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
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
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.
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>
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>
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
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
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>
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>
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
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
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
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
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
213 (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
214 Handle OMP_CLAUSE_LASTPRIVATE.
215 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
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.
224 (GOMP_SIMD_LAST_LANE): New.
225 * omp-low.c: Include target.h.
226 (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
228 (check_omp_nesting_restrictions): Same.
230 (lower_rec_simd_input_clauses): New.
231 (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
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
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
266 (vect_analyze_data_refs): Handle simd loops.
267 * tree-vect-loop.c (vectorizable_live_operation): Handle
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
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>
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>
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
387 2013-08-26 Jan Hubicka <jh@suse.cz>
389 * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
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
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
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
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>
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
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
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>
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
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
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
605 * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
606 pretty printer initialization.
607 * coretypes.h (diagnostic_context): Remove superflous type alias
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
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)
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__,
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,
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
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.
711 (*movxi_internal_avx512f): Ditto.
712 (*movdi_internal): Replace constraint "x" with the new constraint "v".
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):
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
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,
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.
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...
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>
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:
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.
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,
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
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
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,
972 * gengtype.c (open_base_files) <ifiles>: Add context.h.
974 2013-08-20 Alan Modra <amodra@gmail.com>
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
1002 2013-08-19 Aldy Hernandez <aldyh@redhat.com>
1004 * doc/invoke.texi (-fcilkplus): Clarify that implementation is
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):
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
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>
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
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>
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
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
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>
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.
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>
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):
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>
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
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.
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
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>
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
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):
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
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
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;
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
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.
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
1559 * config/sparc/sparc.opt (LEON, LEON3): New masks.
1560 * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
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
1618 * cgraphunit.c (expand_function): Remove all references after
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
1627 * ipa-split.c (split_function): Remove references in split function.
1629 2013-08-08 Richard Earnshaw <rearnsha@arm.com>
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>
1638 * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
1639 suggested mode for the assignment isn't compatible with the
1642 2013-08-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
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>
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
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
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
1757 * cgraph.h (cgraph_get_body): Declare.
1758 * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
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
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>
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>
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>
1783 * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
2803 (make_pass_branch_target_load_optimize1): ...new function.
2804 (pass_thread_prologue_and_epilogue): Replace declaration with that
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
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
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
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
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
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
3263 2013-08-05 Jan Hubicka <jh@suse.cz>
3266 * cgraph.c (verify_cgraph_node): Accept local flags from other
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
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):
3375 2013-08-03 Gabriel Dos Reis <gdr@integrable-solutions.net>
3377 * pretty-print.h (pp_underscore): New.
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
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
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
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"
3497 * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
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
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
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
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
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
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
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
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
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
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
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
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
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
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
3793 (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
3794 (UCLIBC_DYNAMIC_LINKER32): Undefine macro first. Handle
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
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>
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
3851 (generate_skeleton): Likewise.
3852 (break_out_comdat_types): Move recursive call to break out nested
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
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>
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
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
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
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  Mark Kettenis  <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
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>
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.
3974 * toplev.c (general_init): Create the singleton gcc::context instance.
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
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
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
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.
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
4128 * i386.c (ix86_pad_returns): Exit loop after setting replace.
4130 * cfgloopmanip.c (remove_path): Exit loop after setting
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
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
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
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.
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
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.
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.
4286 (atomic_test_and_set): 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
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
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
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
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
4404 (cortex_a53_alu_shift): Likewise.
4405 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
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):
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>
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
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
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>
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>
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>
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
4564 (reload_pseudo_compare_func): Check nregs first for reload
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"
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.
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
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
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
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
4730 (cortex_a5_alu_shift): Likewise.
4731 * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
4733 (cortex_a53_alu_shift): Likewise.
4734 * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
4736 (cortex_a7_alu_reg): Likewise.
4737 (cortex_a7_alu_shift): Likewise.
4738 * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
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
4747 * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
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
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>
4788 * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
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
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
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):
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.
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
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
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
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>
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
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>
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
4998 * doc/tm.texi: Regenerate.
5000 2013-07-10 Paolo Carlini <paolo.carlini@oracle.com>
5003 * doc/invoke.texi: Document -Wconditionally-supported.
5005 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
5008 * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
5011 2013-07-10 Georg-Johann Lay <avr@gjlay.de>
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>
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>
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
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
5076 * config/s390/s390.h: Define *_REGNUM macros for floating point
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>
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
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>
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>
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>
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>
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.
