md: Add a new bypass.
12143 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12145         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
12146         New pattern.
12147         (*subs_<optab><mode>_multp2): New pattern.
12148         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12149         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
12151 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12153         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
12154         (*subs_mul_imm_<mode>): New pattern.
12156 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
12158         PR target/56948
12159         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
12160         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
12161         (vsx_movti_32bit): Change j->wa to O->wa.
12163 2013-04-16  Richard Biener  <rguenther@suse.de>
12165         PR rtl-optimization/56921
12166         * loop-init.c (pass_rtl_move_loop_invariants): Add
12167         TODO_do_not_ggc_collect to todo_flags_finish.
12168         (pass_rtl_unswitch): Same.
12169         (pass_rtl_unroll_and_peel_loops): Same.
12170         (pass_rtl_doloop): Same.
12172 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
12174         * config/arm/arm.c (emit_multi_reg_push): New declaration
12175         for an existing function.
12176         (arm_emit_strd_push): New function.
12177         (arm_expand_prologue): Used here.
12178         (arm_emit_ldrd_pop): New function.
12179         (arm_expand_epilogue): Used here.
12180         (arm_get_frame_offsets): Update condition.
12181         (arm_emit_multi_reg_pop): Add a special case for load of a single
12182         register with writeback.
12184 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
12186         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
12187         description.
12189 2013-04-16  Richard Biener  <rguenther@suse.de>
12191         PR tree-optimization/56756
12192         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
12193         (first_mem_ref_loc): New.
12194         (execute_sm): Place the load temporarily before a previous
12195         access instead of in the latch edge to ensure its SSA dependencies
12196         are defined at points dominating the load.
12198 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
12200         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
12201         correct fix by moving header and footer insn to the footer of
12202         the merged basic block.  Clear BB_END of the merged-away block.
12204         PR middle-end/43631
12205         * emit-rtl.c (make_note_raw): New function.
12206         (link_insn_into_chain): New static inline function.
12207         (add_insn): Use it.
12208         (add_insn_before, add_insn_after): Factor insn chain linking code...
12209         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
12210         using link_insn_into_chain.
12211         (note_outside_basic_block_p): New helper function for emit_note_after
12212         and emit_note_before.
12213         (emit_note_after): Use nobb variant of add_insn_after if the note
12214         should not be contained in a basic block.
12215         (emit_note_before): Use nobb variant of add_insn_before if the note
12216         should not be contained in a basic block.
12217         (emit_note_copy): Use make_note_raw.
12218         (emit_note): Likewise.
12219         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
12220         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
12221         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
12222         the moved barrier the tail of the basic block it follows.
12223         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
12225 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
12227         PR tree-optimization/56962
12228         * gimple-ssa-strength-reduction.c (record_increment): Only set
12229         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
12230         either rhs1 or rhs2 is equal to c->base_expr.
12232 2013-04-15  Richard Biener  <rguenther@suse.de>
12234         PR tree-optimization/56933
12235         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
12236         member.
12237         (GROUP_READ_WRITE_DEPENDENCE): Remove.
12238         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
12239         * tree-vect-data-refs.c (vect_analyze_group_access): Move
12240         dependence check ...
12241         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
12242         ... here.
12243         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
12244         GROUP_READ_WRITE_DEPENDENCE.
12246 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12248         * emit-rtl.c (reset_all_used_flags): New function.
12249         (verify_rtl_sharing): Call reset_all_used_flags before and after
12250         performing the checks.
12252 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12254         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
12255         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
12256         * config/arm/constraints.md (De): New constraint.
12257         * config/arm/neon.md (anddi3_neon): Delete.
12258         (neon_vand<mode>): Expand to standard anddi3 pattern.
12259         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
12260         Move earlier in the file.
12261         (neon_inv_logic_op2): Likewise.
12262         (arm_anddi_operand_neon): New predicate.
12264 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
12266         * configure.ac (gcc_cv_ld_as_needed): Set
12267         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
12268         Use -z ignore, -z record on *-*-solaris2*.
12269         (HAVE_LD_AS_NEEDED): Update comment.
12270         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
12271         * configure: Regenerate.
12272         * config.in: Regenerate.
12273         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
12274         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
12275         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
12276         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
12277         equivalents.  Fix markup.
12278         * doc/tm.texi: Regenerate.
12280 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
12282         * config/i386/i386.opt: New option mstack-protector-guard=.
12283         * config/i386/i386-opts.h: Add enum stack_protector_guard.
12284         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
12285         TARGET_SSP_TLS_GUARD.
12286         * config/i386/i386.c (ix86_option_override_internal): Set
12287         ix86_stack_protector_guard.
12288         * config/i386/i386.md (stack_protect_set): Enable for
12289         TARGET_SSP_TLS_GUARD only.
12290         (stack_protect_set_<mode>): Ditto.
12291         (stack_protect_test): Ditto.
12292         (stack_protect_test_<mode>): Ditto.
12293         * doc/invoke.texi (i386 Option): Document.
12295 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
12297         PR target/56890
12298         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
12299         (S_MODES): Set H_MODE bit.
12300         (SF_MODES): Set only S_MODE and SF_MODE bits.
12301         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
12302         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
12303         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
12304         <MODE_FLOAT>: Likewise.
12306 2013-04-15  Joey Ye  <joey.ye@arm.com>
12308         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
12310 2013-04-15  Joey Ye  <joey.ye@arm.com>
12312         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
12313         for real far jump.
12314         (thumb_far_jump_used_p): Count instruction size and set
12315         far_jump_used.
12317 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
12319         * reorg.c (fill_simple_delay_slots): Reindent block of code.
12320         * resource.c (mark_target_live_regs): Reformat conditional block.
12322 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12324         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
12325         notes, they are emitted only just before final.
12326         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
12328 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
12330         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
12331         * cfgrtl.c (delete_insn): Call it here instead.
12332         * lra-spills.c (lra_final_code_change): Use delete_insn.
12333         * haifa-sched.c (sched_remove_insn): Likewise.
12334         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
12335         returning to the nop pool.
12336         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
12337         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
12339 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12341         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
12342         * doc/tm.texi: Regenerated.
12344 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
12346         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
12347         QImode checks.
12349 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
12351         * df-core.c (df_find_def): Compare register numbers.
12352         (df_find_use): Likewise.
12354 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
12356         PR target/56903
12357         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
12358         lra_in_progress for return.
12360 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12362         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
12363         define_insn into define_insn_and_split and emit movsicc patterns.
12365 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
12367         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
12369 2013-04-12  Richard Biener  <rguenther@suse.de>
12371         * tree-pass.h (TODO_do_not_ggc_collect): New.
12372         * passes.c (execute_one_ipa_transform_pass): Honor
12373         TODO_do_not_ggc_collect.
12374         (execute_one_pass): Likewise.
12376         Revert
12377         2013-04-10  Richard Biener  <rguenther@suse.de>
12379         * passes.c (init_optimization_passes): Remove reload pass.
12380         * ira.c (do_reload): Merge into ...
12381         (ira): ... this.
12382         (rest_of_handle_reload): Remove.
12383         (pass_reload): Likewise.
12384         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12385         of reload for vzeroupper pass placement.
12387 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
12389         PR tree-optimization/56918
12390         PR tree-optimization/56920
12391         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
12392         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
12393         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
12394         use wide_mul_with_sign method.
12396 2013-04-12  Richard Biener  <rguenther@suse.de>
12398         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
12399         not be considered a gimple constant.
12401 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
12403         * fold-const.c (const_binop): Handle vector shifts by a scalar.
12404         (fold_binary_loc): Call const_binop also for mixed vector-scalar
12405         operations.
12407 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
12408             Jakub Jelinek  <jakub@redhat.com>
12410         * opts.c: Include diagnostic-color.h.
12411         (common_handle_option): Handle OPT_fdiagnostics_color_.
12412         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
12413         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
12414         (diagnostic-color.o): New.
12415         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
12416         (diagnostic_color_rule): New enum.
12417         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
12418         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
12419         the location string.
12420         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
12421         either NULL, or color kind.
12422         * diagnostic-color.c: New file.
12423         * diagnostic-color.h: New file.
12424         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
12425         arguments.
12426         * doc/invoke.texi (-fdiagnostics-color): Document.
12427         * pretty-print.h (pp_show_color): Define.
12428         (struct pretty_print_info): Add show_color field.
12429         * diagnostic.c: Include diagnostic-color.h.
12430         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
12431         macros.  Colorize error:, warning: etc. strings and also the location
12432         string.
12433         (diagnostic_show_locus): Colorize the caret line.
12434         * pretty-print.c: Include diagnostic-color.h.
12435         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
12436         inside of %< %> quotes or quoted through q format modifier.
12438 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12440         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
12442 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12444         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
12445         code in CC_NZ mode.
12446         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
12447         pattern.
12449 2013-04-11  Marek Polacek  <polacek@redhat.com>
12451         PR tree-optimization/48184
12452         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
12454 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
12456         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
12457         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
12458         (skip_simple_arithmetic): Tidy up.
12459         * tree.h (skip_simple_constant_arithmetic): Declare.
12461 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
12463         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
12465 2013-04-11  Richard Biener  <rguenther@suse.de>
12467         * tree-vect-loop.c (get_initial_def_for_induction): Properly
12468         generate vector constants.
12470 2013-04-11  Richard Biener  <rguenther@suse.de>
12472         PR tree-optimization/56878
12473         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
12474         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
12475         New function.
12476         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
12477         Prefer to align the DR with the most invariant base address.
12479 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12481         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
12482         comment.
12484 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
12486         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
12487         floating-point vector comparisons against 0.
12489 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
12491         PR tree-optimization/56899
12492         * fold-const.c (extract_muldiv_1): Apply distributive law
12493         only if TYPE_OVERFLOW_WRAPS (ctype).
12495 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
12497         PR target/56124
12498         * ira-costs.c (scan_one_insn): Check whether the source rtx of
12499         loading has side effect.
12501 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
12503         * config/sparc/sparc.c: Include tree-pass.h.
12504         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
12505         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
12506         head of file.  Change return type.  Split off gate function.
12507         (sparc_gate_work_around_errata): New function.
12508         (pass_work_around_errata): New pass definition.
12509         (insert_pass_work_around_errata) New pass insert definition to
12510         insert pass_work_around_errata just after delayed-branch scheduling.
12511         (sparc_option_override): Insert the pass.
12512         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
12514 2013-04-10  David S. Miller  <davem@davemloft.net>
12516         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
12517         or -mcpu=hypersparc.
12519         * target.def (cstore_mode): New hook.
12520         * target.h: Include insn-codes.h
12521         * targhooks.c: Likewise.
12522         (default_cstore_mode): New function.
12523         * targhooks.h: Declare it.
12524         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
12525         * doc/tm.texi: Rebuild.
12526         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
12527         target hook, rather than inspecting the insn_data.
12528         * config/sparc/sparc.c (sparc_cstore_mode): New function.
12529         (TARGET_CSTORE_MODE): Redefine.
12530         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
12531         result patterns.
12532         * config/sparc/predicates.md (cstore_result_operand): New special
12533         predicate.
12534         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
12535         Use it for operand 0.
12536         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
12537         (*snesi_special): Likewise.
12538         (*snesi_zero): Likewise.
12539         (*seqsi_zero): Likewise.
12540         (*sltu_insn): Likewise.
12541         (*sgeu_insn): Likewise.
12542         (*seqdi_special): Make operand 0 and comparison operation be of
12543         DImode.
12544         (*snedi_special): Likewise.
12545         (*snedi_special_vis3): Likewise.
12546         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
12547         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
12548         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
12549         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
12550         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
12551         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
12552         (*sltu_extend_sp64): Likewise.
12553         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
12554         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
12555         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
12556         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
12557         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
12559 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
12561         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
12562         (aarch64_start_file): Use the new function.
12564 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
12565             Jason Merrill  <jason@redhat.com>
12567         * common.opt: Add -gdwarf.
12568         * opts.c (common_handle_option): Handle it.
12569         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
12571 2013-04-10  Richard Biener  <rguenther@suse.de>
12573         * passes.c (execute_todo): Do not call ggc_collect conditional here.
12574         (execute_one_ipa_transform_pass): But unconditionally here.
12575         (execute_one_pass): And here.
12576         (init_optimization_passes): Remove reload pass.
12577         * tree-pass.h (TODO_ggc_collect): Remove.
12578         (pass_reload): Likewise.
12579         * ira.c (do_reload): Merge into ...
12580         (ira): ... this.
12581         (rest_of_handle_reload): Remove.
12582         (pass_reload): Likewise.
12583         * config/i386/i386.c (ix86_option_override): Refer to ira instead
12584         of reload for vzeroupper pass placement.
12585         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
12586         and todo_flags_finish of all passes.
12588 2013-04-10  Richard Biener  <rguenther@suse.de>
12590         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
12591         first_const_oprnd field, rename first_def_type to first_op_type.
12592         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
12593         (vect_get_and_check_slp_defs): Always use the type of the
12594         operand.  Allow mixed vect_external_def, vect_constant_def types.
12595         (vect_get_constant_vectors): Handle mixed vect_external_def,
12596         vect_constant_def types.
12598 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
12600         PR tree-optimization/55524
12601         * tree-ssa-math-opts.c
12602         (convert_mult_to_fma): Don't use an fms construct
12603         when we don't have an fms operation, but fnma, and it looks
12604         likely that we'll be able to use the latter.
12606 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
12608         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
12609         function.
12610         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
12611         inline fail caused by overwritable functions.
12613 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
12615         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
12616         unnecessary bits in the constant power of two case.
12618 2013-04-10  Richard Biener  <rguenther@suse.de>
12620         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
12621         broken code swapping operands.
12622         (vect_build_slp_tree): Do not compute load permutations here.
12623         (vect_analyze_slp_instance): Compute load permutations here,
12624         after building the SLP tree.
12626 2013-04-09  Christian Bruel  <christian.bruel@st.com>
12628         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
12629         of next/prev_real_insn.
12631 2013-04-09  Jan Hubicka  <jh@suse.cz>
12633         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
12634         Drop aliased parameter.
12635         (function_and_variable_visibility): Do not handle alias pairs.
12636         * cgraph.c (varpool_externally_visible_p): Update prototype.
12637         * varpool.c (varpool_add_new_variable): Update.
12639 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12641         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
12643 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
12645         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
12647         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
12649 2013-04-09  Marek Polacek  <polacek@redhat.com>
12651         PR tree-optimization/48762
12652         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
12654 2013-04-09  Richard Biener  <rguenther@suse.de>
12656         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
12657         dealing with cost.
12658         (vect_build_slp_tree): Likewise.
12659         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
12660         calculating the cost of a SLP instance.
12661         (vect_analyze_slp_instance): Use it from here, after building
12662         the SLP tree.
12664 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12666         PR middle-end/56883
12667         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
12668         expand_omp_for_static_chunk): Use simple_p = true in
12669         force_gimple_operand_gsi calls when assigning to addressable decls.
12671 2013-04-09  Jeff Law  <law@redhat.com>
12673         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
12674         when the boolean was created by converting a wider object which
12675         had a boolean range.
12677 2013-04-09  Richard Biener  <rguenther@suse.de>
12679         * tree-vectorizer.h (slp_void_p): Remove.
12680         (slp_tree): Typedef before _slp_tree declaration.
12681         (struct _slp_tree): Use a vector of slp_tree as children.
12682         (vect_get_place_in_interleaving_chain): Remove.
12683         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
12684         Move ...
12685         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
12686         and make static.
12687         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
12688         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
12689         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
12690         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
12691         Use slp_node instead of slp_void_p and adjust.
12693 2013-04-09  Richard Biener  <rguenther@suse.de>
12695         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
12696         work that is not necessary.
12698 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
12700         PR tree-optimization/56854
12701         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
12702         forward into clobber stmts if it would change MEM_REF lhs into
12703         non-MEM_REF.
12705 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
12707         * tree.c (type_hash_lookup, type_hash_add): Make static.
12708         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
12710 2013-04-09  Richard Biener  <rguenther@suse.de>
12712         * tree.h (unsave_expr_now): Remove.
12713         * tree-inline.c (mark_local_for_remap_r): Remove.
12714         (unsave_expr_1): Likewise.
12715         (unsave_r): Likewise.
12716         (unsave_expr_now): Likewise.
12717         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
12718         (propagate_tree_value): Likewise.
12720 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
12722         * doc/rtl.texi (sequence): Rewrite documentation to match the
12723         current use of SEQUENCE rtl objects.
12724         * rtl.def (SEQUENCE): Likewise.
12726         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
12727         Update documentation.
12728         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
12729         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
12731         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
12733 2013-04-08  Teresa Johnson  <tejohnson@google.com>
12735         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
12736         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
12737         methods.
12738         (estimate_edge_size_and_time): Add comment to suggest using rounding
12739         methods.
12740         (estimate_node_size_and_time): Ditto.
12741         (remap_edge_change_prob): Use helper rounding divide methods.
12742         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
12743         (gimple_mod_pow2_value_transform): Ditto.
12744         (gimple_mod_subtract_transform): Ditto.
12745         (gimple_ic_transform): Ditto.
12746         (gimple_stringops_transform): Ditto.
12747         * stmt.c (conditional_probability): Ditto.
12748         (emit_case_dispatch_table): Ditto.
12749         * lto-cgraph.c (merge_profile_summaries): Ditto.
12750         * tree-optimize.c (execute_fixup_cfg): Ditto.
12751         * cfgcleanup.c (try_forward_edges): Ditto.
12752         * cfgloopmanip.c (scale_loop_profile): Ditto.
12753         (loopify): Ditto.
12754         (duplicate_loop_to_header_edge): Ditto.
12755         (lv_adjust_loop_entry_edge): Ditto.
12756         * tree-vect-loop.c (vect_transform_loop): Ditto.
12757         * profile.c (compute_branch_probabilities): Ditto.
12758         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
12759         * lto-streamer-in.c (input_cfg): Ditto.
12760         * gimple-streamer-in.c (input_bb): Ditto.
12761         * ipa-cp.c (update_profiling_info): Ditto.
12762         (update_specialized_profile): Ditto.
12763         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
12764         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
12765         rounding methods.
12766         * sched-rgn.c (compute_dom_prob_ps): Ditto.
12767         (compute_trg_info): Ditto.
12768         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
12769         (purge_dead_edges): Ditto.
12770         * loop-unswitch.c (unswitch_loop): Ditto.
12771         * cgraphclones.c (cgraph_clone_edge): Ditto.
12772         (cgraph_clone_node): Ditto.
12773         * tree-inline.c (copy_bb): Ditto.
12774         (copy_edges_for_bb): Ditto.
12775         (initialize_cfun): Ditto.
12776         (copy_cfg_body): Ditto.
12777         (expand_call_inline): Ditto.
12779 2013-04-08  Kai Tietz  <ktietz@redhat.com>
12781         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
12782         TARGET_CYGWIN64 by TARGET_64BIT.
12784 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12786         * config/epiphany/epiphany.md (GPR_1): New constant.
12787         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
12788         * config/epiphany/epiphany.c (gen_compare_reg):
12789         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
12790         is already in place.
12791         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
12792         Don't require being called during rtl expansion; If y operlaps r0,
12793         return 0.
12794         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
12795         (epiphany_expand_epilogue): Likewise.
12797         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
12798         Don't use CC_FPmode for ORDERED / UNORDERED.
12799         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
12801         * config/epiphany/constraints.md (CnL): New constraint.
12802         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
12803         * config/epiphany/predicates.md (add_operand): Allow 1024.
12805         * config/epiphany/epiphany.md (logical_op): New code iterator.
12806         (op_mnc): New code attribute.
12807         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
12808         (mov_f+1, mov_f+2): New peephole2 patterns.
12810         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
12811         (cstoresi4): Also allow re-use of zero result when doing a NE
12812         comparison to a non-zero operand.
12813         Use (clobber (scratch)) for first insn if the gpr output is not needed.
12815         * config/epiphany/epiphany.md (<insn_opname>v2si3):
12816         Use gen_addsi3_i / gen_subsi3_i.
12818 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12820         PR c++/34949
12821         PR c++/50243
12822         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
12823         contain anything but clobbers, at most one __builtin_stack_restore,
12824         optionally debug stmts and final resx, and if it has at least one
12825         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
12826         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
12827         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
12828         which isn't defaut definition, remove them.
12829         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
12830         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
12831         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
12832         with MEM_REF LHS with SSA_NAME address.
12834 2013-04-08  Jeff Law  <law@redhat.com>
12836         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
12838 2013-04-08  Richard Biener  <rguenther@suse.de>
12840         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
12841         extra newline.
12842         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
12843         determined vector type.
12844         (vect_analyze_data_refs): Likewise.
12845         (vect_get_new_vect_var): Adjust.
12846         (vect_create_destination_var): Preserve SSA name versions.
12847         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
12848         not dump anything here.
12850 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
12852         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
12853         Add member lr_slot_known.
12854         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
12855         if necessary.
12856         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
12857         Remove code that sets lr_slot_offset according to what a previous
12858         version of epiphany_emit_save_restore used to do.
12859         (epiphany_emit_save_restore): When doing an lr save or restore,
12860         set/verify lr_slot_known and lr_slot_offset.
12862 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
12864         PR target/54338
12865         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
12866         in ALL_REGS.
12868 2013-04-08  Richard Biener  <rguenther@suse.de>
12870         * alias.c (find_base_term): Fix thinko in previous change.
12872 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
12874         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
12875         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
12876         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
12877         if possible to compute val.
12878         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
12879         For QImode integers don't require anything about precision.  Use
12880         const_with_all_bytes_same to find out if the constant doesn't have
12881         repeated bytes in it.
12883 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12885         * config/s390/s390.c (s390_expand_insv): Only accept insertions
12886         within mode size.
12888 2013-04-08  Marek Polacek  <polacek@redhat.com>
12890         PR rtl-optimization/48182
12891         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
12892         value to 1.
12894 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
12896         PR target/55487
12897         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
12898         nuses, make sure we have a label.
12900 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
12902         PR target/56843
12903         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
12904         (rs6000_emit_swdiv_low_precision): Remove.
12905         (rs6000_emit_swdiv): Rewrite to handle between one and four
12906         iterations of Newton-Raphson generally; modify required number of
12907         iterations for some cases.
12908         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
12910 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
12912         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
12913         set-but-unused variable.
12915         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
12916         basic blocks of released function bodies garbage-collectable.
12918         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
12919         (struct rtl_opt_pass): Add TODO_df_finish.
12921         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
12923 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12925         * config/arm/constraints.md (q): New constraint.
12926         * config/arm/ldrdstrd.md: New file.
12927         * config/arm/arm.md (ldrdstrd.md) New include.
12928         (arm_movdi): Use "q" instead of "r" constraint
12929         for double-word memory access.
12930         (movdf_soft_insn): Likewise.
12931         * config/arm/vfp.md (movdi_vfp): Likewise.
12932         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
12933         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
12934         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
12935         (mem_ok_for_ldrd_strd): Likewise.
12936         (output_move_double): Update assertion.
12938 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12940         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
12942 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12944         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
12945         define_insn_and_split.
12946         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
12948 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12950         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
12951         define_insn_and_split.
12952         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
12953         (shiftsi3_compare): New pattern.
12954         (rrx): New pattern.
12955         * config/arm/unspecs.md (UNSPEC_RRX): New.
12957 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12959         * config/arm/arm.md (negdi_extendsidi): New pattern.
12960         (negdi_zero_extendsidi): Likewise.
12962 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12964         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
12965         define_insn_and_split.
12966         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
12967         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
12969 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12971         * config/arm/arm.md (arm_subdi3): Convert define_insn into
12972         define_insn_and_split.
12973         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
12974         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
12976 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12978         * config/arm/arm.md (subsi3_carryin): New pattern.
12979         (subsi3_carryin_const): Likewise.
12980         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
12981         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
12983 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12985         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
12987 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
12989         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
12990         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
12992 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12994         * config/arm/arm.c (arm_expand_builtin): Change fcode
12995         type to unsigned int.
12997 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
12999         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
13001 2013-04-04  Ian Lance Taylor  <iant@google.com>
13003         * doc/standards.texi (Standards): The Go frontend supports the Go 1
13004         language standard.
13006 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
13008         PR middle-end/56729
13009         * df-scan.c (df_insn_delete): Disable failing assert.
13011 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13013         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
13014         New function prototype.
13015         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
13016         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
13017         (arm_builtin_vectorized_function): New function.
13019 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13021         * config/arm/arm_neon_builtins.def: New file.
13022         * config/arm/arm.c (neon_builtin_data): Move contents to
13023         arm_neon_builtins.def.
13024         (enum arm_builtins): Include neon builtin definitions.
13025         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
13026         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
13028 2013-04-04  Marek Polacek  <polacek@redhat.com>
13030         PR tree-optimization/48186
13031         * predict.c (maybe_hot_frequency_p): Return false if
13032         HOT_BB_FREQUENCY_FRACTION is 0.
13033         (cgraph_maybe_hot_edge_p): Likewise.
13035 2013-04-04  Richard Biener  <rguenther@suse.de>
13037         PR tree-optimization/56826
13038         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
13039         more accurately.
13041 2013-04-04  Richard Biener  <rguenther@suse.de>
13043         PR tree-optimization/56213
13044         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
13045         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
13047 2013-04-04  Richard Biener  <rguenther@suse.de>
13049         PR tree-optimization/56837
13050         * tree-loop-distribution.c (classify_partition): For non-zero
13051         values require that the value has the same precision as its
13052         mode to be useful as memset value.
13054 2013-04-03  Nick Clifton  <nickc@redhat.com>
13056         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
13057         (fmssf4): Use fmsf.s on E3V5 architectures.
13058         (fnmasf4): Use fnmaf.s on E3V5 architectures.
13059         (fnmssf4): Use fnmsf.s on E3V5 architectures.
13061 2013-04-03  Jeff Law  <law@redhat.com>
13063         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
13064         (lra-eliminations.o): Likewise.
13066 2013-04-03  Teresa Johnson  <tejohnson@google.com>
13068         * gcov-io.c (compute_working_sets): Moved most of body of old
13069         compute_working_sets here from profile.c.
13070         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
13071         (gcov_working_set_t): Moved typedef here from basic-block.h
13072         (compute_working_set): Declare.
13073         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
13074         (get_working_sets): Renamed from compute_working_set,
13075         replace most of body with call to new compute_working_sets.
13076         (get_exec_counts): Replace call to compute_working_sets
13077         to get_working_sets.
13078         * profile.h (get_working_sets): Renamed from compute_working_set.
13079         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
13080         to get_working_sets.
13081         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
13082         * gcov-dump.c (dump_working_sets): New function.
13084 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
13086         * hwint.c (sext_hwi, zext_hwi): New functions.
13087         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
13088         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
13089         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
13090         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
13091         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
13092         (sext_hwi, zext_hwi): New functions.
13094 2013-04-03  Jeff Law  <law@redhat.com>
13096         PR tree-optimization/56799
13097         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
13098         back test for widening conversion erroneously dropped in prior change.
13100 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13102         PR target/56809
13103         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
13104         instead of next_real_insn.
13106 2013-04-03  Marek Polacek  <polacek@redhat.com>
13108         PR sanitizer/55702
13109         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13111 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13113         PR target/56809
13114         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
13115         next_real_insn.
13116         (thumb1_output_casesi): Likewise.
13117         (thumb2_output_casesi): Likewise.
13119 2013-04-03  Richard Biener  <rguenther@suse.de>
13121         PR tree-optimization/56817
13122         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
13123         Split out ...
13124         (tree_unroll_loops_completely_1): ... new function to manually
13125         walk the loop tree, properly defering outer loops of unrolled
13126         loops to later iterations.
13128 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13130         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
13131         (vectorizable_load): Likewise.
13132         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
13133         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
13135 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13137         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
13138         BIT_FIELD_REF.
13140 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
13142         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
13144 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
13146         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
13148 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13150         PR tree-optimization/56790
13151         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
13152         folding.
13154 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
13156         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13157         Handle VEC_MERGE.
13158         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13159         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13160         equal arguments.