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.
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
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
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
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
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>
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>
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>
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
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):
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
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
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.
5627 (smax_m1): 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.
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>
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>
5706 * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
5707 constraints of operand 1 and 2.
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>
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
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
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
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
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>
5845 * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
5846 * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
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>
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
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>
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>
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
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>
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
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>
6074 * lra.c (emit_add3_insn, emit_add2_insn): New functions.
6075 (lra_emit_add): Use the functions. Add comment about Y as an
6078 2013-06-19 David Edelsohn <dje.gcc@gmail.com>
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.
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>
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>
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>
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
6143 * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
6144 (simplify_bitwise_binary): Use it to simpify certain binary ops on
6147 2013-06-19 Sofiane Naci <sofiane.naci@arm.com>
6149 * config/arm/vfp.md: Move VFP instruction classification documentation
6151 * config/arm/arm.md: ... here. Update instruction classification
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>
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
6167 2013-06-19 Paolo Carlini <paolo.carlini@oracle.com>
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
6178 2013-06-19 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
6181 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
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
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
6205 * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
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.
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
6316 (cortex_a15_mult64): Likewise.
6317 (cortex_a15_sdiv): Likewise.
6318 (cortex_a15_udiv): Likewise.
6319 * config/arm/arm1026ejs.md (mult1): Update attribute change.
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>
6388 * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
6390 2013-06-18 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
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
6456 * tree-streamer.c (streamer_tree_cache_add_to_node_array):
6457 Add hash value argument and record that if hashes are recorded
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
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
6506 (mips*-*-vxworks*): Move with_arch default from with_cpu block to
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>
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>):
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>
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.
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
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
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.
6656 2013-06-12 Richard Sandiford <rdsandiford@googlemail.com>
6658 * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
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
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
6692 (aarch64_simd_combine<mode>): New instruction expansion.
6693 * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
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>
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
6730 * symtab.c (symtab_unregister_node): Be lax about missin entries
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>
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
6758 2013-06-11 Michael Meissner <meissner@linux.vnet.ibm.com>
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
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.
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>
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
6852 * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
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.
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.
6918 (floatunssi<mode>2_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
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>
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
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>
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>
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
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
7099 (xscvdpspn): Likewise.
7105 (vpopcnt): Likewise.
7106 (vpopcntb): Likewise.
7107 (vpopcnth): Likewise.
7108 (vpopcntw): Likewise.
7109 (vpopcntd): 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.
7121 (nand_v16qi3): Likewise.
7122 (nand_v8hi3): Likewise.
7123 (nand_v4si3): Likewise.
7124 (nand_v2di3): Likewise.
7125 (nand_v4sf3): Likewise.
7126 (nand_v2df3): 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
7143 (altivec_expand_vec_perm_const): Add support for new power8 merge
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
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
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
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
7200 * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
7203 * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
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
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>
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
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
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"
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
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
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
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
7453 (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
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>):
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
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):
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
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):
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
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
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>
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>
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).
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
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
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>
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>
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>
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>
7703 2013-05-31 Tobias Burnus <burnus@net-b.de>
7706 * tree-ssa-math-opts.c (execute_cse_sincos): Move check
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
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.
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
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
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>
7774 * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
7776 2013-05-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
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
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>
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
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
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
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
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.
7943 (vpkudum): Likewise.
7944 (vpksdss): Likewise.
7945 (vpkudus): Likewise.
7946 (vpksdus): 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.
7965 (vpksdss): Likewise.
7966 (vpksdus): Likewise.
7967 (vpkudum): Likewise.
7968 (vpkudus): 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.
8000 (VI_char): Likewise.
8001 (VI_scalar): Likewise.
8002 (VI_unit): 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
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
8160 (process_references): Simplify.
8161 (symtab_remove_unreachable_nodes): Update; cleanup way function/var
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):
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):
8187 * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
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):
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,
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
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
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
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
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
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
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>
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>
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
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>
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
8417 (udivmod<mode>4_internal): Likewise udivmod<mode>4.
8419 2013-05-25 Richard Sandiford <rdsandiford@googlemail.com>
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>
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
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
8482 (check_and_process_move): Consider mem-reg moves for secondary
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
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>
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>
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>
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>
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>
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
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
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>
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>
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.