13162 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
13164         PR c/19449
13165         * tree.h (force_folding_builtin_constant_p): New decl.
13166         * builtins.c (force_folding_builtin_constant_p): New variable.
13167         (fold_builtin_constant_p): Fold immediately also if
13168         force_folding_builtin_constant_p.
13170 2013-04-03  Richard Biener  <rguenther@suse.de>
13172         PR tree-optimization/56812
13173         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
13174         DRs of the same interleaving chain are independent.
13176 2013-04-02  Jason Merrill  <jason@redhat.com>
13178         * gdbinit.in (pbb): Use debug fn.
13180 2013-04-02  Lawrence Crowl  <crowl@google.com>
13182         * sese.h (struct ivtype_map_elt_s): Remove unused.
13183         (extern debug_ivtype_map): Remove unused.
13184         (extern eq_ivtype_map_elts): Remove unused.
13185         * sese.c (debug_ivtype_map): Removed unused.
13186         (debug_ivtype_map_1): Removed unused.
13187         (debug_ivtype_elt): Remove unused.
13188         (eq_ivtype_map_elts): Remove unused.
13191 2013-04-02  Kai Tietz  <ktietz@redhat.com>
13193         PR target/52790
13194         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
13195         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
13196         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
13197         function.
13198         (legitimize_pe_coff_symbol): Likewise.
13199         (is_imported_p): New helper-function.
13200         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
13201         for Windows x64 targets.
13202         (ix86_expand_prologue): Optimize for pe-coff targets.
13203         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
13204         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
13205         medium/large code-model.
13206         (legitimize_pic_address): Likewise.
13207         (legitimize_tls_address): Likewise.
13208         (ix86_expand_call): Likewise.
13209         (x86_output_mi_thunk): Likewise.
13210         (get_dllimport_decl): Add new beimport argument.
13211         (construct_plt_address): Don't assert for x64 pe-coff targets.
13212         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
13213         targets.
13214         (SYMBOL_FLAG_STUBVAR): New macro.
13215         (SYMBOL_REF_STUBVAR_P): Likewise.
13216         * config/i386/winnt.c (stub_list): New structure.
13217         (stub_head): New local variable.
13218         (i386_pe_record_stub): New function.
13219         (i386_pe_file_end): Emit refptr-stubs.
13221 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
13223         PR rtl-optimization/56745
13224         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
13225         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
13227         PR c++/34949
13228         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
13229         and both of them are MEM_REFs, just compare first argument for
13230         equality and attempt to deal even with differing offsets.
13232         PR c++/34949
13233         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
13234         of gimple_clobber_p to be MEM_REF.
13235         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
13236         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
13237         after gimplification.
13238         * asan.c (get_mem_ref_of_assignment): Don't instrument
13239         gimple_clobber_p stmts.
13240         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
13241         gimple_clobber_p stmt if they have MEM_REF lhs and
13242         are dead because of another gimple_clobber_p stmt.
13243         * tree-ssa-live.c (clear_unused_block_pointer): Treat
13244         gimple_clobber_p stmts like debug stmts.
13245         (remove_unused_locals): Remove clobbers with MEM_REF lhs
13246         that refer to unused VAR_DECLs or uninitialized values.
13247         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
13248         gimple_clobber_p stmts if they refer to removed parameters.
13249         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
13250         formatting.
13252 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
13254         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
13255         using SWI48 mode attribute.
13257 2013-04-02  Wei Mi  <wmi@google.com>
13259         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
13260         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
13261         *<rotate_insn><mode>3_mask in i386.md.
13263 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13265         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
13267 2013-04-02  Richard Biener  <rguenther@suse.de>
13269         PR tree-optimization/56778
13270         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13271         Runtime alias tests are not supported for gather loads.
13272         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
13273         stmts referenced from SSA operands before updating SSA form.
13275 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
13276             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
13278         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
13279         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
13280         * config/arm/cortex-a53.md: New file.
13281         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
13282         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
13283         * config/arm/arm.c (arm_issue_rate): Likewise.
13284         * config/arm/arm-tune.md: Regenerate
13285         * config/arm/arm-tables.opt: Regenerate.
13286         * config/arm/arm-cores.def: Add cortex-a53.
13288 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
13290         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
13291         non-static link.
13293 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13295         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
13296         scalar load/store operations using B/H registers.
13297         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
13299 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
13301         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
13302         scalar move.
13303         * config/aarch64/aarch64.c
13304         (aarch64_simd_scalar_immediate_valid_for_move): New.
13305         * config/aarch64/aarch64-protos.h
13306         (aarch64_simd_scalar_immediate_valid_for_move): New.
13307         * config/aarch64/constraints.md (Dh, Dq): New.
13308         * config/aarch64/iterators.md (hq): New.
13310 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
13312         * reorg.c (get_branch_condition): Deal with conditional returns.
13313         (fill_simple_delay_slots): Remove dead code dealing with jumps.
13315 2013-04-01  Wei Mi  <wmi@google.com>
13317         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
13318         Truncate operand 2 using %b asm operand modifier.
13319         (*<shift_insn><mode>3_mask): Ditto.
13320         (*<rotate_insn><mode>3_mask): Ditto.
13322 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
13324         PR middle-end/56798
13325         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
13327 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
13329         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
13330         of next_real_insn.
13331         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
13333 2013-03-30  Lawrence Crowl  <crowl@google.com>
13335         * dse.c (clear_alias_sets): Remove never set.
13336         (disqualified_clear_alias_sets): Remove never set.
13337         (clear_alias_mode_pool): Remove never set.
13338         (dse_step0): Remove condition that is never true.
13339         (canon_address): Remove condition that is never true.
13340         (dse_step7): Remove condition that is never true.
13341         (rest_of_handle_dse): Remove condition that is never true.
13342         (rest_of_handle_dse::did_global): Remove never read from above.
13343         (dse_step2_spill): Remove never called from above.
13344         (dse_step5_spill): Remove never called from above.
13346 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
13348         * doc/md.texi (Standard Names) <casesi>: Update documentation for
13349         JUMP_TABLE_DATA changes.
13350         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
13351         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
13352         (Insns) <jump_table_data>: New entry.
13353         * doc/tm.texi: Regenerate.
13355         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
13357         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
13358         for table jump at the end of a basic block using tablejump_p.
13359         * targhooks.c (default_invalid_within_doloop): Likewise.
13360         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
13361         target hook implementation that is identical to the default hook.
13362         (rs6000_invalid_within_doloop): Remove.
13364         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
13365         unused variable from tablejump_p call.
13367         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
13368         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
13369         (INSN_DELETED_P): Likewise.
13370         (emit_jump_table_data): New prototype.
13371         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
13372         after 4th as unused.
13373         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
13374         * sched-vis.c (print_insn): Likewise.
13375         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
13376         insn for compatibility with back ends that use next_active_insn to
13377         identify jump table data.
13378         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
13379         (remove_insn): Likewise.
13380         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
13381         to be emitted.
13382         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
13383         (emit_jump_table_data): New function.
13385         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
13386         basic block, a JUMP_TABLE_DATA never is.
13387         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
13388         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
13389         off from code handling real insns.
13390         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
13391         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
13392         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
13393         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
13394         is not a NONDEBUG_INSN_P.
13395         * ira-costs.c (scan_one_insn): Likewise.
13396         * jump.c (mark_all_labels): Likewise.
13397         (mark_jump_label_1): Likewise.
13398         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13399         * lra.c (get_insn_freq): Expect all insns reaching here to be in
13400         a basic block.
13401         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
13402         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
13403         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
13404         JUMP_TABLE_DATA_P insns.
13405         (calculate_elim_costs_all_insns): Likewise.
13406         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
13407         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
13408         (delete_output_reload): Code style fixups.
13409         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
13410         insn flags on this non-insn.
13411         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
13412         as scheduling barriers, for pre-change compatibility.
13413         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
13414         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
13416         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
13417         redundant JUMP_TABLE_DATA_P test.
13418         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
13419         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13420         (frv_for_each_packet): Likewise.
13421         * config/i386/i386.c (min_insn_size): Likewise.
13422         (ix86_avoid_jump_mispredicts): Likewise.
13423         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13424         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
13425         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13426         (mips16_insn_length): Robustify.
13427         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
13428         (mips16_split_long_branches): Likewise.
13429         * config/pa/pa.c (pa_combine_instructions): Likewise.
13430         * config/rs6000/rs6000.c (get_next_active_insn): Treat
13431         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
13432         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
13433         as contributing to pool range lengths.
13434         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
13435         Remove redundant JUMP_TABLE_DATA_P test.
13436         (sh_loop_align): Likewise.
13437         (split_branches): Likewise.
13438         (sh_insn_length_adjustment): Likewise.
13439         * config/spu/spu.c (get_branch_target): Likewise.
13441 2013-03-29  Jan Hubicka  <jh@suse.cz>
13443         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
13444         gcov streaming; stream hot bb threshold to ltrans.
13445         * predict.c (get_hot_bb_threshold): Break out from ....
13446         (maybe_hot_count_p): ... here.
13447         (set_hot_bb_threshold): New function.
13448         * lto-section-in.c (lto_section_name): Add profile.
13449         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
13450         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
13451         and data-streamer.h
13452         (histogram_entry): New structure.
13453         (histogram, histogram_pool): New global vars.
13454         (histogram_hash): New structure.
13455         (histogram_hash::hash): New method.
13456         (histogram_hash::equal): Likewise.
13457         (account_time_size): New function.
13458         (cmp_counts): New function.
13459         (dump_histogram): New function.
13460         (ipa_profile_generate_summary): New function.
13461         (ipa_profile_write_summary): New function.
13462         (ipa_profile_read_summary): New function.
13463         (ipa_profile): Decide on threshold.
13464         (pass_ipa_profile): Add ipa_profile_write_summary and
13465         ipa_profile_read_summary.
13466         * Makefile.in (ipa.o): Update dependencies.
13467         * lto-streamer.h (LTO_section_ipa_profile): New section.
13469 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13471         * tree.h (VAR_P): New.
13473 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
13475         PR lto/56777
13476         * doc/invoke.texi ([-fwhole-program]): Fix typo.
13478 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
13480         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
13481         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13482         (control_flow_insn_p): Likewise.
13483         * cfgrtl.c (duplicate_insn_chain): Likewise.
13484         * final.c (get_attr_length_1): Likewise.
13485         (shorten_branches): Likewise.
13486         (final_scan_insn): Likewise.
13487         * function.c (instantiate_virtual_regs): Likewise.
13488         * gcse.c (insert_insn_end_basic_block): Likewise.
13489         * ira-costs.c (scan_one_insn): Likewise.
13490         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
13491         * lra.c (check_rtl): Likewise.
13492         * reload1.c (elimination_costs_in_insn): Likewise.
13493         * reorg.c (follow_jumps): Likewise.
13495         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
13496         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
13497         (thumb_far_jump_used_p): Likewise.
13498         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
13499         (workaround_speculation): Likewise.
13500         (add_sched_insns_for_speculation): Likewise.
13501         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
13502         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
13503         (frv_for_each_packet): Likewise.
13504         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
13505         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
13506         (final_emit_insn_group_barriers): Likewise.
13507         * config/m32r/m32r.c (m32r_is_insn): Likewise.
13508         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
13509         (mips16_insn_length): Likewise.
13510         * config/pa/pa.c (pa_reorg): Likewise.
13511         (pa_combine_instructions): Likewise.
13512         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
13513         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
13514         (sh_reorg): Likewise.
13515         (split_branches): Likewise.
13516         * config/spu/spu.c (get_branch_target): Likewise.
13518         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
13519         JUMP_TABLE_DATA_P.
13521 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
13523         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
13524         Fix declaration name.
13526 2013-03-28  Lawrence Crowl  <crowl@google.com>
13528         * graphds.h (struct graph.indicies): Remove unused.
13529         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
13530         (SCOP_ORIGINAL_PDDRS): Remove unused.
13531         * sese.h (extern insert_loop_close_phis): Removed unused.
13532         (extern insert_guard_phis): Removed unused.
13533         (extern ivtype_map_elt_info): Removed unused.
13534         (new_ivtype_map_elt): Removed unused.
13535         * sese.c (ivtype_map_elt_info): Removed unused.
13537 2013-03-28  Lawrence Crowl  <crowl@google.com>
13539         * Makefile.in: Add several missing include dependences.
13540         (DUMPFILE_H): New.
13541         (test-dump.o): New.  This object is not added to any executable,
13542         but is present for ad-hoc testing.
13543         * bitmap.c
13544         (debug (const bitmap_head_def &)): New.
13545         (debug (const bitmap_head_def *)): New.
13546         * bitmap.h
13547         (extern debug (const bitmap_head_def &)): New.
13548         (extern debug (const bitmap_head_def *)): New.
13549         * cfg.c
13550         (debug (edge_def &)): New.
13551         (debug (edge_def *)): New.
13552         * cfghooks.c
13553         (debug (basic_block_def &)): New.
13554         (debug (basic_block_def *)): New.
13555         * dumpfile.h
13556         (dump_node (const_tree, int, FILE *)): Correct source file.
13557         * dwarf2out.c
13558         (debug (die_struct &)): New.
13559         (debug (die_struct *)): New.
13560         * dwarf2out.h
13561         (extern debug (die_struct &)): New.
13562         (extern debug (die_struct *)): New.
13563         * gimple-pretty-print.c
13564         (debug (gimple_statement_d &)): New.
13565         (debug (gimple_statement_d *)): New.
13566         * gimple-pretty-print.h
13567         (extern debug (gimple_statement_d &)): New.
13568         (extern debug (gimple_statement_d *)): New.
13569         * ira-build.c
13570         (debug (ira_allocno_copy &)): New.
13571         (debug (ira_allocno_copy *)): New.
13572         (debug (ira_allocno &)): New.
13573         (debug (ira_allocno *)): New.
13574         * ira-int.h
13575         (extern debug (ira_allocno_copy &)): New.
13576         (extern debug (ira_allocno_copy *)): New.
13577         (extern debug (ira_allocno &)): New.
13578         (extern debug (ira_allocno *)): New.
13579         * ira-lives.c
13580         (debug (live_range &)): New.
13581         (debug (live_range *)): New.
13582         * lra-int.h
13583         (debug (lra_live_range &)): New.
13584         (debug (lra_live_range *)): New.
13585         * lra-lives.c
13586         (debug (lra_live_range &)): New.
13587         (debug (lra_live_range *)): New.
13588         * omega.c
13589         (debug (omega_pb_d &)): New.
13590         (debug (omega_pb_d *)): New.
13591         * omega.h
13592         (extern debug (omega_pb_d &)): New.
13593         (extern debug (omega_pb_d *)): New.
13594         * print-rtl.c
13595         (debug (const rtx_def &)): New.
13596         (debug (const rtx_def *)): New.
13597         * print-tree.c
13598         (debug_tree (tree): Move within file.
13599         (debug_raw (const tree_node &)): New.
13600         (debug_raw (const tree_node *)): New.
13601         (dump_tree_via_hooks (const tree_node *, int)): New.
13602         (debug (const tree_node &)): New.
13603         (debug (const tree_node *)): New.
13604         (debug_verbose (const tree_node &)): New.
13605         (debug_verbose (const tree_node *)): New.
13606         (debug_head (const tree_node &)): New.
13607         (debug_head (const tree_node *)): New.
13608         (debug_body (const tree_node &)): New.
13609         (debug_body (const tree_node *)): New.
13610         (debug_vec_tree (tree): Move and reimplement in terms of dump.
13611         (debug (vec<tree, va_gc> &)): New.
13612         (debug (vec<tree, va_gc> *)): New.
13613         * rtl.h
13614         (extern debug (const rtx_def &)): New.
13615         (extern debug (const rtx_def *)): New.
13616         * sbitmap.c
13617         (debug_raw (simple_bitmap_def &)): New.
13618         (debug_raw (simple_bitmap_def *)): New.
13619         (debug (simple_bitmap_def &)): New.
13620         (debug (simple_bitmap_def *)): New.
13621         * sbitmap.h
13622         (extern debug (simple_bitmap_def &)): New.
13623         (extern debug (simple_bitmap_def *)): New.
13624         (extern debug_raw (simple_bitmap_def &)): New.
13625         (extern debug_raw (simple_bitmap_def *)): New.
13626         * sel-sched-dump.c
13627         (debug (vinsn_def &)): New.
13628         (debug (vinsn_def *)): New.
13629         (debug_verbose (vinsn_def &)): New.
13630         (debug_verbose (vinsn_def *)): New.
13631         (debug (expr_def &)): New.
13632         (debug (expr_def *)): New.
13633         (debug_verbose (expr_def &)): New.
13634         (debug_verbose (expr_def *)): New.
13635         (debug (vec<rtx> &)): New.
13636         (debug (vec<rtx> *)): New.
13637         * sel-sched-dump.h
13638         (extern debug (vinsn_def &)): New.
13639         (extern debug (vinsn_def *)): New.
13640         (extern debug_verbose (vinsn_def &)): New.
13641         (extern debug_verbose (vinsn_def *)): New.
13642         (extern debug (expr_def &)): New.
13643         (extern debug (expr_def *)): New.
13644         (extern debug_verbose (expr_def &)): New.
13645         (extern debug_verbose (expr_def *)): New.
13646         (extern debug (vec<rtx> &)): New.
13647         (extern debug (vec<rtx> *)): New.
13648         * sel-sched-ir.h
13649         (_list_iter_cond_expr): Make inline instead of static.
13650         * sreal.c
13651         (debug (sreal &)): New.
13652         (debug (sreal *)): New.
13653         * sreal.h
13654         (extern debug (sreal &)): New.
13655         (extern debug (sreal *)): New.
13656         * tree.h
13657         (extern debug_raw (const tree_node &)): New.
13658         (extern debug_raw (const tree_node *)): New.
13659         (extern debug (const tree_node &)): New.
13660         (extern debug (const tree_node *)): New.
13661         (extern debug_verbose (const tree_node &)): New.
13662         (extern debug_verbose (const tree_node *)): New.
13663         (extern debug_head (const tree_node &)): New.
13664         (extern debug_head (const tree_node *)): New.
13665         (extern debug_body (const tree_node &)): New.
13666         (extern debug_body (const tree_node *)): New.
13667         (extern debug (vec<tree, va_gc> &)): New.
13668         (extern debug (vec<tree, va_gc> *)): New.
13669         * tree-cfg.c
13670         (debug (struct loop &)): New.
13671         (debug (struct loop *)): New.
13672         (debug_verbose (struct loop &)): New.
13673         (debug_verbose (struct loop *)): New.
13674         * tree-dump.c: Add header dependence.
13675         * tree-flow.h
13676         (extern debug (struct loop &)): New.
13677         (extern debug (struct loop *)): New.
13678         (extern debug_verbose (struct loop &)): New.
13679         (extern debug_verbose (struct loop *)): New.
13680         * tree-data-ref.c
13681         (debug (data_reference &)): New.
13682         (debug (data_reference *)): New.
13683         (debug (vec<data_reference_p> &)): New.
13684         (debug (vec<data_reference_p> *)): New.
13685         (debug (vec<ddr_p> &)): New.
13686         (debug (vec<ddr_p> *)): New.
13687         * tree-data-ref.h
13688         (extern debug (data_reference &)): New.
13689         (extern debug (data_reference *)): New.
13690         (extern debug (vec<data_reference_p> &)): New.
13691         (extern debug (vec<data_reference_p> *)): New.
13692         (extern debug (vec<ddr_p> &)): New.
13693         (extern debug (vec<ddr_p> *)): New.
13694         * tree-ssa-alias.c
13695         (debug (pt_solution &)): New.
13696         (debug (pt_solution *)): New.
13697         * tree-ssa-alias.h
13698         (extern debug (pt_solution &)): New.
13699         (extern debug (pt_solution *)): New.
13700         * tree-ssa-alias.c
13701         (debug (_var_map &)): New.
13702         (debug (_var_map *)): New.
13703         (debug (tree_live_info_d &)): New.
13704         (debug (tree_live_info_d *)): New.
13705         * tree-ssa-alias.h
13706         (extern debug (_var_map &)): New.
13707         (extern debug (_var_map *)): New.
13708         (extern debug (tree_live_info_d &)): New.
13709         (extern debug (tree_live_info_d *)): New.
13711 2013-03-28  Jan Hubicka  <jh@suse.cz>
13713         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
13715 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
13717         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
13718         record only when desired or required.
13720 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
13722         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
13723         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
13725 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
13727         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
13728         (*andsi3_compare0_uxtw): New pattern.
13729         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
13730         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
13732 2013-03-28  Jan Hubicka  <jh@suse.cz>
13734         * data-streamer-in.c (streamer_read_gcov_count): New function.
13735         * gimple-streamer-out.c: Include value-prof.h.
13736         (output_gimple_stmt): Output histogram.
13737         (output_bb): Use streamer_write_gcov_count.
13738         * value-prof.c: Include data-streamer.h
13739         (dump_histogram_value): Add HIST_TYPE_MAX.
13740         (stream_out_histogram_value): New function.
13741         (stream_in_histogram_value): New function.
13742         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
13743         (stream_out_histogram_value, stream_in_histogram_value): Declare.
13744         * data-streamer-out.c (streamer_write_gcov_count): New function.
13745         (streamer_write_gcov_count_stream): New function.
13746         * lto-cgraph.c (lto_output_edge): Update counter streaming.
13747         (lto_output_node): Likewise.
13748         (input_node, input_edge): Likewise.
13749         * lto-streamer-out.c (output_cfg): Update streaming.
13750         * lto-streamer-in.c (input_cfg): Likewise.
13751         * data-streamer.h (streamer_write_gcov_count,
13752         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
13753         * gimple-streamer-in.c: Include value-prof.h
13754         (input_gimple_stmt): Input histograms.
13755         (input_bb): Update profile streaming.
13757 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13759         * genmodes.c (emit_max_int): New function.
13760         (emit_insn_modes_h): Added call to emit_max_function.
13761         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
13762         Added doc.
13763         * machmode.def: Fixed comment.
13765 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
13767         * combine.c (try_combine): Removed useless assert.
13768         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
13770 2013-03-28  Marek Polacek  <polacek@redhat.com>
13771             Richard Biener  <rguenther@suse.de>
13773         PR tree-optimization/56695
13774         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
13775         build signed result of a vector comparison.
13776         * tree-cfg.c (verify_gimple_comparison): Check that a result
13777         of a vector comparison has signed type.
13779 2013-03-28  Richard Biener  <rguenther@suse.de>
13781         PR tree-optimization/37021
13782         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
13783         do not restrict gaps between groups.
13784         * tree-vect-stmts.c (vectorizable_load): Properly account for
13785         a gap between groups.
13787 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
13789         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
13790         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
13791         is not enabled.
13793 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
13795         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
13796         * doc/extend.texi (Named Address Spaces): Ditto.
13797         (Variable Attributes): Ditto.
13799 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13801         * config.build: Add support for cygwin x64 target.
13802         * config.gcc: Likewise.
13803         * config.host: Likewise.
13804         * configure.ac: Likewise
13805         * configure: Regenerated.
13807 2013-03-27  Kai Tietz  <ktietz@redhat.com>
13809         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
13810         * config/i386/t-cygwin-w64: New file.
13811         * config/i386/cygwin-w64.h: New file.
13812         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
13813         and add support for x64-cygwin target.
13814         (CPP_SPEC): Likewise.
13815         (CXX_WRAP_SPEC_LIST): Undefine before define.
13816         (LIBGCJ_SONAME): Use 15 as version.
13818 2013-03-27  Richard Biener  <rguenther@suse.de>
13820         PR tree-optimization/56716
13821         * tree-ssa-structalias.c (perform_var_substitution): Adjust
13822         dumping for ref nodes.
13824 2013-03-27  Martin Jambor  <mjambor@suse.cz>
13826         PR tree-optimization/55334
13827         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
13828         restricted pointers to arrays.
13830 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13832         * Makefile.in (.SUFFIXES): Add .cc.
13833         (.c.o): Apply same recipe for implicit rule .cc.o.
13835 2013-03-27  Richard Biener  <rguenther@suse.de>
13837         PR tree-optimization/37021
13838         * tree-vect-data-refs.c (vect_check_strided_load): Allow
13839         REALPART/IMAGPART_EXPRs around the supported refs.
13840         * tree-ssa-structalias.c (find_func_aliases): Assume that
13841         floating-point values are not used to transfer pointers.
13843 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
13845         * gcc/target.def (TARGET_HAS_IFUNC_P): New target hook.
13846         * gcc/doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
13847         * gcc/doc/tm.texi: Regenerate.
13848         * gcc/targhooks.h (default_has_ifunc_p): New.
13849         * gcc/targhooks.c (default_has_ifunc_p): Ditto.
13850         * gcc/config/linux-protos.h: New file.
13851         * gcc/config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of
13852         this hook for linux which disables support of indirect functions in
13853         android.
13854         * gcc/config/linux-android.c: New file.
13855         * gcc/config/t-linux-android.c: Ditto.
13856         * gcc/config.gcc: Added new object file linux-android.o.
13857         * gcc/config/i386/i386.c (ix86_get_function_versions_dispatcher):
13858         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
13859         * gcc/varasm.c (do_assemble_alias): Likewise.
13860         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
13861         doesn't support indirect functions.
13862         * configure: Regenerate.
13864 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
13866         PR target/56102
13867         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
13868         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
13869         mult-word mode.
13871 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13873         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
13875 2013-03-27  Terry Guo  <terry.guo@arm.com>
13877         * config/arm/arm-cores.def: Added core cortex-r7.
13878         * config/arm/arm-tune.md: Regenerated.
13879         * config/arm/arm-tables.opt: Regenerated.
13880         * doc/invoke.texi: Added entry for core cortex-r7.
13882 2013-03-27  Walter Lee  <walt@tilera.com>
13884         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
13885         double-decrement of next_scratch_regno.
13887 2013-03-27  Walter Lee  <walt@tilera.com>
13889         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
13890         input operands.
13891         (insn_v1mulus): Ditto.
13892         (insn_v2muls): Ditto.
13894 2013-03-27  Walter Lee  <walt@tilera.com>
13896         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
13897         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
13899 2013-03-27  Walter Lee  <walt@tilera.com>
13901         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
13902         (*sibcall_value): Ditto.
13904 2013-03-27  Walter Lee  <walt@tilera.com>
13906         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
13907         (insn_mnz_v8qi): ... this ...
13908         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
13909         vector equivalent.
13910         (insn_v<n>mnz): Replaced by ...
13911         (insn_v1mnz): ... this ...
13912         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
13913         equivalent.
13914         (insn_mz_<mode>): Replaced by ...
13915         (insn_mz_v8qi): ... this ...
13916         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
13917         vector equivalent.
13918         (insn_v<n>mz): Replaced by ...
13919         (insn_v1mz): ... this ...
13920         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
13921         equivalent.
13923 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
13925         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
13927 2013-03-26  Roland McGrath  <mcgrathr@google.com>
13929         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
13930         than fprintf with a non-constant, non-format string.
13932 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13934         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
13935         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
13936         operand 0 predicate.
13937         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
13938         attribute.  Use general_x64nomem_operand as operand 1 predicate.
13939         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
13940         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13941         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
13942         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
13943         (mov<mode>_insv_1): Remove expander.  Merge insn with
13944         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
13945         Use general_x64nomem_operand as operand 1 predicate.
13946         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
13947         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
13948         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13949         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
13950         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
13951         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
13952         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
13953         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
13954         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
13955         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
13956         (general_x64nomem_operand): Ditto.
13958 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
13960         * config/rtems.opt: Add -pthread option.
13962 2013-03-26  Richard Biener  <rguenther@suse.de>
13964         * alias.c (find_base_term): Avoid redundant and not used recursion.
13965         (base_alias_check): Get the initial base term from the caller.
13966         (true_dependence_1): Compute and pass base terms to base_alias_check.
13967         (write_dependence_p): Likewise.
13968         (may_alias_p): Likewise.
13970 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
13972         * config/aarch64/aarch64.c (aarch64_classify_address): Support
13973         PC-relative load in SI modes and above only.
13975 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
13977         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
13978         * config/arm/iwmmxt.md (WCGR0): Update.
13979         (WCGR1, WCGR2, WCGR3): Likewise.
13981 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
13983         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
13984         Use x64 and nox64 isa attributes.
13986 2013-03-26  Richard Biener  <rguenther@suse.de>
13988         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
13989         alignment computations and rely on get_object_alignment_1
13990         for the !TYPE_P case.
13991         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
13993 2013-03-26  Walter Lee  <walt@tilera.com>
13995         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
13996         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
13998 2013-03-25  Jeff Law  <law@redhat.com>
14000         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
14001         check for INTEGRAL_TYPE_P that was missing due to checking in
14002         wrong version of prior patch.
14004 2013-03-25  Walter Lee  <walt@tilera.com>
14006         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
14007         TILEGX_INSN_SHUFFLEBYTES1.
14008         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
14009         shufflebytes1.
14010         (tilegx_builtins): Ditto.
14011         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
14013 2013-03-25  Walter Lee  <walt@tilera.com>
14015         * config/tilegx/tilegx.md (floatsisf2): New pattern.
14016         (floatunssisf2): New pattern.
14017         (floatsidf2): New pattern.
14018         (floatunssidf2): New pattern.
14020 2013-03-25  Walter Lee  <walt@tilera.com>
14022         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
14023         tests for constraint J, K, N, P.
14025 2013-03-25  Walter Lee  <walt@tilera.com>
14027         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
14028         Use indirect/pcrel encoding.
14029         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
14030         Ditto.
14032 2013-03-25  Steve Ellcey  <sellcey@mips.com>
14034         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
14035         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
14036         * config/mips/mips.c (mips_option_override): Set IMADD default.
14037         * config/mips/mips.h (PTF_AVOID_IMADD): New.
14038         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
14039         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
14040         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
14041         * doc/invoke.texi (-mimadd/-mno-imadd): New.
14043 2013-03-25  Jeff Law  <law@redhat.com>
14045         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
14046         slightly to avoid creating and folding useless trees.  Simplify
14047         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
14049 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
14051         * config/i386/i386.md (*zero_extendsidi2): Merge with
14052         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
14053         * config/i386/predicates.md (x86_64_zext_operand): Rename from
14054         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
14055         targets.  Clarify comment.
14057 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14059         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
14060         pass-through jump functions differently.
14061         (ipa_read_jump_function): Likewise.  Also use setter functions to set
14062         up jump functions.
14064 2013-03-25  Martin Jambor  <mjambor@suse.cz>
14066         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
14067         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
14068         process it.
14069         (ipa_get_indirect_edge_target): New function.
14070         (devirtualization_time_bonus): New parameter known_aggs, pass it to
14071         ipa_get_indirect_edge_target.  Update all callers.
14072         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
14073         ipa_get_indirect_edge_target_1 instead of calling
14074         ipa_get_indirect_edge_target.
14075         (create_specialized_node): Pass aggvlas to
14076         ipcp_discover_new_direct_edges.
14078 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14080         * config/arm/arm.md (f_sels, f_seld): New types.
14081         (*cmov<mode>): New pattern.
14082         * config/arm/predicates.md (arm_vsel_comparison_operator): New
14083         predicate.
14085 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14087         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
14088         POSIX-printf for mingw-hosted  builds.
14090 2013-03-25  Richard Biener  <rguenther@suse.de>
14092         PR middle-end/56694
14093         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
14094         must-not-throw stmt location.
14096 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14098         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
14099         Emit load-acquire versions when acq is true.
14100         (arm_emit_store_exclusive): Add rel parameter.
14101         Emit store-release versions when rel is true.
14102         (arm_split_compare_and_swap): Use acquire-release instructions
14103         instead.
14104         of barriers when appropriate.
14105         (arm_split_atomic_op): Likewise.
14106         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
14107         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
14108         (VUNSPEC_SLX): Likewise.
14109         (VUNSPEC_LDA): Likewise.
14110         (VUNSPEC_STL): Likewise.
14111         * config/arm/sync.md (atomic_load<mode>): New pattern.
14112         (atomic_store<mode>): Likewise.
14113         (arm_load_acquire_exclusive<mode>): Likewise.
14114         (arm_load_acquire_exclusivesi): Likewise.
14115         (arm_load_acquire_exclusivedi): Likewise.
14116         (arm_store_release_exclusive<mode>): Likewise.
14118 2013-03-25  Catherine Moore  <clm@codesourcery.com>
14120         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
14121         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
14122         * config/mip/predicates.md (lwsp_swsp_operand,
14123         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
14124         sb16_operand, db4_operand, db7_operand, ib3_operand,
14125         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
14126         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
14127         andi16_operand): New predicates.
14128         * config/mips/mips.md (compression): New attribute.
14129         (enabled): New attribute.
14130         (length): Consider compression in computing length.
14131         (shift_compression): New code attribute.
14132         (*add<mode>3): New operands. Record compression.
14133         (sub<mode>3): Likewise.
14134         (one_cmpl<mode>2): Likewise.
14135         (*and<mode>3): Likewise.
14136         (*ior<mode>3): Likewise.
14137         (unnamed pattern for xor): Likewise.
14138         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
14139         (*<optab><mode>3): Likewise.
14140         (*mov<mode>_internal: Likewise.
14141         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
14142         (mips_unsigned_immediate_p): New.
14143         (umips_lwsp_swsp_address_p): New.
14144         (m16_based_address_p): New.
14145         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
14146         (mips_unsigned_immediate_p): New prototype.
14147         (lwsp_swsp_address_p): New prototype.
14148         (m16_based_address_p): New prototype.
14149         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
14150         (mips_signed_immediate_p): New function.
14151         (m16_based_address_p): New function.
14152         (lwsp_swsp_address_p): New function.
14153         (mips_print_operand_punctuation): Recognize short delay slot insns
14154         for microMIPS.add<mode>3"
14156 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14158         PR target/56720
14159         * config/arm/iterators.md (v_cmp_result): New mode attribute.
14160         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
14162 2013-03-25  Richard Biener  <rguenther@suse.de>
14164         PR tree-optimization/56689
14165         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
14166         any edge.
14168 2013-03-25  Richard Biener  <rguenther@suse.de>
14170         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
14171         of bitmap.
14172         (memory_references): Likewise.
14173         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
14174         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
14175         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
14176         (gather_mem_refs_in_loops): Fold into ...
14177         (analyze_memory_references): ... this.  Move initialization
14178         to tree_ssa_lim_initialize.
14179         (fill_always_executed_in): Rename to ...
14180         (fill_always_executed_in_1): ... this.
14181         (fill_always_executed_in): Move contains_call computation to
14182         this new function from ...
14183         (tree_ssa_lim_initialize): ... here.
14184         (tree_ssa_lim): Call fill_always_executed_in.
14186 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14188         * postreload.c (reload_combine): Fix code detecting returns.
14190 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
14192         * function.c (emit_use_return_register_into_block): On cc0 targets,
14193         do not emit the sequence between cc0 setter and user.
14195 2013-03-25  Kai Tietz  <ktietz@redhat.com>
14197         * config/i386/predicates.md (local_symbolic_operand): Interpret
14198         dll-imported symbols as none-local.
14200 2013-03-25  Richard Biener  <rguenther@suse.de>
14202         * tree-ssa-loop-im.c (struct depend): Remove.
14203         (struct lim_aux_data): Make depends a vec of gimples.
14204         (free_lim_aux_data): Adjust.
14205         (add_dependency): Likewise.
14206         (set_level): Likewise.
14208 2013-03-25  Richard Biener  <rguenther@suse.de>
14210         PR middle-end/56434
14211         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
14212         the pointer returned by calls with ECF_MALLOC set.
14214 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14216         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
14218 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
14220         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
14221         using MMXMODE mode iterator.
14222         (*move<mode>_internal): Merge with *movv2sf_internal and
14223         *movv2sf_internal_rex64 using MMXMODE mode iterator.
14225 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14227         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
14228         (record_last_mem_set_info): Likewise.
14230         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
14231         of XNEWVEC followed by memset.
14232         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
14234 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
14236         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
14237         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
14238         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
14239         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
14240         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
14241         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
14242         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
14243         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
14244         BARRIER_P instead of GET_CODE.
14246 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14248         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
14249         inaccuracy in the probing code.
14251         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
14252         (ctrapdi4): Likewise.
14254 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14256         * calls.c (expand_call): Add missing guard to code handling return
14257         of non-BLKmode structures in MSB.
14258         * function.c (expand_function_end): Likewise.
14260 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
14262         * combine.c (try_combine): Adjust comment.  Do not add the set of
14263         insn #0 if the destination indirectly is set or dies in insn #2.
14264         Tidy up code to distribute a new note.
14266 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14268         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
14269         also for alternatives 16 and 17.
14271 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
14273         * config/i386/sse.md (*mov<mode>_internal): Merge with
14274         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
14275         Emit insn template depending on type attribute.  Use
14276         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14277         movd instead of movq mnemonic for interunit moves.  Rewrite mode
14278         attribute calculation.  Remove unit attribute calculation.
14279         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
14280         Set prefix_data16 attribute for DImode ssemov types.
14281         Use Ym instead of y for SSE-MMX conversion alternatives.
14282         Reorder operand constraints.
14284 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
14286         * df.h (df_insn_delete): Adjust prototype.
14287         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
14288         and let it decide whether mark the basic block dirty.
14289         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
14290         * df-scan.c (df_insn_info_delete): New helper function, split
14291         off from df_insn_delete.
14292         (df_scan_free_bb_info): Use it.
14293         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
14294         Likewise.
14295         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
14296         that the insn is actually an insn and it has a non-NULL basic block.
14297         Do not mark basic block dirty if only deleting a DEBUG_INSN.
14299 2013-03-22  Richard Biener  <rguenther@suse.de>
14301         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
14302         dep_ref members.
14303         (mem_ref_alloc): Do not allocate them.
14304         (refs_independent_p): Do not query or maintain a cache.
14306 2013-03-22  Richard Biener  <rguenther@suse.de>
14308         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
14309         (gather_mem_refs_in_loops): Do not compute it.
14310         (analyze_memory_references): Do not allocate it.
14311         (tree_ssa_lim_finalize): Do not free it.
14312         (for_all_locs_in_loop): Do not query all_refs_in_loop.
14314 2013-03-22  Richard Biener  <rguenther@suse.de>
14316         * is-a.h (as_a): Use gcc_checking_assert.
14318 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
14320         * config/aarch64/aarch64.c (aarch64_print_operand): New
14321         format specifier for printing a constant in hex.
14322         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
14323         format specifier for printing second operand.
14325 2013-03-22  Richard Biener  <rguenther@suse.de>
14327         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
14328         bitmaps.
14329         (gather_mem_refs_in_loops): Perform store accumulation here.
14330         (create_vop_ref_mapping_loop): Remove.
14331         (create_vop_ref_mapping): Likewise.
14332         (analyze_memory_references): Initialize refs_stored_in_loop.
14333         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
14334         (record_indep_loop): Remove.
14335         (record_dep_loop): New function.
14336         (ref_indep_loop_p_1): Adjust to only walk over references
14337         in the loop, not its subloops.
14338         (ref_indep_loop_p): Rename to ...
14339         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
14340         maintaining a more fine-grained cache.
14341         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
14342         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
14344 2013-03-22  Richard Biener  <rguenther@suse.de>
14346         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
14347         (struct mem_ref): Make accesses_in_loop a vec of a vec of
14348         aggregate mem_ref_loc.
14349         (free_mem_ref_locs): Inline into ...
14350         (memref_free): ... this and adjust.
14351         (mem_ref_alloc): Adjust.
14352         (mem_ref_locs_alloc): Remove.
14353         (record_mem_ref_loc): Adjust.
14354         (get_all_locs_in_loop): Rewrite into ...
14355         (for_all_locs_in_loop): ... this iterator.
14356         (rewrite_mem_ref_loc): New functor.
14357         (rewrite_mem_refs): Use for_all_locs_in_loop.
14358         (sm_set_flag_if_changed): New functor.
14359         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
14360         (ref_always_accessed): New functor.
14361         (ref_always_accessed_p): Use for_all_locs_in_loop.
14363 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14365         * tree-pass.h (PROP_gimple_lvec): New.
14366         * passes.c (dump_properties): Handle PROP_gimple_lvec.
14367         (init_optimization_passes): Move pass_lower_vector.
14368         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
14369         PROP_gimple_lvec.
14370         (pass_lower_vector): Provide PROP_gimple_lvec.
14371         (pass_lower_vector_ssa): Likewise.
14372         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
14374 2013-03-21  Mark Wielaard  <mjw@redhat.com>
14376         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
14378 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14380         * config/i386/i386.md (*movdi_internal): Disparage slightly
14381         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
14382         conversion alternatives.
14384 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
14386         PR middle-end/48087
14387         * diagnostic.def (DK_WERROR): New kind.
14388         * diagnostic.h (werrorcount): Define.
14389         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
14390         promoted to DK_ERROR, increment DK_WERROR counter instead of
14391         DK_ERROR counter.
14392         * toplev.c (toplev_main): Call print_ignored_options even if
14393         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
14394         even if just werrorcount is non-zero.
14396         PR debug/55608
14397         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
14398         on failure.
14399         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
14400         (string_cst_pool_decl): New function.
14401         (optimize_one_addr_into_implicit_ptr): New function.
14402         (resolve_addr_in_expr): Optimize DWARF location expression
14403         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
14404         which doesn't live in memory, but has DW_AT_location or
14405         DW_AT_const_value, or refers to a string literal, into
14406         DW_OP_GNU_implicit_pointer.
14407         (optimize_location_into_implicit_ptr): New function.
14408         (resolve_addr): If removing DW_AT_location of a variable because
14409         it was DW_OP_addr of address of the variable, but the variable doesn't
14410         live in memory, try to emit const value attribute for the initializer.
14412 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14414         * tree.h (VECTOR_TYPE_P): New macro.
14415         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
14416         TYPE_MODE): Use it.
14417         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
14418         VEC_COND_EXPR cannot be lvalues.
14419         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
14421 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
14423         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
14424         Restrict the transformation to equal modes.
14426 2013-03-21  Richard Biener  <rguenther@suse.de>
14428         PR tree-optimization/39326
14429         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
14430         (MEM_ANALYZABLE): Adjust.
14431         (record_mem_ref_loc): Move bitmap ops ...
14432         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
14433         unanalyzable refs, do not record locations for it.
14434         (analyze_memory_references): Allocate ref zero as shared
14435         unanalyzable ref.
14436         (refs_independent_p): Do not test for unanalyzed mems here.
14437         (ref_indep_loop_p_1): Special-case disambiguation against
14438         the unanalyzed ref.
14439         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
14441 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
14443         * config/arm/arm-protos.h (tune_params): Add
14444         prefer_neon_for_64bits field.
14445         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
14446         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
14447         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
14448         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
14449         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
14450         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
14451         (arm_option_override): Handle -mneon-for-64bits new option.
14452         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
14453         (prefer_neon_for_64bits): Declare new variable.
14454         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
14455         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
14456         (arch_enabled): Handle new arch types. Remove support for onlya8
14457         and nota8.
14458         (one_cmpldi2): Use new arch names.
14459         (zero_extend<mode>di2, extend<mode>di2): Ditto.
14460         * config/arm/arm.opt (mneon-for-64bits): Add option.
14461         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
14462         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
14463         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
14464         of onlya8.
14465         * doc/invoke.texi (-mneon-for-64bits): Document.
14467 2013-03-21  Richard Biener  <rguenther@suse.de>
14469         PR tree-optimization/39326
14470         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
14471         (sort_bbs_in_loop_postorder_cmp): New function.
14472         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
14474 2013-03-21  Richard Biener  <rguenther@suse.de>
14476         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
14477         (vect_insert_into_interleaving_chain): Likewise.
14478         (vect_drs_dependent_in_basic_block): Inline ...
14479         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
14480         split out from ...
14481         (vect_analyze_data_ref_dependence): ... here.  Simplify.
14482         (vect_check_interleaving): Simplify.
14483         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
14484         (vect_slp_analyze_data_ref_dependences): ... this new function.
14485         (dr_group_sort_cmp): New function.
14486         (vect_analyze_data_ref_accesses): Compute data-reference groups
14487         here instead of in vect_analyze_data_ref_dependence.  Use
14488         a more efficient algorithm.
14489         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
14490         vect_slp_analyze_data_ref_dependences.  Call
14491         vect_analyze_data_ref_accesses earlier.
14492         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
14493         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
14494         (vect_slp_analyze_data_ref_dependences): New prototype.
14496 2013-03-21  Richard Biener  <rguenther@suse.de>
14498         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
14499         ref is stored in the loop.
14500         (find_refs_for_sm): Walk only over all stores.
14501         (store_motion_loop): Allocate from lim_bitmap_obstack.
14502         (store_motion): Likewise.
14504 2013-03-21  Richard Biener  <rguenther@suse.de>
14506         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
14507         Update virtual SSA form.
14509 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
14511         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
14512         * configure: Regenerate.
14513         * config.in: Regenerate.
14514         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
14515         if !HAVE_LD_EH_FRAME_CIEV3.
14517 2013-03-21  Richard Biener  <rguenther@suse.de>
14519         * tree-cfg.c (verify_expr_no_block): New function.
14520         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
14521         nor DECL_VALUE_EXPR have locations with associated blocks.
14522         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
14523         (clear_unused_block_pointer): Remove code dealing with
14524         blocks in DECL_DEBUG_EXPR locations.
14526 2013-03-21  Richard Biener  <rguenther@suse.de>
14528         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
14529         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
14530         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
14531         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
14532         instead of DECL_DEBUG_EXPR_IS_FROM.
14533         * gimplify.c (gimplify_modify_expr): Likewise.
14534         * tree-cfg.c (verify_expr_location_1): Likewise.
14535         * tree-complex.c (create_one_component_var): Likewise.
14536         * tree-sra.c (create_access_replacement): Likewise.
14537         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
14538         (clear_unused_block_pointer): Likewise.
14539         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
14540         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
14541         * var-tracking.c (var_debug_decl): Likewise.
14542         (track_expr_p): Likewise.
14543         * tree-inline.c (add_local_variables): Likewise.  Set
14544         DECL_HAS_DEBUG_EXPR_P after copying it.
14545         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
14546         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
14548 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
14550         PR bootstrap/56656
14551         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
14552         * configure: Regenerate.
14553         * config.in: Regenerate.
14554         * config/i386/i386.md (*movdf_internal): Use
14555         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
14556         movd instead of movq mnemonic for interunit moves.
14557         (*movdi_internal): Ditto.
14559 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14561         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
14562         (abd<mode>_3): New pattern.
14563         (aba<mode>_3): New pattern.
14564         (fabd<mode>_3): New pattern.
14566 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
14568         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
14569         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
14570         occurrence of REGISTER_PREFIX as its empty string.
14572 2013-03-20  Jeff Law  <law@redhat.com>
14574         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
14575         addititional equivalences for equality comparisons between an SSA_NAME
14576         and a constant where the SSA_NAME was set from a widening conversion.
14578 2013-03-20  Walter Lee  <walt@tilera.com>
14580         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
14582 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14584         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
14585         depending on type attribute.
14586         (*movti_internal): Ditto.
14587         (*movtf_internal): Ditto.
14588         (*movxf_internal): Ditto.
14589         (*movdf_internal): Ditto.
14590         (*movsf_internal): Ditto.
14592 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
14594         * config/i386/i386.md (*movti_internal): Set prefix attribute to
14595         maybe_vex for sselog1 and ssemov types.
14596         (*movdi_internal): Reorder operand constraints.
14597         (*movsi_internal): Ditto.  Set prefix attribute to
14598         maybe_vex for sselog1 and ssemov types.
14599         (*movtf_internal): Set prefix attribute to maybe_vex
14600         for sselog1 and ssemov types.
14601         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
14602         DImode ssemov types.  Reorder operand constraints.
14603         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
14604         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
14605         attribute for SImode ssemov types.  Reorder operand constraints.
14607 2013-03-20  Martin Jambor  <mjambor@suse.cz>
14609         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
14610         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
14612 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
14614         * config/rs6000/predicates.md (indexed_address, update_address_mem
14615         update_indexed_address_mem): New predicates.
14616         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
14617         attribute for load/store instructions.
14618         * config/rs6000/dfp.md (movsd_store): Likewise.
14619         (movsd_load): Likewise.
14620         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
14621         (unnamed HI->DI extend define_insn): Likewise.
14622         (unnamed SI->DI extend define_insn): Likewise.
14623         (unnamed QI->SI extend define_insn): Likewise.
14624         (unnamed QI->HI extend define_insn): Likewise.
14625         (unnamed HI->SI extend define_insn): Likewise.
14626         (unnamed HI->SI extend define_insn): Likewise.
14627         (extendsfdf2_fpr): Likewise.
14628         (movsi_internal1): Likewise.
14629         (movsi_internal1_single): Likewise.
14630         (movhi_internal): Likewise.
14631         (movqi_internal): Likewise.
14632         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
14633         attribute for load/store instructions.
14634         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
14635         instructions.
14636         (mov<mode>_softfloat): Likewise.
14637         (mov<mode>_hardfloat32): Likewise.
14638         (mov<mode>_hardfloat64): Likewise.
14639         (mov<mode>_softfloat64): Likewise.
14640         (movdi_internal32): Likewise.
14641         (movdi_internal64): Likewise.
14642         (probe_stack_<mode>): Likewise.
14644 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
14646         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
14647         floating point, and decimal floating point to reload iterator.
14649         * config/rs6000/constraints.md (wl constraint): New constraints to
14650         return FLOAT_REGS if certain options are used to reduce the number
14651         of separate patterns that exist in the file.
14652         (wx constraint): Likewise.
14653         (wz constraint): Likewise.
14655         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14656         -mdebug=reg, print wg, wl, wx, and wz constraints.
14657         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14658         Initialize the reload functions for 64-bit binary/decimal floating
14659         point types.
14660         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14661         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14662         create the buffer on the stack to overcome not having a 32-bit
14663         load and store.
14664         (rs6000_emit_move): Likewise.
14665         (rs6000_secondary_memory_needed_rtx): Likewise.
14666         (rs6000_alloc_sdmode_stack_slot): Likewise.
14667         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14668         via xxlxor, just like DFmode 0.0.
14670         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
14671         define as 1 if we are running on a power7 or newer.
14672         (enum r6000_reg_class_enum): Add new constraints.
14674         * config/rs6000/dfp.md (movsd): Delete, combine with binary
14675         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14676         with other moves by using conditional constraits (wg).  Use LFIWZX
14677         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
14678         (movsd splitter): Likewise.
14679         (movsd_hardfloat): Likewise.
14680         (movsd_softfloat): Likewise.
14682         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
14683         binary and decimal floating point moves.
14684         (fmove_ok): New attributes to combine binary and decimal floating
14685         point moves, and to combine power6x (mfpgpr) moves along normal
14686         floating moves.
14687         (real_value_to_target): Likewise.
14688         (f32_lr): Likewise.
14689         (f32_lm): Likewise.
14690         (f32_li): Likewise.
14691         (f32_sr): Likewise.
14692         (f32_sm): Likewise.
14693         (f32_si): Likewise.
14694         (movsf): Combine binary and decimal floating point moves.  Combine
14695         power6x (mfpgpr) moves with other moves by using conditional
14696         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
14697         (mov<mode> for SFmode/SDmode); Likewise.
14698         (SFmode/SDmode splitters): Likewise.
14699         (movsf_hardfloat): Likewise.
14700         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
14701         (movsf_softfloat): Likewise.
14702         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
14704         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
14705         wx and wz constraints.
14707         * config/rs6000/constraints.md (wg constraint): New constraint to
14708         return FLOAT_REGS if -mmfpgpr (power6x) was used.
14710         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
14711         constraint.
14713         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
14714         -mdebug=reg, print wg, wl, wx, and wz constraints.
14715         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
14716         Initialize the reload functions for 64-bit binary/decimal floating
14717         point types.
14718         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
14719         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
14720         create the buffer on the stack to overcome not having a 32-bit
14721         load and store.
14722         (rs6000_emit_move): Likewise.
14723         (rs6000_secondary_memory_needed_rtx): Likewise.
14724         (rs6000_alloc_sdmode_stack_slot): Likewise.
14725         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
14726         via xxlxor, just like DFmode 0.0.
14728         * config/rs6000/dfp.md (movdd): Delete, combine with binary
14729         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
14730         with other moves by using conditional constraits (wg).  Use LFIWZX
14731         and STFIWX for loading SDmode on power7.
14732         (movdd splitters): Likewise.
14733         (movdd_hardfloat32): Likewise.
14734         (movdd_softfloat32): Likewise.
14735         (movdd_hardfloat64_mfpgpr): Likewise.
14736         (movdd_hardfloat64): Likewise.
14737         (movdd_softfloat64): Likewise.
14739         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
14740         64-bit binary and decimal floating point moves.
14741         (FMOVE64X): Likewise.
14742         (movdf): Combine 64-bit binary and decimal floating point moves.
14743         Combine power6x (mfpgpr) moves with other moves by using
14744         conditional constraits (wg).
14745         (mov<mode> for DFmode/DDmode): Likewise.
14746         (DFmode/DDmode splitters): Likewise.
14747         (movdf_hardfloat32): Likewise.
14748         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
14749         (movdf_softfloat32): Likewise.
14750         (movdf_hardfloat64_mfpgpr): Likewise.
14751         (movdf_hardfloat64): Likewise.
14752         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
14753         (movdf_softfloat64): Likewise.
14754         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
14755         (reload_<mode>_load): Move to later in the file so they aren't in
14756         the middle of the floating point move insns.
14757         (reload_<mode>_store): Likewise.
14759         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
14760         constraint.
14762         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
14763         constraint if -mdebug=reg.
14764         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
14765         Enable using dd reload support if needed.
14767         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
14768         binary and decimal floating point moves in rs6000.md.
14769         (movtd_internal): Likewise.
14771         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
14772         decimal floating point moves.
14773         (movtf): Likewise.
14774         (movtf_internal): Likewise.
14775         (mov<mode>_internal, TDmode/TFmode): Likewise.
14776         (movtf_softfloat): Likewise.
14777         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
14779         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
14780         movdi_internal64, using wg constraint for move direct operations.
14781         (movdi_internal64): Likewise.
14783         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
14784         MODES_TIEABLE_P for selected modes.  Print the numerical value of
14785         the various virtual registers. Use GPR/FPR first/last values,
14786         instead of hard coding the register numbers.  Print which modes
14787         have reload functions registered.
14788         (rs6000_option_override_internal): If -mdebug=reg, trace the options
14789         settings before/after setting cpu, target and subtarget settings.
14790         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
14791         and for secondary reload failures in rs6000_secondary_reload_inner.
14792         (rs6000_secondary_reload_fail): Likewise.
14793         (rs6000_secondary_reload_inner): Likewise.
14795         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
14796         macros for first/last GPR and FPR registers.
14797         (LAST_GPR_REGNO): Likewise.
14798         (FIRST_FPR_REGNO): Likewise.
14799         (LAST_FPR_REGNO): Likewise.
14801         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
14802         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
14803         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
14804         (vcond<mode><mode>): Likewise.
14805         (vcondu<mode><mode>): Likewise.
14806         (vector_gtu<mode>): Likewise.
14807         (vector_gte<mode>): Likewise.
14808         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
14809         to prevent the compiler from converting DImode operations to TImode.
14810         (ior<mode>3): Likewise.
14811         (and<mode>3): Likewise.
14812         (one_cmpl<mode>2): Likewise.
14813         (nor<mode>3): Likewise.
14814         (andc<mode>3): Likewise.
14816         * config/rs6000/constraints.md (wt constraint): New constraint
14817         that returns VSX_REGS if TImode is allowed in VSX registers.
14819         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
14820         constant under VSX.
14822         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
14823         similar to TImode, but it is restricted to being in the GPRs.
14825         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
14826         TImode to occupy a single VSX register.
14828         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
14829         -mvsx-timode for power7/power8.