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
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
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
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
8723 (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
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
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>
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>
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>
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>
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>
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
8919 * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
8920 * cgraphunit.c (handle_alias_pairs): Store target of unresolved
8922 (output_weakrefs): Update.
8924 2013-05-17 Po-Chun Chang <pchang9@cs.wisc.edu>
8925 Martin Jambor <mjambor@suse.cz>
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
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
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
9038 2013-05-17 Jakub Jelinek <jakub@redhat.com>
9040 * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
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>
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>
9130 * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
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>
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>
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
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
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>
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
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
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
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
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
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
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>
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):
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>
9423 * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
9424 variable for the shift amount. Check that we shift by non-negative
9427 2013-05-14 Chung-Lin Tang <cltang@codesourcery.com>
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
9438 2013-05-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
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.
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>
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>
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>
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
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>
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
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
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
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
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
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
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>
9716 * varasm.c (default_elf_select_section): Move !DECL_P check..
9717 (get_named_section): ..to here before calling get_section_name.
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
9743 * tree.c (build_constructor_va): New.
9744 * tree.h: Declare it.
9746 2013-05-09 Martin Jambor <mjambor@suse.cz>
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>
9756 * symtab.c (symtab_make_decl_local): Do not add private names.
9758 2013-05-09 Jan Hubicka <jh@suse.cz>
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
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
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
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
9812 2013-05-08 Diego Novillo <dnovillo@google.com>
9817 2012-08-17 Diego Novillo <dnovillo@google.com>
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>
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>
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
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
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
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
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>
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.
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>
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
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>
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>
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
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>
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
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):
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
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"
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"
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>
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
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
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>
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>
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
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
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
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):
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>
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.
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.
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.
10514 (smax): Update for V2SF, V4SF and V2DF modes.
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
10528 (vmin<q>_f<32,64>): Likewise.
10529 * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10531 (su): Add mappings for smax, smin, umax, umin.
10533 (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
10534 (FMAXMIN): Rename as...
10535 (FMAXMIN_UNS): ...This.
10537 (fmaxminv): Likewise.
10538 (fmaxmin): Likewise.
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.
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...
10580 (cmhi): Rename to...
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
10590 (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
10591 (COMPARISONS): New.
10592 (UCOMPARISONS): Likewise.
10593 (optab): Add missing comparisons.
10599 (VCMP_S): Likewise.
10600 (VCMP_U): Likewise.
10601 (V_cmp_result): Add DF, SF modes.
10602 (v_cmp_result): 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
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):
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>
10753 * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
10756 2013-04-29 Vladimir Makarov <vmakarov@redhat.com>
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>
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
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
10792 (move_stmt_r): Remove redundant checking.
10794 2013-04-29 Teresa Johnson <tejohnson@google.com>
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>
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>
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
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
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...
10998 (frintp): Rename to...
11000 (frintm): Rename to...
11002 (frinti): Rename to...
11003 (nearbyint): ...this.
11004 (frintx): Rename to...
11006 (frinta): Rename to...
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.
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
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>
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
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
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
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
11178 * omp-low.c (expand_parallel_call): Properly add new basic-blocks
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
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
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
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
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>
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>
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
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
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.
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>
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>
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>
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
11865 (tilepro-linux): Likewise.
11867 2013-04-22 James Greenhalgh <james.greenhalgh@arm.com>
11869 * config/aarch64/aarch64-builtins.c
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.
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>
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
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
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
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>
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.
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
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):
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):
12025 * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
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
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>
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.
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
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):
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>
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
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
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):
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>
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
12314 (thumb_far_jump_used_p): Count instruction size and set
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
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>
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.
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 ...
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
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
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
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
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):
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
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>
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
12532 * config/sparc/predicates.md (cstore_result_operand): New special
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
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 ...
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
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
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
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):
12685 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
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
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
12738 (estimate_edge_size_and_time): Add comment to suggest using rounding
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.
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
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,
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>
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
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
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>
12865 * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_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
12888 2013-04-08 Marek Polacek <polacek@redhat.com>
12890 PR rtl-optimization/48182
12891 * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
12894 2013-04-06 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
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>
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
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
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>
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>
13109 * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
13111 2013-04-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13114 * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
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):
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
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
13154 2013-04-03 Marc Glisse <marc.glisse@inria.fr>
13156 * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
13158 (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
13159 for masks. Test for side effects. Handle nested VEC_MERGE. Handle
13162 2013-04-03 Jakub Jelinek <jakub@redhat.com>
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>
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
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
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.