14830         (power7 cpu): Likewise.
14831         (power8 cpu): Likewise.
14833         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
14834         sure that TFmode/TDmode take up two registers if they are ever
14835         allowed in the upper VSX registers.
14836         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
14837         registers.
14838         (rs6000_init_hard_regno_mode_ok): Likewise.
14839         (rs6000_debug_reg_global): Add debugging for PTImode and wt
14840         constraint.  Print if LRA is turned on.
14841         (rs6000_option_override_internal): Give an error if -mvsx-timode
14842         and VSX is not enabled.
14843         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
14844         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
14845         to reg+offset addressing.  Use PTImode when checking offset
14846         addresses for validity.
14847         (reg_offset_addressing_ok_p): Likewise.
14848         (rs6000_legitimate_offset_address_p): Likewise.
14849         (rs6000_legitimize_address): Likewise.
14850         (rs6000_legitimize_reload_address): Likewise.
14851         (rs6000_legitimate_address_p): Likewise.
14852         (rs6000_eliminate_indexed_memrefs): Likewise.
14853         (rs6000_emit_move): Likewise.
14854         (rs6000_secondary_reload): Likewise.
14855         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
14856         reloads to fpr registers to continue to use reg+offset addressing,
14857         but 64-bit reloads to altivec registers need reg+reg addressing.
14858         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
14859         it.  Treat LO_SUM like a PLUS operation.
14860         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
14861         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
14862         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
14863         registers to share a register with a smaller sized type, since VSX
14864         puts scalars in the upper 64-bits.
14865         (print_operand): Add support for PTImode.
14866         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
14867         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
14868         registers, but don't have arithmetic support.
14869         (rs6000_memory_move_cost): Add test for VSX.
14870         (rs6000_opt_masks): Add -mvsx-timode.
14872         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
14873         for TImode.
14874         (VSs): Likewise.
14875         (VSr): Use wt constraint for TImode.
14876         (VSv): Drop TImode support.
14877         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
14878         (vsx_movti_64bit): Likewise.
14879         (vsx_movti_32bit): Likewise.
14880         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
14881         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
14882         one '?' on the appropriate output constraint.  Do not allow TImode
14883         logical operations on 32-bit systems.
14884         (vsx_ior<mode>3): Likewise.
14885         (vsx_xor<mode>3): Likewise.
14886         (vsx_one_cmpl<mode>2): Likewise.
14887         (vsx_nor<mode>3): Likewise.
14888         (vsx_andc<mode>3): Likewise.
14889         (vsx_concat_<mode>): Likewise.
14890         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
14892         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
14893         OPTION_MASK_VSX_TIMODE.
14894         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
14895         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
14897         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
14898         (TI2 iterator): New iterator for TImode, PTImode.
14899         (wd mode attribute): Add values for vector types.
14900         (movti_string): Replace TI move operations with operations for TImode
14901         and PTImode.  Add support for TImode being allowed in VSX registers.
14902         (mov<mode>_string, TImode/PTImode): Likewise.
14903         (movti_ppc64): Likewise.
14904         (mov<mode>_ppc64, TImode/PTImode): Likewise.
14905         (TI mode splitters): Likewise.
14907         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
14908         constraint.
14910 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
14912         PR tree-optimization/56355
14913         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
14914         Also handle integers with undefined overflow.
14916 2013-03-20  Catherine Moore  <clm@codesourcery.com>
14917             Maciej W. Rozycki  <macro@codesourcery.com>
14918             Tom de Vries  <tom@codesourcery.com>
14919             Nathan Sidwell  <nathan@codesourcery.com>
14920             Iain Sandoe  <iain@codesourcery.com>
14921             Nathan Froyd  <froydnj@codesourcery.com>
14922             Chao-ying Fu  <fu@mips.com>
14924         * doc/extend.texi: (micromips, nomicromips, nocompression):
14925         Document new function attributes.
14926         * doc/invoke.texi (minterlink-compressed, mmicromips,
14927         m14k, m14ke, m14kec): Document new options.
14928         (minterlink-mips16): Update documentation.
14929         * doc/md.texi (ZC, ZD): Document new constraints.
14930         * configure.ac (gcc_cv_as_micromips): Check if linker
14931         supports the .set micromips directive.
14932         * configure: Regenerate.
14933         * config.in: Regenerate.
14934         * config/mips/mips-tables.opt: Regenerate.
14935         * config/mips/micromips.md: New file.
14936         * constraints.md (ZC, ZD): New constraints.
14937         * config/mips/predicates.md (movep_src_register): New predicate.
14938         (movep_src_operand): New predicate.
14939         (non_volatile_mem_operand): New predicate.
14940         * config/mips/mips.md (multimem): New type.
14941         (length): Differentiate between 17-bit and 18-bit branch offsets.
14942         (MOVEP1, MOVEP2): New mode iterator.
14943         (mov_<load>l): Use ZC constraint.
14944         (mov_<load>r): Likewise.
14945         (mov_<store>l): Likewise.
14946         (mov_<store>r): Likewise.
14947         (*branch_equality<mode>_inverted): Add microMIPS support.
14948         (*branch_equality<mode>): Likewise.
14949         (*jump_absolute): Likewise.
14950         (indirect_jump_<mode>): Likewise.
14951         (tablejump_<mode>): Likewise.
14952         (<optab>_internal): Likewise.
14953         (sibcall_internal): Likewise.
14954         (sibcall_value_internal): Likewise.
14955         (prefetch): Use constraint ZD.
14956         * config/mips/mips.opt (minterlink-compressed): New option.
14957         (minterlink-mips16): Now an alias for minterlink-compressed.
14958         (mmicromips): New option.
14959         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
14960         (compare_and_swap_12): Likewise.
14961         (sync_add<mode>): Likewise.
14962         (sync_<optab>_12): Likewise.
14963         (sync_old_<optab>_12): Likewise.
14964         (sync_new_<optab>_12): Likewise.
14965         (sync_nand_12): Likewise.
14966         (sync_old_nand_12): Likewise.
14967         (sync_new_nand_12): Likewise.
14968         (sync_sub<mode>): Likewise.
14969         (sync_old_add<mode>): Likewise.
14970         (sync_old_sub<mode>): Likewise.
14971         (sync_new_add<mode>): Likewise.
14972         (sync_new_sub<mode>): Likewise.
14973         (sync_<optab><mode>): Likewise.
14974         (sync_old_<optab><mode>): Likewise.
14975         (sync_new_<optab><mode>): Likewise.
14976         (sync_nand<mode>): Likewise.
14977         (sync_old_nand<mode>): Likewise.
14978         (sync_new_nand<mode>): Likewise.
14979         (sync_lock_test_and_set<mode>): Likewise.
14980         (test_and_set_12): Likewise.
14981         (atomic_compare_and_swap<mode>): Likewise.
14982         (atomic_exchange<mode>_llsc): Likewise.
14983         (atomic_fetch_add<mode>_llsc): Likewise.
14984         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
14985         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
14986         (umips_save_restore_pattern_p): Likewise.
14987         (umips_load_store_pair_p): Likewise.
14988         (umips_output_load_store_pair): Likewise.
14989         (umips_movep_target_p): Likewise.
14990         (umips_12bit_offset_address_p): Likewise.
14991         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
14992         (mips_base_mips16): Rename this...
14993         (mips_base_compression_flags): ...to this. Update all uses.
14994         (mips_attribute_table): Add micromips, nomicromips and nocompression.
14995         (mips_mips16_decl_p): Delete.
14996         (mips_nomips16_decl_p): Delete.
14997         (mips_get_compress_on_flags): New function.
14998         (mips_get_compress_off_flags): New function.
14999         (mips_get_compress_mode): New function.
15000         (mips_get_compress_on_name): New function.
15001         (mips_get_compress_off_name): New function.
15002         (mips_insert_attributes): Support multiple compression types.
15003         (mips_merge_decl_attributes): Likewise.
15004         (umips_12bit_offset_address_p): New function.
15005         (mips_start_function_definition): Emit .set micromips directive.
15006         (mips_call_may_need_jalx_p): New function.
15007         (mips_function_ok_for_sibcall): Add microMIPS support.
15008         (mips_print_operand_punctuation): Support short delay slots and
15009         compact jumps.
15010         (umips_swm_mask, umips_swm_encoding): New.
15011         (umips_build_save_restore): New function.
15012         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
15013         (was_mips16_p): Remove.
15014         (old_compression_mode): New.
15015         (mips_set_compression_mode): New function.
15016         (mips_set_current_function): Add microMIPS support.
15017         (mips_option_override): Likewise.
15018         (umips_save_restore_pattern_p): New function.
15019         (umips_output_save_restore): New function.
15020         (umips_load_store_pair_p_1): New function.
15021         (umips_load_store_pair_p): New function.
15022         (umips_output_load_store_pair_1): New function.
15023         (umips_output_load_store_pair): New function.
15024         (umips_movep_target_p) New function.
15025         (mips_prepare_pch_save): Add microMIPS support.
15026         * config/mips/mips.h (TARGET_COMPRESSION): New.
15027         (TARGET_CPU_CPP_BUILTINS): Update macro
15028         to use new compression flags and to support microMIPS.
15029         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
15030         (MIPS_ARCH_FLOAT_SPEC): Likewise.
15031         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
15032         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
15033         (ASM_SPEC): Support mmicromips and mno-micromips.
15034         (M16STORE_REG_P): New macro.
15035         (MIPS_CALL): Support TARGET_MICROMIPS.
15036         (MICROMIPS_J): New macro.
15037         (mips_base_mips16): Rename this...
15038         (mips_base_compression_flags): ...to this.
15039         (UMIPS_12BIT_OFFSET_P): New macro.
15040         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
15041         (MULTILIB_DIRNAMES): Likewise.
15042 2013-03-20  Richard Biener  <rguenther@suse.de>
15044         PR tree-optimization/56661
15045         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
15046         the result does not have to be distinct.
15048 2013-03-20  Richard Biener  <rguenther@suse.de>
15050         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
15051         remap_gimple_op_r.
15053 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
15054             Steven Bosscher  <steven@gcc.gnu.org>
15056         PR rtl-optimization/56605
15057         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
15059 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
15061         PR bootstrap/56656
15062         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
15063         that require movd instead of movq.
15065 2013-03-20  Richard Biener  <rguenther@suse.de>
15067         * tree-ssa-structalias.c (struct variable_info): Add pointer
15068         to the first field of an aggregate with sub-vars.  Make
15069         this and the pointer to the next subfield its ID.
15070         (vi_next): New function.
15071         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
15072         storedanything_id, integer_id): Increment by one.
15073         (new_var_info, get_call_vi, lookup_call_clobber_vi,
15074         get_call_clobber_vi): Adjust.
15075         (solution_set_expand): Simplify and speedup.
15076         (solution_set_add): Inline into ...
15077         (set_union_with_increment): ... this.  Adjust accordingly.
15078         (do_sd_constraint): Likewise.
15079         (do_ds_constraint): Likewise.
15080         (do_complex_constraint): Simplify.
15081         (build_pred_graph): Adjust.
15082         (solve_graph): Likewise.  Simplify and speedup.
15083         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
15084         get_constraint_for_component_ref, get_constraint_for_1,
15085         first_vi_for_offset, first_or_preceding_vi_for_offset,
15086         create_function_info_for, create_variable_info_for_1,
15087         create_variable_info_for, intra_create_variable_infos): Adjust.
15088         (init_base_vars): Push NULL for ID zero.
15089         (compute_points_to_sets): Adjust.
15091 2013-03-20  Richard Biener  <rguenther@suse.de>
15093         * cfgloop.c (verify_loop_structure): Streamline and avoid
15094         ICEing on corrupt loop tree.
15095         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
15096         loop tree.
15098 2013-03-20  Richard Biener  <rguenther@suse.de>
15100         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
15101         check whether an SSA update is needed.
15103 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
15105         * config/mips/constraints.md (T): Rename to...
15106         (Yf): ...this.
15107         (U): Rename to...
15108         (Yd): ...this.
15109         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
15110         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
15112 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15114         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
15115         (*subsi3_carryin_uxtw): Likewise.
15117 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15119         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
15120         (*rorsi3_insn_uxtw): Likewise.
15122 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
15124         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
15125         (*extrsi5_insn_uxtw): Likewise.
15127 2013-03-19  Richard Biener  <rguenther@suse.de>
15129         PR tree-optimization/56273
15130         * passes.c (init_optimization_passes): Move second VRP after DOM.
15132 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
15134         * config/i386/i386.md (*movti_internal): Merge from
15135         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
15136         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
15137         nox64 isa attributes.
15139 2013-03-18  Richard Biener  <rguenther@suse.de>
15141         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
15142         (unite): Likewise.
15143         (merge_node_constraints): Likewise.
15144         (build_succ_graph): Likewise.
15145         (valid_graph_edge): Inline into single caller.
15146         (unify_nodes): Likewise.  Use bitmap_set_bit return value
15147         and cache varinfo.
15148         (scc_visit): Fix formatting and variable use.
15149         (do_sd_constraint): Use gcc_checking_assert.
15150         (do_ds_constraint): Likewise.
15151         (do_complex_constraint): Likewise.
15152         (condense_visit): Likewise.  Cleanup.
15153         (dump_pred_graph): New function.
15154         (perform_var_substitution): Dump the pred-graph before
15155         variable substitution.
15156         (find_equivalent_node): Use gcc_checking_assert.
15157         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
15159 2013-03-18  Richard Biener  <rguenther@suse.de>
15161         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
15162         Remove cond_expr_stmt_list argument and do not gimplify the
15163         built expression.
15164         (vect_loop_versioning): Adjust.
15165         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
15166         Cleanup to use less temporaries.
15167         (vect_create_data_ref_ptr): Cleanup.
15169 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15171         PR tree-optimization/56635
15172         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
15173         require types_compatible_p types.
15175 2013-03-18  Nick Clifton  <nickc@redhat.com>
15177         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
15178         spurious backslash.
15180         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
15181         Add missing line to comment describing function.
15183 2013-03-18  Richard Biener  <rguenther@suse.de>
15185         PR tree-optimization/56210
15186         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15187         Handle string / character search functions.
15188         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
15190 2013-03-18  Richard Biener  <rguenther@suse.de>
15192         PR middle-end/56483
15193         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
15194         and implement properly.
15195         * gimple.h (gimple_cond_single_var_p): Remove.
15197 2013-03-18  Richard Biener  <rguenther@suse.de>
15199         * tree-data-ref.h (find_data_references_in_loop): Declare.
15200         * tree-data-ref.c (get_references_in_stmt): Use a stack
15201         vector pre-allocated in the callers.
15202         (find_data_references_in_stmt): Adjust.
15203         (graphite_find_data_references_in_stmt): Likewise.
15204         (create_rdg_vertices): Likewise.
15205         (find_data_references_in_loop): Export.
15206         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15207         Compute dependences here...
15208         (vect_analyze_data_refs): ...not here.  When we encounter
15209         a non-vectorizable data reference in basic-block vectorization
15210         truncate the data reference vector.  Do not bother to
15211         fixup data-dependence information for gather loads.
15212         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
15213         of data references, as reported.
15215 2013-03-18  Richard Biener  <rguenther@suse.de>
15217         PR tree-optimization/3713
15218         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
15219         has_constants and expr.
15220         (stmt_has_constants): Properly valueize SSA names when deciding
15221         whether the stmt has constants.
15223 2013-03-18  Richard Biener  <rguenther@suse.de>
15225         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
15226         whole function when there is nothing to do.
15227         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
15228         * tree-vectorizer.c (vectorize_loops): Update virtual and
15229         loop-closed SSA once.
15230         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
15232 2013-03-18  Richard Biener  <rguenther@suse.de>
15234         PR middle-end/56113
15235         * domwalk.c (bb_postorder): New global static.
15236         (cmp_bb_postorder): New function.
15237         (walk_dominator_tree): Replace scheme imposing an order for
15238         visiting dominator sons by one sorting them at the time they
15239         are pushed on the stack.
15241 2013-03-18  Richard Biener  <rguenther@suse.de>
15243         PR tree-optimization/39326
15244         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
15245         (struct mem_ref): Replace mem member with ao_ref typed member.
15246         (MEM_ANALYZABLE): Adjust.
15247         (memref_eq): Likewise.
15248         (mem_ref_alloc): Likewise.
15249         (gather_mem_refs_stmt): Likewise.
15250         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
15251         (execute_sm_if_changed_flag_set): Adjust.
15252         (execute_sm): Likewise.
15253         (ref_always_accessed_p): Likewise.
15254         (refs_independent_p): Likewise.
15255         (can_sm_ref_p): Likewise.
15257 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
15259         PR c/56566
15260         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
15261         return 1 even for !unsignedp.
15263 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15265         * config/i386/i386.md (isa): Add x64 and nox64.
15266         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
15267         (*pushtf): Enable *roF alternative for x64 isa only.
15268         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
15269         mode attribute of integer alternatives to DImode for TARGET_64BIT.
15270         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
15271         (*movtf_internal): Merge from *movtf_internal_rex64 and
15272         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
15273         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
15274         nox64 isa attributes.
15275         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
15276         nox64 isa attributes.
15277         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
15279 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
15281         * config/alpha/alpha.c (TARGET_LRA_P): New define.
15283 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
15285         PR target/56640
15286         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
15287         class names.  Remove trailing comma after "ALL_REGS".
15289 2013-03-16  Jan Hubicka  <jh@suse.cz>
15291         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
15292         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
15293         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
15294         of cgraph_get_create_node.
15295         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15297 2013-03-16  Jason Merrill  <jason@redhat.com>
15299         PR debug/49090
15300         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
15301         with DW_AT_default_value.
15303 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
15305         * BASE-VER: Set to 4.9.0.
15307 2013-03-14  Andi Kleen  <ak@linux.intel.com>
15309         PR target/56619
15310         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
15311         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
15312         Document _x* TSX intrinsics.
15314 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
15315             David Holsgrove  <david.holsgrove@xilinx.com>
15317         * configure.ac: Add MicroBlaze TLS support detection.
15318         * configure: Regenerate.
15319         * config/microblaze/microblaze-protos.h
15320         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
15321         symbol_mentioned_p, label_mentioned_p): Add prototypes.
15322         * config/microblaze/microblaze.c (microblaze_address_type): Add
15323         ADDRESS_TLS and tls_reloc address types.
15324         (microblaze_address_info): Add tls_reloc.
15325         (TARGET_HAVE_TLS): Define.
15326         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
15327          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
15328          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
15329         load_tls_operand,  microblaze_call_tls_get_addr,
15330         microblaze_legitimize_tls_address): New functions.
15331         (microblaze_classify_unspec): Handle UNSPEC_TLS.
15332         (get_base_reg): Use microblaze_tls_symbol_p.
15333         (microblaze_classify_address): Handle TLS.
15334         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
15335         label_mentioned_p and microblaze_tls_referenced_p.
15336         (microblaze_legitimize_address): Handle TLS.
15337         (microblaze_address_insns): Handle ADDRESS_TLS.
15338         (pic_address_needs_scratch): Handle TLS.
15339         (print_operand_address): Handle TLS.
15340         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
15341         (microblaze_expand_move): Handle TLS.
15342         (microblaze_legitimate_constant_p): Check
15343         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
15344         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
15345         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
15346         (PIC_OFFSET_TABLE_REGNUM): Set.
15347         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
15348         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
15349         (addsi3, movsi_internal2, movdf_internal): Update constraints
15350         * config/microblaze/predicates.md (arith_plus_operand): Define
15351         (move_operand): Redefine as move_src_operand,
15352         check microblaze_tls_referenced_p.
15354 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15356         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
15357         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
15359 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
15361         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
15362         CC mode for AND.
15364 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
15366         PR tree-optimization/53265
15367         * common.opt (Waggressive-loop-optimizations): New option.
15368         * tree-ssa-loop-niter.c: Include tree-pass.h.
15369         (do_warn_aggressive_loop_optimizations): New function.
15370         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
15371         if number_of_latch_executions returned constant.
15372         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
15373         early.  If number_of_latch_executions returned constant, set
15374         nb_iterations_upper_bound back to it.
15375         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
15376         field.
15377         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
15378         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
15380         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
15381         (MULTILIB_OSDIRNAMES): Set.
15382         * genmultilib: If defaultosdirname doesn't start with :: , set
15383         defaultosdirname2 instead, clear it and emit two . multilib_raw
15384         entries instead of just one.
15386 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
15388         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15389         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
15390         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
15391         (SUBTARGET_OVERRIDE_OPTIONS): New.
15393 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
15395         PR target/49880
15396         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
15397         (musermode): Convert to Var(TARGET_USERMODE).
15398         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
15399         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
15400         * config/sh/sh.c (sh_option_override): Use
15401         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
15402         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
15403         condition.
15404         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
15405         TARGET_SH4.
15406         (udivsi3_i4_single, divsi3_i4_single): Use
15407         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
15409 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
15411         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
15412         default setting.
15414 2013-03-13  Richard Biener  <rguenther@suse.de>
15416         PR tree-optimization/56608
15417         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
15418         calls when vectorizing basic-blocks.
15420 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
15422         PR plugins/45078
15423         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
15424         tm_file.
15426 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
15428         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
15430 2013-03-11  Jan Hubicka  <jh@suse.cz>
15432         PR lto/56557
15433         * lto-streamer-out.c (output_symbol_p): Skip references from
15434         constructors of external variables.
15436 2013-03-11  Jan Hubicka  <jh@suse.cz>
15438         PR middle-end/56571
15439         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
15440         from pseudos.
15441         * emit-rtl.c (verify_rtx_sharing): Likewise.
15442         (copy_insn_1): Likewise.
15443         * rtl.c (copy_rtx): Likewise.
15445 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
15447         PR target/56591
15448         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
15449         output_operand_lossage message.
15451 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
15453         PR target/56470
15454         * arm.c (shift_op): Validate RTL pattern on the fly.
15455         (arm_print_operand, case 'S'): Don't use shift_operator to validate
15456         the RTL.
15458 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
15460         PR target/56347
15461         * config/pa/pa.md (call_value): Check for calls to powf and direct to
15462         new call patterns that clobber %fr12.
15463         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
15464         split and postreload patterns.
15465         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
15466         registers %fr12 and %fr12R as call used.
15468 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
15470         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
15471         (canon_address, record_store, replace_read, check_mem_read_rtx,
15472         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
15473         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
15474         rest_of_handle_dse): Likewise.
15476 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
15478         PR middle-end/56524
15479         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
15480         Add base_optabs.
15481         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
15482         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
15483         (save_optabs_if_changed): Replace with...
15484         (init_tree_optimization_optabs): ...this.
15485         * optabs.c (save_optabs_if_changed): Rename to...
15486         (init_tree_optimization_optabs): ...this.  Take the optimization node
15487         as argument.  Do nothing if the base optabs are already correct.
15488         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
15489         to recompute optabs.
15490         * function.h (function): Remove optabs field.
15491         * function.c (invoke_set_current_function_hook): Call
15492         init_tree_optimization_optabs.  Use the result to initialize
15493         this_fn_optabs.
15495 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
15497         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
15498         if GTMA_HAS_NO_INSTRUMENTATION.
15499         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
15500         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
15501         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
15502         * gimple-pretty-print.c (dump_gimple_transaction): Handle
15503         GTMA_HAS_NO_INSTRUMENTATION.
15505 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
15507         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
15508         libasan_preinit.o.
15510 2013-03-08  Marek Polacek  <polacek@redhat.com>
15511             Jakub Jelinek  <jakub@redhat.com>
15513         PR tree-optimization/56478
15514         * predict.c (is_comparison_with_loop_invariant_p): Change the
15515         type of loop_step to tree.
15516         (predict_loops): Adjust.
15517         (predict_iv_comparison): Perform the computations on double_ints.
15519 2013-03-08  Richard Biener  <rguenther@suse.de>
15521         PR tree-optimization/56570
15522         * tree-cfg.c (verify_expr_location_1): Verify locations for
15523         DECL_DEBUG_EXPR.
15524         * tree-sra.c (create_access_replacement): Strip locations
15525         from DECL_DEBUG_EXPRs.
15527 2013-03-08  Richard Biener  <rguenther@suse.de>
15529         * tree-inline.c (expand_call_inline): Do not associate
15530         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
15531         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
15533 2013-03-08  Richard Biener  <rguenther@suse.de>
15535         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
15536         or block changes with -Og.  Fix for location / block encoding
15537         changes and PHI arguments with locations.
15539 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
15541         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
15542         for all counters.
15543         (struct output_info): Likewise.
15544         (register_overhead): Remove bad gcc_assert.
15545         (bitmap_find_bit): If there is only a single bitmap element, do not
15546         count a miss as a search.
15547         (print_statistics): Update for counter type changes.
15548         (dump_bitmap_statistics): Likewise.  Print headers such that they
15549         are properly lined up with the printed counters.
15551 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
15553         PR tree-optimization/56559
15554         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
15555         check that it has only a single use.
15557 2013-03-07  Richard Biener  <rguenther@suse.de>
15559         * doc/invoke.texi (fwhole-program): Discourage use in combination
15560         with -flto.
15562 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
15564         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
15566         PR tree-optimization/56539
15567         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
15568         instead of GSI_CONTINUE_LINKING as last argument to
15569         force_gimple_operand_gsi.  Adjust function comment.
15571         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
15572         aarch64-cores.def.
15574         PR middle-end/56548
15575         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
15576         promoted mode, convert the result back to the original mode.
15578 2013-03-06  Richard Biener  <rguenther@suse.de>
15580         PR middle-end/56294
15581         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
15582         (insert_updated_phi_nodes_compare_uids): New function.
15583         (update_ssa): Sort symbols_to_rename after UID before
15584         traversing it to insert PHI nodes.
15586 2013-03-06  Richard Biener  <rguenther@suse.de>
15588         PR middle-end/50494
15589         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
15590         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
15592         Revert
15593         2013-02-13  Richard Biener  <rguenther@suse.de>
15595         PR lto/50494
15596         * varasm.c (output_constant_def_1): Get the decl representing
15597         the constant as argument.
15598         (output_constant_def): Wrap output_constant_def_1.
15599         (make_decl_rtl): Use output_constant_def_1 with the decl
15600         representing the constant.
15601         (build_constant_desc): Optionally re-use a decl already
15602         representing the constant.
15603         (tree_output_constant_def): Adjust.
15605 2013-03-06  Joey Ye  <joey.ye@arm.com>
15607         PR lto/50293
15608         * gcc.c (convert_white_space): New function.
15609         (main): Handles white space in function name.
15611 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
15613         PR target/56529
15614         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
15615         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
15616         to SH_DIV_CALL_TABLE for TARGET_SH2.
15617         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
15618         list.
15619         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
15620         call-table options.
15622 2013-03-05  Sterling Augustine  <saugustine@google.com>
15623             Cary Coutant  <ccoutant@google.com>
15625         PR debug/55364
15626         * dwarf2out.c (resolve_addr): Don't call
15627         remove_loc_list_addr_table_entries a second time for the same
15628         expression.
15630 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15632         PR debug/56510
15633         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
15634         (avoid_complex_debug_insns): New function.
15635         (expand_debug_locations): Call it.
15637         PR rtl-optimization/56484
15638         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
15639         lifetimes of hard registers on small register class machines.
15641 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15643         * config/microblaze/microblaze-protos.h: Rename
15644         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
15645         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
15646         fast_interrupt.
15647         (microblaze_fast_interrupt_function_p): New function.
15648         (microblaze_is_interrupt_handler): Rename to
15649         microblaze_is_interrupt_variant and add fast_interrupt check.
15650         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
15651         (save_restore_insns): Likewise.
15652         (compute_frame_size): Likewise.
15653         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
15654         (microblaze_globalize_label): Likewise.
15655         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
15656         * config/microblaze/microblaze.md: Use wrapper
15657         microblaze_is_interrupt_variant.
15659 2013-03-05  Kai Tietz  <ktietz@redhat.com>
15661         * sdbout.c (sdbout_one_type): Switch to current function's section
15662         supporting cold/hot.
15664 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
15666         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
15667         -mxl-reorder.