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.
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
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
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
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
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
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
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>
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
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.
13541 (test-dump.o): New. This object is not added to any executable,
13542 but is present for ad-hoc testing.
13544 (debug (const bitmap_head_def &)): New.
13545 (debug (const bitmap_head_def *)): New.
13547 (extern debug (const bitmap_head_def &)): New.
13548 (extern debug (const bitmap_head_def *)): New.
13550 (debug (edge_def &)): New.
13551 (debug (edge_def *)): New.
13553 (debug (basic_block_def &)): New.
13554 (debug (basic_block_def *)): New.
13556 (dump_node (const_tree, int, FILE *)): Correct source file.
13558 (debug (die_struct &)): New.
13559 (debug (die_struct *)): New.
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.
13570 (debug (ira_allocno_copy &)): New.
13571 (debug (ira_allocno_copy *)): New.
13572 (debug (ira_allocno &)): New.
13573 (debug (ira_allocno *)): New.
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.
13580 (debug (live_range &)): New.
13581 (debug (live_range *)): New.
13583 (debug (lra_live_range &)): New.
13584 (debug (lra_live_range *)): New.
13586 (debug (lra_live_range &)): New.
13587 (debug (lra_live_range *)): New.
13589 (debug (omega_pb_d &)): New.
13590 (debug (omega_pb_d *)): New.
13592 (extern debug (omega_pb_d &)): New.
13593 (extern debug (omega_pb_d *)): New.
13595 (debug (const rtx_def &)): New.
13596 (debug (const rtx_def *)): New.
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.
13614 (extern debug (const rtx_def &)): New.
13615 (extern debug (const rtx_def *)): New.
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.
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.
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.
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.
13649 (_list_iter_cond_expr): Make inline instead of static.
13651 (debug (sreal &)): New.
13652 (debug (sreal *)): New.
13654 (extern debug (sreal &)): New.
13655 (extern debug (sreal *)): New.
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.
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.
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.
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.
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.
13695 (debug (pt_solution &)): New.
13696 (debug (pt_solution *)): New.
13698 (extern debug (pt_solution &)): New.
13699 (extern debug (pt_solution *)): New.
13701 (debug (_var_map &)): New.
13702 (debug (_var_map *)): New.
13703 (debug (tree_live_info_d &)): New.
13704 (debug (tree_live_info_d *)): New.
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):
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
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
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>
13867 * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
13868 rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
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
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
13910 (insn_v<n>mnz): Replaced by ...
13911 (insn_v1mnz): ... this ...
13912 (insn_v2mnz): ... and this. Replace (const_int 0) with the vector
13914 (insn_mz_<mode>): Replaced by ...
13915 (insn_mz_v8qi): ... this ...
13916 (insn_mz_v4hi): ... and this. Replace (const_int 0) with the
13918 (insn_v<n>mz): Replaced by ...
13919 (insn_v1mz): ... this ...
13920 (insn_v2mz): ... and this. Replace (const_int 0) with the vector
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
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):
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
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
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
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
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>
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
14168 2013-03-25 Richard Biener <rguenther@suse.de>
14170 * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
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):
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
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
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
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
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.
14397 * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
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
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
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
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,
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>
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
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
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
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
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
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
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
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
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.
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
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
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
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
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
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>
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
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...
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.
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
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
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>
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>
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>
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>
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
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
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>
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
15404 (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
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
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>
15423 * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
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>
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
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>
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>
15454 * arm.c (shift_op): Validate RTL pattern on the fly.
15455 (arm_print_operand, case 'S'): Don't use shift_operator to validate
15458 2013-03-10 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
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.
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
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
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
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
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
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
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.
15593 2013-02-13 Richard Biener <rguenther@suse.de>
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>
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>
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
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>
15626 * dwarf2out.c (resolve_addr): Don't call
15627 remove_loc_list_addr_table_entries a second time for the same
15630 2013-03-05 Jakub Jelinek <jakub@redhat.com>
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
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,
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
15692 2013-03-05 Steven Bosscher <steven@gcc.gnu.org>
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
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
15754 2013-03-05 Richard Biener <rguenther@suse.de>
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.