15669 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15671         PR middle-end/56461
15672         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
15673         if VALGRIND_GET_VBITS is defined, temporarily make object
15674         memory all defined, and restore previous valgrind addressability
15675         and definability afterwards.  Free this_object at the end.
15677         PR middle-end/56461
15678         * lra.c (lra): Call lra_clear_live_ranges if live_p,
15679         right before calling lra_create_live_ranges, also call it
15680         when clearing live_p.  Only call lra_clear_live_ranges
15681         at the end if live_p.
15683         PR middle-end/56461
15684         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
15686 2013-03-05  Richard Biener  <rguenther@suse.de>
15688         PR tree-optimization/56521
15689         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
15690         value-id.
15692 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
15694         PR c++/55135
15695         * except.h (remove_unreachable_eh_regions): New prototype.
15696         * except.c (remove_eh_handler_splicer): New function, split out
15697         of remove_eh_handler.
15698         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
15699         warning about running it on many EH regions one at a time.
15700         (remove_unreachable_eh_regions_worker): New function, walk the
15701         EH tree in depth-first order and remove non-marked regions.
15702         (remove_unreachable_eh_regions): New function.
15703         * tree-eh.c (mark_reachable_handlers): New function, split out
15704         from remove_unreachable_handlers.
15705         (remove_unreachable_handlers): Use mark_reachable_handlers and
15706         remove_unreachable_eh_regions.
15707         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
15708         and remove_unreachable_eh_regions.
15710 2013-03-05  Richard Biener  <rguenther@suse.de>
15712         PR middle-end/56525
15713         * loop-init.c (fix_loop_structure): Remove loops in two stages,
15714         not freeing them until the end.
15716 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15718         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
15720 2013-03-05  Richard Biener  <rguenther@suse.de>
15722         PR tree-optimization/56270
15723         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
15724         of loads after scheduling an SLP instance.
15726 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15728         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
15729         tic6x.exp.
15730         (check_gcc_parallelize): Run guality.exp as a separate job from
15731         vect.exp with unsorted.exp and $(dg_target_exps) separately from
15732         struct-layout-1.exp with stackalign.exp.
15734         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
15736         PR middle-end/56461
15737         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
15738         load_index sbitmap even if some bit in it isn't set.
15740         PR middle-end/56461
15741         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
15742         (discover_iteration_bound_by_body_walk): Change queues to
15743         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
15744         spelling in comment.  Call safe_push on queues[bound_index] directly.
15745         Release queues[queue_index] in every iteration unconditionally.
15746         Release bounds vector.
15748         PR middle-end/56461
15749         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
15750         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
15751         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
15752         inner_phis vector.
15754 2013-03-05  Richard Biener  <rguenther@suse.de>
15756         PR lto/56515
15757         * tree-inline.c (remap_blocks_to_null): New function.
15758         (expand_call_inline): When expanding a call stmt without
15759         an associated block inline remap all callee blocks to NULL.
15761 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
15763         PR rtl-optimization/56494
15764         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
15765         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
15766         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
15768         PR middle-end/56461
15769         * sel-sched-ir.c (free_sched_pools): Release
15770         succs_info_pool.stack[succs_info_pool.max_top] vectors too
15771         if succs_info_pool.max_top isn't -1.
15773         PR bootstrap/56509
15774         * opts.c (opts_obstack, opts_concat): Moved to...
15775         * opts-common.c (opts_obstack, opts_concat): ... here.
15777 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15779         PR middle-end/56461
15780         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
15782 2013-03-04  Martin Jambor  <mjambor@suse.cz>
15784         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
15785         all appropriate places.
15787 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
15789         PR tree-optimization/56424
15790         * ipa-split.c (split_function): Do not set the RSO flag if result is
15791         not by reference and its type is a register type.
15793 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
15795         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
15796         (microblaze_legitimate_pic_operand): Likewise
15797         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
15798         new function microblaze_legitimate_pic_operand
15799         * config/microblaze/microblaze-protos.h
15800         (microblaze_legitimate_pic_operand): Declare.
15802 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
15804         * config/microblaze/predicates.md (call_insn_simple_operand):
15805         New predicate for supported rtx code types.
15806         * config/microblaze/microblaze.md (call_internal1): Use
15807         call_insn_simple_operand predicate.
15809 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
15811         PR middle-end/56461
15812         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
15813         partitions.ordered_remove.
15815         PR middle-end/56461
15816         * tree-vect-stmts.c (vectorizable_conversion): Don't call
15817         vec_oprnds0.create (1) for modifier == NONE.
15819         PR middle-end/56461
15820         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
15821         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
15822         vec_oprnds1 right before pushing anything to it for
15823         scalar_shift_arg.
15825         PR middle-end/56461
15826         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
15827         set nbbs to 0 instead of having separate code path.
15828         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
15829         instead of false as last argument if returning NULL.
15831 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
15833         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
15834         the attribute is now called "target" instead of "option".
15835         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
15836         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
15837         attribute/pragma name for TARGET_OPTION_VALID_P and
15838         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
15839         * doc/tm.texi: Regenerated.
15841 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
15843         * config/microblaze/microblaze.c:
15844         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
15845         * config/microblaze/microblaze.h: Add -mxl-reorder to
15846         DRIVER_SELF_SPECS.
15847         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
15848         instructions emitted if TARGET_REORDER.
15849         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
15850         or 0 for -m/-mno case, but initialises as 2 to detect default use case
15851         separately.
15853 2013-03-01  Xinliang David Li  <davidxl@google.com>
15855         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
15856         walk length.
15858 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
15860         PR middle-end/56461
15861         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
15862         vector even when returning true.  Fix up function comment formatting.
15864         PR middle-end/56461
15865         * ira-build.c (ira_loop_nodes_count): New variable.
15866         (create_loop_tree_nodes): Initialize it.
15867         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
15869         PR middle-end/56461
15870         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
15871         method on dr_chain and result_chain.
15872         * tree-vect-stmts.c (vectorizable_store): Only call
15873         result_chain.create if j == 0.
15875         PR middle-end/56461
15876         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
15877         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
15878         before overwriting it.
15880 2013-03-01  Tobias Burnus  <burnus@net-b.de>
15882         * doc/extended.texi (C Extensions): Change order in @menu
15883         to match @node.
15884         (Other MIPS Built-in Functions): Move last MIPS entry before
15885         "picoChip Built-in Functions".
15886         (SH Built-in Functions): Move after RX Built-in Functions.
15887         * doc/gcc.texi (Introduction): Change order in @menu
15888         to match @node.
15889         * doc/md.texi (Constraints): Ditto.
15890         * gty.texi (Type Information): Ditto.
15891         (User-provided marking routines for template types): Make
15892         subsection.
15893         * doc/invoke.texi (AArch64 Options): Move before
15894         "Adapteva Epiphany Options".
15896 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
15897             Jakub Jelinek  <jakub@redhat.com>
15899         PR sanitizer/56454
15900         * asan.c (gate_asan): Lookup no_sanitize_address instead of
15901         no_address_safety_analysis attribute.
15902         * doc/extend.texi (no_address_safety_attribute): Rename to
15903         no_sanitize_address attribute, mention no_address_safety_analysis
15904         attribute as deprecated alias.
15906 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
15908         PR middle-end/56461
15909         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
15910         type to vec<vec<tree> > *.
15911         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
15912         to be vec<tree> instead of vec<tree> *, set vec_defs
15913         to vNULL and call vec_defs.create (number_of_vects), adjust other
15914         uses of vec_defs.
15915         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
15916         vectorizable_condition): Adjust vect_get_slp_defs callers.
15918 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15920         * config/aarch64/aarch64.c
15921         (aarch64_float_const_representable): Remove unused variable.
15923 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15925         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
15927 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15929         * config/aarch64/aarch64-builtins.c
15930         (aarch64_init_simd_builtins): Make static.
15932 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
15934         * config/aarch64/aarch64.c
15935         (aarch64_simd_make_constant): Make static.
15937 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15939         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
15940         with no initialization to the RHS of debug statements.
15942 2013-02-28  Martin Jambor  <mjambor@suse.cz>
15944         PR tree-optimization/56294
15945         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
15946         Adjust dumping.
15947         (get_access_replacement): Do not call create_access_replacement.
15948         Assert a replacement exists.
15949         (get_repl_default_def_ssa_name): Create the replacement declaration
15950         itself.
15952 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
15954         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
15955         final_end_function.
15957 2013-02-28  Marek Polacek  <polacek@redhat.com>
15959         PR rtl-optimization/56466
15960         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
15961         if we're changing a loop.
15962         (peel_loops_completely): Likewise.
15964 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
15966         PR c++/55813
15967         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15969 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15971         PR target/56445
15972         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
15973         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
15974         INTX_FTYPE_FX, FX_FTYPE_INTX.
15975         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
15977 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
15979         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
15980         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
15981         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
15982         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
15983         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
15984         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
15985         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
15986         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
15987         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
15988         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
15989         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
15990         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
15991         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
15992         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
15993         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
15994         (avrxmega6): Increase max flash segments from 5 to 6.
15995         * config/avr/t-multilib: Regenerate.
15996         * config/avr/avr-tables.opt: Regenerate.
15997         * doc/avr-mmcu.texi: Regenerate.
15999 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
16001         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
16002         (avr_device_to_arch): Rename to avr_device_to_ld.
16003         (avr_device_to_as): New prototype.
16004         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
16005         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
16006         * config/avr/driver-avr.c (avr_device_to_as): New.
16007         (avr_device_to_arch): Rename to avr_device_to_ld.
16009 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16011         PR middle-end/56461
16012         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
16013         method on dr_chain and result_chain.
16015         PR middle-end/56461
16016         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
16017         pointer_set_destroy on not_executed_last_iteration.
16019         PR middle-end/56461
16020         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
16022         PR middle-end/56461
16023         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
16024         FOR_EACH_DEFINED_FUNCTION when freeing state.
16026         PR middle-end/56461
16027         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
16028         pool_free.
16029         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
16030         overwriting it.
16032         PR middle-end/56461
16033         * ipa-cp.c (decide_whether_version_node): Call vec_free on
16034         known_aggs[i].items and release known_aggs vector.
16036         PR middle-end/56461
16037         * ipa-reference.c (propagate): Free node_info even for alias nodes.
16039 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16041         * config/microblaze/microblaze.c (microblaze_emit_compare):
16042         Use xor for EQ/NE comparisions.
16043         * config/microblaze/microblaze.md (cstoresf4): Add constraints
16044         (cbranchsf4): Adjust operator to comparison_operator.
16046 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16048         PR middle-end/56461
16049         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
16050         vector.
16051         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
16052         vec_safe_push, always update *slot.
16053         (redirect_edge_var_map_clear): Use vec_free.
16054         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
16055         (free_var_map_entry): Use vec_free.
16056         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
16057         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
16059 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
16061         PR middle-end/45472
16062         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
16063         when the may_trap_p bit of the exprs being merged differs.
16064         Reorder tests for speculativeness in the logical and operator.
16066 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
16068         * incpath.c (add_standard_paths): Use reconcat instead of concat
16069         where appropriate and avoid leaking memory.
16071         * opts.h: Include obstack.h.
16072         (opts_concat): New prototype.
16073         (opts_obstack): New declaration.
16074         * opts.c (opts_concat): New function.
16075         (opts_obstack): New variable.
16076         (init_options_struct): Call gcc_init_obstack on opts_obstack.
16077         (finish_options): Use opts_concat instead of concat
16078         and XOBNEWVEC instead of XNEWVEC.
16079         * opts-common.c (generate_canonical_option, decode_cmdline_option,
16080         generate_option): Likewise.
16081         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
16082         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
16084         PR target/56455
16085         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
16086         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
16088 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16090         PR middle-end/56461
16091         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
16093 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16095         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
16096         (arm_block_move_unaligned_straight): Likewise.
16097         (arm_adjust_block_mem): Likewise.
16099 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
16101         PR target/48901
16102         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
16103         temp, cond and label.
16104         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
16106         PR target/52500
16107         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16108         * config/c6x/c6x.h (dbx_register_map): Update declaration.
16110         PR target/52501
16111         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16112         of prologue/epilogue functions.
16114         PR target/52550
16115         * config/tilegx/tilegx.c (tilegx_expand_prologue):
16116         Remove unused variable cfa_offset.
16117         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16119         PR target/54639
16120         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16121         type promotion to unsigned.
16123         PR target/54640
16124         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
16125         for HOST_WIDE_INT of 32 bit / same size as int.
16126         (arm_block_move_unaligned_straight): Likewise.
16127         (arm_adjust_block_mem): Likewise.
16129         PR target/54662
16130         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
16131         ALL_CFLAGS.
16133 2013-02-26  Marek Polacek  <polacek@redhat.com>
16135         PR tree-optimization/56426
16136         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
16138 2013-02-26  Richard Biener  <rguenther@suse.de>
16140         PR target/56444
16141         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
16142         unused variable loops.
16144 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
16146         PR tree-optimization/56448
16147         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
16148         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
16149         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
16150         later operands of the references, or even first operand for
16151         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
16153         PR tree-optimization/56443
16154         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
16155         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
16156         to type_for_mode langhook.
16158 2013-02-25  Matt Turner  <mattst88@gmail.com>
16160         * doc/invoke.texi: Document r4700.
16162 2013-02-25  Richard Biener  <rguenther@suse.de>
16164         PR tree-optimization/56175
16165         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
16166         split out from ...
16167         (simplify_bitwise_binary): ... here.  Also guard the conversion
16168         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
16170 2013-02-25  Catherine Moore  <clm@codesourcery.com>
16172         Revert:
16173         2013-02-24  Catherine Moore  <clm@codesourcery.com>
16174             Maciej W. Rozycki  <macro@codesourcery.com>
16175             Tom de Vries  <tom@codesourcery.com>
16176             Nathan Sidwell  <nathan@codesourcery.com>
16177             Iain Sandoe  <iain@codesourcery.com>
16178             Nathan Froyd  <froydnj@codesourcery.com>
16179             Chao-ying Fu  <fu@mips.com>
16181         * doc/extend.texi: (micromips, nomicromips, nocompression):
16182         Document new function attributes.
16183         * doc/invoke.texi (minterlink-compressed, mmicromips,
16184         m14k, m14ke, m14kec): Document new options.
16185         (minterlink-mips16): Update documentation.
16186         * doc/md.texi (ZC, ZD): Document new constraints.
16187         * configure.ac (gcc_cv_as_micromips): Check if linker
16188         supports the .set micromips directive.
16189         * configure: Regenerate.
16190         * config.in: Regenerate.
16191         * config/mips/mips-tables.opt: Regenerate.
16192         * config/mips/micromips.md: New file.
16193         * constraints.md (ZC, AD): New constraints.
16194         * config/mips/predicates.md (movep_src_register): New predicate.
16195         (movep_src_operand): New predicate.
16196         (non_volatile_mem_operand): New predicate.
16197         * config/mips/mips.md (multimem): New type.
16198         (length): Differentiate between 17-bit and 18-bit branch offsets.
16199         (MOVEP1, MOVEP2): New mode iterator.
16200         (mov_<load>l): Use ZC constraint.
16201         (mov_<load>r): Likewise.
16202         (mov_<store>l): Likewise.
16203         (mov_<store>r): Likewise.
16204         (*branch_equality<mode>_inverted): Add microMIPS support.
16205         (*branch_equality<mode>): Likewise.
16206         (*jump_absolute): Likewise.
16207         (indirect_jump_<mode>): Likewise.
16208         (tablejump_<mode>): Likewise.
16209         (<optab>_internal): Likewise.
16210         (sibcall_internal): Likewise.
16211         (sibcall_value_internal): Likewise.
16212         (prefetch): Use constraint ZD.
16213         * config/mips/mips.opt (minterlink-compressed): New option.
16214         (minterlink-mips16): Now an alias for minterlink-compressed.
16215         (mmicromips): New option.
16216         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16217         (compare_and_swap_12): Likewise.
16218         (sync_add<mode>): Likewise.
16219         (sync_<optab>_12): Likewise.
16220         (sync_old_<optab>_12): Likewise.
16221         (sync_new_<optab>_12): Likewise.
16222         (sync_nand_12): Likewise.
16223         (sync_old_nand_12): Likewise.
16224         (sync_new_nand_12): Likewise.
16225         (sync_sub<mode>): Likewise.
16226         (sync_old_add<mode>): Likewise.
16227         (sync_old_sub<mode>): Likewise.
16228         (sync_new_add<mode>): Likewise.
16229         (sync_new_sub<mode>): Likewise.
16230         (sync_<optab><mode>): Likewise.
16231         (sync_old_<optab><mode>): Likewise.
16232         (sync_new_<optab><mode>): Likewise.
16233         (sync_nand<mode>): Likewise.
16234         (sync_old_nand<mode>): Likewise.
16235         (sync_new_nand<mode>): Likewise.
16236         (sync_lock_test_and_set<mode>): Likewise.
16237         (test_and_set_12): Likewise.
16238         (atomic_compare_and_swap<mode>): Likewise.
16239         (atomic_exchange<mode>_llsc): Likewise.
16240         (atomic_fetch_add<mode>_llsc): Likewise.
16241         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16242         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16243         (umips_save_restore_pattern_p): Likewise.
16244         (umips_load_store_pair_p): Likewise.
16245         (umips_output_load_store_pair): Likewise.
16246         (umips_movep_target_p): Likewise.
16247         (umips_12bit_offset_address_p): Likewise.
16248         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16249         (mips_base_mips16): Rename this...
16250         (mips_base_compression_flags): ...to this. Update all uses.
16251         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16252         (mips_mips16_decl_p): Delete.
16253         (mips_nomips16_decl_p): Delete.
16254         (mips_get_compress_on_flags): New function.
16255         (mips_get_compress_off_flags): New function.
16256         (mips_get_compress_mode): New function.
16257         (mips_get_compress_on_name): New function.
16258         (mips_get_compress_off_name): New function.
16259         (mips_insert_attributes): Support multiple compression types.
16260         (mips_merge_decl_attributes): Likewise.
16261         (umips_12bit_offset_address_p): New function.
16262         (mips_start_function_definition): Emit .set micromips directive.
16263         (mips_call_may_need_jalx_p): New function.
16264         (mips_function_ok_for_sibcall): Add microMIPS support.
16265         (mips_print_operand_punctuation): Support short delay slots and
16266         compact jumps.
16267         (umips_swm_mask, umips_swm_encoding): New.
16268         (umips_build_save_restore): New function.
16269         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16270         (was_mips16_p): Remove.
16271         (old_compression_mode): New.
16272         (mips_set_compression_mode): New function.
16273         (mips_set_current_function): Add microMIPS support.
16274         (mips_option_override): Likewise.
16275         (umips_save_restore_pattern_p): New function.
16276         (umips_output_save_restore): New function.
16277         (umips_load_store_pair_p_1): New function.
16278         (umips_load_store_pair_p): New function.
16279         (umips_output_load_store_pair_1): New function.
16280         (umips_output_load_store_pair): New function.
16281         (umips_movep_target_p) New function.
16282         (mips_prepare_pch_save): Add microMIPS support.
16283         * config/mips/mips.h (TARGET_COMPRESSION): New.
16284         (TARGET_CPU_CPP_BUILTINS): Update macro
16285         to use new compression flags and to support microMIPS.
16286         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16287         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16288         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16289         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16290         (ASM_SPEC): Support mmicromips and mno-micromips.
16291         (M16STORE_REG_P): New macro.
16292         (MIPS_CALL): Support TARGET_MICROMIPS.
16293         (MICROMIPS_J): New macro.
16294         (mips_base_mips16): Rename this...
16295         (mips_base_compression_flags): ...to this.
16296         (UMIPS_12BIT_OFFSET_P): New macro.
16297         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16298         (MULTILIB_DIRNAMES): Likewise.
16300 2013-02-25  Tom de Vries  <tom@codesourcery.com>
16302         PR rtl-optimization/56131
16303         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
16304         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
16305         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
16307 2013-02-25  Tobias Burnus  <burnus@net-b.de>
16309         * doc/invoke.texi (-fsanitize=): Move from optimization
16310         to debugging options.
16312 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16314         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
16316 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
16317             Alexander Monakov  <amonakov@ispras.ru>
16319         PR middle-end/56077
16320         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
16321         flush pending lists also on non-jumps.  Adjust comment.
16323 2013-02-24  Catherine Moore  <clm@codesourcery.com>
16324             Maciej W. Rozycki  <macro@codesourcery.com>
16325             Tom de Vries  <tom@codesourcery.com>
16326             Nathan Sidwell  <nathan@codesourcery.com>
16327             Iain Sandoe  <iain@codesourcery.com>
16328             Nathan Froyd  <froydnj@codesourcery.com>
16329             Chao-ying Fu  <fu@mips.com>
16331         * doc/extend.texi: (micromips, nomicromips, nocompression):
16332         Document new function attributes.
16333         * doc/invoke.texi (minterlink-compressed, mmicromips,
16334         m14k, m14ke, m14kec): Document new options.
16335         (minterlink-mips16): Update documentation.
16336         * doc/md.texi (ZC, ZD): Document new constraints.
16337         * configure.ac (gcc_cv_as_micromips): Check if linker
16338         supports the .set micromips directive.
16339         * configure: Regenerate.
16340         * config.in: Regenerate.
16341         * config/mips/mips-tables.opt: Regenerate.
16342         * config/mips/micromips.md: New file.
16343         * constraints.md (ZC, AD): New constraints.
16344         * config/mips/predicates.md (movep_src_register): New predicate.
16345         (movep_src_operand): New predicate.
16346         (non_volatile_mem_operand): New predicate.
16347         * config/mips/mips.md (multimem): New type.
16348         (length): Differentiate between 17-bit and 18-bit branch offsets.
16349         (MOVEP1, MOVEP2): New mode iterator.
16350         (mov_<load>l): Use ZC constraint.
16351         (mov_<load>r): Likewise.
16352         (mov_<store>l): Likewise.
16353         (mov_<store>r): Likewise.
16354         (*branch_equality<mode>_inverted): Add microMIPS support.
16355         (*branch_equality<mode>): Likewise.
16356         (*jump_absolute): Likewise.
16357         (indirect_jump_<mode>): Likewise.
16358         (tablejump_<mode>): Likewise.
16359         (<optab>_internal): Likewise.
16360         (sibcall_internal): Likewise.
16361         (sibcall_value_internal): Likewise.
16362         (prefetch): Use constraint ZD.
16363         * config/mips/mips.opt (minterlink-compressed): New option.
16364         (minterlink-mips16): Now an alias for minterlink-compressed.
16365         (mmicromips): New option.
16366         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
16367         (compare_and_swap_12): Likewise.
16368         (sync_add<mode>): Likewise.
16369         (sync_<optab>_12): Likewise.
16370         (sync_old_<optab>_12): Likewise.
16371         (sync_new_<optab>_12): Likewise.
16372         (sync_nand_12): Likewise.
16373         (sync_old_nand_12): Likewise.
16374         (sync_new_nand_12): Likewise.
16375         (sync_sub<mode>): Likewise.
16376         (sync_old_add<mode>): Likewise.
16377         (sync_old_sub<mode>): Likewise.
16378         (sync_new_add<mode>): Likewise.
16379         (sync_new_sub<mode>): Likewise.
16380         (sync_<optab><mode>): Likewise.
16381         (sync_old_<optab><mode>): Likewise.
16382         (sync_new_<optab><mode>): Likewise.
16383         (sync_nand<mode>): Likewise.
16384         (sync_old_nand<mode>): Likewise.
16385         (sync_new_nand<mode>): Likewise.
16386         (sync_lock_test_and_set<mode>): Likewise.
16387         (test_and_set_12): Likewise.
16388         (atomic_compare_and_swap<mode>): Likewise.
16389         (atomic_exchange<mode>_llsc): Likewise.
16390         (atomic_fetch_add<mode>_llsc): Likewise.
16391         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
16392         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
16393         (umips_save_restore_pattern_p): Likewise.
16394         (umips_load_store_pair_p): Likewise.
16395         (umips_output_load_store_pair): Likewise.
16396         (umips_movep_target_p): Likewise.
16397         (umips_12bit_offset_address_p): Likewise.
16398         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
16399         (mips_base_mips16): Rename this...
16400         (mips_base_compression_flags): ...to this. Update all uses.
16401         (mips_attribute_table): Add micromips, nomicromips and nocompression.
16402         (mips_mips16_decl_p): Delete.
16403         (mips_nomips16_decl_p): Delete.
16404         (mips_get_compress_on_flags): New function.
16405         (mips_get_compress_off_flags): New function.
16406         (mips_get_compress_mode): New function.
16407         (mips_get_compress_on_name): New function.
16408         (mips_get_compress_off_name): New function.
16409         (mips_insert_attributes): Support multiple compression types.
16410         (mips_merge_decl_attributes): Likewise.
16411         (umips_12bit_offset_address_p): New function.
16412         (mips_start_function_definition): Emit .set micromips directive.
16413         (mips_call_may_need_jalx_p): New function.
16414         (mips_function_ok_for_sibcall): Add microMIPS support.
16415         (mips_print_operand_punctuation): Support short delay slots and
16416         compact jumps.
16417         (umips_swm_mask, umips_swm_encoding): New.
16418         (umips_build_save_restore): New function.
16419         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
16420         (was_mips16_p): Remove.
16421         (old_compression_mode): New.
16422         (mips_set_compression_mode): New function.
16423         (mips_set_current_function): Add microMIPS support.
16424         (mips_option_override): Likewise.
16425         (umips_save_restore_pattern_p): New function.
16426         (umips_output_save_restore): New function.
16427         (umips_load_store_pair_p_1): New function.
16428         (umips_load_store_pair_p): New function.
16429         (umips_output_load_store_pair_1): New function.
16430         (umips_output_load_store_pair): New function.
16431         (umips_movep_target_p) New function.
16432         (mips_prepare_pch_save): Add microMIPS support.
16433         * config/mips/mips.h (TARGET_COMPRESSION): New.
16434         (TARGET_CPU_CPP_BUILTINS): Update macro
16435         to use new compression flags and to support microMIPS.
16436         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
16437         (MIPS_ARCH_FLOAT_SPEC): Likewise.
16438         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
16439         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
16440         (ASM_SPEC): Support mmicromips and mno-micromips.
16441         (M16STORE_REG_P): New macro.
16442         (MIPS_CALL): Support TARGET_MICROMIPS.
16443         (MICROMIPS_J): New macro.
16444         (mips_base_mips16): Rename this...
16445         (mips_base_compression_flags): ...to this.
16446         (UMIPS_12BIT_OFFSET_P): New macro.
16447         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
16448         (MULTILIB_DIRNAMES): Likewise.
16450 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
16452         PR target/52555
16453         * target-globals.c (save_target_globals): For init_reg_sets and
16454         target_reinit remporarily set this_fn_optabs to this_target_optabs.
16456 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
16458         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
16459         * config/aarch64/t-aarch64
16460         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
16462 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
16464         PR inline-asm/56148
16465         * lra-constraints.c (process_alt_operands): Reload operand
16466         conflicting with earlier clobber only if no more other conflicting
16467         operands.
16469 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
16471         PR sanitizer/56393
16472         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
16473         if not linking a shared library.
16475 2013-02-22  Seth LaForge  <sethml@google.com>
16477         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
16479 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
16481         * config/arm/arm.md (split for extendsidi): Update condition.
16482         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
16483         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
16484         (qhs_zextenddi_cstr): Likewise.
16486 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16488         PR middle-end/56420
16489         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
16490         avoid signed wrapping.
16491         (expand_mult): Handle properly multiplication by
16492         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
16493         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
16494         in the compiler if coeff is HOST_WIDE_INT_MIN.
16495         (expand_divmod): Don't make ext_op1 static, change it's type to
16496         uhwi.  Avoid undefined behavior in -INTVAL (op1).
16498         PR rtl-optimization/50339
16499         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
16500         field.
16501         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
16502         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
16503         into splitting_ashiftrt field.
16504         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
16505         ASHIFTRT.
16506         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
16507         choices.
16509 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
16511         PR middle-end/56108
16512         * trans-mem.c (execute_tm_mark): Do not expand transactions that
16513         are sure to go irrevocable.
16515 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
16517         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
16518         scalars are valid operands.
16520 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16522         PR tree-optimization/56310
16523         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
16524         only matching indices and non-negative final offsets.
16525         (intersect_aggregates_with_edge): Pass src_idx to
16526         agg_replacements_to_vector.  Pass src_idx insstead of index to
16527         intersect_with_agg_replacements.
16529 2013-02-21  Martin Jambor  <mjambor@suse.cz>
16531         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
16532         instead of hard-wired defaults.
16534 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
16536         * doc/invoke.texi (MIPS Options): Update documentation of the
16537         floating-point multiply-accumulate instruction restrictions.
16539 2013-02-21  Kostya Serebryany  <kcc@google.com>
16541         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
16542         asan_shadow_offset on x86_64 linux.
16544 2013-02-21  Richard Biener  <rguenther@suse.de>
16546         PR tree-optimization/56415
16547         Revert
16548         2013-02-11  Richard Biener  <rguenther@suse.de>
16550         PR tree-optimization/56273
16551         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
16552         first VRP run.
16554 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
16556         PR bootstrap/56258
16557         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
16558         instead of @itemx.
16560         PR inline-asm/56405
16561         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
16562         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
16564 2013-02-20  Jan Hubicka  <jh@suse.cz>
16566         PR tree-optimization/56265
16567         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
16568         when target is referenced for first time.
16570 2013-02-20  Richard Biener  <rguenther@suse.de>
16572         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
16573         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
16574         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
16575         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
16576         not return anything.
16577         (rename_ssa_copies): Do not remove unused locals.
16578         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
16579         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
16580         * passes.c (execute_function_todo): Do not schedule unused locals
16581         removal if cleanup_tree_cfg did something.
16582         * tree-ssa-live.c (remove_unused_locals): Dump statistics
16583         about the number of removed locals.
16585 2013-02-20  Richard Biener  <rguenther@suse.de>
16587         PR tree-optimization/56398
16588         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
16590 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16592         PR tree-optimization/55334
16593         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
16594         restricted pointers to arrays.
16596 2013-02-20  Richard Biener  <rguenther@suse.de>
16597             Jakub Jelinek  <jakub@redhat.com>
16599         PR tree-optimization/56396
16600         * tree-ssa-ccp.c (n_const_val): New static variable.
16601         (get_value): Return NULL for SSA names we don't have a lattice
16602         entry for.
16603         (ccp_initialize): Initialize n_const_val.
16604         * tree-ssa-copy.c (n_copy_of): New static variable.
16605         (init_copy_prop): Initialize n_copy_of.
16606         (get_value): Return NULL_TREE for SSA names we don't have a
16607         lattice entry for.
16609 2013-02-20  Martin Jambor  <mjambor@suse.cz>
16611         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
16613 2013-02-20  Richard Biener  <rguenther@suse.de>
16615         * genpreds.c (write_lookup_constraint): Do not compare first
16616         letter of the constraint again.
16618 2013-02-20  Richard Biener  <rguenther@suse.de>
16620         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
16621         and ceil_log2.
16622         (get_use_iv_cost): Terminate hashtable walk when coming across
16623         an empty entry.
16625 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
16627         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
16628         reassociation for avx2 targets.
16630 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16632         * config/microblaze/microblaze.c: microblaze_has_clz = 0
16633         Add version check for v8.10.a to enable microblaze_has_clz
16634         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
16635         version and TARGET_PATTERN_COMPARE check
16636         * config/microblaze/microblaze.md: New clzsi2 instruction
16638 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16640         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
16641         function before branching.
16643 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16645         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
16646         DUMP_INSN_RTX_UID.
16647         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
16649 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
16651         PR middle-end/55889
16652         * sel-sched.c: Include ira.h.
16653         (implicit_clobber_conflict_p): New function.
16654         (moveup_expr): Use it.
16655         * Makefile.in (sel-sched.o): Depend on ira.h.
16657 2013-02-19  Richard Biener  <rguenther@suse.de>
16659         PR tree-optimization/56384
16660         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
16661         (vn_hash_type): Split out from ...
16662         (vn_hash_constant_with_type): ... here.
16663         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
16664         (vn_phi_eq): Compare types from vn_phi_s structure.
16665         (vn_phi_lookup): Populate vn_phi_s type.
16666         (vn_phi_insert): Likewise.
16668 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
16670         PR tree-optimization/56350
16671         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
16672         if haven't found reduction or nested cycle operand, rather than
16673         asserting we must find it.
16675         PR tree-optimization/56381
16676         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
16677         to fold_build3.
16679 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
16680             Jakub Jelinek  <jakub@redhat.com>
16682         PR target/52555
16683         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
16684         (swap_optab_enable): Same.
16685         (init_all_optabs): Use argument instead of global.
16686         * tree.h (struct tree_optimization_option): New field target_optabs.
16687         * expr.h (init_all_optabs): Add argument to prototype.
16688         (TREE_OPTIMIZATION_OPTABS): New.
16689         (save_optabs_if_changed): Protoize.
16690         * optabs.h: Declare this_fn_optabs.
16691         * optabs.c (save_optabs_if_changed): New.
16692         Declare this_fn_optabs.
16693         (init_optabs): Add argument to init_all_optabs() call.
16694         * function.c (invoke_set_current_function_hook): Handle per
16695         function optabs.
16696         * function.h (struct function): New field optabs.
16697         * config/mips/mips.c (mips_set_mips16_mode): Handle when
16698         optimization_current_node has changed.
16699         * target-globals.h (save_target_globals_default_opts): Protoize.
16700         * target-globals.c (save_target_globals_default_opts): New.
16702 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
16704         PR target/56347
16705         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16706         registers %fr12 and %fr12R as call used.
16708         PR target/56214
16709         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16710         and HImode, require all displacements to be an integer multiple of
16711         their mode size.
16712         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
16713         only allow QImode and HImode when reload is in progress and strict is
16714         true.  Likewise for symbolic addresses.  Use base14_operand to check
16715         displacements in REG+BASE addresses.
16717 2013-02-18  Richard Biener  <rguenther@suse.de>
16719         PR tree-optimization/56366
16720         * tree-vect-loop.c (get_initial_def_for_induction): Properly
16721         handle sign-conversion of outer-loop initial induction value.
16723 2013-02-18  Richard Biener  <rguenther@suse.de>
16725         PR middle-end/56349
16726         * cfghooks.c (merge_blocks): If we merge a latch into another
16727         block adjust references to it.
16728         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
16729         (verify_loop_structure): Verify that a recorded latch is in fact
16730         a latch.
16732 2013-02-18  Richard Biener  <rguenther@suse.de>
16734         PR tree-optimization/56321
16735         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
16736         order SSA name release and virtual operand unlinking.
16738 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16740         * config/microblaze/microblaze.md (save_stack_block): Define.
16741         (restore_stack_block): Likewise.
16743 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16745         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
16746         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
16747         * config/microblaze/microblaze.c (microblaze_option_override):
16748         Bail out early for PIC modes when target does not support PIC.
16750 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
16752         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
16753         Replace with a microblaze version.
16754         (microblaze_trampoline_init): Adapt for microblaze.
16755         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
16756         microblaze.
16758 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
16759             Dodji Seketeli  <dodji@redhat.com>
16761         PR asan/56330
16762         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
16763         (instrument_mem_region_access): Do not forget to always put
16764         instrumentation of the of 'base' and 'base + len' in a "if (len !=
16765         0) statement, even for cases where either 'base' or 'base + len'
16766         are not instrumented -- because they have been previously
16767         instrumented.  Simplify the logic by putting all the statements
16768         instrument 'base + len' inside a sequence, and then insert that
16769         sequence right before the current insertion point.  Then, to
16770         instrument 'base + len', just get an iterator on that statement.
16771         And do not forget to update the pointer to iterator the function
16772         received as argument.
16774 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
16776         PR rtl-optimization/56348
16777         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
16779 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
16781         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
16782         (clean_graph_dump_file): Pass base to start_graph_dump.
16784 2013-02-14  Richard Henderson  <rth@redhat.com>
16786         PR target/55941
16787         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
16789 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
16791         * collect2-aix.h: Define F_LOADONLY.
16793 2013-02-14  Richard Biener  <rguenther@suse.de>
16795         PR lto/50494
16796         * varasm.c (output_constant_def_1): Get the decl representing
16797         the constant as argument.
16798         (output_constant_def): Wrap output_constant_def_1.
16799         (make_decl_rtl): Use output_constant_def_1 with the decl
16800         representing the constant.
16801         (build_constant_desc): Optionally re-use a decl already
16802         representing the constant.
16803         (tree_output_constant_def): Adjust.
16805 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
16807         Fix an asan crash
16808         * asan.c (instrument_builtin_call):  Really put the length of the
16809         second source argument into src1_len.
16811 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16813         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
16814         argument.  If it is false, don't create edge from then_bb to
16815         fallthru_bb.
16816         (insert_if_then_before_iter): Pass true to it.
16817         (build_check_stmt): Pass false to it.
16818         (transform_statements): Flush hash table only on extended basic
16819         block boundaries, rather than at the beginning of every bb.
16820         Don't flush hash table on nonfreeing_call_p calls.
16821         * tree-flow.h (nonfreeing_call_p): New prototype.
16822         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
16824 2013-02-13  David S. Miller  <davem@davemloft.net>
16826         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
16828 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
16830         PR target/56184
16831         * ira.c (max_regno_before_ira): Move from ...
16832         (ira): ... here.
16833         (fix_reg_equiv_init): Use max_regno_before_ira instead of
16834         vec_safe_length.
16836 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
16838         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
16840 2013-02-13  Richard Biener  <rguenther@suse.de>
16842         PR lto/56295
16843         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
16844         globals in MEM_REFs.
16846 2013-02-13  Richard Biener  <rguenther@suse.de>
16848         * loop-init.c (loop_optimizer_init): Clear loop state when
16849         re-initializing preserved loops.
16850         * loop-unswitch.c (unswitch_single_loop): Return whether
16851         we unswitched the loop.  Do not verify loop state here.
16852         (unswitch_loops): When we unswitched a loop discover new loops.
16854 2013-02-13  Kostya Serebryany  <kcc@google.com>
16856         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
16857         on x86_64 linux.
16858         * sanitizer.def: Rename __asan_init to __asan_init_v1.
16860 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
16862         Avoid instrumenting duplicated memory access in the same basic block
16863         * Makefile.in (asan.o): Add new dependency on hash-table.h
16864         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
16865         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
16866         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
16867         (free_mem_ref_resources, has_mem_ref_been_instrumented)
16868         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
16869         (get_mem_ref_of_assignment): New functions.
16870         (get_mem_refs_of_builtin_call): Extract from
16871         instrument_builtin_call and tweak a little bit to make it fit with
16872         the new signature.
16873         (instrument_builtin_call): Use the new
16874         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
16875         of is_gimple_builtin_call.
16876         (instrument_derefs, instrument_mem_region_access): Insert the
16877         instrumented memory reference into the hash table.
16878         (maybe_instrument_assignment): Renamed instrument_assignment into
16879         this, and change it to advance the iterator when instrumentation
16880         actually happened and return true in that case.  This makes it
16881         homogeneous with maybe_instrument_assignment, and thus give a
16882         chance to callers to be more 'regular'.
16883         (transform_statements): Clear the memory reference hash table
16884         whenever we enter a new BB, when we cross a function call, or when
16885         we are done transforming statements.  Use
16886         maybe_instrument_assignment instead of instrumentation.  No more
16887         need to special case maybe_instrument_assignment and advance the
16888         iterator after calling it; it's now handled just like
16889         maybe_instrument_call.  Update comment.
16891 2013-02-13  Richard Biener  <rguenther@suse.de>
16893         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
16894         Fix loop discovery code.
16896 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
16898         PR inline-asm/56148
16899         * lra-constraints.c (process_alt_operands): Match early clobber
16900         operand with itself.  Check conflicts with earlyclobber only if
16901         the operand is not reloaded.  Prefer to reload conflicting operand
16902         if earlyclobber and matching operands are the same.
16904 2013-02-12  Richard Biener  <rguenther@suse.de>
16906         PR lto/56297
16907         * lto-streamer-out.c (write_symbol): Do not output symbols
16908         for hard register variables.
16910 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
16912         PR target/54222
16913         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
16914         (umulsidi3_insn, mulsidi3_insn): New insns.
16916 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
16918         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
16919         (struct tune_params): Add vec_costs field.
16920         * config/arm/arm.c (arm_builtin_vectorization_cost)
16921         (arm_add_stmt_cost): New functions.
16922         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
16923         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
16924         (arm_default_vec_cost): New struct of type cpu_vec_costs.
16925         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
16926         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
16927         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
16928         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
16930 2013-02-12  Richard Biener  <rguenther@suse.de>
16932         PR lto/56295
16933         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
16934         decls again if possible.
16936 2013-02-12  Richard Biener  <rguenther@suse.de>
16938         PR middle-end/56288
16939         * tree-ssa.c (verify_ssa_name): Fix check, move
16940         SSA_NAME_IN_FREE_LIST check up.
16942 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
16943             Steven Bosscher   <steven@gcc.gnu.org>
16945         PR rtl-optimization/56151
16946         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
16947         equal to op0 or op1, and last_insn pattern is CODE operation
16948         with MEM dest and one of the operands matches that MEM.
16950 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
16952         * doc/extend.texi: Document Function Multiversioning and "default"
16953         parameter string to target attribute.
16954         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
16955         target attribute parameter is "default".
16956         (ix86_compare_version_priority): Remove checks for target attribute.
16957         (ix86_mangle_function_version_assembler_name): Change error to sorry.
16958         Remove check for target attribute equal to NULL. Add assert.
16959         (ix86_generate_version_dispatcher_body): Change error to sorry.
16961 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
16962             Jack Howarth  <howarth@bromo.med.uc.edu>
16963             Patrick Marlier  <patrick.marlier@gmail.com>
16965         PR libitm/55693
16966         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
16967         define ENDFILE_SPEC as TM_DESTRUCTOR.
16968         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
16970 2013-02-11  Alexander Potapenko  <glider@google.com>
16971             Jack Howarth  <howarth@bromo.med.uc.edu>
16972             Jakub Jelinek  <jakub@redhat.com>
16974         PR sanitizer/55617
16975         * config/darwin.c (cdtor_record): Rename ctor_record.
16976         (sort_cdtor_records): Rename sort_ctor_records.
16977         (finalize_dtors): New routine to sort destructors by
16978         priority before use in assemble_integer.
16979         (machopic_asm_out_destructor): Use finalize_dtors if needed.
16981 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
16983         PR rtl-optimization/56275
16984         * simplify-rtx.c (avoid_constant_pool_reference): Check that
16985         offset is non-negative and less than cmode size before
16986         calling simplify_subreg.
16988 2013-02-11  Richard Biener  <rguenther@suse.de>
16990         PR tree-optimization/56264
16991         * cfgloop.h (fix_loop_structure): Adjust prototype.
16992         * loop-init.c (fix_loop_structure): Return the number of
16993         newly discovered loops.
16994         * tree-cfgcleanup.c (repair_loop_structures): When new loops
16995         are discovered, do a full loop-closed SSA rewrite.
16997 2013-02-11  Richard Biener  <rguenther@suse.de>
16999         PR tree-optimization/56273
17000         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
17001         first VRP run.
17002         (check_array_ref): Fix missing newline in dumps.
17003         (search_for_addr_array): Likewise.
17005 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
17007         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
17009 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
17011         PR target/56256
17012         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
17014 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
17016         PR rtl-optimization/56246
17017         * lra-constraints.c (simplify_operand_subreg): Try to reuse
17018         reload pseudo.
17019         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
17020         constraints are satisfied.
17022 2013-02-08  Jeff Law  <law@redhat.com>
17024         PR debug/53948
17025         * emit-rtl.c (reg_is_parm_p): New function.
17026         * regs.h (reg_is_parm_p): New prototype.
17027         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
17028         callee-clobbered registers.
17030 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
17032         PR target/56043
17033         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
17034         If there is no implicit builtin declaration, just return NULL.
17036 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17038         * config/i386/sse.md (FMAMODEM): New mode iterator.
17039         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
17040         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
17042 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17044         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
17045         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
17046         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
17048 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17050         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
17051         (microblaze*-*-elf): Likewise.
17052         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
17053         LINK_SPEC.
17054         * config/microblaze/microblaze-c.c: Add builtin defines for
17055         _LITTLE_ENDIAN and _BIG_ENDIAN.
17056         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
17057         add to TARGET_DEFAULT flags.
17058         Expand ASM_SPEC and LINK_SPEC.
17059         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
17060         * config/microblaze/microblaze.md: Update extendsidi2 and
17061         movdi_internal instructions to use low-order / high-order reg
17062         print_operands.
17063         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
17064         options and inversemask / mask of LITTLE_ENDIAN.
17065         * config/microblaze/t-microblaze: Expand multilib options to
17066         include mlittle-endian (le) and update exceptions patterns.
17068 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
17070         PR rtl-optimization/56195
17071         * lra-constraints.c (get_reload_reg): Don't reuse regs
17072         if they have smaller mode than requested, if they have
17073         wider mode than requested, try to return a SUBREG.
17075         PR tree-optimization/56250
17076         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
17077         if type is unsigned and code isn't MULT_EXPR.
17079 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17081         PR tree-optimization/56064
17082         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
17083         bits according to mode.
17084         * fixed-value.h (fixed_from_double_int)
17085         (const_fixed_from_double_int): Adjust comments.
17087 2013-02-08  Richard Biener  <rguenther@suse.de>
17089         PR lto/56231
17090         * lto-streamer.h (struct data_in): Remove current_file, current_line
17091         and current_col members.
17092         * lto-streamer-out.c (lto_output_location): Stream changed bits
17093         en-block for efficiency.
17094         * lto-streamer-in.c (clear_line_info): Remove.
17095         (lto_input_location): Cache current file, line and column
17096         globally via local statics.  Read changed bits en-block.
17097         (input_function): Do not call clear_line_info.
17098         (lto_read_body): Likewise.
17099         (lto_input_toplevel_asms): Likewise.
17101 2013-02-08  Michael Matz  <matz@suse.de>
17103         PR tree-optimization/52448
17104         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
17105         (nt_call_phase): New static.
17106         (add_or_mark_expr): Only mark accesses with newer phase than any
17107         call seen.
17108         (nonfreeing_call_p): New.
17109         (nt_init_block): Update nt_call_phase, mark blocks as visited.
17110         (nt_fini_block): Keep blocks marked as visited.
17111         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
17113 2013-02-08  Richard Biener  <rguenther@suse.de>
17115         * ira.c (ira): Free broken dominator information.
17117 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
17119         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
17121 2013-02-08  Marek Polacek  <polacek@redhat.com>
17123         * cfgloop.c (verify_loop_structure): Add more checking of headers.
17125 2013-02-08  Richard Biener  <rguenther@suse.de>
17127         PR middle-end/56181
17128         * cfgloop.h (flow_loops_find): Adjust.
17129         (bb_loop_header_p): Declare.
17130         * cfgloop.c (bb_loop_header_p): New function split out from ...
17131         (flow_loops_find): ... here.  Adjust function signature,
17132         support incremental loop structure update.
17133         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
17134         * cfgloopmanip.c (fix_loop_structure): Move ...
17135         * loop-init.c (fix_loop_structure): ... here.
17136         (apply_loop_flags): Split out from ...
17137         (loop_optimizer_init): ... here.
17138         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
17139         in incremental mode, only remove dead loops here.
17141 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
17143         PR target/54222
17144         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
17145         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
17146         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
17147         (*round<mode>3.libgcc): New insns for fixed-modes.
17148         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
17149         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
17150         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
17151         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
17152         implementations.  Define to __builtin_avr_absFX,
17153         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
17154         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
17155         __builtin_avr_countlsFX, respectively.
17156         * config/avr/avr-c.c (target.h): Include it.
17157         (enum avr_builtin_id): New enum.
17158         (avr_resolve_overloaded_builtin): New static function.
17159         (avr_register_target_pragmas): Use it to set
17160         targetm.resolve_overloaded_builtin.
17161         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
17162         tree nodes used by DEF_BUILTIN.
17163         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
17164         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
17165         <AVR_BUILTIN_xxBITS>: Same.
17167 2013-02-08  Richard Biener  <rguenther@suse.de>
17169         * cfgloop.c (verify_loop_structure): Properly handle
17170         a loop exiting to another loop header.
17171         * ira-int.h (ira_loops): Remove.
17172         * ira.c (ira_loops): Remove.
17173         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
17174         (do_reload): Use loop_optimizer_finalize.
17175         * ira-build.c (create_loop_tree_nodes): Use get_loops and
17176         number_of_loops to access the loop tree.
17177         (more_one_region_p): Likewise.
17178         (finish_loop_tree_nodes): Likewise.
17179         (rebuild_regno_allocno_maps): Likewise.
17180         (mark_loops_for_removal): Likewise.
17181         (mark_all_loops_for_removal): Likewise.
17182         (remove_unnecessary_regions): Likewise.
17183         (ira_build): Likewise.
17184         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
17186 2013-02-08  Richard Biener  <rguenther@suse.de>
17188         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
17189         * ipa-pure-const.c (analyze_function): Avoid calling
17190         mark_irreducible_loops twice.
17191         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
17193 2013-02-07  David S. Miller  <davem@davemloft.net>
17195         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
17196         on 'reg'.
17197         * var-tracking.c (vt_add_function_parameter): Test the presence of
17198         HAVE_window_save properly and do not remap argument registers when
17199         we have a leaf function.
17201 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
17203         PR bootstrap/56227
17204         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
17205         instead of "ll".
17206         * config/i386/i386.c (ix86_print_operand): Ditto.
17208 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17210         * lra-constraints.c (process_alt_operands): Fix recently added comment.
17212 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
17214         PR rtl-optimization/56225
17215         * lra-constraints.c (process_alt_operands): Check that reload hard
17216         reg can hold value for strict_low_part.
17218 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17220         PR debug/56154
17221         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
17222         dwarf2out_end_function.
17223         (in_first_function_p, maybe_at_text_label_p,
17224         first_loclabel_num_not_at_text_label): New variables.
17225         (dwarf2out_var_location): In the first function find out
17226         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
17227         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
17228         functions.
17230 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
17232         PR rtl-optimization/56178
17233         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
17234         SUBREG of a register.  Tidy up related block of code.
17235         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
17236         note if the source is a register or a SUBREG of a register.
17238 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
17240         PR target/56228
17241         * config/rs6000/rs6000.md (ptrm): New mode attr.
17242         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
17243         call_value_indirect_aix<pttrsize>,
17244         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
17245         m in constraints.
17247 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17249         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
17250         if -bnortl. Convert to strcmp and strncmp.
17252 2013-02-07  Alan Modra  <amodra@gmail.com>
17254         PR target/54009
17255         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
17256         addresses won't wrap when offsetting.
17257         (rs6000_secondary_reload): Provide secondary reloads needed for
17258         wrapping LO_SUM addresses.
17260 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
17262         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
17263         MACH, just __MACH__.
17265 2013-02-06  Richard Biener  <rguenther@suse.de>
17267         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
17268         instead of calling fix_loop_structure.
17270 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
17272         PR middle-end/56217
17273         * omp-low.c (use_pointer_for_field): Return false if
17274         lower_send_shared_vars doesn't generate any copy-out code.
17276 2013-02-06  Tom de Vries  <tom@codesourcery.com>
17278         PR rtl-optimization/56131
17279         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
17280         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
17281         of the label is NULL.  Add comment.
17283 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
17285         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
17287         PR sanitizer/55374
17288         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
17289         (STATIC_LIBTSAN_LIBS): Likewise.
17290         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
17291         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
17292         is defined, don't add anything else beyond that.
17293         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
17294         (LINK_COMMAND_SPEC): Use them.
17296         PR tree-optimization/56205
17297         * tree-stdarg.c (check_all_va_list_escapes): Return true if
17298         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
17299         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
17301 2013-02-05  Richard Biener  <rguenther@suse.de>
17303         PR tree-optimization/53342
17304         PR tree-optimization/53185
17305         * tree-vectorizer.h (vect_check_strided_load): Remove.
17306         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
17307         not disallow peeling for vectorized strided loads.
17308         (vect_check_strided_load): Make static and simplify.
17309         (vect_analyze_data_refs): Adjust.
17310         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
17311         correctly when vectorizing strided loads.
17313 2013-02-05  Richard Biener  <rguenther@suse.de>
17315         * doc/install.texi: Refer to ISL, not PPL.
17317 2013-02-05  Jan Hubicka  <jh@suse.cz>
17319         PR tree-optimization/55789
17320         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
17322 2013-02-05  Jan Hubicka  <jh@suse.cz>
17324         PR tree-optimization/55789
17325         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
17326         the dead call anyway.
17328 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
17330         PR sanitizer/55374
17331         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
17333 2013-02-04  Alexander Potapenko  <glider@google.com>
17334             Jack Howarth  <howarth@bromo.med.uc.edu>
17335             Jakub Jelinek  <jakub@redhat.com>
17337         PR sanitizer/55617
17338         * config/darwin.c (sort_ctor_records): Stabilized qsort
17339         on constructor priority by using original position.
17340         (finalize_ctors): New routine to sort constructors by
17341         priority before use in assemble_integer.
17342         (machopic_asm_out_constructor): Use finalize_ctors if needed.
17344 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
17346         PR libstdc++/54314
17347         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
17348         about visibility on artificial decls.
17349         * config/sol2.c (solaris_assemble_visibility): Likewise.
17351 2013-02-04  Kai Tietz  <ktietz@redhat.com>
17353         PR target/56186
17354         * config/i386/i386.c (function_value_ms_64): Add additional valtype
17355         argument and improve checking of return-argument types for 16-byte
17356         modes.
17357         (ix86_function_value_1): Add additional valtype argument on call
17358         of function_value_64.
17359         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
17360         handling infunction_value_64 function.
17362 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
17364         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
17366 2013-02-04  Richard Biener  <rguenther@suse.de>
17368         PR tree-optimization/56188
17369         * tree-ssa-structalias.c (label_visit): Consider case with
17370         initially non-empty points-to set.
17371         (perform_var_substitution): Dump node mapping and clean up.
17373 2013-02-04  Richard Guenther  <rguenther@suse.de>
17375         PR lto/56168
17376         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
17377         node prevail as last resort.
17378         (lto_symtab_merge_decls): Remove guard on LTRANS here.
17379         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
17381 2013-02-04  Richard Biener  <rguenther@suse.de>
17383         PR tree-optimization/56113
17384         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
17385         Merge into ...
17386         (equiv_class_lookup_or_add): ... this.
17387         (label_visit): Adjust and fix error in previous patch.
17388         (perform_var_substitution): Adjust.
17390 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
17392         * config/sh/divtab.c: Fix formatting and comments throughout the file.
17393         * config/sh/sh4-300.md: Likewise.
17394         * config/sh/sh4a.md: Likewise.
17395         * config/sh/constraints.md: Likewise.
17396         * config/sh/sh.md: Likewise.
17397         * config/sh/netbsd-elf.h: Likewise.
17398         * config/sh/predicates.md: Likewise.
17399         * config/sh/sh-protos.h: Likewise.
17400         * config/sh/ushmedia.h: Likewise.
17401         * config/sh/linux.h: Likewise.
17402         * config/sh/sh.c: Likewise.
17403         * config/sh/superh.h: Likewise.
17404         * config/sh/elf.h: Likewise.
17405         * config/sh/sh4.md: Likewise.
17406         * config/sh/sh.h: Likewise.
17408 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17410         * config/pa/constraints.md: Adjust unused letters.  Change "T"
17411         constraint to match_test floating_point_store_memory_operand().
17412         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
17413         (base14_operand): New.
17414         (floating_point_store_memory_operand): New.
17415         (integer_store_memory_operand): Revise to use base14_operand and
17416         reg_plus_base_memory_operand.
17417         (move_dest_operand): Allow symbolic_memory_operands.