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
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
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
15853 2013-03-01 Xinliang David Li <davidxl@google.com>
15855 * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
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
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
15889 * doc/md.texi (Constraints): Ditto.
15890 * gty.texi (Type Information): Ditto.
15891 (User-provided marking routines for template types): Make
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>
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
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.
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
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>
15967 * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
15969 2013-02-28 Georg-Johann Lay <avr@gjlay.de>
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
16029 (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
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
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.
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>
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.
16107 * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
16108 * config/c6x/c6x.h (dbx_register_map): Update declaration.
16111 * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
16112 of prologue/epilogue functions.
16115 * config/tilegx/tilegx.c (tilegx_expand_prologue):
16116 Remove unused variable cfa_offset.
16117 * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
16120 * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
16121 type promotion to unsigned.
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.
16130 * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
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>
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,
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>
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
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
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>
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
16469 2013-02-22 Jakub Jelinek <jakub@redhat.com>
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
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
16506 (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
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
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
16554 2013-02-21 Jakub Jelinek <jakub@redhat.com>
16557 * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
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
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
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
16622 (get_use_iv_cost): Terminate hashtable walk when coming across
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
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
16679 2013-02-18 Aldy Hernandez <aldyh@redhat.com>
16680 Jakub Jelinek <jakub@redhat.com>
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
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>
16705 * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
16706 registers %fr12 and %fr12R as call used.
16709 * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
16710 and HImode, require all displacements to be an integer multiple of
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
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
16758 2013-02-16 Jakub Jelinek <jakub@redhat.com>
16759 Dodji Seketeli <dodji@redhat.com>
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>
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>
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>
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
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>
16831 * ira.c (max_regno_before_ira): Move from ...
16833 (fix_reg_equiv_init): Use max_regno_before_ira instead of
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>
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
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
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>
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>
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>
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>
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>
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
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>
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
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>
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>
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
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
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>
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
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>
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
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>
17204 * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
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>
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
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>
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
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>
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.
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>
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>
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>
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>
17354 * config/i386/i386.c (function_value_ms_64): Add additional valtype
17355 argument and improve checking of return-argument types for 16-byte
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>
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):
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.
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
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
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
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>
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>
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.
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
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
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>
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>
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>
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,
17619 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
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>
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>
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
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
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>
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
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
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>
17809 * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
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
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>
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>
17866 * config.gcc (microblaze*-linux*): Add tmake_file to allow building
17869 2013-01-26 Edgar E. Iglesias <edgar.iglesias@gmail.com>
17871 * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
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
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
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.
17918 2013-01-25 Georg-Johann Lay <avr@gjlay.de>
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.
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>
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>
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>
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.
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>
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
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>
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>
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
18155 * config/i386/i386.md (btver2_decode): New type attributes.
18156 * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
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>
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>
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
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>
18252 PR rtl-optimization/55547
18253 PR rtl-optimization/53827
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
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>
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
18314 * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
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
18368 2013-01-16 Christian Bruel <christian.bruel@st.com>
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
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
18399 * alias.c (memrefs_conflict_p): Set sizes to negative after
18402 2013-01-15 Jakub Jelinek <jakub@redhat.com>
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>
18418 * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
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>
18443 * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
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>
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>
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>
18522 * config/i386/sync.md (atomic_store<mode>_1): New pattern.
18523 (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
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
18561 * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
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
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
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>
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>
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
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>
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
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>
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>
18705 * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
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
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):
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
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
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
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):
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>
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
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>
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,
18908 2013-01-07 Steve Ellcey <sellcey@mips.com>
18911 * config/mips/mips.opt: Change mad to mmad to match documentation.
18913 2013-01-07 Georg-Johann Lay <avr@gjlay.de>
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>
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
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
18943 (place_block_symbol): Adjust size for asan protected STRING_CSTs if
18944 TREE_CONSTANT_POOL_ADDRESS_P. Increase alignment for asan protected
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>
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>
18970 * doc/install.texi (Cross-Compiler-Specific Options): Document
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
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>
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):
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
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
19113 2013-01-03 John David Anglin <dave.anglin@nrc-cnrc.gc.ca>
19116 * config/pa/pa.md (movsi): Revert previous change.
19117 * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
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>
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>
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
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>
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>
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>
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
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
19253 (fold_builtin_cpu): Do not share cpu model decls across statements.
19255 2013-01-02 Jason Merrill <jason@redhat.com>
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
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>
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.