17418         (symbolic_memory_operand): Check for LO_SOM.
17419         (symbolic_operand): Change default case to break.
17420         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
17421         CONST_DOUBLE values to be reloaded by putting them into memory when
17422         the destination is a floating point register.
17423         (movdf): Remove code to handle CONST_DOUBLE.
17424         (movsf): Likewise.
17425         (reload_indf_r1): New.
17426         (reload_insf_r1): New.
17427         Consistently use "Q" and "T" constraints with integer and floating
17428         point move instructions, respectively.
17429         (movdi): Remove FAIL.
17430         Change predicate for source operand unamed DImode move from
17431         general_operand to move_src_operand.
17432         (umulsidi3): Change predicate for destination operand to
17433         register_operand.
17434         Likewise for similar unamed patterns.
17435         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
17436         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
17437         (hppa_legitimize_address): Simplify mask calculation.
17438         (pa_emit_move_sequence): Revised handling of secondary reloads from
17439         REG+D addresses for floating point loads and stores.  Directly handle
17440         loading CONST0_RTX (mode) to a floating point register.
17441         (pa_secondary_reload): Handle reloading DF and SFmode constant values
17442         to floating point registers.  Don't restrict secondary reloads to
17443         floating point registers to integer modes.  Revise some comments and
17444         cleanup some code.
17445         (TARGET_LEGITIMATE_ADDRESS_P): Define.
17446         (pa_legitimate_address_p): New.
17447         (pa_legitimize_reload_address): New.
17448         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
17449         (STRICT_REG_OK_FOR_BASE_P): New.
17450         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
17451         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
17453 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
17454             Andrew Dixie  <andrewd@gentrack.com>
17456         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
17457         flag set.
17459 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
17461         * expmed.c (extract_bit_field_1): Pass the full width of the
17462         structure to get_best_reg_extraction_insn.
17464 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17466         PR target/54601
17467         * configure.ac (use_cxa_atexit): Add AIX.
17468         * configure: Regenerate.
17470         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
17472 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
17474         PR debug/54793
17475         * final.c (need_profile_function): New variable.
17476         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
17477         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
17478         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
17479         notes, targetm.asm_out.function_prologue doesn't emit anything,
17480         HAVE_prologue and profiler should be emitted before prologue,
17481         set need_profile_function instead of emitting it.
17482         (final_scan_insn): If need_profile_function, emit
17483         profile_function on the first NOTE_INSN_BASIC_BLOCK or
17484         NOTE_INSN_FUNCTION_BEG note.
17486 2013-02-01  Richard Henderson  <rth@redhat.com>
17488         * config/rs6000/rs6000.md (smulditi3): New.
17489         (umulditi3): New.
17491         * config/alpha/alpha.md (umulditi3): New.
17493 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
17495         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
17496         (ASM_OUTPUT_ALIGNED_LOCAL): New.
17498 2013-02-01  Richard Biener  <rguenther@suse.de>
17500         PR tree-optimization/56113
17501         * tree-ssa-structalias.c (label_visit): Reduce work for
17502         single-predecessor nodes.
17504 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
17506         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
17507         range isn't testing for zero.
17509 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
17511         PR middle-end/56113
17512         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
17514 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
17515             Nick Clifton  <nickc@redhat.com>
17517         * config/v850/constraints.md (Q): Define as a memory constraint.
17518         * config/v850/predicates.md (label_ref_operand): New predicate.
17519         (e3v5_shift_operand): New predicate.
17520         (ior_operator): New predicate.
17521         * config/v850/t-v850: Add e3v5 multilib.
17522         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
17523         (v850_gen_movdi): Prototype.
17524         * config/v850/v850.c: Add support for e3v5 architecture.
17525         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
17526         TARGET_V850E_UP.
17527         (construct_save_jarl): Add e3v5 long JARL support.
17528         (v850_adjust_insn_length): New function.  Adjust length of call
17529         insns when using e3v5 instructions.
17530         (v850_gen_movdi): New function: Generate instructions to move a
17531         DImode value.
17532         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
17533         (CPP_SPEC): Define __v850e3v5__ as appropriate.
17534         (TARGET_USE_FPU): Enable for e3v5.
17535         (CONST_OK_FOR_W): New macro.
17536         (ADJUST_INSN_LENGTH): Define.
17537         * config/v850/v850.md (UNSPEC_LOOP): Define.
17538         (attr cpu): Add v850e3v5.
17539         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
17540         (movdi): New pattern.
17541         (movdi_internal): New pattern.
17542         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
17543         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
17544         (cstoresf4): Likewise.
17545         (cstoredf4): Likewise.
17546         (insv): New pattern.
17547         (rotlso3_a): New pattern.
17548         (rotlsi3_b): New pattern
17549         (rotlsi3_v850e3v5): New pattern.
17550         (doloop_begin): New pattern.
17551         (fix_loop_counter): New pattern.
17552         (doloop_end): New pattern.
17553         (branch_normal): Add e3v5 long branch support.
17554         (branch_invert): Likewise.
17555         (branch_z_normal): Likewise.
17556         (branch_z_invert): Likewise.
17557         (branch_nz_normal): Likewise.
17558         (branch_nz_invert): Likewise.
17559         (call_internal_short): Add e3v5 register-indirect JARL support.
17560         (call_internal_long): Likewise.
17561         (call_value_internal_short): Likewise.
17562         (call_value_internal_long): Likewise.
17563         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
17564         (mloop): New option.
17565         * config.gcc: Add support for configuring v840e3v5 target.
17566         * doc/invoke.texi: Document new v850 specific command line options.
17568 2013-01-31  Paul Koning  <ni1d@arrl.net>
17570         PR debug/55059
17571         PR debug/54508
17572         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
17573         children if parent is a class.
17574         (prune_unused_types_prune): Don't add DW_AT_declaration.
17576 2013-01-31  Richard Biener  <rguenther@suse.de>
17578         PR tree-optimization/56157
17579         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
17580         match up operand with SLP child.
17582 2013-01-31  Jason Merrill  <jason@redhat.com>
17584         PR debug/54410
17585         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
17586         parameters the first time.
17587         (gen_scheduled_generic_parms_dies): Check completeness here.
17589 2013-01-31  Richard Biener  <rguenther@suse.de>
17591         PR middle-end/53073
17592         * common.opt (faggressive-loop-optimizations): New flag,
17593         enabled by default.
17594         * doc/invoke.texi (faggressive-loop-optimizations): Document.
17595         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
17596         infer_loop_bounds_from_undefined by it.
17598 2013-01-31  Richard Biener  <rguenther@suse.de>
17600         PR tree-optimization/56150
17601         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
17602         visit virtual operands.
17603         (find_uses_to_rename_bb): Likewise.
17605 2013-01-31  Richard Biener  <rguenther@suse.de>
17607         PR tree-optimization/56150
17608         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
17609         mixed store non-store stmts.
17611 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
17613         PR sanitizer/55374
17614         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
17615         LIBASAN_EARLY_SPEC is defined.
17616         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
17617         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
17618         before %o.
17619         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
17621         PR c++/55742
17622         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
17623         invalid args instead of ICEing on it.
17624         (ix86_valid_target_attribute_tree): Return error_mark_node if
17625         ix86_valid_target_attribute_inner_p failed.
17626         (ix86_valid_target_attribute_p): Return false only if
17627         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
17628         target("default") attribute.
17629         (sorted_attr_string): Change argument from const char * to tree,
17630         merge in all target attribute arguments rather than just one.
17631         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
17632         instead of free.  Avoid using strcat.
17633         (ix86_mangle_function_version_assembler_name): Mangle
17634         target("default") as if no target attribute is present.  Adjust
17635         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
17636         instead of xmalloc and XDELETEVEC instead of free.
17637         (ix86_function_versions): Don't return true if one of the decls
17638         doesn't have target attribute.  If they don't and one of the decls
17639         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
17640         sorted_attr_string caller.  Use XDELETEVEC instead of free.
17641         (ix86_supports_function_versions): Remove.
17642         (make_name): Fix up formatting.
17643         (make_dispatcher_decl): Remove resolver_name and its initialization.
17644         Avoid leaking memory.
17645         (is_function_default_version): Return true if there is
17646         target("default") attribute rather than no target attribute at all.
17647         (make_resolver_func): Avoid leaking memory.
17648         (ix86_generate_version_dispatcher_body): Likewise.
17649         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
17650         * target.def (supports_function_versions): Remove.
17651         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
17652         * doc/tm.texi: Regenerated.
17654 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
17656         PR rtl-optimization/56144
17657         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
17658         for values with side effects.
17660 2013-01-30  Richard Biener  <rguenther@suse.de>
17662         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
17663         (sparseset_pop): Likewise.
17664         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
17665         to be able to use quick_push in the worker loop.
17667 2013-01-30  Marek Polacek  <polacek@redhat.com>
17669         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
17671 2013-01-30  Richard Biener  <rguenther@suse.de>
17673         PR lto/56147
17674         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
17676 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
17678         PR tree-optimization/56064
17679         * fixed-value.c (fixed_from_double_int): New function.
17680         * fixed-value.h (fixed_from_double_int): New prototype.
17681         (const_fixed_from_double_int): New static inline function.
17682         * fold-const.c (native_interpret_fixed): New static function.
17683         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
17684         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
17685         (native_encode_fixed): New static function.
17686         (native_encode_expr) <FIXED_CST>: Use it.
17687         (native_interpret_int): Move double_int worker code to...
17688         * double-int.c (double_int::from_buffer): ...this new static method.
17689         * double-int.h (double_int::from_buffer): Prototype it.
17691 2013-01-30  Richard Biener  <rguenther@suse.de>
17693         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
17694         New pointer-map and obstack.
17695         (init_alias_vars): Allocate pointer-map and obstack.
17696         (delete_points_to_sets): Free them.
17697         (find_what_var_points_to): Cache result.
17698         (find_what_p_points_to): Adjust for changed interface of
17699         find_what_var_points_to.
17700         (compute_points_to_sets): Likewise.
17701         (ipa_pta_execute): Likewise.
17703 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17705         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
17706         * configure: Regenerate.
17707         * config.in: Regenerate.
17708         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
17709         #nobits/#progbits if supported.
17711 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
17713         PR target/56121
17714         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
17715         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
17716         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
17718 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17720         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
17721         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
17723 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17725         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
17726         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
17728 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17730         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
17731         declaration.
17732         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
17733         * config/arm/cortex-a7.md: New bypasses using
17734         arm_mac_accumulator_is_result.
17736 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17738         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
17739         (cortex_a7_neon_mla): Likewise.
17740         (cortex_a7_fpfmad): New reservation.
17741         (cortex_a7_fpmacs): Use ffmas and update required units.
17742         (cortex_a7_fpmuld): Update required units and latency.
17743         (cortex_a7_fpmacd): Likewise.
17744         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
17745         (cortex_a7_neon). Likewise.
17746         (bypass) Update participating units.
17748 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
17750         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
17751         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
17752         from fmac to ffma.
17753         * config/arm/vfp11.md (vfp_farith): Use ffmas.
17754         (vfp_fmul): Use ffmad.
17755         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
17756         (cortex_r4_fmacd): Use ffmad.
17757         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
17758         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
17759         (cortex_a9_fmacd): Use ffmad.
17760         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
17761         (cortex_a8_vfp_macd): Use ffmad.
17762         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
17763         (cortex_a5_fpmacd): Use ffmad.
17764         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
17765         (cortex_a15_vfp_macd): Use ffmad.
17766         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
17768 2013-01-29  Jason Merrill  <jason@redhat.com>
17770         PR libstdc++/54314
17771         * varasm.c (default_assemble_visibility): Don't warn about
17772         visibility on artificial decls.
17774 2013-01-29  Richard Biener  <rguenther@suse.de>
17776         PR tree-optimization/56113
17777         * tree-ssa-structalias.c (equiv_class_lookup): Also return
17778         the bitmap leader.
17779         (label_visit): Free duplicate bitmaps and record the leader instead.
17780         (perform_var_substitution): Adjust.
17782 2013-01-29  Richard Biener  <rguenther@suse.de>
17784         PR tree-optimization/55270
17785         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
17786         the CFG, schedule loops for fixup.
17788 2013-01-29  Nick Clifton  <nickc@redhat.com>
17790         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
17791         SP_REG.
17793 2013-01-28  Leif Ekblad  <leif@rdos.net>
17795         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
17796         * config/i386/i386.h (TARGET_RDOS): New macro.
17797         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
17798         * config/i386/i386.c (ix86_option_override_internal): For 64bit
17799         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
17800         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
17801         DEFAULT_LARGE_SECTION_THRESHOLD.
17802         * config/i386/i386.md (R14_REG, R15_REG): New constants.
17803         * config/i386/rdos.h: New file.
17804         * config/i386/rdos64.h: New file.
17806 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
17808         PR other/54814
17809         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
17810         TEST_HARD_REG_BIT.
17812 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17814         PR rtl-optimization/56117
17815         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
17816         call cselib_lookup_from_insn on the MEM before calling
17817         add_insn_mem_dependence.
17819 2013-01-28  Richard Biener  <rguenther@suse.de>
17821         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
17822         to a stmt that didn't have one.
17823         (copy_phis_for_bb): Likewise for PHI arguments.
17824         (copy_debug_stmt): Likewise for debug stmts.
17826 2013-01-28  Richard Biener  <rguenther@suse.de>
17828         PR tree-optimization/56034
17829         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
17830         (partition_builtin_p): Adjust.
17831         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
17832         it is the last partition.
17833         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
17834         up the vertex for the definition.
17835         (classify_partition): Classify whether a partition is a
17836         PKIND_REDUCTION, thus has uses outside of the loop.
17837         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
17838         Merge all PKIND_REDUCTION partitions into the last partition.
17839         (tree_loop_distribution): Seed partitions from reductions as well.
17841 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
17843         PR tree-optimization/56125
17844         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
17845         pow(x,c) into sqrt(x) * powi(x, n/2) or
17846         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
17847         optimizing for size.
17848         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
17849         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
17850         integer.
17852         PR tree-optimization/56094
17853         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
17854         to UNKNOWN_LOCATION while gimplifying expr.
17856 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
17858         PR target/56114
17859         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
17860         operand 0 in movabs insn template for -masm=intel asm alternative.
17861         (*movabs<mode>_2): Ditto for operand 1.
17863 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
17865         PR target/54663
17866         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
17867         of microblaze-c.o
17869 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
17871         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
17872         tm_file.
17874 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17876         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
17877         Undef to avoid warning.
17879 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
17881         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
17882         * configure: Regenerate.
17884 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
17886         PR tree-optimization/56098
17887         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
17888         for stmts with volatile ops.
17889         (cond_store_replacement): Don't optimize if assign has volatile ops.
17890         (cond_if_else_store_replacement_1): Don't optimize if either
17891         then_assign or else_assign have volatile ops.
17892         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
17893         volatile ops.
17895 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17897         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
17899 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17901         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
17902         missing ':' in asm example.
17904 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
17906         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
17907         entries into lane and laneq entries.
17908         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
17909         Remove AdvSIMD scalar modes.
17910         (aarch64_sq<r>dmulh_laneq<mode>): New.
17911         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
17912         modes.
17913         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
17914         builtin implementations to relfect changes in RTL in aarch64-simd.md.
17915         * config/aarch64/iterators.md (VCOND): New.
17916         (VCONQ): New.
17918 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
17920         PR target/54222
17921         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
17922         Add NULL LIBNAME argument to existing definitions.
17923         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
17924         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
17925         * config/avr/avr.c (DEF_BUILTIN): Same.
17926         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
17927         (avr_expand_builtin): Expand to a vanilla call if a libgcc
17928         implementation is available (DECL_ASSEMBLER_NAME is set).
17929         (avr_fold_absfx): New static function.
17930         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
17931         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
17932         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
17933         AVR_BUILTIN_ABSLLK.
17934         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
17935         (abshk, absk, abslk, absllk): Provide as static inline functions.
17937 2013-01-25  Marek Polacek  <polacek@redhat.com>
17939         PR tree-optimization/56035
17940         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
17942 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
17944         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
17945         (*movtf_internal_rex64): Add (!o,C) alternative
17946         (*movxf_internal_rex64): Ditto.
17947         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
17949 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
17951         * doc/invoke.texi: fix typo.
17952         * doc/objc.texi: fix typo.
17954 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
17956         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
17957         for the first two alternatives.
17959 2013-01-24  Diego Novillo  <dnovillo@google.com>
17961         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
17962         (ggc-zone.o): Remove.
17963         * configure.ac: Remove option --with-gc.
17964         * configure: Re-generate.
17965         * doc/install.texi: Remove documentation for --with-gc.
17966         * gengtype.c (write_enum_defn): Remove.  Update all users.
17967         (write_Types_process_field): Remove generation of gt_e_* argument.
17968         (output_type_enum): Remove.  Update all users.
17969         (write_enum_defn): Remove.  Update all users.
17970         (enum alloc_zone): Remove.  Update all users.
17971         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
17972         * ggc-common.c (ggc_splay_alloc): Remove first argument.
17973         Update all callers.
17974         (struct ptr_data): Remove field TYPE.  Update all users.
17975         (gt_pch_note_object): Remove argument TYPE.  Update all users.
17976         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
17977         Update all users.
17978         * ggc-none.c (ggc_alloc_typed_stat): Remove.
17979         (struct alloc_zone): Remove.
17980         (ggc_internal_alloc_zone_stat): Remove.
17981         (ggc_internal_cleared_alloc_zone_stat): Remove.
17982         * ggc-page.c (ggc_alloc_typed_stat): Remove.
17983         (ggc_pch_count_object): Remove last argument.  Update all users.
17984         (ggc_pch_alloc_object): Remove last argument.  Update all users.
17985         (struct alloc_zone): Remove.
17986         * ggc-zone.c: Remove.
17987         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
17988         (struct alloc_zone): Remove.
17989         (ggc_alloc_typed_stat): Remove.
17990         (ggc_alloc_typed): Remove.
17991         (ggc_splay_alloc): Remove first argument.
17992         (rtl_zone): Remove.  Update all users.
17993         (tree_zone): Remove.  Update all users.
17994         (tree_id_zone): Remove.  Update all users.
17995         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
17996         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
17997         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
17998         * tree-ssanames.c: Remove references to zone allocator in comments.
18000 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
18002         * config/avr/avr.c (avr_out_fract): Make register numbers that
18003         might be outside of source operand signed.
18005 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
18007         * config/i386/constraints.md (Yf): New constraint.
18008         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
18009         of f constraint to conditionaly disable x87 register preferences.
18010         (*movdf_internal): Ditto.
18011         (*movsf_internal): Ditto.
18013 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
18015         PR inline-asm/55934
18016         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
18017         that have operands with impossible constraints.
18018         Add a FIXME for a speed-up opportunity.
18019         * lra-constraints.c (process_alt_operands): Verify that a class
18020         selected from constraints on asms is valid for the operand mode.
18021         (curr_insn_transform): Remove incorrect comment.
18023 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
18025         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
18026         TOC operand is a valid symbol ref in the constant pool.
18028 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
18030         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
18032 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
18034         PR target/54222
18035         * config/avr/stdfix.h: New file.
18036         * t-avr (stdfix-gcc.h): New rule to build it.
18037         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
18039 2013-01-23  Kostya Serebryany  <kcc@google.com>
18041         * config/darwin.h: remove dependency on
18042         CoreFoundation (asan on Mac OS).
18044 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
18046         PR target/49069
18047         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
18048         instead of cmpdi_operand for first comparison operand.
18049         Don't assert that comparison operands aren't both constants.
18051 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18053         * doc/install.texi (Downloading the Source): Update references to
18054         downloading separate components.
18056 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
18058         * doc/extend.texi (__int128): Improve grammar.
18060 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
18062         PR target/56028
18063         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
18064         alternative to (o,r).
18065         (*movdi_internal_rex64): Remove (!o,n) alternative.
18066         (DImode immediate->memory splitter): Remove.
18067         (DImode immediate->memory peephole2): Remove.
18068         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
18069         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
18070         alternative to (!o,*r).
18071         (*movtf_internal_sse): New pattern.
18072         (*movxf_internal_rex64): New pattern.
18073         (*movxf_internal): Disable for TARGET_64BIT.
18074         (*movdf_internal_rex64): Remove (!o,F) alternative.
18076 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
18078         PR middle-end/56074
18079         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
18080         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
18081         * tree-vect-loop-manip.c (find_loop_location): Also ignore
18082         stmt locations where LOCATION_LOCUS of the stmt location is
18083         UNKNOWN_LOCATION or BUILTINS_LOCATION.
18085         PR target/55686
18086         * config/i386/i386.md (UNSPEC_STOS): New.
18087         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
18088         *strsetqi_1): Add UNSPEC_STOS.
18090 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
18092         PR c++/56067
18093         * doc/invoke.texi: Remove left over -Wsynth example.
18095 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18097         PR tree-optimization/56051
18098         * fold-const.c (fold_binary_loc): Don't fold
18099         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
18100         a narrowing conversion, or widening conversion from signed
18101         to unsigned.
18103 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18105         PR rtl-optimization/56023
18106         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
18107         dependent on debug instruction.
18109 2013-01-21  Martin Jambor  <mjambor@suse.cz>
18111         PR middle-end/56022
18112         * function.c (allocate_struct_function): Call
18113         invoke_set_current_function_hook earlier.
18115 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
18117         * reload1.c (init_reload): Only initialize reload_obstack
18118         during the first call.
18120 2013-01-21  Marek Polacek  <polacek@redhat.com>
18122         * cfgloop.c (verify_loop_structure): Fix up grammar.
18124 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
18126         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
18127         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
18129 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18131         PR target/56058
18132         * config/arm/marvell-pj4.md: Update copyright year.
18133         Fix up use of alu to alu_reg and simple_alu_imm.
18135 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
18137         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
18139 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
18141         PR target/55433
18142         * lra-constraints.c (curr_insn_transform): Don't reuse original
18143         insn for secondary memory move when memory mode should be different.
18145 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18147         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
18148         atomic_storedi_1): New patterns.
18150 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18152         btver2 pipeline descriptions.
18153         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
18154         descriptions.
18155         * config/i386/i386.md (btver2_decode): New type attributes.
18156         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
18157         type attributes.
18158         * config/i386/btver2.md: New file describing btver2 pipelines.
18160 2013-01-19  Andrew Pinski  <apinski@cavium.com>
18162         PR tree-optimization/52631
18163         * tree-ssa-sccvn (visit_use): Before looking up the original
18164         statement, try looking up the simplified expression.
18166 2013-01-19  Anthony Green  <green@moxielogic.com>
18168         * config/moxie/moxie.c (moxie_expand_prologue): Set
18169         current_function_static_stack_size.
18171 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18173         PR tree-optimization/56029
18174         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
18175         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
18177 2013-01-18  Sharad Singhai  <singhai@google.com>
18179         PR tree-optimization/55995
18180         * dumpfile.c (dump_loc): Print location only if available.
18181         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
18183 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
18185         PR target/55433
18186         * lra-constraints.c (curr_insn_transform): Reuse original insn for
18187         secondary memory move.
18188         (inherit_reload_reg): Use rclass instead of cl for
18189         check_secondary_memory_needed_p.
18191 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
18193         PR middle-end/56015
18194         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
18195         the case where writing real complex part of target modifies op1.
18197 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
18199         * config/aarch64/aarch64-simd.md
18200         (aarch64_vcond_internal<mode>): Handle unordered cases.
18201         * config/aarch64/iterators.md (v_cmp_result): New.
18203 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
18204             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
18206         * config/arm/marvell-pj4.md: New file.
18207         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
18208         * config/arm/arm.md (generic_sched): Add marvell_pj4.
18209         (generic_vfp): Likewise.
18210         * config/arm/arm-cores.def: Add marvell-pj4.
18211         * config/arm/arm-tune.md: Regenerate.
18212         * config/arm/arm-tables.opt: Regenerate.
18213         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
18214         * doc/invoke.texi: Document marvell-pj4.
18216 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
18218         * config/aarch64/arm_neon.h: Map scalar types to standard types.
18220 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18222         PR debug/54114
18223         PR debug/54402
18224         PR debug/49888
18225         * var-tracking.c (negative_power_of_two_p): New.
18226         (global_get_addr_cache, local_get_addr_cache): New.
18227         (get_addr_from_global_cache, get_addr_from_local_cache): New.
18228         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
18229         heading comment.
18230         (vt_stack_offset_p): Remove.
18231         (vt_canon_true_dep): Always canonicalize loc's address.
18232         (clobber_overlapping_mems): Make sure we have a MEM.
18233         (local_get_addr_clear_given_value): New.
18234         (val_reset): Clear local cached entries.
18235         (compute_bb_dataflow): Create and release the local cache.
18236         Disable duplicate MEMs clobbering.
18237         (emit_notes_in_bb): Clobber MEMs likewise.
18238         (vt_emit_notes): Create and release the local cache.
18239         (vt_initialize, vt_finalize): Create and release the global
18240         cache, respectively.
18241         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
18243 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18245         PR libmudflap/53359
18246         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
18247         not found in the symtab.
18249 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
18251         PR debug/56006
18252         PR rtl-optimization/55547
18253         PR rtl-optimization/53827
18254         PR debug/53671
18255         PR debug/49888
18256         * alias.c (offset_overlap_p): New, factored out of...
18257         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
18258         the conservative special case for symbolic constants.  Don't
18259         adjust zero sizes on alignment.
18261 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
18263         PR rtl-optimization/52573
18264         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
18265         REG_UNUSED for the same register.
18267 2013-01-17  Richard Biener  <rguenther@suse.de>
18268             Marek Polacek  <polacek@redhat.com>
18270         PR rtl-optimization/55833
18271         * loop-unswitch.c (unswitch_loops): Move loop verification...
18272         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
18273         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
18274         Set it to true when we're removing a loop from hierarchy tree in
18275         an irreducible region.
18276         (fix_bb_placements): Adjust caller.
18277         (fix_loop_placements): Likewise.
18279 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
18281         * config/avr/builtins.def (DEF_BUILTIN): Factor out
18282         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
18283         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
18284         Remove ID.  Adjust comments.
18285         * config/avr/avr-c.c (avr_builtin_name): Remove.
18286         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
18287         * config/avr/avr.c (avr_tolower): New static function.
18288         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
18289         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
18290         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
18291         default expansion.
18293 2013-01-17  Jan Hubicka  <jh@suse.cz>
18295         PR tree-optimization/55273
18296         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
18298 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
18300         PR target/55981
18301         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
18302         store through atomic_store<mode>_1.
18303         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
18305 2013-01-17  Martin Jambor  <mjambor@suse.cz>
18307         PR tree-optimizations/55264
18308         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
18309         for virtual methods.
18310         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
18311         virtual methods before inlining is over.
18312         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
18313         virtual functions.
18314         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
18315         non-virtual.
18317 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
18319         PR rtl-optimization/56005
18320         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
18321         pending reads for prefetch.
18323 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
18325         * config/aarch64/aarch64.md
18326         (*cstoresi_neg_uxtw): New pattern.
18327         (*cmovsi_insn_uxtw): New pattern.
18328         (*<optab>si3_uxtw): New pattern.
18329         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
18330         (*<optab>si3_insn_uxtw): New pattern.
18331         (*bswapsi2_uxtw): New pattern.
18333 2013-01-16  Richard Biener  <rguenther@suse.de>
18335         * tree-inline.c (tree_function_versioning): Remove set but
18336         never used variable.
18338 2013-01-16  Richard Biener  <rguenther@suse.de>
18340         PR tree-optimization/55964
18341         * tree-flow.h (rename_variables_in_loop): Remove.
18342         (rename_variables_in_bb): Likewise.
18343         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
18344         (copy_loop_before): Adjust and delete update-ssa status.
18345         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
18346         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
18347         (rename_variables_in_loop): Remove.
18348         (slpeel_update_phis_for_duplicate_loop): Likewise.
18349         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
18350         use available cfg machinery instead of duplicating it.
18351         Update PHI nodes and perform poor-mans SSA update here.
18352         (slpeel_tree_peel_loop_to_edge): Adjust.
18354 2013-01-16  Richard Biener  <rguenther@suse.de>
18356         PR tree-optimization/54767
18357         PR tree-optimization/53465
18358         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
18359         (vrp_visit_phi_node): For PHI arguments coming via backedges
18360         drop all symbolical range information.
18361         (execute_vrp): Compute backedges.
18363 2013-01-16  Richard Biener  <rguenther@suse.de>
18365         * doc/install.texi: Update CLooG and ISL requirements to
18366         0.18.0 and 0.11.1.
18368 2013-01-16  Christian Bruel  <christian.bruel@st.com>
18370         PR target/55301
18371         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
18372         (broken_move): Handle UNSPECV_SP_SWITCH_B.
18373         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
18375 2013-01-16  DJ Delorie  <dj@redhat.com>
18377         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
18378         (UNSPECV_SP_SWITCH_E): New.
18379         (sp_switch_1): Change to an unspec.
18380         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
18381         replace $r15.
18383 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
18385         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
18386         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
18387         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
18388         (expand_mem_thread_fence): Ditto.
18389         (expand_mem_signal_fence): Ditto.
18390         (expand_atomic_load): Ditto.
18391         (expand_atomic_store): Ditto.
18393 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
18395         PR rtl-optimization/55547
18396         PR rtl-optimization/53827
18397         PR debug/53671
18398         PR debug/49888
18399         * alias.c (memrefs_conflict_p): Set sizes to negative after
18400         AND adjustments.
18402 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18404         PR target/55940
18405         * function.c (thread_prologue_and_epilogue_insns): Always
18406         add crtl->drap_reg to set_up_by_prologue.set, even if
18407         stack_realign_drap is false.
18409 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
18411         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
18412         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
18413         *call): Fix indention.
18415 2013-01-15  Tom de Vries  <tom@codesourcery.com>
18417         PR target/55876
18418         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
18419         Update comment.
18421 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
18423         PR rtl-optimization/55153
18424         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
18426 2013-01-15  Martin Jambor  <mjambor@suse.cz>
18428         PR tree-optimization/55920
18429         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
18430         accesses as grp_to_be_debug_replaced.
18432 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18434         PR tree-optimization/55920
18435         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
18436         there is non-useless type conversion needed from debug rhs to lhs,
18437         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
18439 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
18440             Mikael Pettersson  <mikpe@it.uu.se>
18442         PR target/43961
18443         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
18444         Thumb.
18445         (ASM_OUTPUT_CASE_LABEL): Remove.
18446         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
18447         * final.c (shorten_branches): Update alignment of labels before
18448         jump tables if CASE_VECTOR_SHORTEN_MODE.
18450 2013-01-15  Richard Biener  <rguenther@suse.de>
18452         PR bootstrap/55961
18453         * system.h: Do not include gmp.h for building host tools.
18455 2013-01-15  Richard Biener  <rguenther@suse.de>
18457         PR middle-end/55882
18458         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
18459         account for bitpos when computing alignment.
18461 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
18463         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
18464         (ix86_target_macros_internal): Likewise.
18466         * config/i386/i386.c (m_CORE2I7): Removed.
18467         (m_CORE_HASWELL): New macro.
18468         (m_CORE_ALL): Likewise.
18469         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
18470         (initial_ix86_arch_features): Likewise.
18471         (processor_target_table): Initializations for Core avx2.
18472         (cpu_names): New names "core-avx2".
18473         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
18474         PROCESSOR_CORE_HASWELL.
18475         (ix86_issue_rate): New case.
18476         (ia32_multipass_dfa_lookahead): Likewise.
18477         (ix86_sched_init_global): Likewise.
18479         * config/i386/i386.h (TARGET_HASWELL): New macro.
18480         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
18481         (processor_type): New PROCESSOR_HASWELL.
18483 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
18485         PR tree-optimization/55955
18486         * tree-vect-loop.c (vectorizable_reduction): Give up early on
18487         *SHIFT_EXPR and *ROTATE_EXPR codes.
18489         PR tree-optimization/48766
18490         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
18491         -ftrapv disable -fwrapv.
18493 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18495         PR target/55974
18496         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
18497         etc. to 1 and not to __flash.
18498         Use LL suffix for __INT24_MAX__ with -mint8.
18499         Use ULL suffix for __UINT24_MAX__ with -mint8.
18501 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18503         * config/avr/avr-arch.h
18504         (struct base_arch_s): Use typedef avr_arch_t instead.
18505         (struct arch_info_s): Use typedef avr_arch_info_t instead.
18506         (struct mcu_type_s): Use typedef avr_mcu_t instead.
18507         * config/avr/avr.c: Same.
18508         * config/avr/avr-devices.c: Same.
18509         * config/avr/driver-avr.c: Same.
18510         * config/avr/gen-avr-mmcu-texi.c: Same.
18511         * config/avr/avr-mcus.def: Adjust comment.
18513 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
18515         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
18516         * config/aarch64/iterators.md (VALLDI): New.
18518 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
18519             Andi Kleen  <ak@linux.intel.com>
18521         PR target/55948
18522         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18523         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
18524         memmodel flag.
18526 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18528         * config/avr/avr-stdint.h: Remove trailing blanks.
18529         * config/avr/avr-log.h: Same.
18530         * config/avr/avr-arch.h: Same.
18531         * config/avr/avr-devices.c: Same.
18532         * config/avr/avr-dimode.md: Same.
18533         * config/avr/predicates.md: Same.
18534         * config/avr/avr-c.c: Same.  And fix typo.
18536         * config/avr/avr-protos.h: Same.  And:
18537         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
18538         (init_cumulative_args): Rename to avr_init_cumulative_args.
18539         (expand_prologue): Rename to avr_expand_prologue.
18540         (expand_epilogue): Rename to avr_expand_epilogue.
18541         (adjust_insn_length): Rename to avr_adjust_insn_length.
18542         (notice_update_cc): Rename to avr_notice_update_cc.
18543         (final_prescan_insn): Rename to avr_final_prescan_insn.
18544         * config/avr/avr.c: Same.
18545         * config/avr/avr.h: Same.
18546         * config/avr/avr.md: Remove trailing blanks.
18547         (prologue): Use avr_expand_prologue.
18548         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
18550 2013-01-14  Richard Biener  <rguenther@suse.de>
18552         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
18553         verify_location, collect_subblocks): New functions.
18554         (verify_gimple_in_cfg): Verify that locations only reference
18555         BLOCKs in the functions BLOCK tree.
18557 2013-01-14  Richard Biener  <rguenther@suse.de>
18559         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
18560         PHI argument.
18561         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
18562         unshare reference.
18563         (insert_out_of_ssa_copy_on_edge): Likewise.
18564         (rewrite_close_phi_out_of_ssa): Likewise.
18565         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
18566         debug expressions.
18567         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
18568         propagated constants.
18569         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
18570         can not be shared.
18572 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
18574         * config/avr/avr-modes.def: Add GPL copyright notice.
18576 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
18578         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
18579         MEMMODEL_MASK to determine memory model.
18580         (atomic_store<mode>): Ditto from operands[2].
18581         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
18583 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
18585         PR fortran/55935
18586         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
18587         (fold_gimple_assign): Don't call unshare_expr here.
18588         (fold_ctor_reference): Call unshare_expr.
18590 2013-01-13  Terry Guo  <terry.guo@arm.com>
18592         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
18593         * doc/fragments.texi: Document MULTILIB_REUSE.
18594         * gcc.c (multilib_reuse): New internal spec.
18595         (set_multilib_dir): Also search multilib from multilib_reuse.
18596         * genmultilib (tmpmultilib3): Refactor code.
18597         (tmpmultilib4): Ditto.
18598         (multilib_reuse): New multilib argument.
18600 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
18602         * Makefile.in: Update copyright.
18604 2013-01-12  Tom de Vries  <tom@codesourcery.com>
18606         PR middle-end/55890
18607         * calls.c (expand_call): Check if arg_nr is valid.
18609 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
18611         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
18612         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
18613         documentation.  Add missing '__' in front of
18614         __builtin_ia32_packssdw256.
18616 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18618         PR target/55719
18619         * config/s390/s390.c (s390_preferred_reload_class): Do not return
18620         NO_REGS for larl operands.
18621         (s390_reload_larl_operand): Use s390_load_address instead of
18622         emit_move_insn.
18624 2013-01-11  Richard Biener  <rguenther@suse.de>
18626         * tree-cfg.c (verify_node_sharing_1): Split out from ...
18627         (verify_node_sharing): ... here.
18628         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
18630 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
18632         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
18633         Substitute TREECHECKING.
18634         * configure: Regenerate.
18635         * Makefile.in (TREECHECKING): New.
18637 2013-01-11  Richard Guenther  <rguenther@suse.de>
18639         PR tree-optimization/44061
18640         * tree-vrp.c (extract_range_basic): Compute zero as
18641         value-range for __builtin_constant_p of function parameters.
18643 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
18645         Update copyright years.
18647 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
18649         PR rtl-optimization/55672
18650         * lra-eliminations.c (mark_not_eliminable): Permit addition with
18651         const to be eliminable.
18653 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
18655         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
18656         * configure: Regenerate.
18658 2013-01-10  Richard Biener  <rguenther@suse.de>
18660         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
18662 2013-01-10  Richard Biener  <rguenther@suse.de>
18664         PR bootstrap/55792
18665         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
18666         locations for virtual PHI arguments.
18667         (rewrite_update_phi_arguments): Likewise.
18669 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
18671         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
18672         on to assembler.
18674 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
18676         PR tree-optimization/55921
18677         * tree-complex.c (expand_complex_asm): New function.
18678         (expand_complex_operations_1): Call it for GIMPLE_ASM.
18680 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
18682         PR target/55718
18683         * config/s390/s390.c (s390_symref_operand_p)
18684         (s390_loadrelative_operand_p): Merge the two functions.
18685         (s390_check_qrst_address, print_operand_address): Add parameters
18686         to s390_loadrelative_operand_p invokation.
18687         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
18688         (s390_reload_larl_operand, s390_secondary_reload): Use
18689         s390_loadrelative_operand_p instead of s390_symref_operand_p.
18690         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
18692 2013-01-09  Mike Stump  <mikestump@comcast.net>
18694         * dse.c (record_store): Remove unnecessary assert.
18696 2013-01-09  Jan Hubicka  <jh@suse.cz>
18698         PR tree-optimization/55569
18699         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
18700         * cfgloop.h (scale_loop_profile): Likewise.
18702 2013-01-09  Jan Hubicka  <jh@suse.cz>
18704         PR lto/45375
18705         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
18706         functions.
18707         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
18709 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
18711         PR middle-end/55114
18712         * expr.h (maybe_emit_group_store): Declare.
18713         * expr.c (maybe_emit_group_store): New function.
18714         * builtins.c (expand_builtin_int_roundingfn): Call it.
18715         (expand_builtin_int_roundingfn_2): Likewise.
18717 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
18719         PR rtl-optimization/55829
18720         * lra-constraints.c (match_reload): Add code for absent output.
18721         (curr_insn_transform): Add code for reloads of matched inputs
18722         without output.
18724 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
18726         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
18727         attribute of movddup insn to DF.
18728         (*vec_interleave_lowv2df): Ditto.
18729         (vec_dupv2df): Ditto.
18731 2013-01-09  Jan Hubicka  <jh@suse.cz>
18733         PR tree-optimiation/55875
18734         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
18735         EVERY_ITERATION parameter.
18736         (number_of_iterations_exit): Check if exit is executed every iteration.
18737         (idx_infer_loop_bounds): Similarly here.
18738         (n_of_executions_at_most): Simplify
18739         to only test for cases where statement is dominated by the
18740         particular bound; handle correctly the "postdominance" test.
18741         (scev_probably_wraps_p): Use max loop iterations info
18742         as a global bound first.
18744 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
18745             Nick Clifton  <nickc@redhat.com>
18747         * config/v850/v850.md (cbranchsf4): New pattern.
18748         (cstoresf4): New pattern.
18749         (cbranchdf4): New pattern.
18750         (cstoredf4): New pattern.
18751         (movsicc): Disallow floating point comparisons.
18752         (cmpsf_le_insn): Fix order of operators.
18753         (cmpsf_lt_insn): Likewise.
18754         (cmpsf_eq_insn): Likewise.
18755         (cmpdf_le_insn): Likewise.
18756         (cmpdf_lt_insn): Likewise.
18757         (cmpdf_eq_insn): Likewise.
18758         (cmpsf_ge_insn): Use LE comparison.
18759         (cmpdf_ge_insn): Likewise.
18760         (cmpsf_gt_insn): Use LT comparison.
18761         (cmpdf_gt_insn): Likewise.
18762         (cmpsf_ne_insn): Delete pattern.
18763         (cmpdf_ne_insn): Delete pattern.
18764         * config/v850/v850.c (v850_gen_float_compare): Use
18765         gen_cmpdf_eq_insn for NE comparison.
18766         (v850_float_z_comparison_operator)
18767         (v850_float_nz_comparison_operator): Move from here ...
18768         * config/v850/predicates.md: ... to here.  Move GT and GE
18769         comparisons into v850_float_z_comparison_operator.
18770         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
18771         Delete prototype.
18772         (v850_float_nz_comparison_operator): Likewise.
18774 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18776         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
18777         with calls to gen_insvsi/gen_insvdi.
18779 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
18781         * config/i386/i386.c (initial_ix86_tune_features): Set up
18782         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
18784 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
18785             Jakub Jelinek  <jakub@redhat.com>
18787         PR tree-optimization/48189
18788         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
18789         If nitercst is 0, don't predict the exit edge.
18791 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
18793         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
18794         in asm_fprintf with reg_names.
18795         (aarch64_print_operand_address): Likewise.
18796         (aarch64_return_addr): Likewise.
18797         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
18799 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18801         * config/pa/pa.h (VAL_U6_BITS_P): Define.
18802         (INT_U6_BITS): Likewise.
18803         * config/pa/predicates.md (uint6_operand): New predicate.
18804         (shift5_operand, shift6_operand): Likewise.
18805         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
18806         arith32_operand.
18807         (lshrdi3): Use shift6_operand.
18808         (shrpsi4, shrpdi4): New insn patterns.
18809         (extzv): Delete expander.
18810         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
18811         predicates in unamed zero extract patterns.  Tighten common constraint.
18812         (extv): Delete expander.
18813         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
18814         predicates in unamed sign extract patterns.  Tighten common constraint.
18815         (insv): Delete expander.
18816         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
18817         predicates in unamed insert patterns.  Tighten common constraint.
18818         Change uint32_operand predicate to uint6_operand predicate in unamed
18819         DImode pattern to insert constant values of type 1...1xxxx.
18821 2013-01-04  Jan Hubicka  <jh@suse.cz>
18823         PR tree-optimization/55823
18824         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
18825         issue.
18827 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18828             Uros Bizjak  <ubizjak@gmail.com>
18830         PR rtl-optimization/55845
18831         * df-problems.c (can_move_insns_across): Stop scanning at
18832         volatile_insn_p source instruction or give up if
18833         across_from .. across_to range contains any volatile_insn_p
18834         instructions.
18836 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18838         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
18839         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
18840         Declare.
18841         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
18842         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
18844 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18846         PR fortran/55341
18847         * asan.c (asan_clear_shadow): New function.
18848         (asan_emit_stack_protection): Use it.
18850 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
18852         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
18853         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
18854         with tab instead of space.
18856 2013-01-08  Nick Clifton  <nickc@redhat.com>
18858         * config/rl78/rl78.c (rl78_expand_prologue): Always select
18859         register bank 0 at the start of an interrupt handler.
18860         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
18861         MDBH registers.
18863 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18865         * config/aarch64/aarch64-simd.md
18866         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
18867         (aarch64_simd_bsl): Likewise.
18868         (aarch64_vcond_internal<mode>): Likewise.
18869         (vcond<mode><mode>): Likewise.
18870         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
18871         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
18873 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
18875         * config/aarch64/aarch64-builtins.c
18876         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
18878 2013-01-08  Martin Jambor  <mjambor@suse.cz>
18880         PR debug/55579
18881         * tree-sra.c (analyze_access_subtree): Return true also after
18882         potentially creating a debug-only replacement.
18884 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18886         PR middle-end/55890
18887         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
18889         PR tree-optimization/54120
18890         * tree-vrp.c (range_fits_type_p): Don't allow
18891         src_precision < precision from signed vr to unsigned_p
18892         if vr->min or vr->max is negative.
18893         (simplify_float_conversion_using_ranges): Test can_float_p
18894         against CODE_FOR_nothing.
18896 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
18897             Richard Biener  <rguenther@suse.de>
18899         PR middle-end/55851
18900         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
18901         types instead of just INTEGER_TYPE types.
18903 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
18905         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
18906         TF_SIZE): Define.
18908 2013-01-07  Steve Ellcey  <sellcey@mips.com>
18910         PR target/42661
18911         * config/mips/mips.opt: Change mad to mmad to match documentation.
18913 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18915         PR target/55897
18916         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
18917         .progmemx.data now.
18919 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18921         PR target/55897
18922         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
18923         (avr_addrspace_t): Add .section_name field.
18924         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
18925         array size.
18926         (avr_addrspace): Same.  Initialize .section_name.  Remove last
18927         NULL entry.  Put __memx into .progmemx.data.
18928         (progmem_section_prefix): Remove.
18929         (avr_asm_init_sections): No need to initialize progmem_section.
18930         (avr_asm_named_section): Use avr_addrspace[].section_name to get
18931         section name prefix.
18932         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
18933         retrieve the progmem section.
18934         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
18935         boundary to run over avr_addrspace[].
18936         (avr_register_target_pragmas): Ditto.
18938 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
18940         * varasm.c (output_constant_def_contents): For asan_protect_global
18941         protected strings, adjust DECL_ALIGN if needed, before testing for
18942         anchored symbols.
18943         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18944         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
18945         normal decls.
18946         (output_object_block): For asan protected decls, emit asan padding
18947         after their contents.
18948         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
18949         (asan_finish_file): Test it here instead.
18951 2013-01-07  Nick Clifton  <nickc@redhat.com>
18952             Matthias Klose  <doko@debian.org>
18953             Doug Kwan  <dougkwan@google.com>
18954             H.J. Lu  <hongjiu.lu@intel.com>
18956         PR driver/55470
18957         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
18959         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
18961         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
18963         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
18965         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
18967 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
18969         PR target/54461
18970         * doc/install.texi (Cross-Compiler-Specific Options): Document
18971         --with-avrlibc.
18973 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
18975         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
18976         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
18977         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
18978         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
18979         vqmovun_high_s64): Fix source operand number and update copyright.
18981 2013-01-07  Richard Biener  <rguenther@suse.de>
18983         PR middle-end/55890
18984         * gimple.h (gimple_call_builtin_p): New overload.
18985         * gimple.c (validate_call): New function.
18986         (gimple_call_builtin_p): Likewise.
18987         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18988         Use gimple_call_builtin_p.
18989         (find_func_clobbers): Likewise.
18990         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
18991         (strlen_optimize_stmt): Likewise.
18993 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
18995         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
18996         (vld1q_dup_*): Likewise.
18997         (vld1_*): Likewise.
18998         (vld1q_*): Likewise.
18999         (vld1_lane_*): Likewise.
19000         (vld1q_lane_*): Likewise.
19002 2013-01-07  Richard Biener  <rguenther@suse.de>
19004         * lto-streamer.h (LTO_minor_version): Bump to 2.
19006 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
19008         * config/aarch64/aarch64-protos.h
19009         (aarch64_const_double_zero_rtx_p): Rename to...
19010         (aarch64_float_const_zero_rtx_p): ...this.
19011         (aarch64_float_const_representable_p): New.
19012         (aarch64_output_simd_mov_immediate): Likewise.
19013         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
19014         move immediate case.
19015         * config/aarch64/aarch64.c
19016         (aarch64_const_double_zero_rtx_p): Rename to...
19017         (aarch64_float_const_zero_rtx_p): ...this.
19018         (aarch64_print_operand): Allow printing of new constants.
19019         (aarch64_valid_floating_const): New.
19020         (aarch64_legitimate_constant_p): Check for valid floating-point
19021         constants.
19022         (aarch64_simd_valid_immediate): Likewise.
19023         (aarch64_vect_float_const_representable_p): New.
19024         (aarch64_float_const_representable_p): Likewise.
19025         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
19026         (aarch64_output_simd_mov_immediate): New.
19027         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
19028         (*movdf_aarch64): Likewise.
19029         * config/aarch64/constraints.md (Ufc): New.
19030         (Y): call aarch64_float_const_zero_rtx.
19031         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
19033 2013-01-07  Richard Biener  <rguenther@suse.de>
19035         PR tree-optimization/55888
19036         PR tree-optimization/55862
19037         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
19038         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
19039         not if it is contained therein.
19041 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19043         * config/avr/t-avr: Typo.
19045 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
19047         PR55243
19048         * config/avr/t-avr: Don't automatically rebuild
19049         $(srcdir)/config/avr/t-multilib
19050         $(srcdir)/config/avr/avr-tables.opt
19051         $(srcdir)/doc/avr-mmcu.texi
19052         (avr-mcus): New phony target to build them on request.
19053         (s-avr-mlib, s-avr-mmcu-texi): Remove.
19054         * avr/avr-mcus.def: Adjust comments.
19056 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
19058         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
19060 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19062         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
19064 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
19066         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
19068 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
19070         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
19071         to generate profiling.
19072         * config/rs6000/aix64.h (LIB_SPEC): Same.
19074 2013-01-04  Andrew Pinski  <apinski@cavium.com>
19076         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
19077         New function.
19078         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
19080 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
19082         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
19083         unconditionally.
19084         (ix86_expand_move): Ditto.
19085         (ix86_zero_extend_to_Pmode): Ditto.
19086         (ix86_expand_call): Ditto.
19087         (ix86_expand_special_args_builtin): Ditto.
19088         (ix86_expand_builtin): Ditto.
19090 2013-01-04  Richard Biener  <rguenther@suse.de>
19092         PR tree-optimization/55862
19093         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
19094         translating them through PHI nodes.
19096 2013-01-04  Martin Jambor  <mjambor@suse.cz>
19098         PR tree-optimization/55755
19099         * tree-sra.c (sra_modify_assign): Do not check that an access has no
19100         children when trying to avoid producing a VIEW_CONVERT_EXPR.
19102 2013-01-04  Marek Polacek  <polacek@redhat.com>
19104         PR middle-end/55859
19105         * opts.c (default_options_optimization): Clarify error message.
19107 2013-01-04  Richard Biener  <rguenther@suse.de>
19109         PR middle-end/55863
19110         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
19111         reassociation.
19113 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19115         PR target/53789
19116         * config/pa/pa.md (movsi): Revert previous change.
19117         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
19118         references.
19120 2013-01-03  Richard Henderson  <rth@redhat.com>
19122         * config/i386/i386.c (ix86_expand_move): Always assign to op1
19123         after eliminating TLS symbols.
19125 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
19127         PR bootstrap/50167
19128         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
19129         * graphite-poly.c (debug_gmp_value): Likewise.
19131 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
19133         PR target/55712
19134         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
19135         selected code model, define __code_mode_small__, __code_model_medium__,
19136         __code_model_large__, __code_model_32__ or __code_model_kernel__.
19137         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
19138         xchg temporary register with %k.  Declare temporary register as
19139         early clobbered.
19140         [__x86_64__]: For medium and large code models, preserve %rbx register.
19142 2013-01-03  Richard Biener  <rguenther@suse.de>
19144         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
19145         (dump_subscript): Adjust.
19146         (finalize_ddr_dependent): Do not dump redundant info.
19147         (analyze_siv_subscript): Adjust.
19148         (subscript_dependence_tester): Likewise.
19149         (compute_affine_dependence): Likewise.
19151 2013-01-03  Richard Biener  <rguenther@suse.de>
19153         Revert
19154         2013-01-03  Richard Biener  <rguenther@suse.de>
19156         PR tree-optimization/55857
19157         * tree-vect-stmts.c (vectorizable_load): Do not setup
19158         re-alignment for invariant loads.
19160         2013-01-02  Richard Biener  <rguenther@suse.de>
19162         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19163         invariant load do not generate a vector load from the scalar location.
19165 2013-01-03  Richard Biener  <rguenther@suse.de>
19167         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
19168         for not vectorizing.
19169         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
19170         not build INDIRECT_REFs, call get_name once only.
19171         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
19172         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
19174 2013-01-03  Richard Biener  <rguenther@suse.de>
19176         PR tree-optimization/55857
19177         * tree-vect-stmts.c (vectorizable_load): Do not setup
19178         re-alignment for invariant loads.
19180 2013-01-03  Richard Biener  <rguenther@suse.de>
19182         PR lto/55848
19183         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
19184         prefer a built-in decl.
19186 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19188         * gcc.c (process_command): Update copyright notice dates.
19189         * gcov.c (print_version): Likewise.
19190         * gcov-dump.c (print_version): Likewise.
19192         PR rtl-optimization/55838
19193         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
19194         iv0.step, iv1.step and step.
19196 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19197             Marc Glisse  <marc.glisse@inria.fr>
19199         PR tree-optimization/55832
19200         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
19201         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
19202         integer_{one,zero}_node.
19204 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
19206         PR debug/54402
19207         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
19208         * var-tracking.c (reverse_op): Don't add reverse ops to
19209         VALUEs that have already
19210         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
19212 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
19214         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
19216 2013-01-02  Teresa Johnson  <tejohnson@google.com>
19218         * dumpfile.c (dump_loc): Print filename with location.
19219         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
19220         new location_t parameter to emit complete unroll message with
19221         new dump framework.
19222         (canonicalize_loop_induction_variables): Compute loops location
19223         and pass to try_unroll_loop_completely.
19224         * loop-unroll.c (report_unroll_peel): New function.
19225         (peel_loops_completely): Use new dump format with location
19226         for main dumpfile message, and invoke report_unroll_peel on success.
19227         (decide_unrolling_and_peeling): Ditto.
19228         (decide_peel_once_rolling): Remove old dumpfile message subsumed
19229         by report_unroll_peel.
19230         (decide_peel_completely): Ditto.
19231         (decide_unroll_constant_iterations): Ditto.
19232         (decide_unroll_runtime_iterations): Ditto.
19233         (decide_peel_simple): Ditto.
19234         (decide_unroll_stupid): Ditto.
19235         * cfgloop.c (get_loop_location): New function.
19236         * cfgloop.h (get_loop_location): Declare.
19238 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19240         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
19241         NULL.
19243 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
19245         PR middle-end/55198
19246         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
19247         BLKmode objects when EXPAND_MEMORY is specified.
19249 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
19251         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
19252         in loop predicate.
19253         (fold_builtin_cpu): Do not share cpu model decls across statements.
19255 2013-01-02  Jason Merrill  <jason@redhat.com>
19257         PR c++/55804
19258         * tree.c (build_array_type_1): Revert earlier change.
19260 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
19262         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
19263         "cortex-a57".
19264         * config/aarch64/aarch64-tune.md: Re-generate.
19266 2013-01-02  Richard Biener  <rguenther@suse.de>
19268         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
19269         invariant load do not generate a vector load from the scalar location.
19271 2013-01-02  Richard Biener  <rguenther@suse.de>
19273         PR bootstrap/55784
19274         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
19275         * configure: Regenerate.
19277 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19279         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
19280         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
19281         (expand_builtin_int_roundingfn_2): Keep the original target around
19282         for the fallback case.
19284 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
19286         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
19287         to be clear for sign changes.
19289 2013-01-01  Jan Hubicka  <jh@suse.cz>
19291         * ipa-inline-analysis.c: Fix formatting.
19293 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
19295         PR tree-optimization/55831
19296         * tree-vect-loop.c (get_initial_def_for_induction): Use
19297         gsi_after_labels instead of gsi_start_bb.
19299 Copyright (C) 2013 Free Software Foundation, Inc.
19301 Copying and distribution of this file, with or without modification,
19302 are permitted in any medium without royalty provided the copyright
19303 notice and this notice are preserved.